A short one today, with a more detailed breakdown in the video below.

In the latest industry price data, for April to June 2026, lab-grown diamond wholesale prices were about 13% lower than a year earlier. In the very next quarter, July to September, International Gemological Institute (NSE: IGIL) earned 37% more from certifying those stones. It mines none of them, cuts none of them and sells none of them. It grades them, and by company and press estimates about two in every three lab-grown diamonds that carry a grading report carry its name.

That pairing is the whole puzzle, and it resolves once you look at how IGI gets paid. Revenue is the number of reports issued multiplied by the fee per report. In the 15 months to March 2026, IGI earned ₹1,598 Cr of revenue at a 61% EBITDA margin, and its purchases were ₹2 Cr. It holds no diamonds, carries no inventory and takes no view on carat prices. The economics sit closer to CDSL or CAMS than to a jeweller: a fee on a flow, earned by a team of skilled graders.

Cheaper stones, more traffic. When a diamond gets cheaper, more of them are grown, set and sold, and each one still needs a report. IGI issued 8.3 million reports in calendar 2023, 10.6 million in 2024 and 14.8 million in the twelve months to September 2026, while lab-grown wholesale prices fell about 26% in 2025 alone. The report itself costs roughly 2 to 3% of the retail price, so nobody in the chain shops around for it. Retailers name the lab their customers trust, and growers send their stones there. GIA’s 2025 decision to stop giving lab-grown stones full 4C grades has only pushed more of that traffic towards IGI.

A bombed-out IPO with profits intact. SOIC has a useful mental model for Indian listings: hype, a post-listing peak, a sharp fall, a long stretch of boredom, and sometimes a reversal 12 to 24 months later. IGI fits it closely. It listed at ₹510 in December 2024 against a ₹417 issue price, touched ₹642 on 1 January 2025 at about 60 times calendar-2024 earnings, fell 43% by early March after a soft quarter, and spent most of the next 18 months in the ₹300s. Over the same period PAT rose 56%, from ₹427 Cr in calendar 2024 to ₹666 Cr in the twelve months to September 2026, so the multiple did all the falling. In September’s quarter, revenue grew 31%, PAT grew 43% and management raised its full-year guidance.

The line that argues against me. Management says fees haven’t changed since 2024, and on list price that is accurate. Rebates and discounts, the gap between what IGI bills and what it books, have risen from 2.6% of standalone billings in 2023 to 5.6% in the latest audited period, while the top ten customers grew from 40.7% to 46.5% of sales. A few large growers are negotiating harder. Add Blackstone, which still owns 76.55% after recovering about 96% of its cost, and you have the two reasons the stock trades where it does. At ₹381, a reverse DCF implies roughly 13% annual profit growth for a decade, against 25%+ delivered so far.

The full 26-minute walk-through covers the five revenue engines, the cash conversion record, management’s walk-the-talk scorecard, the risks in one place, and what I’m watching over the next four quarters. It’s below.

Disclosure: The analyst and family hold positions in International Gemological Institute Ltd (NSE: IGIL). SEBI Reg. No. INH000023807.

GET TWO FREE IPR SAMPLE REPORTS The analysis you just read, built from cash conversion, customer concentration and promoter structure rather than the headline growth rate, is the foundation of IPR’s research method. Two free sample reports show the full framework applied to individual companies: Gravita India (Future Titans) and Kalyani Cast Tech (Microcap Mavericks). [Get Free Samples] → https://www.inflectionpointresearch.in/free-sample-research-reports No payment required. SEBI Reg. INH000023807.

DISCLAIMER: Karnik Shah is a SEBI-registered Research Analyst (Registration No. INH000023807). This content is for educational and informational purposes only and should not be construed as investment advice or a recommendation to buy, sell, or hold any security. Readers should conduct their own due diligence and consult a qualified financial advisor before making any investment decisions. The analyst and family hold positions in International Gemological Institute Ltd (NSE: IGIL). Past performance does not guarantee future results. Investments in securities are subject to market risks. Read all related documents carefully.

Hello friends,

Look around the room you are sitting in.

The phone in your hand has a metal frame milled to within a few hundredths of a millimetre. The chair has a gas cylinder with a turned steel rod inside it. The fan above you spins on a bearing whose races were ground smooth by an abrasive wheel. The car in the parking lot has an engine block that was bored, faced and drilled on a production line, and a gearbox full of gears whose teeth were cut, one after another, by a machine that never got tired.

Every manufactured object around you was made by a machine. And the machines that made them were themselves made by machines. Follow that chain back far enough, past the factory, past the assembly line, past the component supplier, and you eventually arrive at a very particular kind of equipment: the machine tool. A machine that cuts, drills, bores, grinds or shapes metal into the precise parts that every other machine is built from.

The industry has a lovely name for these. It calls them mother machines.

Here is the uncomfortable number at the heart of this piece. In FY26, India consumed ₹36,419 crore of machine tools. Of that, ₹21,628 crore, or 59.4%, was imported. Twenty-five years ago the import share was 29%. It peaked at 75% in 2008-09. For the last fifteen years it has sat stubbornly between 55% and 65%, through Make in India, through two Capital Goods Schemes, through PLI, through a decade of speeches about manufacturing self-reliance. India is the world’s fourth-largest consumer of machine tools and only its ninth-largest producer.

In a country that wants to become a manufacturing superpower, that is a strange place to be. It is a little like a country that wants to be a great cricketing nation importing most of its bats.

Now here is the more interesting fact. On the edge of Rajkot, in the Metoda industrial estate in Lodhika taluka, sit two listed companies that are trying to close that gap. They are neighbours in the most literal sense: both list their registered addresses in the same GIDC belt.

One of them, Jyoti CNC Automation, is India’s largest listed CNC machine builder. It did ₹2,093 crore of revenue in FY26 at a 25.2% EBITDA margin, owns a French aerospace machine brand called Huron, sells to HAL and ISRO and Airbus and Safran, and is worth about ₹22,680 crore.

The other, Macpower CNC Machines, did ₹333 crore of revenue at a 16.2% margin, sells mostly ₹20 lakh lathes and machining centres to small job shops, has no subsidiaries and no meaningful debt, and is worth about ₹1,926 crore.

If you had to guess which of these two is the better business, you would guess Jyoti, and you would be right. It is bigger, more profitable, more technologically advanced, and it plays in the part of the market that India most needs to localise.

But I want to ask a slightly different question in this piece: which of these two businesses has turned its success into cash for its shareholders, and which one is asking you to underwrite the bigger risk at today’s price? Those two questions do not have the same answer, and the gap between them is what makes this one of the most interesting comparisons in Indian capital goods right now.

There is an old story about Hannibal, the Carthaginian general who marched an army and a herd of elephants over the Alps and then beat Rome in battle after battle: Trebia, Trasimene, and finally Cannae, still studied in military academies. He was the finest general of his age. He spent fifteen years in Italy. And he never took Rome, because the reinforcements, siege equipment and supplies he needed to turn victories into a conquered city never arrived in the quantities he needed. Rome eventually wore him down, and he lost the war at Zama.

Keep Hannibal in the back of your mind. We will come back to him at the end, and you can decide for yourself whether he belongs in this story.

This is a long one. Pour something cold.

The elevator pitch

India’s machine tool market is large (₹36,419 crore in FY26, +15%), import-dependent (59.4%) and, on IMTMA’s estimates, headed towards about ₹54,000 crore by 2030. The imports are concentrated at the high end, where machines are expensive, qualification is slow and margins are fat. China is now the largest supplier, which makes the low end a price war.

Jyoti CNC is the more capable business: scale, deep vertical integration, a real 5-axis franchise through Huron and an aerospace customer list no Indian peer matches. It is also carrying an open-ended French judicial export-control investigation at Huron, three years of negative cumulative operating cash flow on ₹803 crore of profit, and nearly half its standalone net worth tied to the French group.

Macpower CNC is the simpler business: smaller, lower margin, a volume player in the part of the market most exposed to Chinese competition, with sloppy disclosure and weak cash conversion of its own. But it has no debt, no subsidiaries, no legal overhang, and its valuation asks for less growth than it has been delivering.

The better business and the cleaner balance sheet sit in different companies. This piece is about how to think about that trade-off.

You can also watch a video I made on this topic here:

Part 1: What a CNC Machine Actually Is

From a craftsman’s hands to a computer’s instructions

For most of industrial history, cutting metal precisely was a skill that lived in a person. A machinist stood at a lathe or a milling machine, turned handwheels, watched the chips curl away, measured, adjusted, and measured again. A good one could hold remarkable tolerances. But he could only make one part at a time, he got tired, he had bad days, and the next machinist down the line did it slightly differently.

The leap came out of the American aerospace industry in the late 1940s. A Michigan contractor called John T. Parsons, who made helicopter rotor blades, worked out that the complex curves he needed could be cut far more accurately if a machine’s movements were driven by a sequence of numbers rather than a human hand. With US Air Force funding, the Servomechanisms Laboratory at MIT built on his idea and, in 1952, demonstrated the first numerically controlled milling machine. Its instructions came on punched paper tape.

That was NC, numerical control. The “C” in CNC, computer, arrived over the next two decades as hardwired logic gave way to minicomputers and then microprocessors. A Japanese engineering team inside Fujitsu started work on NC systems in the 1950s; it later spun out as Fanuc, which went on to become the dominant name in CNC controllers worldwide. Germany’s Siemens became the other.

How a part gets made today, in five steps

A modern CNC machine tool is a metal-cutting or metal-forming machine in which every motion is commanded by a computer running a part program. A part designed on a screen can be reproduced to within a few microns, thousands of times, without a skilled hand on the wheels. Here is how it works:

1. Design. An engineer models the part in CAD software.

2. Programming. CAM software converts that geometry into toolpaths and writes G-code and M-code: the coordinates, feed rates, spindle speeds, tool changes and coolant commands.

3. Control. The CNC controller, the machine’s brain, reads the program and issues motion commands. Globally this market is dominated by Fanuc (Japan) and Siemens (Germany), with Mitsubishi and Heidenhain behind them.

4. Actuation. Servo drives and motors turn ball screws that ride on linear guideways to move each axis. A spindle motor rotates either the cutting tool (in milling) or the workpiece (in turning).

5. Feedback. Encoders and linear scales report the actual position back to the controller, which corrects any deviation in real time.

That last step, the closed loop, is what separates a CNC machine from a manual or cam-driven one. The machine knows where it is, compares that with where it should be, and fixes the difference continuously.

What is inside the box, and why it decides the margin

This is the most important technical idea in the whole piece, so it is worth slowing down for. A CNC machine is a bill of materials, and an Indian builder’s economics depend almost entirely on how much of that bill it makes itself.

Two things follow from that table, and they will matter for everything that comes later.

First, gross margin is a vertical integration score. A builder that casts its own beds, makes its own spindles, turrets and tool changers, and buys only the controller and precision motion parts keeps far more of every rupee than one that assembles a kit of bought-out subsystems. Hold that thought when you see Jyoti’s 55.1% gross margin next to Macpower’s 38.5%.

Second, the brain is foreign almost everywhere. Even India’s best builders buy their controllers from Fanuc, Siemens or Mitsubishi. That creates two risks investors rarely discuss: currency (a weaker rupee against the yen or euro squeezes the margin) and licensing (high-end controllers used in 5-axis machines are subject to export controls, which, as you will see, is about to become very relevant).

Why the “mother machine” is a leading indicator

Because a machine tool is a capital purchase, customers can defer it. When an auto component maker is confident about next year’s volumes, he orders machines. When he is not, he waits. That makes machine tool demand one of the cleanest real-time reads on manufacturing capex in any economy, and it makes order intake, more than last year’s revenue, the variable to watch in any machine tool company.

Part 2: The Family Tree, From ₹20 Lakh Lathes to ₹20 Crore Five-Axis Centres

CNC machine tools split along two lines: whether the tool or the workpiece rotates (milling versus turning), and how many axes the machine can move at once. Price, margin and customer type all climb with axis count and part size.

The practical takeaway is that this industry has two ends, and they behave like different businesses.

The volume end is turning centres and VMCs, sold mostly to small and mid-sized job shops, where price, delivery time and local service win. This is where most Indian builders live. It is also the end most exposed to Chinese competition.

The value end is 5-axis, HMC and large-format machines, sold to aerospace, defence, railways and large OEMs, where qualification cycles are long, imports have historically dominated, and margins are far higher.

Jyoti already earns a large share of its revenue at the value end through Huron. Macpower is still predominantly a volume player trying to climb. Both companies launched a double-column machine in Q1 FY27, which tells you exactly where each of them wants to go next.

Part 3: Who Actually Buys a CNC Machine

The five demand engines

1. Auto and auto components. Still the largest buyer, now adding EV parts. Also the most cyclical. When two-wheeler and passenger vehicle volumes wobble, tier-1 and tier-2 suppliers stop ordering machines within a quarter.

2. Defence and aerospace. Driven by indigenisation, HAL and private aerostructure programmes, and European primes localising supply chains in India. Long qualification cycles, high prices, and an astonishing statistic: by an IMTMA internal study cited by MMI, domestic builders meet only about 4% of India’s aerospace and defence machine tool needs.

3. Electronics manufacturing (EMS). PLI and the ₹40,000 crore components scheme could need very large numbers of drill-tap and machining centres. But orders have lagged announcements for two years, a theme you will see in Jyoti’s numbers.

4. General engineering, die and mould, and MSME job shops. The volume base. Thousands of small shops, financed by bank loans and state capital subsidies, buying one or two machines at a time. This is Macpower’s core and Jyoti’s fastest-growing segment.

5. Railways and heavy engineering. Bogies and Vande Bharat components, the target market for both companies’ new double-column machines.

The job shop economics that drive the volume end

To understand Macpower, and the entry-level half of Jyoti, you need to picture the customer. He runs a shop with six to twenty people somewhere in Rajkot, Pune, Coimbatore, Ludhiana or Faridabad. He machines parts for a bigger supplier. When he wins a new contract he buys a machine, almost always with a bank loan, often with a state capital subsidy attached.

That has three consequences. First, he buys locally and he buys on demonstration: he wants to watch his own part being cut on a machine at a technology centre near his shop before he signs. Second, dispatch timing depends on when his loan is sanctioned and his margin money is ready, which makes quarterly revenue lumpy. Third, when credit tightens or demand slows, he simply stops buying.

The cycle is real

Both companies have been through it. In the FY20 to FY21 downturn, Jyoti’s revenue fell 45% and Macpower’s fell 42%. Today both managements describe record order intake and no slowdown, and FY26 industry data supports that. But Jyoti’s Q1 FY27 order intake was about 15% below Q4 FY26. A machine tool company in a good year can look like a compounder. It remains a cyclical capital goods business, and the share price will remind you of that eventually.

Part 4: The ₹36,419 Crore Market That India Mostly Imports

Exhibit 1

Exhibit 1: Domestic production grew faster than imports in FY26, but a jump in exports meant home demand still leaned more on imports.

Here are the FY26 numbers, from IMTMA’s IMTEX 2027 press release of 13 October 2026:

Notice the oddity. Domestic production grew faster than imports in FY26, 19% against 16%. And yet the share of home demand met by imports still edged up, from 58.8% to 59.4%. The reason is that exports jumped 76%, so more of what Indian builders made left the country. And the largest destination for those exports, per IMTMA data reported by MMI, was Russia, at 34% of India’s machine tool exports (48% in Q1 FY26). Hold on to that detail too. It will coWhat amme back in an unexpected place.

Exhibit 2

Exhibit 2: The import share has not moved below 55% in fifteen years of policy support.

Where the imports come from, and what they are

Exhibit 3

Exhibit 3: China is now the largest supplier, which changes the competitive problem at the low end.

China supplied 30% of India’s FY26 machine tool imports, overtaking Japan (21%) during the year, with Germany at 12%. By product, presses were 16% of imports, VMCs 12% and lathes 10%.

That mix is the whole strategic picture in one sentence. Indian builders have been trying for years to push up into the high end that Japanese and German machines dominate. Meanwhile, Chinese builders are arriving at the low end, which is exactly the volume tier that pays the bills for most Indian makers.

India in the world

By Gardner’s World Machine Tool Survey for full-year 2025, China led both production and consumption, producing about US$30.7 billion and consuming about US$26.4 billion of machine tools. The United States was the second-largest consumer at US$11.4 billion, and Germany third at about US$5.1 billion. India’s FY26 consumption of ₹36,419 crore is roughly US$4.1 billion, which is consistent with its position as the fourth-largest consumer in the CY2024 survey, behind those three. On production it ranked ninth.

Where it could go

Macpower’s CMD, citing IMTMA on the Q4 FY26 call, put Indian consumption at about ₹54,000 crore by 2030. From ₹36,419 crore, that is roughly 10% a year. I treat this as an industry body’s projection rather than a forecast I would build a model on, but the direction is consistent with a decade of data.

My take: the import gap is real and large, but its structure matters more than its size. The opportunity is biggest exactly where Indian builders are weakest (high-end, aerospace, large format), and the competitive threat is growing exactly where they are strongest (entry-level volume). Any company in this industry has to be judged on how it handles both at once.

Part 5: Sixty Years in Three Acts

India’s machine tool story runs in three acts: a state-led build-out after independence, two decades in which imports took most of the growth, and a private-sector revival over the last ten years that is still catching up with demand. The thread running through all three is that India has always consumed far more machine tools than it makes.

Act One: The State Builds (1946 to 1990)

1946. The Indian Machine Tool Manufacturers’ Association (IMTMA) is founded in New Delhi with 19 members. It still publishes the industry’s official data.

1953. Hindustan Machine Tools (HMT) is incorporated in Bengaluru, with Oerlikon of Switzerland supplying technology for centre lathes. Units follow at Pinjore, Kalamassery, Hyderabad and Ajmer through 1975. HMT becomes the anchor of a licensed, protected industry.

1962 to 1965. CMTI, the government machine tool R&D institute, is set up in Bengaluru with Czechoslovak support (sources differ on the year). Its engineers will later found private builders.

1969. The first IMTEX exhibition is held in Mumbai. It becomes Asia’s flagship machine tool show, held at BIEC Bengaluru since 2007.

1972 to 1979. HMT forms its first NC lathe design team. A slant-bed CNC lathe wins the first “Best Design” award at IMTEX 1979. Indigenous CNC capability begins.

1979. Three ex-CMTI engineers found Ace Designers. It enters CNC lathes in 1986 and grows into today’s largest private builder, the Ace Micromatic group.

1984 to 1986. HMT collaborates with Siemens on CNC controls and sets up a CNC Systems Division. Japanese builders enter through licensing. CNC goes mainstream, but the controls stay foreign, a pattern that has never changed.

1989 to 1991. Jyoti’s founders start a partnership in Rajkot, incorporated in 1991.

Act Two: The Doors Open (1991 to 2013)

1991 to 2002. Liberalisation. All physical import barriers on machine tools are removed by 1992. Domestic production stagnates while the PSUs lose ground. The private builders that survive are technocrat-led firms competing on price and service.

2000 to 2011. The auto boom buys foreign CNC machines. The import share of consumption rises from 29% in 2000-01 to a peak of 75% in 2008-09, and is still 65% in 2010-11. This is the import dependence that persists today.

2003. Macpower is founded in Rajkot.

2007. Jyoti acquires Huron Graffenstaden of Strasbourg, France, becoming the first Indian builder to own a European 5-axis brand. Jyoti’s own revenue at the time was about ₹50 crore.

Act Three: The Private Revival (2014 to today)

2014. Capital Goods Scheme Phase I is approved (about ₹931 to ₹996 crore), funding centres of excellence and the 530-acre Tumakuru Machine Tool Park. It is the first targeted policy support since the HMT era.

2016. The government closes HMT’s watch, bearing and tractor units. HMT Machine Tools survives as a small PSU. The state-led model is over.

2018 to 2024. Public capital enters the sector. Macpower lists on NSE Emerge (2018) and moves to the main board (2020). Jyoti’s IPO raises ₹1,000 crore in January 2024.

2022. Capital Goods Scheme Phase II is notified with a ₹1,207 crore outlay.

2026. Three things happen that define the current moment. In January, the machinery technical regulation (a quasi quality-control order) is rescinded. On 26 August, DN Solutions, the former Doosan machine tools business, opens a ₹600 crore plant in Bengaluru, initially for VMCs and turning centres. And IMTMA reports record FY26 production.

What I find most interesting: two lessons from sixty years carry directly into the investment case. Protection built capacity but not technology; the industry only modernised when private builders had to compete. And imports have never been displaced at the high end. Even after a decade of policy support, controllers come from Japan and Germany, and 5-axis, large-format and high-speed machines are still mostly imported. That is the opportunity both companies pitch. It is also the reason most of their growth is concentrated in simpler machines.

Part 6: The Battlefield

Who competes

The thing to notice is that the largest Indian builder files no quarterly results. Ace Micromatic, which is private, reportedly expected FY25 turnover above ₹2,400 crore, more than Jyoti’s entire consolidated FY25 revenue of ₹1,818 crore. That matters for investors: the listed universe gives you a partial view of the competitive field, and the strongest domestic competitor files no quarterly results.

No moat from the government

On policy, the industry gets support but no protection. There are two Capital Goods Scheme phases, state capital subsidies and PLI-linked demand. But no quality-control order applies to machine tools after January 2026, no anti-dumping case covers Chinese CNC machines, and the 2026-27 Budget left duties unchanged.

Compare that with, say, optical fibre, where anti-dumping duties on Chinese fibre give Indian makers a structural shield. Machine tool builders get no such shield. With China now the largest import source, entry-level pricing pressure is the main threat to both companies’ volume tiers. Any moat in this industry has to be built by the company itself.

With that context, let us meet the two neighbours.

Part 7: Jyoti CNC Automation, the Champion With a French Problem

NSE: JYOTICNC | CMP ₹997 | Market cap about ₹22,680 crore | P/E 70x trailing (all as of 16 Oct 2026)

From a Rajkot partnership to a French 5-axis brand

Jyoti began in 1989 as a partnership, Jyoti Enterprise, in Rajkot. It was incorporated in 1991 as AMB Engineering and started making CNC machines in 2002. The promoter and CMD is Parakramsinh Jadeja, and the promoter group holds 62.55%, a figure unchanged for eight quarters.

The defining decision in the company’s history came in 2007, when Jyoti bought Huron Graffenstaden, a Strasbourg-based maker of 5-axis machines for aerospace and defence. Jyoti’s own revenue at the time was about ₹50 crore. A ₹50 crore Rajkot machine builder buying a European technology brand was an extraordinarily ambitious move, and for more than a decade it looked like overreach. Between FY18 and FY23, Jyoti reported a net loss in five of six years, a cumulative net loss of about ₹175 crore. Revenue fell 45% from FY19 to FY21, and in FY22 operating profit did not even cover interest.

Then came the listing. The January 2024 IPO raised about ₹1,000 crore, used largely to repay debt, and profitability arrived alongside the deleveraging.

What Jyoti makes and how it makes it

Jyoti makes the full range, from 2-axis turning centres to simultaneous 5-axis and multi-tasking machines: 200+ variants across 44 product series, priced from under ₹50 lakh to more than ₹20 crore. It employs 4,500+ people including 140+ in R&D, and runs 29 sales and service centres in India.

What sets it apart is how much of the machine it makes itself. Jyoti runs its own foundry, machine shop, sheet-metal shop, paint shop and assembly, and makes spindles, turrets, rotary tables, pallet changers and tool changers in-house. It imports the controllers, drives, motors, precision bearings and guideways, mainly from Germany and Japan. Management puts those at about 30% of purchases by value; the annual report puts direct imports at about 12% of material cost.

Go back to the bill-of-materials table in Part 1. Jyoti owns almost every row except the brain and the precision motion parts. That is why its gross margin was 55.1% in FY26, against 38.5% at Macpower.

Two engines

India standalone earned FY26 revenue of about ₹1,949 crore at a 28.9% EBITDA margin. It sells mostly entry and mid-range machines to more than 15,000 customers, plus high-end machines to HAL, ISRO, Tata and the defence PSUs. Capacity at its two Rajkot plants is 6,000 machines a year, expanding to 16,000.

Huron, in Strasbourg, is the 5-axis specialist. Its capacity doubled in November 2025, and management values it at up to about €75 million of annual output. Management says about 70% of the material in a Huron machine now comes from India, which is the clever part of the 2007 acquisition: Rajkot’s cost base feeding a European brand’s price point. Listed customers include Airbus, Safran, GE and Rolls-Royce.

How big is Huron? That depends on which document you read. The CMD put Huron’s FY26 standalone revenue at about ₹248 crore, including ₹105 crore of intra-group business. The annual report shows ₹92 crore of sales from Jyoti to Huron and only ₹9 crore in the other direction, and puts Jyoti SAS revenue at ₹264 crore on an Ind AS basis. I have not reconciled these, and I will come back to why that matters.

Who buys from Jyoti

The order book stood at ₹4,848 crore at 30 June 2026, 2.3 times FY26 revenue, with 38% from aerospace and defence. For a machine tool company that is unusually deep visibility.

But look at the machine mix underneath. In Q1 FY27 Jyoti sold 1,406 machines, of which 1,349 (96%) were entry-level, against 89% a year earlier. Mid-range units fell from 104 to 33. Management put FY26 average realisation at ₹35.6 lakh and now guides towards about ₹30 lakh once the new plant is full. The new capacity is overwhelmingly for simpler machines.

The numbers

Exhibit 4: The FY24 inflection was real. FY26 is where growth halved and costs ran ahead of revenue.

Three things stand out about FY26.

Growth halved. Revenue grew 15.2%, against management’s own ambition of 30 to 35% stated as late as the Q2 FY26 call in November 2025.

Costs ran ahead. Operating profit grew only ₹36 crore on ₹275 crore of extra revenue, because employee costs rose 28% and other expenses 49%, including “miscellaneous” expenses up 129% to ₹54 crore with no explanation.

Profit growth rested on currency. FY26 PAT grew 6.3% to ₹336 crore, but other income included a ₹43 crore foreign-exchange gain, mostly unrealised and largely on euro loans and advances to the loss-making French subsidiary. Excluding FX gains, consolidated PAT fell about 3%.

Here is the important nuance, and it is fair to Jyoti: the Indian business is healthy. Standalone PAT grew 26% in FY26 and 21% in Q1 FY27, at a 27% EBITDA margin. The consolidated drag is Huron. Which brings us to the issue that changes the risk profile of the whole company.

The Huron investigation

On 12 April 2026, Jyoti disclosed to the exchanges that French authorities had opened a judicial investigation into suspected violations of EU dual-use export controls at Huron. The filing disclosed a temporary restriction of the director general’s duties, an interim seizure of about €4 million in bank accounts (€3.02 million per the annual report) and of two residential properties owned by Jyoti SAS. The stock fell about 15% on 13 April (exchange price data, T3).

Analysts on the Q4 call linked it to alleged re-routing of machines to Russia via India in 2022 to 2024. The CMD denied any re-routing, said Jyoti itself shipped to Russia in 2022 and 2023 and not since, and described the probe as one of more than 50 industry-wide European investigations. You may remember from Part 4 that Russia is the largest destination for Indian machine tool exports. To be clear, nothing in the filings links the investigation to that industry data; the two facts sit side by side, and I am not drawing a line between them.

The accounting consequences are already visible:

• ₹67 crore of Huron revenue recognised under percentage-of-completion was reversed in Q4 FY26, an EBITDA hit of about ₹40 crore.

• A further ₹35 crore was not recognised in Q1 FY27 because end-user certificates and export licences now take far longer.

• About ₹100 crore (7 to 8 machines) sits in work in progress awaiting licences.

• ₹32.9 crore of bank balance is classified as under regulatory seizure and is not usable cash.

Huron’s own economics are poor. Jyoti SAS earned ₹264 crore of revenue and lost ₹56.7 crore in FY26, and its net assets in the consolidation are negative ₹111 crore. Revenue earned outside India collapsed from ₹623 crore (34% of the total) to ₹244 crore (12%).

The auditor noticed. The FY26 audit report carries an emphasis of matter citing Huron’s accumulated losses, eroded net worth, the judicial investigation and machines under customs control as a material uncertainty over the subsidiary’s ability to continue as a going concern. The opinion itself is unqualified, but that is serious language.

And the parent’s exposure is larger than the headline suggests.

Exhibit 7: Nearly half of Jyoti’s standalone net worth is tied to the French group, with no impairment booked.

The parent holds ₹341.6 crore of equity and ₹160.1 crore of loans in Jyoti SAS (the CARO annexure notes no stipulated repayment of principal or interest), plus ₹170.6 crore of “advances to supplier”, ₹118.9 crore of receivables from Huron, ₹90.4 crore of uncollected interest and commission, and a ₹264.3 crore SBLC and letter of comfort. That is ₹882 crore on balance sheet, 36% of the ₹2,454 crore standalone net worth, or ₹1,146 crore (47%) including the guarantee. The supplier advance alone is 18 times FY26 purchases from Huron, which reads more like funding than trade.

How long will this take? Management will not give a timeline. An analyst on the call noted that comparable European cases have taken two to three years, and the CMD agreed it could be “2 years, 3 years, 5 years.”

My take on Huron: the investigation is unlikely to stop the Indian business, which earns most of the profit. The risk lives in the tail. An adverse finding could bring fines, an impairment of up to ₹882 crore of parent exposure, reputational damage with exactly the Western aerospace primes Jyoti lists as customers, and tighter licensing on any 5-axis machine using Siemens or Fanuc controllers, which Jyoti’s India-built machines also need. And it sits inside a French court process that no Indian investor can monitor quarter by quarter.

The cash that never quite arrives

The Huron issue would matter less if the Indian business were throwing off cash. It has not been.

Exhibit 6: Three years of record profits, negative cumulative operating cash flow.

Over FY24 to FY26, Jyoti reported cumulative PAT of ₹803 crore and cumulative operating cash flow of minus ₹99 crore. FY26 CFO was ₹54 crore, 16% of PAT on a consolidated basis and 26% standalone.

Where did the profit go? Into working capital. At March 2026, inventory of ₹1,166 crore plus receivables of ₹599 crore plus unbilled revenue of ₹613 crore added up to ₹2,378 crore, more than a full year of revenue. The cash conversion cycle stretched to about 360 days from 304. Customer advances were only 1.7% of the order book. Standalone unbilled revenue rose 35% while revenue rose 21%, even though the auditor’s key audit matter describes revenue as recognised at a point in time on dispatch; which contracts use percentage-of-completion is not disclosed.

Meanwhile, borrowings are back to about ₹850 crore (net debt about ₹700 crore), with ₹439 crore due within a year and about ₹260 crore of floating-rate euro debt with no forward cover. The ₹400 to ₹450 crore Rajkot expansion is largely debt-funded through term loans running to 2032-33, and FY27 capex is guided at ₹200 to ₹250 crore.

The smaller flags, in one place

Guidance against delivery

The pattern is clear enough: margin guidance has broadly held; growth, timing and cash promises have repeatedly slipped.

What could go right

To be balanced, there is a genuinely large prize here.

• The 16,000-machine Rajkot plant. If it fills, FY27 production would move towards the guided 8,000+ machines from 5,550 in FY26.

• Backward integration into electronics. On 17 August 2026 MeitY approved a ₹1,020.65 crore, five-year electronics components investment proposal (with up to a 25% capex incentive) for in-house controllers, drives and motors. Management says this could lift gross margin by 4 to 5 points once commercial in about two years. If Jyoti genuinely builds its own controller, it would own the one row of the bill of materials no Indian builder owns today. The overlap with the existing capex plan has not been disclosed.

• EMS. A 13% share of the order book is EMS, against single digits of revenue. If electronics customers get their factories ready, that converts.

• Huron resolution. A clean close of the investigation would release WIP, seized cash and, probably, the multiple.

Verdict on Jyoti, as a business

Jyoti owns the better franchise in Indian machine tools: scale, deep integration, a real 5-axis capability and an aerospace customer list no Indian peer can match. It is also simultaneously absorbing an open-ended French judicial investigation, three years of negative cumulative operating cash flow, rising debt and visible mix-driven margin pressure. At about 60 times forward earnings, the price is assuming the new plant fills, margins hold at 25% while realisations fall, the cash cycle normalises and Huron’s investigation ends quietly. All four have to go right.

Part 8: Macpower CNC Machines, the Neighbour That Kept It Simple

NSE: MACPOWER | CMP ₹1,925 | Market cap about ₹1,926 crore | P/E 49x trailing (all as of 16 Oct 2026)

From the “Turner” lathe to the main board

Macpower was founded in 2003 in Rajkot as a lathe maker, selling under the “Turner” brand. The CMD, Rupesh Mehta, has 29 years in machine tools and per company disclosures is an IMTMA executive director; his brother Nikesh Mehta is CEO. The Mehta family holds 73.2%, with no pledge. The company listed on NSE Emerge in 2018 and moved to the main board in August 2020.

It is, in structure, the opposite of Jyoti: a single entity, no subsidiaries, one segment, on 8 acres at Metoda GIDC with a capacity of 2,500 machines a year (since 10 November 2025) and 950+ employees.

What Macpower makes and who buys it

Macpower sells 383+ variants across turning, twin-spindle, VMC, HMC, VTL, drill-tap, turn-mill, double-column, grinders and 5-axis, with a premium basket branded NEXA. But the business is overwhelmingly turning centres and VMCs, sold to small job shops, die-and-mould makers and tier-3 and tier-4 suppliers. Average realisation is about ₹20 lakh per machine (low end ₹12-20 lakh, mid ₹20-30 lakh, high ₹30 lakh to ₹2 crore), against Jyoti’s ₹35 lakh.

The go-to-market is exactly the job-shop model from Part 3: 10 branches, 8 technology centres, presence in 39 cities and 13,875+ installations. A prospect watches his own part being cut at a nearby tech centre. Management says more than half the order book is repeat business. No customer exceeds 10% of revenue, and exports were just 0.6% of FY26 sales.

Exhibit 10: The mix is climbing, slowly. VMC and HMC/double-column machines delivered about 70% of FY26’s incremental product revenue.

Defence and aerospace get a lot of space in the marketing (200+ machines to 35 defence factories, HAL and ISRO references), but they are still a single-digit share of revenue. The defence order book is about ₹20 crore, with ₹304 crore of tenders under evaluation, some open for two years. And on 5-axis, management was refreshingly blunt in July 2026: it “will not be my business model”, citing controller export licensing (BAFA, Germany’s export control office) and end-use compliance. It expects to sell one or two a year.

Read that again in the light of the Huron story. Macpower looked at the high end, saw the export-control regime, and chose not to build a business there. Whether that is wisdom or lack of ambition depends on your time horizon.

Backward integration is real but partial: about 40 to 45% of components by count are made in-house (spindles, sheet metal, powder coating, machining), with a stated goal of 75 to 80%. Spindle and sheet-metal capacity of roughly 1,300 to 2,000 sets a year sits below the 2,500-machine assembly capacity, which is why job work rose from ₹9.5 crore to ₹13.4 crore in FY26.

The numbers

Exhibit 5: The margin step-up came in FY24. Operating leverage since then has mostly been absorbed by job work, transport, marketing and rent.

Over a longer arc, FY14 revenue was ₹41 crore, a 19% CAGR to FY26. FY26 revenue grew 27.3% to ₹333 crore, EBITDA was ₹53.9 crore (16.2%) and PAT grew 33% to ₹33.9 crore. Q1 FY27 revenue rose 56% to ₹95.2 crore and PAT 110% to ₹9.6 crore, on a soft base. The order book was ₹456 crore at 30 June 2026, up 32%, about 1.4 times FY26 revenue, with 40% from NEXA.

Two cautions. The 13-year record includes a 42% revenue collapse in FY20, the clearest reminder that this is a cyclical, SME-financed capital good. And FY26 EBITDA margin of 16.2% undershot management’s own aim of about 18%.

Guidance against delivery

Revenue promises are credible. Margin, land and timing promises are not.

The weaknesses, in one place

Macpower is cleaner than Jyoti, but it is a long way from clean. Here is everything I found, in one table, because a reader should see the warts before forming a view.

My take: nothing here suggests fabricated profits. The anomalies are small in rupees, and a business that sells at the point of dispatch, with no subsidiaries and no debt, leaves very little room to hide. But the disclosure is sloppy, the presentations overstate, and the accounts need to be read line by line. The investor’s consolation is that every one of these weaknesses shows up in a half-yearly filing, where it can be checked.

Where Macpower goes next

The next leg is a ₹50 crore assembly and backward-integration project on a 13-acre, 30-year lease near Metoda, targeted for about mid-2027, eligible for a 25% capital subsidy (paid over five years) and 7% interest subvention under the Viksit Gujarat Industrial Policy 2026, funded by internal accruals plus some debt.

This matters because existing capacity tops out at roughly ₹450 to ₹500 crore of revenue (2,500 machines at about ₹20 lakh). FY27 guidance fits inside today’s plant. FY28 growth depends on the new site arriving on time, and this management’s record on land and timelines is the weakest part of its file. The larger government-land project remains unallotted, and its description keeps shifting between a 30-acre greenfield and a 60-acre defence-policy allotment, with three different capex budgets in the same annual report.

Verdict on Macpower, as a business

Macpower is the simpler, cleaner and less fragile business: no subsidiaries, no debt, no regulatory overhang, a 26 to 27% ROCE and revenue guidance it has met. Its weaknesses are cash conversion, inventory, receivable ageing, the land timeline and the gap between presentation claims and reality. It is also a ₹1,926 crore company with a free float of roughly ₹515 crore and under 2% institutional ownership (FII 0.65%, DII 0.97% at September 2026), whose retail shareholder count rose 30% in the September quarter alone. Liquidity and momentum ownership are part of the risk.

Part 9: Head to Head

Exhibit 8: Both cash cycles lengthened in FY26. Jyoti’s is twice Macpower’s.

The scoreboard

Let me compress all of that into five tests. I set these before I tallied them, and I have deliberately included tests Jyoti wins.

It is a draw. Two each. Neither company passes the cash test.

But look at which tests each company wins. Jyoti wins on what the business is: capability and margin. Macpower wins on what can go wrong and what you pay: balance sheet, legal tail and price. The scoreboard tells you these two stocks are bets on different things, and that is more useful than any single ranking.

Five observations on what the table means:

1. Quality of earnings favours Macpower, but neither is a cash machine. Jyoti generated negative cumulative operating cash flow over three years of record profits; Macpower converted less than half. Both argue that inventory is strategic. Only Macpower funds it without debt.

2. Jyoti’s margin premium is real but narrowing. Its 9-point EBITDA advantage comes from integration and high-end mix. The incremental capacity is overwhelmingly entry-level, realisation per machine is falling, and management itself guides towards ₹30 lakh and 25% margins.

3. Visibility favours Jyoti on paper, Macpower in practice. Jyoti’s order book is deeper, but converting it depends on a 10,000-machine plant ramp, Huron export licences and EMS customers getting their factories ready. Macpower’s is shorter but turns into revenue within months; its constraint arrives in FY28, when the existing plant fills.

4. The tail risks are asymmetric. Macpower’s worst plausible year looks like FY20: a cyclical demand drop that hurts earnings but not the balance sheet. Jyoti’s worst plausible outcome combines a demand slowdown with ₹850 crore of debt, a write-down of up to ₹882 crore of exposure to the French group, and reputational damage with Western aerospace customers.

5. The market charges about 40% more per rupee of forward earnings for the business with the legal overhang and the worse cash conversion.

Part 10: What You Have to Believe

Here is the cleanest way I know to think about the price. It is illustrative arithmetic, not a forecast or a target.

Take each company’s forward P/E on FY27E earnings at the 16 October close. Assume an investor wants to earn 12% a year for five years, ignores dividends, and assumes the stock exits at 30 times earnings, which is roughly where Macpower’s own five-year median P/E (about 31x) sits. Then ask: how fast would earnings per share need to grow for that to work?

Exhibit 9: Illustrative arithmetic. Jyoti’s price asks for more growth than it delivered last year; Macpower’s asks for less.

ILLUSTRATIVE ARITHMETIC. IPR calculation on 16 Oct 2026 closes. FY27E assumes Jyoti meets 25-30% revenue growth at about 25% EBITDA margin and Macpower meets 30% revenue growth at a 16.5-17% margin, per management commentary. Not company forecasts.

Read the two rows together. Jyoti’s price needs about 29% compound EPS growth for five years, after a year in which consolidated profit barely grew. Its Indian business is growing much faster than that consolidated number suggests, so the number is achievable if Huron stops subtracting from it. Macpower’s price needs about 20%, and it is currently growing faster than that. But Macpower also trades at about 1.6 times its own five-year median P/E on trailing earnings, after more than doubling from its 52-week low of ₹761, and its growth beyond FY27 depends on a factory that does not exist yet.

Neither stock is cheap. One simply asks for less heroism than the other.

Part 11: What I Think

Let me separate the industry, the businesses and the stocks, because they deserve different answers.

The industry. India’s machine tool opportunity is real and probably early. ₹21,628 crore of annual imports, a 4% domestic share of aerospace and defence needs, an industry body projecting ₹54,000 crore of consumption by 2030, and record production in FY26. But the opportunity is largest exactly where Indian builders are weakest, and the competition is growing fastest exactly where they are strongest. There is no tariff wall coming to help.

Jyoti, the business. The best-built machine tool franchise in listed India: integrated, profitable, with a 5-axis brand and aerospace relationships that took nearly two decades to build. If it fills the 16,000-machine plant, builds its own controllers and puts Huron behind it, it could become something genuinely rare: an Indian capital goods company competing with Japan and Germany at the high end.

Macpower, the business. A good small company with ordinary disclosure. It is a volume player that is climbing slowly, earns better returns on capital than its bigger neighbour, and has chosen, explicitly, to stay away from the part of the market where export controls live.

The stocks. At 16 October prices, the market is paying the bigger premium for the business with the bigger tail. That is the single sentence I would want a reader to take away.

Which brings me back to Hannibal. He won Trebia, Trasimene and Cannae. He was, by any measure, the better general. He lost because victories in the field did not turn into supplies, reinforcements and a captured city. Jyoti has won its battles too: the Huron acquisition that looked like overreach, the IPO, the margin inflection, the aerospace customers. What it has not yet done is turn those victories into cash. Its operations consumed cash across three years of record profit, and its supply line now runs through a French court. Whether Jyoti turns out to be Hannibal depends on whether the cash and the court case come good. I do not know the answer. Nobody outside that courtroom does.

What I’m doing:

I hold Macpower CNC. I do not hold Jyoti CNC. My reasons are the ones in this piece: a debt-free balance sheet, no legal tail, and weaknesses I can check in every half-yearly filing. The things that would make me change my mind are listed below.

What I’m Watching

1. H1 FY27 cash flow, for both (November 2026). Q2 results come with the half-year balance sheet and cash flow statement. For Macpower: is operating cash flow at least half of PAT, and are inventory plus receivables growing more slowly than revenue? For Jyoti: is it on track for management’s guide of CFO at about 50% of EBITDA?

2. The Huron investigation. Any closure, penalty, release of the ₹32.9 crore seizure, export licences for the ~₹100 crore of machines in WIP, or a decision to sell, ring-fence or write down the French business.

3. Jyoti’s new Rajkot plant. Commissioning confirmation (due end-September, not confirmed in filings I had seen by 16 October), production heading towards 8,000+ machines in FY27, and whether EBITDA margin holds above 23% as the mix shifts entry-level.

4. Macpower’s 13-acre site. Lease signed July 2026, completion targeted for mid-2027. If construction has not visibly started by March 2027, FY28 revenue is effectively capped near ₹500 crore.

5. IMTEX 2027, Bengaluru, 21 to 27 January 2027. The industry’s order-booking barometer, and the best read on whether FY27’s record demand is holding up, and on how aggressively Chinese and localising foreign builders are pricing.

See you at the next inflection point.

Karnik

Inflection Point Research

Get Two Free IPR Sample Reports

India’s machine tool import gap is the kind of structural opportunity IPR extends into specific, forensically screened names for subscribers. Two free sample reports show that format applied to individual companies: Gravita India (Future Titans) and Kalyani Cast Tech (Microcap Mavericks).

Get Free Samples

No payment required. SEBI Reg. INH000023807.

Portfolio Disclosure

The author and his family hold positions in Macpower CNC. No positions are held in Jyoti CNC by either the author or his family. The author has received no compensation from either company.

Figures are from company annual reports (FY26), investor presentations and earnings call transcripts (Q4 FY26 and Q1 FY27), exchange filings, IMTMA and Gardner’s World Machine Tool Survey, as stated. Aggregator-sourced figures are labelled T3. FY27E earnings, forward multiples and implied growth rates are IPR’s translation of management guidance and illustrative arithmetic, not company forecasts. Market data as of the 16 Oct 2026 close.

SEBI Disclaimer

DISCLAIMER: Karnik Shah is a SEBI-registered Research Analyst (Registration No. INH000023807). This content is for educational and informational purposes only and should not be construed as investment advice or a recommendation to buy, sell, or hold any security. Readers should conduct their own due diligence and consult a qualified financial advisor before making any investment decisions. The author may or may not hold positions in the securities discussed. Past performance does not guarantee future results. Investments in securities are subject to market risks. Read all related documents carefully.

Hello friends,

Picture a road developer in Rajasthan. He has spent four years and a few hundred crore building a four-lane highway. The tolls are now flowing, the traffic counts are better than his original estimate, and the asset is, by any measure, a success.

He wants to sell it.

He has nothing against the road. A finished toll road is a slow, steady, boring cash machine, and his business is building things. Every rupee of equity sitting in a completed highway is a rupee he cannot put into the next one. So he looks for a buyer: someone who wants twenty years of steady toll receipts, knows how to run a road better than a construction company does, and has patient capital to write a cheque of several hundred crore.

Now picture a mid-sized chemicals promoter who needs ₹500 crore to buy out a private equity partner. His bank will lend him against a factory. It will not lend him against his own shares to fund a buyout. Or a Pune developer whose housing project is 70% complete and has run out of construction finance. Or a steel company that went through insolvency, came out the other side with positive EBITDA, and still cannot get a bank to look at it.

In all four cases, somebody has to show up with money that banks will not or cannot provide.

EAAA India Alternatives, the alternatives arm of the Edelweiss group, is that somebody. Its funds own 4,586 lane-kilometres of Indian roads, 1,835 circuit-kilometres of power transmission lines, 1.73 GW of solar plants and 3.10 million square feet of offices. Its credit funds have made 190 investments and fully exited 140 of them. It manages ₹95,345 Cr of commitments for Canadian, German, Swedish and American pension funds and insurers, Indian family offices and thousands of HNIs. By its own commissioned industry report, it is the largest alternatives platform in India.

On 26 October it comes to the market. The price band is ₹1,585 to ₹1,665 a share, valuing the company at up to ₹10,765 Cr. The entire ₹1,200 Cr issue is a sale of shares by the Edelweiss holding company; EAAA itself raises nothing.

This piece is a full company analysis. I am going to do something slightly different from my usual forensic-first format, because this company deserves to be understood as a business before it is judged as an IPO. We will start with why Indian private capital is growing at all, go through how an alternatives manager actually makes money, walk through EAAA’s 18-year history, its two engines, its clients, its people and its competitors, and then look at where it could go. Only after all of that will we get to the price, the risks and the red flags. There are some, and one of them is a sentence the prospectus leaves out entirely.

If you want the elevator version first, here it is.

EAAA is a genuine franchise: the largest Indian alternatives platform, with 17 years of fund series, real exits, real carry received in cash and a 26% return on equity. The structural tailwinds behind private credit and infrastructure yield in India are real and probably early. The questions are about the quality of the recent growth, a regulatory file that is still open, and a price that asks you to believe the best version of the story.

Pour something cold. This is a long one.

Part 1: The Money Banks Left Behind

Every business in this piece exists because of a gap. To understand EAAA, you have to understand the gap.

The banks went retail

Between FY18 and FY26, the share of corporate loans in Indian bank credit fell from 69% to 55.3%. For NBFCs, the fall was from 80% to 62.5% (RHP p.216, citing the CARE report). After the twin balance sheet crisis, the IL&FS collapse and the NBFC freeze of 2018 to 2019, Indian lenders did the rational thing: they moved towards home loans, credit cards, gold loans and personal loans, where the data is better, the tickets are smaller and the losses are more predictable.

The companies did not stop needing money. A promoter funding an acquisition, a developer finishing a project, a mid-sized company refinancing a lumpy maturity, a business coming out of insolvency: these are exactly the kinds of loans a bank’s credit committee finds hardest to approve. They need structure, security over shares or cash flows, and a lender who will spend time understanding one company deeply instead of approving a thousand loans through a scorecard.

That is private credit. In India it is about 0.6% of GDP, against 4% to 4.5% in developed markets. CARE projects Indian private credit AUM to grow from US$26.7 bn in CY26 to US$56 to 65 bn by CY30 (p.215).

The developers want their money back

The second gap is in infrastructure. Private developers spent roughly ₹13 lakh Cr on core infrastructure between FY17 and FY23 (p.249). Much of that is now operating: roads collecting tolls, transmission lines earning availability payments, solar parks selling power under long contracts. The developers would rather recycle that equity into new projects. CARE puts the asset recycling opportunity at ₹25 lakh Cr (p.249).

Somebody has to buy those assets. Since 2019, India has built the plumbing to make that easier: InvITs and REITs give a listed exit for a portfolio of operating assets, which means a fund can buy assets one at a time, run them well, bundle them and sell the bundle to the public markets.

The stressed pile and the domestic money

Two more forces complete the picture. The Insolvency and Bankruptcy Code created a functioning market for stressed assets, which CARE sizes at US$82 to 85 bn (p.246). Banks selling bad loans need buyers; turnarounds need fresh money.

And the investor base has changed. Indian HNIs and family offices have discovered alternatives in a big way. SEBI data shows total AIF commitments reached ₹15.74 lakh Cr by December 2025, up about 21% in a year (SEBI data via Cafemutual, March 2026). Domestic insurers and pension funds, which manage around ₹135 lakh Cr, are being given more room to invest in AIFs, slowly (p.240).

Put it together and CARE projects India’s “yield and income” strategies (private credit plus real assets) to grow from US$50.6 bn in CY25 to US$112.9 bn by CY30, at 16% to 18% a year, roughly twice the projected growth rate of Indian private equity (p.234).

One caveat before we go any further, and it applies to every market figure in this piece. The CARE report was commissioned and paid for by EAAA for this IPO (p.228). The broad direction is consistent with what you can see independently in SEBI’s AIF data and in bank credit mix. The precise forecasts deserve a pinch of salt.

My take: the structural case for Indian private credit and infrastructure yield is one of the more durable themes in Indian financial services. Unlike many IPO stories, this tailwind is visible in regulator data and in the behaviour of banks, and it does not depend on a single policy or a single cycle.

You can watch a video I made on this topic, here:

Part 2: How a Fund Manager Gets Paid (A Two-Minute Primer)

You cannot value EAAA without understanding how an alternatives manager earns money, so let me take two minutes on it.

Closed-ended funds. An investor commits, say, ₹10 Cr to a fund. The manager draws that money down as it finds deals, invests for a few years, and returns money as loans are repaid or assets are sold. EAAA’s investors are locked in for 4.5 to 10 years depending on the fund (p.241). The lock-in is why this business has unusually sticky revenue.

Three income streams.

1. The management fee. A percentage of committed, invested or NAV capital, charged every year regardless of performance. EAAA can charge up to 2.25%; in practice it has earned 1.43% to 1.49% of fee-paying assets (p.233, p.254). Think of it as the toll on every vehicle that uses the road.

2. Carried interest. If a fund beats a hurdle return for its investors (typically 8% to 12% a year), the manager earns a share of the profit above it, up to 20% for EAAA (p.254). EAAA calls this Variable Additional Return, or VAR. The rest of the world calls it carry. It is a performance bonus that arrives years after the money is invested.

3. Returns on its own money. Managers invest alongside their clients; SEBI requires at least 2.5% or ₹5 Cr for most AIFs (p.45). EAAA calls the return on this “GP income”.

AUM versus FPAUM. This is the single most important distinction in the piece. AUM is total commitments: ₹95,345 Cr at June 2026. FPAUM (fee-paying AUM) is the slice that pays a management fee today: ₹48,623 Cr (p.228). The difference comes from older funds in harvest mode (which charge on a shrinking NAV), funds whose fee base is invested rather than committed capital, and newer products such as InvITs and a fund of funds. For anything to do with earnings, FPAUM is the number that matters.

The catch with carry. Carry is earned in cash only when a fund exits investments. Accounting rules let a manager book expected carry earlier, based on the fund’s current valuation. That creates a gap between profit and cash. It is normal for the industry. How big the gap is, and how fast it closes, is one of the things we will look at.

With that in hand, let us meet the company.

Part 3: From a Credit Desk to India’s Largest Alternatives Platform

Chapter 1: Inside Edelweiss (2008 to 2012)

Edelweiss was founded in 1995 by Rashesh Shah and Venkat Ramaswamy and listed in 2007 (context, not from the RHP; the RHP confirms Shah has been associated with the group since 1995, p.279). Through the 2000s it grew from an investment bank into a sprawling financial conglomerate: broking, wealth management, NBFC lending, an asset reconstruction company, insurance and a mutual fund.

EAAA began as one corner of that empire. It was incorporated in May 2008 as Edelweiss Alternative Asset Advisors, wholly owned by Edelweiss Capital, with a Singapore arm set up the same year (p.116, p.273). Its first performing credit fund, US$230 mn, closed in FY10 (p.250), at a time when “Indian private credit” was barely a category. A special situations fund followed in FY12.

Chapter 2: The global money arrives (2013 to 2019)

In 2015, Swedish insurer SEB Trygg Liv bought a stake at ₹270 a share, or ₹135 adjusted for the later split (p.116), becoming EAAA’s first outside investor.

The real step change came in FY17, with the first significant commitments from North American institutions and the raise of EISAF II, a special assets fund that would eventually gather ₹13,143 Cr of commitments and distribute ₹12,872 Cr back to investors (p.236). Real estate credit had launched in FY16; infrastructure yield and investment-grade credit followed in FY18 with EIYP, the first infrastructure yield fund.

Not everything went to plan. The 2016 to 2018 real estate credit vintages ran straight into demonetisation, GST, RERA, the NBFC crisis and then COVID. The RHP says plainly that many of those projects saw “cost and time overruns” (p.29).

Chapter 3: Scaling through the cycle (2020 to 2023)

During COVID, when most of the world’s capital was running for cover, EAAA raised US$890 mn in performing credit in FY21 (p.271). In FY23 it raised over ₹9,500 Cr from Indian investors and launched its InvIT platform (p.229, p.271).

Behind the scenes, the group was reorganising. Between 2020 and 2022, EAAA’s shares moved between group entities at ₹0.04 a share (₹0.02 split-adjusted), ending up with Edelweiss Securities and Investments Pvt Ltd (ESIPL), the holding company that is selling in this IPO (p.118 to 120). The group’s wealth business was sold to PAG and demerged as Nuvama in 2023 (context, not from the RHP).

Chapter 4: The storm and the IPO preparation (2024 to 2025)

FY24 brought the biggest infrastructure raise in the company’s history, over ₹8,000 Cr for Infrastructure Yield Plus II, and the launch of Rental Yield Plus, an office fund (p.271). ESIPL put ₹340 Cr of fresh capital into EAAA through convertible debentures (p.117).

Then, in May 2024, the Reserve Bank of India imposed business restrictions on two other Edelweiss group companies, ECL Finance and Edelweiss ARC. We will come back to that in the risks section, because the wording matters. The restrictions were lifted in December 2024.

In August 2024 the company was renamed EAAA India Alternatives, with Rashesh Shah as Chairman and Venkat Ramaswamy as MD and CEO (p.270, p.282). The first draft prospectus was filed in December 2024 and returned by SEBI in March 2025, reason undisclosed (p.551). The awards kept coming: IVCA’s “Best Overall Performance: Private Credit Fund” in both 2024 and 2025, and Private Debt Investor’s “Top Private Debt Fund Raisers” five years running (p.231).

Chapter 5: The record quarter (2026)

In March 2026, a group company sold 4.42% of EAAA to 36 outside investors at ₹1,322 a share (p.122 to 123). In the June 2026 quarter the company had the best fundraising period in its history: ₹20,380 Cr raised in three months, against ₹10,855 Cr in the whole of FY26 (p.232). That included the listing of the Citius Transnet InvIT and a ₹7,500 Cr fund of funds, ISAF IV (p.271).

The prospectus was filed on 19 October. The price band, announced on 20 October, is ₹1,585 to ₹1,665. The issue was trimmed from the ₹1,500 Cr proposed in the January 2026 draft to ₹1,200 Cr (press reports, T3).

The Edelweiss group has been explicit about why: value unlocking through stake sales, listings and demergers, and bringing down corporate debt. Per a summary of the group’s Q3FY26 investor materials, group corporate net debt was ₹6,520 Cr, with a stated aim of taking it below ₹3,000 Cr within 18 months (T3 summary; verify against the EFSL investor presentation). EAAA’s IPO is one of the pieces in that plan.

Part 4: Two Engines

EAAA runs 26 active funds, including two InvITs, organised into two verticals (p.229).

The table tells you the most important thing about EAAA’s recent history: the growth has come from real assets. Real Assets FPAUM nearly doubled from ₹14,380 Cr in FY24 to ₹28,040 Cr in June 2026, while Private Credit grew far more slowly. The company that started as a credit desk is now, measured by fee-paying assets, mainly an infrastructure and real estate manager.

The RHP does not disclose margins or profit by vertical, so I cannot tell you which engine is more profitable. I can tell you how each one works.

Engine 1: Real Assets, or “buy, fix, bundle, sell”

The company describes the strategy as acquiring “quality infrastructure assets with low counterparty risk, long residual tenures ... and low operating costs” (p.249). It has four sub-strategies.

Infrastructure Yield (since FY18). Buys operating roads, transmission lines and power assets from developers who want to recycle capital. This is the Rajasthan highway from the opening.

Commercial Real Estate Yield (since FY24). Buys operating office buildings in metros, upgrades them, and sells them to REITs or institutions. Rental Yield Plus, the first fund, has ₹4,158 Cr of commitments.

Perpetual Capital (since FY23). Listed InvITs that hold stable energy and transport assets for the long term: Anzen India Energy Yield Plus Trust and Citius Transnet Investment Trust, listed in Q1FY27 (p.249 to 250, p.271).

Energy Transition (since FY26). Greenfield renewables, green transmission and energy efficiency, through the India Energy Transition Fund, with a US$300 mn target and up to US$60 mn committed by EIB Global, the European Investment Bank’s development arm (p.248).

What makes this engine unusual in India is Sekura, EAAA’s 58-person in-house operating subsidiary. When a fund buys a road, Sekura runs it: maintenance, toll collection, safety, the control room. It operates across 16 states, with engineers and safety specialists on site 24x7, ISO certifications for asset management and safety, and a NABL-accredited transport lab in Turbhe (p.230, p.242). The roads have won MoRTH’s National Highways Excellence Awards three years running (p.242). This is a real capability. Plenty of fund managers own infrastructure on a spreadsheet; EAAA also owns it through people in hard hats.

The full loop looks like this. A fund buys operating assets one at a time. Sekura improves them. After a few years of clean cash flow, the assets are bundled into an InvIT and listed, and the fund exits at a higher valuation. EAAA often stays on as the InvIT’s paid manager, which converts a closed-ended fund into something closer to permanent capital.

The proof that the loop works is EIYP, the first infrastructure yield fund, raised in FY18. It has distributed ₹5,972 Cr on ₹3,281 Cr of commitments, and is nearly wound down with ₹5 Cr of NAV left (p.236). It returned more than 1.8 times what its investors committed. Its successor, IYP II, is a ₹9,044 Cr fund that is still investing, with ₹7,863 Cr of NAV and ₹4,407 Cr of capital yet to be called.

Since inception, Real Assets has raised ₹38,425 Cr, invested ₹28,744 Cr, realised ₹10,128 Cr, and completed 38 investments with 27 exits (p.249).

Engine 2: Private Credit, or “lending where banks will not”

The company offers “structured credit solutions to corporates, projects or sponsors that are relatively underserved by traditional credit providers for their stake enhancement, acquisition, completion and revival” (p.250). Four sub-strategies again.

Performing Credit (since FY10). Loans to profitable mid and large companies for growth, acquisitions and promoter stake purchases, secured on shares, property and cash flows. This is the chemicals promoter.

Special Situations (since FY12). Buys stressed loans from banks and funds one-time settlements and turnarounds of “inherently viable” companies with positive EBITDA (p.250). This is the steel company out of insolvency.

Real Estate Credit (since FY16). Construction and completion finance for mid-income and affordable housing in Mumbai, Pune, Bengaluru, Hyderabad, Ahmedabad, Kolkata, Chennai and NCR. This is the Pune developer.

IG Credit (since FY18). Higher-yield lending to companies that sit just outside what banks and debt mutual funds will touch.

The track record here is longer and deeper. Since inception, Private Credit has raised ₹52,146 Cr, invested ₹42,408 Cr and realised ₹43,973 Cr, across 190 investments and 140 full exits (p.250). The flagship special situations series is now in its fourth generation: EISAF II (₹13,143 Cr, FY17 vintage, ₹12,872 Cr distributed), ISAF III (₹11,903 Cr, FY22 vintage, ₹6,545 Cr distributed and still investing) and ISAF IV, a ₹7,500 Cr fund of funds raised in Q1FY27 (p.236, p.271). ESOF III, the special opportunities fund, has distributed ₹7,127 Cr on ₹8,429 Cr of commitments with ₹2,096 Cr of NAV still to come.

Why does Private Credit carry 56% of AUM but only 40% of FPAUM? Because big older funds such as EISAF II and ESOF III are in harvest mode. Their fees are now charged on a small remaining NAV (EISAF II’s FPAUM is just ₹517 Cr against ₹13,143 Cr of commitments). As those funds wind down, the fee base has to be replaced with new series. That is the treadmill every closed-ended manager runs on.

What I find most interesting: EAAA’s two engines have different personalities. Private Credit is the older, deeper franchise with the longer exit record, and it is where the special situations expertise lives. Real Assets is the growth engine, and it is where EAAA has built something genuinely hard to copy in Sekura. The company has quietly rotated from the first towards the second over the last three years.

The money, stream by stream

Hold on to that “fees as % of total” row. We will come back to it.

Part 5: Who Writes the Cheques

An asset manager’s customers are the investors in its funds, the limited partners. EAAA does not name them, but the RHP says a lot about who they are.

The global institutions. EAAA started with large foreign institutions and built its governance around their requirements. Global institutions have committed about US$4.96 bn (p.244). The top 10 clients, who account for 26.2% of FPAUM, include Indian family offices and insurers and pension funds in Germany, Canada, the USA and Sweden (p.24). Before a Canadian pension fund commits, it spends months on investment and operational due diligence: deal process, conflicts policy, valuation, cash controls, KYC, cybersecurity, often with a global consultant in tow (p.240). Once a manager passes that test, and keeps passing it across fund series, it is very hard to dislodge. The RHP calls this model difficult to replicate, and on this point I agree with it.

The Indian HNIs. Over the last three years, the mix has shifted sharply towards Indian money:

That shift was driven by distributors. EAAA went from 13 distribution partners in FY24 to 53 by June 2026 (p.240), and distributors raised 80.8% of FY26 commitments (p.27). It worked: the company now has 31 clients who have each committed more than ₹500 Cr (p.245).

The stickiness. 1,493 clients have invested in more than one fund, and 77% of AUM comes from repeat clients (p.228). The headline “11,009 client relationships” includes 5,295 InvIT unitholders added with the Citius listing (p.253), so I would use the repeat-client numbers as the better measure of loyalty.

My take: the client franchise is the real moat. Relationships with large pension funds, built over a decade and renewed through third and fourth fund series, are the hardest thing in this industry to buy or copy. The shift towards Indian HNI money is a sensible response to where wealth is growing, with one cost: distributor-sourced money is more expensive to raise, and we will see that cost in the numbers.

Part 6: The People

The independent directors include Neeta Mukerji (ex ICICI Bank, ARCIL, RBL Bank), Sampa Bhasin (ex SBI Capital Markets, EY) and Sunil Kakar (ex IDFC First Bank, Max Life) (p.279 to 280).

The broader team is 277 people: 85 investment professionals including 14 managing directors and 10 investment directors averaging over 21 years of experience, 58 at Sekura, 13 in client coverage and 108 in corporate functions (p.251). Senior investment people share in the carry (p.244), which aligns them with fund performance.

Two observations. First, the CEO’s background is telling. Amit Agarwal came up through ARCIL, India’s first asset reconstruction company, which fits a platform whose deepest expertise is in stressed and special situations credit. Second, the leadership has changed more than you would like in the run-up to an IPO. Venkat Ramaswamy, the co-founder who was MD and CEO, stepped down in September 2025 to become Vice-Chairman of the parent. One of two co-CEOs resigned in June 2026, and the head of real estate credit left in April 2026 (p.282, p.294). I will come back to this in the risks.

Part 7: The Competition

Here is how EAAA compares with other Indian alternatives managers, per the CARE report using Preqin data (p.200):

Among Indian managers, EAAA is roughly 1.7 times the size of the next player and has raised half as much again as 360 ONE over the last decade. CARE notes that the top three managers took about 75% of all fundraising in the Indian alternatives market in the last decade (p.241). Scale begets scale: large LPs prefer large, established platforms.

But this league table leaves out the real competition for deals and for global LP money. Blackstone manages about US$50 bn of India money, Brookfield about US$30 bn and KKR about US$10 bn (p.200). In Indian infrastructure and office real estate, those are the buyers EAAA bids against, and they have cheaper capital and deeper global relationships. Banks and NBFCs, flush with deposits, are also starting to expand wholesale lending again (p.223).

Listed comparables

No listed Indian company does exactly what EAAA does. The closest listed pure-play alternatives manager is Gaja Alternative Asset Management, much smaller, with FY26 total income of ₹158 Cr. 360 ONE WAM combines wealth management with a large alternatives business. Nuvama is wealth-led. The mutual fund AMCs (HDFC AMC, ICICI Prudential AMC, Nippon Life, Aditya Birla Sun Life, UTI) run a very different business: open-ended, fee-only, with almost no carry or balance-sheet income.

What EAAA says sets it apart, and how much I believe it

Part 8: Reading the Numbers

Now to the financials. The headline story is attractive:

• Fee-paying AUM grew from ₹30,398 Cr (FY24) to ₹44,710 Cr (FY26), 21% a year, and to ₹48,623 Cr by June 2026 (p.230).

• Management fees grew 20.6% a year from ₹398 Cr to ₹579 Cr.

• PAT grew 23% a year from ₹175 Cr to ₹265 Cr.

• ROE has been steady at 26% to 27%, and debt to equity is down to 0.34x (p.233).

• Q1FY27 PAT rose 45%, from ₹55.7 Cr to ₹80.7 Cr.

And here is the bull fact I want to state plainly, because it matters: carry is turning into cash. EAAA received ₹769 Cr of carry in cash across FY26 and Q1FY27 (p.354). Operating cash flow was ₹235 Cr in FY26 and ₹130 Cr in Q1FY27 after two negative years. The fund marks behind the carry proved real for the funds that exited.

Three engines, not one

To understand the quality of that growth, it helps to split the profit into the three income streams from Part 2, each net of the costs that go with it. This is my own decomposition from the RHP, and it reconciles exactly to reported profit before tax.

• Fee-related earnings (FRE): management fees minus the fixed costs of running the business (fixed staff cost, stock options, other expenses including amortised distribution cost, and depreciation). This is the annuity.

• Net carry: carry minus the bonus paid to staff out of carry.

• Balance-sheet income: returns on EAAA’s own money in its funds, minus interest cost.

Three things stand out.

First, the annuity is growing more slowly than the fees. FRE grew 10.9% a year from FY24 to FY26, while fees grew 20.6%. The FRE margin slipped from 37.0% to 31.2%. The reasons are visible: distribution cost rose from 12.9% of fees in FY24 to 16.1% in FY26, as the HNI channel grew; fixed staff costs rose about 20% a year; and stock option costs jumped from ₹0.8 Cr to ₹31.6 Cr. In Q1FY27, distribution cost reached 29.4% of fees against 14.0% a year earlier, the price of that record ₹20,380 Cr quarter. To be fair to the company, Q1 is seasonally weak for this measure: the FRE margin was 20.2% in Q1FY26 as well. The annual trend is the cleaner signal, and it is still down.

Second, carry is a growing but shared prize. Net carry rose from ₹10 Cr to ₹129 Cr between FY24 and FY26. But the bonus paid to staff out of carry has been large: 89% of carry in FY24, 69% in FY25, 55% in FY26 and 76% in Q1FY27 (p.34). That is how the industry retains talent. It also means shareholders keep less of each rupee of carry than the headline suggests.

Third, Q1FY27 was a balance-sheet quarter. The 45% PAT growth did not come from fees (FRE rose ₹3 Cr) or from carry (net carry fell ₹19 Cr). It came from ₹58.5 Cr of fair value gains on investments, mostly on EAAA’s own fund units, debt and pass-through certificates, of which ₹42.1 Cr was unrealised (p.354). Balance-sheet income was 47% of the quarter’s pre-tax profit, against 7% a year earlier.

This is still a good business, and the decomposition simply shows which part of it is doing the work: the “23% PAT growth” headline is built from three quite different engines, and the most predictable of the three has been the slowest.

Part 9: Where It Could Go

Here are the six avenues the RHP points to, each with evidence and an honest read of how far along it is.

1. Turning the record raise into fees (high visibility). FPAUM is ₹48,623 Cr, and IYP II and ISAF III still have ₹9,915 Cr of uncalled capital to put to work (p.236). At the recent fee yield, that gives a visible fee floor of about ₹690 Cr for FY27, against ₹579 Cr in FY26 (IPR estimate). This one is close to locked in.

2. The next carry wave (medium visibility). ISAF III, IYP II and Rental Yield Plus, the 2022 to 2024 vintages, hold ₹17,428 Cr of NAV between them (p.236). As they cross the three-year mark and start exiting assets, they can produce the next round of carry. Timing depends on exits.

3. The InvIT machine (medium). Each infrastructure fund can end in a listed InvIT that EAAA keeps managing, turning closed-ended capital into permanent capital. Citius Transnet and Anzen are the first two. A third and fourth would prove the model.

4. Energy transition (optionality). The US$300 mn India Energy Transition Fund, backed by EIB Global, is small relative to the base, but it opens the door to climate and development finance capital.

5. Indian institutions (optionality, multi-year). If Indian insurers and pension funds are given meaningfully more room in AIFs, EAAA’s institutional-grade processes and track record put it near the front of the queue.

6. Special situations (medium, and regulation-sensitive). Bank loan sales and IBC resolutions keep the pipeline full. This is also the strategy most exposed to regulatory scrutiny, for reasons we will get to.

What success would look like: FPAUM compounding at 18% or more without distribution cost rising faster than fees; the FRE margin back above 30%; carry received in cash keeping pace with carry booked; a third InvIT; clean closure of the SEBI matters; and, ideally, disclosure of fund-level returns.

Part 10: The Price

What the band asks for

Against listed peers

At the upper band, EAAA asks for about a 27% premium to the listed peer median on trailing earnings. Its economics, though, look much more like 360 ONE and Nuvama (costs around two-thirds of income, PAT margins in the 20s) than like the mutual fund AMCs, whose 50% to 60% margins come almost entirely from fees. The only listed pure-play alternatives peer trades at 20x.

The price ladder

The seller’s weighted average cost is ₹54.45 a share (p.134). The band sits 20% to 26% above what sophisticated private investors paid seven months ago, and 71% to 80% above the group’s own September 2025 transfer price. Because the ₹1,322 sale was 4.42% of the company, just under a 5% threshold, it does not appear in the SEBI table comparing past acquisition prices with the price band, which shows only ₹925 (p.151). In my view ₹1,322 is the more relevant reference.

What you have to believe

Here is the cleanest way I know to think about the price. Take the upper band of ₹1,665. Ask what FY29 profit EAAA would need to earn for an investor to make a modest 12% a year to March 2029, including dividends, at different exit multiples. Then compare it with three illustrative scenarios.

Exit P/E in FY29FY29 PAT needed for 12% a year from ₹1,665Multiple of FY26 PAT28x (where pure-play peers and our FRE work point)about ₹508 Cr1.9x35x (near 360 ONE today)about ₹406 Cr1.5x41x (today’s multiple holds)about ₹347 Cr1.3x

Read the two tables together. If EAAA keeps its 41x multiple, the base case does the job. If the multiple drifts towards 28x, you need the everything-goes-right case just to earn 12%. The IPO is, in effect, a bet that the market will keep valuing EAAA like a premium fee compounder rather than like the fee-plus-carry-plus-balance-sheet hybrid it is.

One more lens. If you value the fee business, normal carry and net investments separately on FY26 numbers (FRE after tax at 25x to 30x, normalised net carry at 10x to 12x, net investments at book), you get roughly ₹4,600 Cr to ₹5,430 Cr, or 43% to 50% of the market cap at the upper band. The other half of the price is paying for growth that has not happened yet.

Which brings me back to the toll. A toll booth is worth the traffic you can see, multiplied by the toll, minus the cost of running the booth. EAAA’s traffic is growing fast. The toll rate is steady at around 1.45%. The cost of running the booth has been rising faster than the traffic. At ₹1,665, the market is being asked to pay for the traffic it expects in 2029, at a multiple that assumes the running costs come back under control.

Part 11: Risks and Red Flags

I promised the forensic material would come last. Here it is, compressed. Each of these is drawn from the RHP; where something comes from outside it, I say so.

1. The sentence the prospectus leaves out

In May 2024, the RBI restricted ECL Finance and Edelweiss ARC, two group companies, citing group entities acting in concert “for evergreening stressed exposures of ECL, using the platform of EARCL and connected AIFs” (RBI statement of 29 May 2024, as reported by Business Standard/PTI). EAAA is the group’s AIF manager. The restrictions were lifted in December 2024.

The RHP describes the RBI matter as concluded. It does not say what the RBI found, or whether any EAAA fund was among the “connected AIFs”. On my full-text search, the words “evergreen” and “connected AIFs” do not appear anywhere in its 566 pages. The RBI release names no fund, and I have seen nothing that links an EAAA fund to those transactions. The issue is narrower: an investor cannot tell from the document either way, and for a business whose moat is institutional trust, that silence matters.

Separately, EAAA’s own regulatory file is still open:

• a SEBI show cause notice on its stressed-asset fund ESTAR, settled in September 2025 for ₹61.4 lakh without admission (p.35);

• SEBI administrative warnings on 16 July 2026 covering investment processes, investor disclosure and policy implementation, after a 2025 inspection flagged inadequate conflict-of-interest disclosures (p.422 to 423);

• settlement applications still pending; and

• a fresh SEBI inspection notice dated 16 September 2026, a month before the RHP (p.35).

2. The fee margin

Covered in Part 8. FRE margin down from 37.0% to 31.2% over two years, distribution cost at 29.4% of fees in Q1FY27. If the cost of raising HNI money keeps outrunning fees, the most valuable part of the business shrinks as a share of profit.

3. Profit runs ahead of cash

Cumulative PAT from FY24 to Q1FY27 was ₹751 Cr. Cumulative operating cash flow was ₹232 Cr, or 0.31x (p.315). Part of this is the normal timing gap on carry, and the gap has started to close: operating cash flow over FY26 and Q1FY27 was ₹366 Cr against PAT of ₹346 Cr. Part of it is structural, which brings us to the next point.

4. ₹668 Cr of commissions on the balance sheet, and a tax dispute

EAAA capitalises upfront and trail commissions paid to distributors and spreads them over each fund’s life. At June 2026, ₹668 Cr of such prepaid expenses sat on the balance sheet, 58% of net worth (p.326, p.345). Another ₹444 Cr of trail commission is committed but not yet on the books (p.368). For tax, the company deducts these costs when it pays them. For AY2023-24, the tax officer disallowed ₹154 Cr of them as capital in nature, the same logic the company’s own books use, raised a ₹34 Cr demand and started penalty proceedings. The company has appealed (p.423).

5. Cash out before the sale

EAAA paid no dividends in FY24 or FY25. It then paid ₹205.7 Cr in FY26 (78% of PAT) and declared another ₹61.1 Cr in July 2026: ₹267 Cr declared in twelve months, more than FY26 profit of ₹265 Cr, almost all of it to the promoter group (p.303). The company keeps ₹399 Cr of borrowings and ₹497 Cr of uncalled commitments to its own funds, and receives nothing from the IPO.

6. Group linkages and a borrowed brand

Three of my promoter-footprint checks fail. Crisil rates EAAA by combining the whole Edelweiss group (rating rationale of 7 June 2024, outside the RHP). Other group entities, including the selling shareholder, also sponsor and manage AIFs (p.297, p.434). And the “Edelweiss Alternatives” brand belongs to the parent, used under no-objection letters, with a royalty that “may be” charged in future (p.50 to 51). Group AMCs commonly use the parent’s brand, so I would weigh the third check lightly. The first two matter more here, because of the RBI’s language.

7. People, opacity and supply

• Leadership churn: the MD and CEO (September 2025), a co-CEO (June 2026), the head of real estate credit (April 2026) and an independent director (July 2026) have all left within 13 months. Fund documents let investors suspend capital calls if key people leave (p.47).

• No fund returns: the RHP gives no net IRR or DPI for any fund. The only performance figures are two misses: ESTAR-1 returned 10.71% against a 12% hurdle (p.42), and ESOF II returned below its 8% USD hurdle (p.24).

• Pay: key managerial remuneration was ₹85.9 Cr in FY26, 32% of PAT (p.362).

• Supply: the promoter group will still own about 84% after the issue and has to get public shareholding to 25%, so more selling is coming. Six-month lock-ins on the ₹1,322 buyers end around April 2027.

Part 12: What I Think

Let me separate the business from the IPO, because they deserve different answers.

The business. EAAA is the best-established Indian platform in two of the most structurally attractive corners of Indian finance. It has 17 years of fund series, a deep special situations franchise, an operating capability in Sekura that very few Indian managers have, global pension fund relationships that took a decade to build, and ₹769 Cr of carry received in cash in five quarters. If India’s private credit and infrastructure yield markets grow anywhere near CARE’s forecasts, EAAA will be one of the main beneficiaries.

The growth. The quality of recent growth is weaker than the headline. The predictable part of the business, the fee engine, has grown at 11% a year while costs ran ahead, and the IPO-quarter jump came mostly from marking up EAAA’s own fund units.

The regulatory overhang. The SEBI file is open, the RHP is silent on what the RBI meant by “connected AIFs”, and in this business trust is the product.

The price. At ₹1,585 to ₹1,665, the market is being asked to pay 39x to 41x trailing earnings, a premium to listed peers whose earnings are cleaner, at a level where the everything-goes-right scenario is needed unless the premium multiple holds for years.

My take: I like the business more than I expected to when I opened the prospectus, and I like the issue less. For my own money, I am not applying in this IPO. EAAA goes on my watchlist. What would change my mind: SEBI settlement orders passed and disclosed with no bar on fundraising and the September inspection closed; two consecutive half-years of fee-related margin at or above 30% with distribution cost below 18% of fees; and a price that does not require the best case. Listed, with quarterly disclosure, this becomes a much easier company to follow.

What I’m Watching

1. The first listed results (H1FY27). Fee-related margin and distribution cost as a share of fees. This is the single number that tells you whether the fee engine is getting more or less profitable.

2. Carry booked against carry received. The carry asset fell from ₹663 Cr to ₹313 Cr as cash came in. I want to see cash keep pace with bookings, and no clawbacks.

3. The SEBI outcomes. The pending settlement orders and the September 2026 inspection. Any non-monetary terms, or any finding linking EAAA funds to group evergreening, would change the picture.

4. Deployment and the next InvIT. ₹9,915 Cr of uncalled capital in IYP II and ISAF III, and whether a third InvIT lists.

5. Promoter sell-down. The path from about 84% to 75% promoter holding, and the April 2027 lock-in expiry.

See you at the next inflection point.

Karnik

Inflection Point Research

Get Two Free IPR Sample Reports

The kind of analysis you just read, splitting reported profit into the engines that actually produce it, is the foundation of IPR’s research method. Two free sample reports show the full framework applied to individual companies: Gravita India (Future Titans) and Kalyani Cast Tech (Microcap Mavericks).

Get Free Samples

No payment required. SEBI Reg. INH000023807.

Portfolio Disclosure

The author does not hold a position in EAAA India Alternatives Ltd, which is unlisted at the time of writing, and has not applied in the IPO. The author’s family holds a position in Nuvama Wealth Management Ltd. The author has received no compensation from the issuer, the selling shareholder or the book running lead managers.

Figures are from the Red Herring Prospectus dated 19 October 2026 (printed page references) unless stated. Outside sources: RBI statement of 29 May 2024 (as reported by Business Standard/PTI); Crisil rating rationale of 7 June 2024; SEBI AIF data via Cafemutual (March 2026); price band and issue details per press reports of 20 to 21 October 2026 and Chittorgarh. The earnings decomposition, valuation arithmetic and scenarios are IPR estimates.

SEBI Disclaimer

DISCLAIMER: Karnik Shah is a SEBI-registered Research Analyst (Registration No. INH000023807). This content is for educational and informational purposes only and should not be construed as investment advice or a recommendation to buy, sell, or hold any security. Readers should conduct their own due diligence and consult a qualified financial advisor before making any investment decisions. The author may or may not hold positions in the securities discussed. Past performance does not guarantee future results. Investments in securities are subject to market risks. Read all related documents carefully.

Hello friends,

Tonbo is the Japanese word for dragonfly. A dragonfly is, more or less, a pair of eyes with wings attached. Its compound eyes carry tens of thousands of individual lenses and wrap almost all the way around its head, which is why it is one of the most effective hunters in the insect world.

It is a fitting name for a company that builds eyes for soldiers. Tonbo Imaging makes the thermal sights, binoculars, vehicle camera rings, drone gimbals and missile seekers that let a soldier, a tank or a missile see in total darkness, through smoke and through camouflage. Its lightest rifle sight weighs 280 grams; the comparable sights in its own industry report weigh around 800 to 900. Over the two years to March 2026, it supplied about half of all the thermal imaging units the Indian government bought through GeM, the government’s online procurement portal. And it did this as a design house in Bengaluru, without owning a factory.

This week Tonbo comes to market. The IPO opens on 15 October, and it is one of the more interesting defence listings of the year, for two reasons that pull in opposite directions. The technology and the Indian Army franchise are real, and I want to give them their full due, because most IPO coverage skips straight to the red flags. The issue itself, though, is a 100% offer for sale in which the company receives nothing, and the largest sellers are two holding entities whose history with the company is the most important thing to understand before you form a view.

So this piece runs in two halves. The first is the company: the backdrop, the technology, the products, the history, the customers, the people and the growth avenues. The second is the numbers and the prospectus: what the financials say, who is selling, what the price assumes and what I will be watching after listing.

Pour something warm. This one is long.

If you would rather watch a video I made on this company, you can do so here:

Part 1: Why defence electronics is having a moment

Start with the money. India’s Ministry of Defence budget for FY27 is ₹7.85 lakh Cr, with ₹1.85 lakh Cr earmarked for capital acquisition, up about 24% (p.182). Of that capital acquisition budget, 75% is reserved for domestic industry (p.186). Five positive indigenisation lists covering 509 items now bar imports of the listed equipment (p.186). Defence exports reached ₹38,424 Cr in FY26, against a target of ₹50,000 Cr by FY30 (p.190).

Then look at how wars are now fought. The industry report frames the shift in three substitutions (p.165): missiles giving way to loitering munitions, fighter jets to unmanned combat drones, and heavy armour to nimble unmanned ground vehicles. Ukraine made the point vividly, with drones carrying thermal turrets guiding artillery in real time (p.204). Closer to home, Operation Sindoor in 2025 triggered emergency procurement (p.181).

Every one of those substitutions needs the same thing: cheap, small, smart eyes. A loitering munition is useless without a seeker. A drone without a stabilised camera is a toy. A counter-drone system must first find the drone. That is the product thesis Tonbo has built around.

The market sizes in the RHP’s industry report:

Two honest caveats from the same report, because they matter later. Tactical EO/IR, the soldier-carried category that is still Tonbo’s largest product line, is the slowest-growing EO/IR segment globally at 5.1% a year (p.206). And the report itself calls emergency procurement “episodic” (p.164).

Part 2: A two-minute primer on thermal imaging

The whole business rests on one piece of physics, so it is worth a minute.

Everything warm glows in infrared. People, engines and gun barrels give off heat radiation the human eye cannot see. A thermal camera converts that heat into a picture. Because it reads emitted heat rather than reflected light, it works in pitch darkness and sees through smoke, dust and camouflage (p.203 to 204). It is also passive: unlike a searchlight or a laser, it gives nothing away about the person using it.

The detector is the heart. At the centre of every thermal camera sits a heat-sensing chip called a focal plane array. There are two families (p.254):

• Cooled detectors chill the sensor with a tiny refrigerator called a cryocooler. They give the best range and sensitivity, so they go into long-range surveillance and high-end targeting. They are expensive, power-hungry and mechanically delicate.

• Uncooled detectors (microbolometers) work at room temperature. They are cheaper, smaller, more reliable and use far less power.

Tonbo’s central engineering bet has been to push uncooled cameras up to defence-grade performance, so they can do jobs that used to need cooled ones (p.254). Hold that thought; it explains both the rifle sight and the missile seeker below.

Where the value sits. Around the detector, someone has to design the lenses (often germanium, which transmits infrared), the image-processing electronics, the software that cleans and sharpens the picture, and the mechanics that hold the image steady on a moving vehicle or drone. That integration layer is Tonbo’s business. The high-end detectors themselves are mostly imported across Indian industry, and global supply is concentrated in the US, France and Israel (p.209). Lynred of France and SCD of Israel both appear in Tonbo’s top-ten supplier list (p.25).

India’s own journey here is short. For decades the country imported its thermal gear: Israeli Litening targeting pods, French Thales sights for Army tanks. Since about 2014, DRDO laboratories, Bharat Electronics and private firms like Tonbo have been localising it (p.204).

Part 3: From a Princeton lab outpost to an IPO

Tonbo’s history is unusual for an Indian defence company, and it explains a great deal about the structure we will meet in Part 9.

2003 to 2008: a research outpost. The company was incorporated in Bengaluru in December 2003 as Sarnoff Innovative Technologies, the Indian subsidiary of Sarnoff Corporation, a US research company doing R&D and IP design work (p.247, 290). For context (not from the RHP), Sarnoff was the Princeton research lab descended from RCA Laboratories, later folded into SRI International. In 2005, within three months of each other, the three people who run the company today joined: Arvind Kondangi Lakshmikumar in April, Ankit Kumar and Cecilia D’Souza in June (p.302 to 303).

2008 to 2012: the management takeover. In August 2008 the company was renamed Serial Innovations India “due to a change in management” (p.290). In 2012 the promoters bought out the business and changed course from selling IP services to building products (p.247, 292). The holding layer that later became CEAQ Singapore and CEAQ India sits above the Indian company from this point (p.117).

2012 to 2018: becoming a product company. First production order in 2015 (p.292). Renamed Tonbo Imaging India in May 2016 (p.290). By 2018 it had crossed ₹30 Cr of export orders from Israel and the Philippines (p.292). Selling thermal gear into Israel, one of the most demanding defence buyers in the world, became the root of the company’s pitch: prove it abroad, then win at home (p.254).

2019 to 2023: the indigenous champion. In 2019 it won the Defence Ministry’s iDEX innovation challenge to design “see-through armour”, a 360-degree camera system for armoured vehicles (p.292). That became the Wolfpack product. India’s indigenisation push gave it a home platform, and it qualified as an IDDM supplier (Indigenously Designed, Developed and Manufactured), the top preference category in Indian defence procurement (p.247). HBL Engineering, a listed company, invested in April 2023 (p.393).

FY24 to FY25: the breakout. A USD 55 million (₹452.7 Cr) order from the government of Armenia (p.434) took revenue from about ₹97 Cr in FY23 to ₹428.2 Cr in FY24. In 2024 Tonbo integrated a seeker into an Indian government anti-tank missile, set up an in-house clean room and bought its core technology IP from its Singapore group entity. In 2025 it deployed a long-range missile tracking system, crossed ₹450 Cr of revenue and won a second iDEX challenge, this time for a high-power microwave counter-drone weapon (p.292 to 293).

FY26 to now: the pivot home. The Armenian order wound down without a repeat, and revenue from India more than doubled, helped by post-Sindoor procurement. The company converted to a public company in September 2025, appointed three independent directors and filed its first draft prospectus in December 2025. In late September 2026, a roster of marquee institutions bought shares from the holding entities at ₹700 (p.117 to 119). The RHP was filed on 10 October.

Part 4: The products

Tonbo describes its portfolio as an “intelligence stack” with four layers: seeing, understanding, communicating and controlling (p.253). Its stated mission is to “assist, augment and replace humans on the battlefield”. In practice the products fall into three verticals.

Tactical systems: what a soldier carries

These are self-contained products used directly by a soldier, paramilitary or special forces team (p.256, 261).

• Spartan and Spartan-S: a compact uncooled thermal sight that mounts on a rifle, on a helmet or works handheld, flipping between observation and aiming (p.264).

• ARJUN: a rugged long-wave infrared sight for rifles, machine guns and sniper rifles, built to the US military 810G standard (p.264).

• Bloodhound: thermal binoculars on a “shutterless” sensor core with glove-friendly controls and wireless video (p.263).

• Gazehound: a cooled, high-resolution imager for long-range border and coastal surveillance (p.263).

• Geo UC: a day camera, thermal imager, laser rangefinder, GPS and compass in one unit. A forward observer spots a target, gets its exact coordinates and sends them to the artillery, cutting the time to fire (p.264 to 265).

• Wolfpack: the 360-degree camera ring born of the 2019 iDEX win, letting a tank crew see all around without opening the hatch (p.263).

• Minion: a driver vision system for armoured vehicles at night (p.261).

The industry report benchmarks Spartan-S against two unnamed competing sights:

Weight is the whole game here. Under 300 grams is what makes a sight light enough to swap between a helmet and a weapon, and a soldier carrying fewer, lighter devices is a soldier who can move.

Platform systems: the fastest-growing part of the company

These are subsystems sold to whoever builds the platform: tank makers, drone makers, missile makers and weapon-station integrators (p.256, 261). This is where Tonbo’s growth has come from:

That is roughly 11x in two years.

• ELPEOS: a targeting head for remote-controlled gun turrets, combining a cooled thermal camera, HD day camera and laser rangefinder on a gyro-stabilised mount, with auto-tracking (p.265).

• AGS30 Infinity: a thermal fire-control sight for the AGS-30 automatic grenade launcher (p.265).

• Avenger-S gimbals: stabilised airborne camera balls in 8 kg, 25 kg and 50 kg classes for drones, helicopters, aircraft and aerostats (p.265 to 266).

• TRAP missile seeker: a two- or three-band infrared seeker that locks onto a target for “fire-and-forget” guidance in anti-tank missiles, glide bombs and loitering munitions, and can retrofit older missiles (p.266).

TRAP is the clearest expression of the uncooled bet. Missile seekers have traditionally used cooled detectors. TRAP uses an uncooled one:

Same range, a quarter of the failure rate, a tenth of the weight and less than half the power, at a fraction of the cost according to the report. The report adds that India, like many countries, has faced shortages of anti-tank guided missiles (p.233). In 2024 Tonbo integrated a multimode imaging seeker into an anti-tank missile developed by the Indian government (p.292).

The next act: counter-drone and directed energy

Nothing here earns revenue yet, but it is where the company is pointing.

• WaveStrike: a high-power microwave weapon designed to disable drones, including several at once, by frying their electronics. A tracker called TRex finds the target and hands it over (p.262, 266 to 267). Tonbo says it holds the IP for a system that could potentially neutralise drone swarms beyond 3 km (p.259). It won an iDEX challenge in 2025, was selected for an iDEX grant (amount not disclosed) and is “in the process of signing a contract for a prototype system” with a customer (p.259, 293).

• Loitering munitions: Tonbo says it is “now investing in the development” of a loitering munition built around its own seeker (p.256, 261). Its arms licence and objects clause were widened in 2026 to cover loitering munitions, remote weapon stations and projectile launchers (p.291, 454).

Why a microwave weapon matters: it costs almost nothing per shot and can hit many drones at once, which is exactly the problem militaries are trying to solve. The industry report cites Indian Army requests for information for 90 high-power microwave systems and Navy requests for 40 (p.222).

The technology underneath

The building blocks Tonbo designs in-house include polymer and diffractive lenses meant to cut dependence on scarce germanium, “shutterless” image correction, micro-scanning to squeeze more resolution out of a sensor, pixel-level fusion of visible and thermal images on low-power chips, laser ranging, inertial stabilisation precise to under 5 microradians, and edge AI that detects and classifies targets on the device itself (p.251, 253, 269). These are built as modules that recombine into different products, which is how a 130-person design team supports a catalogue this broad. Products also integrate with ATAK, the connected-soldier software used by NATO armies (p.257).

Part 5: How Tonbo makes money

Design-heavy, factory-light. Tonbo keeps design, IP, product engineering, testing and the customer relationship in-house, builds pilot units at its Bengaluru R&D centre and hands volume production to contract manufacturers (p.251, 270 to 271). Kaynes Technology built 45.3% of FY26 units and Avalon Technology 27.7%; in Q1FY27, Avalon built 87.1% (p.272). It owns no immovable property; everything is leased, including a new 39,519 sq ft engineering and operations site at Anekal on the edge of Bengaluru (p.278 to 280).

People and R&D. 339 employees at June 2026, of whom 130 work in product design and development with an average of 10 years’ experience (p.268, 276). R&D spend rose from ₹4.2 Cr in FY24 to ₹9.1 Cr in FY25 and ₹25.9 Cr in FY26 (p.254).

How it wins business. In India, through competitive tenders scored on technical and financial criteria, where the lowest compliant bidder usually wins. Tonbo submitted 275 bids between FY24 and Q1FY27 and won 97, a 35.3% win rate (p.270). Abroad, it mostly sells through resellers, distributors and system integrators. Two models stand out:

• White label: Tonbo supplies kits and thermal “cores” for local defence firms to assemble and brand, so the final product counts as domestically made in the buyer’s country. Margins per unit “may be slightly lower”, but volume is meant to compensate (p.232, 260).

• Co-development: where a subsystem is developed jointly with a DRDO lab or defence PSU, the aim is to become the default nominated supplier without an open tender (p.260).

The business is seasonal: 78.6% of FY26 revenue came in the second half (p.413), which is typical of Indian defence procurement and worth remembering when you read any single quarter.

Part 6: Who buys

The Indian government is now the core customer: 71% of FY26 revenue and 87% of Q1FY27 (p.250). Buyers include the Army, paramilitary forces, DRDO laboratories and defence PSUs.

The market share claim, stated carefully. Over April 2024 to March 2026, Tonbo supplied 51% of the thermal imaging units the Indian government bought through GeM, against 16% for Tata Advanced Systems (p.234). That is a units figure, on one procurement channel, in a report Tonbo paid for. It is still a striking number.

Programmes that show where the business is heading:

• DRDO’s Centre for Airborne Systems: Tonbo won a global tender to design a precision-stabilised multi-sensor gimbal for a surveillance aircraft, to be manufactured in India (p.268). Beating global optronics names on an airborne programme is a real credential.

• DRDO’s ITR test range: an electro-optical tracking system contract in FY26 (p.204).

• Anti-tank missile seekers: the 2024 integration into an Indian government missile (p.292) and an uncooled seeker offered to a state-run missile maker (p.268).

• Africa: a recent order for telescopic weapon sights for assault rifles and machine guns (p.260).

Abroad. The company says it has over 28,000 systems deployed across 24 countries (p.246), and names Armenia, Romania, Slovenia, the USA, Morocco, South Korea, the UAE, the Philippines, Singapore, North Macedonia, Australia and Taiwan among countries supplied over FY24 to Q1FY27 (p.249). The industry report lists US Navy SEALs, the Israel Defense Forces, NATO and the Indian Army among users (p.231). There is no revenue split by country, and US revenue was ₹0.3 Cr in FY26 (p.255), so I treat the user list as a credential rather than a revenue line.

What is in hand. The order book at 30 June 2026 was ₹314.2 Cr across 36 firm orders: 30 domestic orders worth ₹231.9 Cr and 6 overseas orders worth ₹82.4 Cr. New orders of ₹118.2 Cr followed in July to September 2026 (p.250). Combined, that is about ₹432 Cr of visible business against FY26 revenue of ₹362.6 Cr.

How concentrated it is. Defence sales are lumpy by nature, and Tonbo’s are lumpier than most:

Part 7: The people

The striking thing is continuity. The same three people have run this company for over 20 years, through three names, a buyout and a full change of business model. That is a long time to keep a deep-tech team together in Bengaluru.

Pre-IPO shareholding (p.126 to 127): CEAQ Singapore 25.47%, HBL Engineering 14.25%, CEAQ India 8.73%, Arvind 8.67%, Ankit 5.30%, Florintree 5.17%, Amansa 4.25%, Yali Deeptech Fund 4.19%, Cecilia 3.45%, ESOP trust 3.33%.

Part 8: The competition, and where Tonbo stands

The industry report places Tonbo as a “tiered supplier”: it makes the electro-optical eyes that bigger platform makers and integrators put on drones, tanks and surveillance systems (p.229). For tactical sights it sells directly to the end user.

Global benchmarks include Teledyne FLIR (US, USD 3.2 bn revenue), Hensoldt (Germany, USD 2.8 bn), Theon International (Greece), Controp (Israel) and Epirus (US), the best-known high-power microwave counter-drone company (p.236 to 237).

How Tonbo describes its edge (p.232 to 239):

1. It owns its electro-optical IP rather than building under technology transfer from a foreign partner.

2. It is ITAR-free: no US-controlled components that would let Washington veto a sale, which opens “non-aligned and export sensitive markets”.

3. Miniaturisation and low power, with Spartan-S and TRAP as the proof points.

4. It benchmarks its technology against the US and Israel and prices for cost-conscious buyers in India, South-East Asia and Africa.

5. It is asset-light, so it can scale output through contract manufacturers without building factories.

Fair counterweights from the same report: tactical EO/IR grows at 5.1% a year globally (p.206), the report rates Tonbo’s export presence “Low-Medium” (p.211), and the high-end detectors at the heart of every product are imported (p.209). The IP point also needs a footnote, which I come to in Part 9.

My read on the moat: the defensible assets are qualification and incumbency in Indian Army thermal sights, plus a design team that has shipped real products for a decade. Those are worth a lot in a procurement system where requalification takes years.

Part 9: Where it could go

These are the avenues the RHP itself points to, each with an honest note on how far along it is.

1. Re-equipping the Indian Army. The Army’s own technology roadmap (TPCR-2025) lists requirements where Tonbo already has products (p.234 to 235):

This is a deep pipeline in the categories Tonbo already leads. Indian procurement is still tender by tender, usually at the lowest compliant price.

2. Platform systems becoming the main business. Platform revenue went from ₹12.7 Cr to ₹144.8 Cr in two years and is now 40% of sales. A design win on a platform tends to repeat for years, which makes this the most valuable kind of growth Tonbo can have.

3. Missile seekers. If TRAP is adopted in Indian anti-tank missile programmes, volumes run into tens of thousands of missiles. The 2024 integration and the state-run missile maker engagement are early steps; no production order is disclosed.

4. Counter-drone and directed energy. The fastest-growing defence market in the report, at 45% a year. If WaveStrike moves from prototype to contract, Tonbo moves from making sensors to making weapons, a far bigger revenue unit per system.

5. Loitering munitions and unmanned platforms. Licensed and under development; early stage.

6. Exports, take two. The plan is ITAR-free products, the white-label model, new salespeople across South America, South-East Asia, Africa, the Middle East and Europe, and talks with European integrators (p.260). The overseas order book is ₹82.4 Cr. The challenge: FY25 exports came overwhelmingly from one customer, and the EU’s new defence funding rules require 65% European content in certain procurements (p.180).

7. Civil and dual use. Wildfire tracking, disaster response, driver assistance for heavy vehicles. No revenue yet.

If platforms keep compounding, the seeker gets a production order and WaveStrike moves from prototype to contract, Tonbo becomes a much bigger company than the one in the FY26 accounts. That is the bull story, and it is a coherent one.

Part 10: The numbers

The honest bull read first. Revenue from India more than doubled in FY26, from ₹157.0 Cr to ₹337.5 Cr (p.255). Platform systems grew about 11x in two years. Gross margins of around 50% and EBITDA margins in the high 20s in a good year are what a design-led defence electronics company should earn. The core Indian franchise is growing.

The headline read. Total revenue fell 22.7% in FY26 and Q1FY27 was loss-making, because the business is in transition from one large export customer to a broader domestic base. The rest of this piece is about how much that transition, and the way the IPO is structured, should matter to you.

Part 11: The other half of the prospectus

I said I would give the company its due before the fine print, and I have tried to. Here is the fine print, kept to what I think genuinely matters.

1. Who is selling. The company receives nothing from this IPO (p.84, 134). Of the 1.22 crore shares on offer, 75% belong to CEAQ Singapore and CEAQ India, formerly Tonbo Imaging Pte Ltd and Tonbo Imaging Pvt Ltd: the holding layer that has sat above the Indian company since the 2012 buyout (p.3, 117, 125).

The same Singapore entity sold the Indian company its core IP for ₹183.2 Cr in March 2024, paid largely out of fresh investor money (p.277 to 278, 437), and it remains a top-ten supplier at 26% of FY26 expenses (p.94). In late September the two CEAQ entities also sold 59,11,200 shares to institutions at ₹700, about ₹414 Cr (p.117 to 119). Add the pre-IPO sale to the CEAQ share of this offer and the two entities take out about ₹1,055 Cr. For what it is worth, the offer shrank from about 1.81 crore shares in the August draft to 1.22 crore in the RHP, a difference that matches the pre-IPO sale exactly (draft size per IPO portals, T3).

The three promoters formally exited their CEAQ Singapore stakes in November to December 2025, weeks before the first draft prospectus, and the RHP does not disclose who owns CEAQ Singapore today (p.320). After the offer, CEAQ Singapore will hold 18.2% with board rights, more than the three promoters’ combined 14.0%. Indian control matters for IDDM eligibility (p.319), which is part of why the structure deserves attention.

2. One order made two years. The Armenian order booked ₹188.9 Cr in FY24 and ₹267.7 Cr in FY25, 51% of the two years’ revenue, and there has been no repeat order (p.27, 434). The “global OEM” years were, in revenue terms, one foreign government.

3. Profit has not turned into cash. From FY24 to Q1FY27, cumulative PAT was ₹174 Cr and cumulative operating cash flow was minus ₹136 Cr (p.330 to 332). Inventory reached ₹320 Cr at June 2026, larger than the entire ₹314 Cr order book, and working capital lines were 97.8% drawn by 31 August 2026 (p.353, 410, 445). Management attributes the inventory build to a deferred export order and planned readiness for FY27 (p.413). Because the IPO is a pure offer for sale, none of the money raised reaches this balance sheet.

4. Two earnings-quality points. FY26 profit includes a net warranty provision write-back that added ₹18.8 Cr to EBITDA and ₹14.2 Cr to PBT, about 20% of FY26 PBT (p.31, 430); strip it out and the FY26 EBITDA margin is 23.7% rather than 28.8%. Over FY24 to FY26, no warranty claim was charged against the provision (p.429). Separately, liquidated damages for late delivery (₹21.82 Cr in FY24) match “debtors written off” in other expenses almost to the rupee (p.26, 373). Penalties of that kind usually reduce revenue; booked as bad debts, they leave revenue and gross margin reading higher. The RHP does not state the treatment; this is my reconciliation, and a question I would put to management.

5. Indigenous content and open disputes. The imported share of critical-component purchases rose from 51% in FY24 to 82% in FY26 (computed from p.26). IDDM status is measured on indigenous content by contract value, at least 50%, and draft DAP 2026 proposes 60% (p.72, 319), so this is a pressure point to watch rather than a breach. Anonymous complaints to SEBI, the Ministry of Defence and DPIIT allege Chinese components routed through CEAQ Singapore and a false IDDM claim; the company denies them (p.38 to 42). A suit has also been commenced in the Singapore High Court naming CEAQ Singapore, the company and its MD (p.448).

These points sit alongside a capable business rather than cancelling it out. Taken together, they describe an IPO that sells the 2024 to 2025 peak on behalf of a holding structure that has been paid at several points along the way, while the operating numbers are still in transition.

Part 12: What the price assumes

At the top of the band, Tonbo is valued at ₹4,010 Cr.

And the listed peers:

The issue is priced close to the peer median P/E of 83.9x. Every listed peer grew revenue at 13% to 37% a year over FY24 to FY26; Tonbo’s revenue shrank 8% a year over the same period. Indian defence is richly valued across the board, so the real question is how much growth the price already assumes.

Illustrative arithmetic, not a forecast or a target. Suppose the market values Tonbo at 45x earnings in FY29, which is the lowest P/E in the RHP’s own peer set (Bharat Electronics). For a buyer at ₹700 to simply get their money back by then, FY29 PAT would need to be about ₹89 Cr. For a 15% annual return it would need about ₹126 Cr, and for 20% a year about ₹141 Cr. For scale, Tonbo’s best year so far, FY25, produced ₹72.8 Cr of PAT, on the back of the Armenian order. (Computation: market cap of ₹4,010 Cr compounded over 2.5 years, divided by 45; ignores dividends.)

So the price is already asking for several of the Part 9 avenues to come good. That is possible. The platform growth, the Army roadmap and the counter-drone work are all real. It is a lot to ask of a company whose cash flow has been negative for three straight periods and which will not receive any of the IPO money to fund that growth.

What I’m watching

Signs the bull story is arriving:

• Platform systems revenue continuing to compound and overtaking tactical revenue.

• A named production order for the TRAP seeker or the DRDO airborne gimbal.

• A signed WaveStrike prototype contract, then a first subsystem sale.

• A second large export customer, so exports stop depending on one government.

• The order book growing faster than revenue (book-to-bill above 1).

Signs the concerns are live:

• Operating cash flow in H1FY27, the first listed print, and whether the ₹320 Cr inventory starts to unwind.

• Any further warranty write-backs, or a “warranty claims” line appearing for the first time.

• The DAP 2026 indigenous-content threshold and any Ministry of Defence response to the complaints.

• Progress in the Singapore High Court suit.

• Lock-in expiries around April 2027, and whether CEAQ Singapore sells further.

The bottom line

A dragonfly is mostly eyes, and the eyes are extraordinary. What decides how it fares is everything else: whether it can feed itself, and what it is attached to.

Tonbo’s eyes are genuinely good. A 280-gram sight that out-ranges heavier rivals, a seeker that does a cooled detector’s job at a fraction of the weight and power, a 360-degree vision ring that grew out of an innovation challenge, half of the government’s GeM thermal units, and a platform business up 11x in two years. A Bengaluru team built that over two decades, and India’s defence budget is pointed squarely at what it makes.

The rest of the animal is where my questions sit: cash that has not followed profit, two years of revenue that came from one customer, and an IPO in which the company receives nothing while the entities that sold it its own IP are the largest sellers. For me, the evidence that would change the picture is specific: two half-years of positive operating cash flow, inventory coming down, and full disclosure of who owns CEAQ Singapore. I will be reading the first listed results with that list in hand.

Disclosures

Holding disclosure: the author does not hold a position in Tonbo Imaging India Ltd, which is unlisted at the time of writing, and has not applied for shares in the issue. This piece analyses a publicly filed offer document (Red Herring Prospectus dated 10 Oct 2026) and is not a solicitation to subscribe to the issue.

DISCLAIMER: Karnik Shah is a SEBI-registered Research Analyst (Registration No. INH000023807). This content is for educational and informational purposes only and should not be construed as investment advice or a recommendation to buy, sell, or hold any security. Readers should conduct their own due diligence and consult a qualified financial advisor before making any investment decisions. The author may or may not hold positions in the securities discussed. Past performance does not guarantee future results. Investments in securities are subject to market risks. Read all related documents carefully.

GET TWO FREE IPR SAMPLE REPORTS

The kind of analysis you just read, giving the business its full due and then reading the filings line by line, is the foundation of IPR’s research method. Two free sample reports show the full framework applied to individual companies: Gravita India (Future Titans) and Kalyani Cast Tech (Microcap Mavericks).

Get Free Samples

No payment required. SEBI Reg. INH000023807.

Hello friends,

On the morning of 9 November 2016, the day after five hundred and one thousand rupee notes stopped being money, a company most Indians knew as a place to recharge a prepaid phone took out full-page newspaper advertisements congratulating the Prime Minister. Within weeks, vegetable vendors in Mumbai had laminated QR codes taped to their carts. A wallet app that had spent six years looking for a reason to exist had been handed one by the Reserve Bank’s printing presses.

Almost ten years later, on 15 September 2026, a far quieter document did something that may matter just as much. The National Payments Corporation of India announced that from 15 October, merchants receiving a UPI payment above ₹2,000 will pay a fee of 0.4 per cent on it. It is the first charge on bank-account UPI since the government made the system free in January 2020.

In between those two dates sits one of the strangest industrial stories anywhere in the world. India built a payments rail that now carries 24.51 billion transactions worth ₹29.82 lakh crore in a single month. It made that rail free to use. It then watched a generation of companies raise billions of dollars to compete on it, list on the stock exchange, and discover that it is extraordinarily hard to charge money for something the government has decided should cost nothing.

I think of UPI as a national expressway. The state built it, the state maintains it, and for six years the state waived the toll. The companies you can buy on the stock exchange do not own the road. They own the dhabas, the petrol pumps and the loading docks along it. The traffic is enormous, but a dhaba does not get rich because a million trucks pass it. It gets rich because some of those trucks stop, and buy something.

Once you see the industry that way, a lot of confusing things become simple. Why Paytm, the company that lost the consumer payments war with about 8 per cent of UPI, now carries a market value of ₹1.13 lakh crore, above the $9 to $10.5 billion range PhonePe, the company that won it, was reportedly discussing before it shelved its IPO in March. Why a 40 basis point fee can move two payment stocks in opposite directions on the same morning. And why the most important number in Indian fintech is not transactions, users, or GMV, but the thin, stubborn gap between reported profit and operating cash.

Pour something. This is a long one.

If you would rather watch the video I made for this piece, you can do so here-

Part 1: A Short History of Making Money Free

Every industry has a founding myth. Indian fintech has three, and they happened within eight months of each other. But to understand why they mattered, you have to start with what came before.

The age of the card, and the petrol pump

For most of the last three decades, accepting a digital payment in India meant a card, a bank, and a terminal. The merchant rented a machine, the card networks and banks split a fee that routinely ran between 1.5 and 2 per cent, and the whole arrangement worked well for the top of the pyramid: the airline, the mall, the department store. For the kirana store and the chaiwala, it was irrelevant. Cash was free, instant, and anonymous. The card was none of those things.

Here is a detail I love. Pine Labs, one of the two companies at the heart of this piece, was founded in 1998 not as a payments company at all, but to run smart-card loyalty programmes at Indian Oil petrol pumps. It only became a card terminal company later, when it realised that the box sitting at the counter was more valuable than the loyalty card in the customer’s pocket. Remember that instinct. It is the entire Pine Labs strategy in one sentence, and it has not changed in twenty-eight years.

The plumbing years, 2008 to 2014

What made the next decade possible was not a company. It was a sequence of public infrastructure projects that nobody outside a few ministries paid much attention to.

The National Payments Corporation of India was set up in 2008 as a not-for-profit owned by banks. Aadhaar began enrolment in 2010, giving hundreds of millions of people a verifiable identity for the first time. IMPS in 2010 made instant interbank transfers possible. RuPay arrived in 2012 as a domestic card network. And in August 2014, the Jan Dhan scheme opened bank accounts for people who had never had one.

None of this was glamorous. All of it was load-bearing. By 2015 India had, almost by accident, a population with identities, bank accounts and mobile numbers that could be tied together. What it did not have was a reason to use them.

The big bang, 2016

Then came the three founding myths, in quick succession.

UPI went live on 11 April 2016. It let anyone send money to anyone else instantly, using a phone number or a virtual address, directly from a bank account, at no cost.

In September 2016, Reliance Jio launched and collapsed the price of mobile data. A payment app is useless without a data connection, and suddenly the data connection was nearly free.

And on 8 November 2016, demonetisation removed roughly 86 per cent of the currency in circulation by value overnight.

Paytm was the first great beneficiary, because in 2016 the dominant form of digital money for small merchants was not UPI. It was the wallet, and Paytm owned it. But wallets had a structural flaw: money had to be loaded into them first. UPI did not. Over the next four years, UPI quietly ate the wallet, and in doing so it rearranged the entire industry.

UPI processed 1.2 billion transactions in FY18. It now processes roughly that many every day and a half. That scale is the reason the rest of this piece exists, and also the reason it has been so hard to make money from it.

The price goes to zero, 2020

In January 2020, the government amended the Payment and Settlement Systems Act and the Income-tax Act to set the merchant discount rate on UPI and RuPay debit cards to zero. Before that, person-to-merchant UPI transactions had attracted an MDR of up to 0.30 per cent.

It is hard to overstate what this did. It was the policy equivalent of abolishing the toll on the expressway. Traffic exploded. QR codes spread to every cart and kiosk in the country. And the companies building on the rail found themselves in a business with enormous volume and no price.

The state tried to fill the gap with subsidies. The incentive scheme for low-value UPI and RuPay transactions paid ₹1,389 crore in FY22, ₹2,210 crore in FY23 and ₹3,631 crore in FY24. Meanwhile NPCI’s own estimate of what it costs to run UPI, including servers, fraud prevention and bank technical support, is about ₹20,000 crore a year. The subsidy never came close to the bill. Somebody was always going to have to pay for the road.

The consumer side, meanwhile, consolidated into a duopoly. As of mid-2026, PhonePe is reported to handle about 46 per cent of UPI transactions and Google Pay about 33 per cent, with Paytm a distant third at around 8 per cent. NPCI has had a 30 per cent market share cap on the books since 2020, and has deferred enforcement twice. The current deadline is 31 December 2026.

The reckoning, 2021 to 2025

If 2016 was the big bang, 2021 was the gold rush. Paytm listed in November 2021 at ₹2,150 a share in what was then India’s largest IPO. It fell sharply on debut and kept falling.

What followed was a slow, grinding lesson in the difference between a fintech and a bank. The RBI barred Paytm Payments Bank from onboarding new customers in March 2022, citing material supervisory concerns. In January and February 2024, it went much further, stopping fresh deposits, credits and top-ups across accounts, wallets and FASTags. Separately, Paytm’s own payment aggregator application, first filed in 2020, was rejected in November 2022 because of foreign investment rules around its Chinese shareholder.

That period closed in stages. China’s Ant Financial sold its final 5.84 per cent stake in August 2025 for about ₹3,803 crore. Within days, Paytm’s payments subsidiary received in-principle approval as an online payment aggregator, and final authorisation followed on 26 November 2025. Two weeks before that, Pine Labs listed on the exchanges at ₹242, having priced its IPO at ₹221 and valued itself at about $2.9 billion, against the roughly $5 billion it had commanded privately in 2022.

And then, on 24 April 2026, the RBI cancelled Paytm Payments Bank’s licence outright, stating that the bank’s affairs had been conducted in a manner detrimental to depositors and that the general character of its management was prejudicial to depositors and the public interest. The bank is now being wound up.

The toll, 2026

Which brings us back to September. The new framework, finalised by NPCI’s UPI and Services Steering Committee, is more nuanced than the headlines, and the nuance is where the money is. Here is what it actually says.

Two numbers sit in tension inside that table, and they are the most useful thing I can give you.

The Finance Ministry says roughly 96 per cent of merchant transactions remain untouched. That is true. But Emkay Global estimates that about 67 per cent of person-to-merchant UPI value sits above ₹2,000. Both are true at once, because a small number of large payments carry most of the rupees. The fee was designed so that nobody buying chai notices it, and so that it still reaches most of the money.

Ten years after the toll was abolished, it is back. Only on the trucks, not the scooters.

Part 2: Where the Money Actually Goes

Now to the part that almost every piece of coverage I have read this week gets wrong. Forty basis points is not forty basis points to a listed company.

Half the fee goes to banks by construction. The acquirer leg is paid to the acquiring bank, and non-bank aggregators negotiate their share out of it. A consumer app earns its 8 bps only on transactions initiated from its own app, which is why market share on the consumer side still matters.

Follow the rupee. A customer pays ₹10,000 at a large-format store. The merchant bears ₹40. Of that, ₹16 goes to the bank that holds the customer’s account. ₹12 goes to the acquiring side, which is where Pine Labs and Paytm’s merchant business sit, alongside their partner banks. ₹8 goes to the app the customer tapped, and ₹4 to that app’s bank partner.

So the fee was not designed to make fintechs rich. It was designed to make the rail pay for itself. Jefferies sizes the industry pool at ₹15,000 to ₹18,000 crore a year. Set that against NPCI’s ₹20,000 crore estimate of what UPI costs to run and the logic of the policy becomes obvious. This is a maintenance budget with a commercial wrapper. The government, for its part, has publicly rejected claims that the move was a response to a 2026 US Trade Representative report on UPI and American payment providers, and says it will continue to support domestic players.

Which raises the more important question: if the road barely pays, what does?

The three ways to make money on a free road

After spending a lot of time with these companies’ filings, I have come to think there are only three durable ways to make money in Indian payments, and every listed company is some mixture of them.

Rent the counter. Charge merchants a monthly subscription for the device that sits at the till: the Soundbox, the card terminal, the smart POS. This is annuity revenue that does not care whether the payment itself is free. Paytm has 1.57 crore subscription merchants. Subscription fees from Pine Labs’ digital checkout points are now 29 per cent of its revenue.

Take a cut of the expensive rails. UPI was free, but cards, EMI, online payment gateways and prepaid instruments never were. Whoever controls the moment when a customer chooses to pay by credit card, or split a ₹90,000 television into nine instalments, earns a real fee. This is Pine Labs’ heartland.

Use the payment to sell something else. The payment is the reason the customer or merchant shows up. The money is in what you sell once they are there: a merchant loan, a personal loan, a mutual fund, an insurance policy, a gift card programme. This is where Paytm has rebuilt itself.

Put those together and you get the real shape of the industry, which is a very large revenue pool with a startlingly small profit pool.

One industry report cited by Business Standard in July puts Indian fintech revenue at ₹1.03 lakh crore in FY25 and industry net profit at just ₹2,300 crore. Even its FY30 projection has profit at under 2 per cent of revenue. The same report calls FY27 a likely year of earnings reset because of new rules on incentives and payment revenues. Fintech is a scale business that has not yet proven it is a margin business.

The tailwinds, ranked by how much I believe them

1. A price for the rail, at last. The MDR framework is the single most important structural change in Indian payments since 2020. The direct revenue is modest for most listed names. The signal is not. As Emkay put it, UPI acquiring now carries a recurring, transaction-linked commercial model rather than relying on an annual subsidy. A business with a price can be underwritten. A business waiting for a budget allocation cannot.

2. Credit riding the rail. The next wave is not payments, it is credit delivered through payments: credit lines on UPI, RuPay credit cards on UPI, merchant loans underwritten on transaction history. Paytm’s financial services distribution revenue grew 45 per cent year on year in Q1FY27. This is where the profit pool is migrating.

3. The mid-market goes digital. The top of the pyramid digitised a decade ago. The bottom digitised through free QR codes. The middle, the regional chain, the multi-outlet restaurant, the ₹50 crore distributor, is still being built out, and it is the segment that needs software, reconciliation, and credit rather than just a QR code. Pine Labs grew its mid-market merchant base about 40 per cent last year.

4. Licences as moats. For a decade, the problem with Indian payments was that anyone could enter. The RBI’s payment aggregator regime, and the slow, painful approvals it requires, has turned regulatory permission into a genuine barrier. Paytm waited five years for its licence. That is a cost to Paytm, and a moat against the next entrant.

5. Issuing and prepaid. Gift cards, corporate expense cards, employee benefits and incentive programmes are unglamorous, float-rich and sticky. Pine Labs’ issuing and acquiring platform grew 31 per cent in Q1FY27. Zaggle is built entirely on this layer.

6. UPI leaves India. Cross-border UPI acceptance in the Gulf, Southeast Asia and beyond is real but early. I would treat it as optionality, not thesis.

Part 3: Indian Payments in 2036

I want to be careful here, because a ten-year forecast in this industry has a humbling track record. In 2016, almost nobody predicted that UPI would kill the wallet, or that the government would make merchant payments free four years later. So treat what follows as a set of reasoned bets, each with my honest confidence attached, rather than a map.

Payments become a utility. I am highly confident of this. The act of moving money will look increasingly like electricity: universal, regulated, cheap, and priced to cover its cost rather than to generate a profit. The MDR does not reverse this. It formalises it. The listed winners of 2036 will be the ones who stopped thinking of themselves as payment companies somewhere around 2027.

The profit pool moves to credit and software. Highly confident. A payment is a data point. A merchant who has taken ten thousand payments through your device has, in effect, handed you a live cash-flow statement. Whoever turns that into underwriting, for working capital, for inventory, for consumer credit at the till, captures the margin. The payment is the customer acquisition cost. Credit is the product.

Credit on UPI eats into card EMI. Moderately confident, and this matters a great deal for one of our two companies. If a bank can extend a pre-approved credit line that a customer spends through UPI, the old logic of “swipe your card on a terminal and convert to EMI” gets a serious competitor. Pine Labs knows this, which is why it is building credit-on-UPI primitives itself. Whether it can be the one to own that transition, rather than the one disrupted by it, is probably the single most important question in its ten-year story.

Banks strike back. Moderately confident. The MDR framework hands issuing banks the biggest slice. Banks have the balance sheets, the deposit franchises, and now a reason to care about UPI economics. Expect them to push harder into merchant acquiring and consumer apps, which compresses what fintech intermediaries can charge.

Consolidation into a handful of scaled merchant platforms. Moderately confident. Merchant acquiring has network effects on the software side and brutal economics on the hardware side. I would expect three or four scaled platforms to dominate a decade from now, with the rest acquired, niche, or gone.

Agents start paying. Low confidence on timing, high confidence on direction. Both Pine Labs and Paytm are already talking about AI agents initiating and completing purchases. When software, not a human, clicks “pay”, the valuable position is whoever the agent trusts to authenticate, orchestrate and settle. It is too early to model. It is not too early to notice who is building for it.

The wildcard is the state itself. The e-rupee, a revived market share cap, a change of heart on the MDR, a new data localisation rule. The single most consistent lesson of the last decade is that in Indian payments, the regulator is not a referee. It is a player, and it moves the goalposts.

Hold the three money-making models from Part 2 in your mind. Now let us look at two companies that chose very different combinations of them.

Part 4: Paytm, the Company That Lost the Consumer and Found the Merchant

One 97 Communications Ltd (NSE: PAYTM, BSE: 543396)

A history in four acts

Act one, the phone. One 97 Communications was incorporated in 2000 as a mobile content and value-added services business. Paytm, the brand, arrived in 2010 as a way to recharge prepaid phones online.

Act two, the wallet. Through the mid-2010s, Paytm built India’s dominant digital wallet and attracted a cap table that reads like a history of global tech investing: Alibaba and Ant Financial, SoftBank, and, briefly and memorably, Berkshire Hathaway. Demonetisation made it a household name.

Act three, the fall. UPI took the consumer. The 2021 IPO priced the company as though it had not. The regulator took the payments bank, first its growth in 2022, then its deposits in 2024, then its licence in April 2026. Revenue fell from ₹9,978 crore in FY24 to ₹6,900 crore in FY25, a 31 per cent collapse, as the bank’s restrictions rippled through the group.

Act four, the merchant. What is left, and what has quietly rebuilt, is a merchant payments and financial services distribution business. Revenue recovered 22 per cent to ₹8,437 crore in FY26, and the company reported its first full year of profit, ₹552 crore, against a ₹663 crore loss the year before.

The business today

Paytm now has two engines and a smaller third.

The payments engine earns from device subscriptions (1.57 crore subscription merchants), from processing margins on cards, EMI and online payments, and increasingly from consumer UPI, where value grew 45 per cent. Merchant GMV was ₹7.1 lakh crore in the quarter, up 31 per cent.

The financial services engine distributes merchant loans, personal loans, equity broking and mutual funds. Crucially, Paytm does not lend from its own balance sheet. Loans are booked by partner lenders, and Paytm earns distribution and collection fees. That makes it capital-light, and it makes it dependent on partners’ appetite for risk.

Two details from the quarter deserve more attention than they received. First, the government’s Payments Infrastructure Development Fund incentive has ended: Paytm received ₹8 crore from it in Q1FY27 against ₹54 crore a year earlier. On a comparable basis, EBITDA rose from ₹18 crore to ₹195 crore. The improvement is not subsidy-driven. Second, the board chose not to proceed with a proposed bonus issue, which I read as a small but welcome sign of a management team less interested in share-price theatre than it once was.

In the September 2024 quarter, Paytm lost ₹404 crore at the operating line. It now earns ₹203 crore. Whatever else you think of this company, that is a real operating turnaround, and it is the reason the stock has rerated.

Growth triggers

• Management’s margin target. Paytm has said it is confident of reaching a 15 to 20 per cent EBITDA margin over the next two to three years. From 8 per cent today, that is the single biggest driver of the equity story.

• The MDR. More on this below, because Paytm is structurally the bigger beneficiary.

• Financial services scale. Growing 45 per cent, with more than half of merchant loan disbursements coming from repeat borrowers.

• A wallet licence. Paytm Payments Services has applied for one, which would restore a product the payments bank saga took away.

• AI-led operating leverage. In Q1FY27, indirect expenses grew only about 6 per cent while revenue grew 28 per cent.

How the MDR benefits Paytm

Look back at Exhibit 2. Paytm is the only listed company in this letter that can touch two legs of the waterfall: the acquirer leg through its merchant business and the consumer app leg through the Paytm app. That is the cleanest structural reason the same announcement is worth more to Paytm per rupee of volume, and it holds regardless of what you think of its management.

The brokers have tried to size it:

Source: brokerage views as reported in the press, 16 September 2026. These are the brokerages’ estimates, not mine.

Now the nuance, because the range above is enormous, from ₹440 crore to ₹1,400 crore of EBITDA. Three things will decide where Paytm lands.

Its merchant base skews small. Paytm’s 5 crore registered merchants are, overwhelmingly, the kirana and street-level businesses whose tickets sit below ₹2,000 and which may qualify for small-merchant exemptions. A large share of its GMV simply will not be eligible. The eligible volume is more likely to come through its online payment gateway and larger merchants.

Its consumer app share is 8 per cent, not 46. The 8 bps app leg is earned on transactions initiated from the app. Paytm gets its share of that leg, not all of it.

Pass-through is a negotiation, not a formula. Goldman itself cautioned that competitive pressure could limit how much of the MDR is actually passed on to merchants. A payments market with PhonePe, Google Pay, banks and a dozen aggregators competing for merchants does not easily hold a price.

My honest read is that the MDR is a genuine and material earnings tailwind for Paytm, most plausibly toward the lower-to-middle of that range in year one, and that the market has already paid for a good part of it.

The risks

The profit is mostly interest. On a trailing basis, Paytm’s profit before tax is ₹701 crore. Other income, largely treasury yield on its cash pile, is ₹625 crore of that. Take it out, and the core business earned about ₹76 crore before tax over twelve months. That is not hidden and it is not improper. It is on the face of the P&L. But it means the “Paytm is profitable now” narrative is doing work the operating business has not yet done.

Profit did not become cash. In FY26, Paytm reported ₹552 crore of profit and negative ₹743 crore of operating cash flow. Free cash flow was negative ₹1,211 crore. The most plausible explanation is working capital absorbed by the lending distribution business, collections and merchant settlement timing, rather than anything sinister, but I have not yet reconciled it line by line to the FY26 cash flow note, and I will not pretend otherwise. Until that reconciliation is done, it stays a flag.

The regulator has long memory. The RBI’s language when it cancelled the payments bank licence was unusually pointed about management. Paytm the listed company is a separate entity, and the bank is being wound up. But in a business that exists at the regulator’s pleasure, reputation with that regulator is an asset, and it has been spent.

Foreign institutions have been leaving. FII holding has fallen from 60.9 per cent in September 2023 to 48.1 per cent in June 2026, with domestic institutions absorbing it, rising from 4.1 to 24.9 per cent. The retail shareholder count has fallen from 12.3 lakh to 7.8 lakh.

Dependence on partners. Capital-light lending means Paytm’s fastest-growing engine runs on other institutions’ balance sheets and risk appetites. Those can change quickly.

Valuation, briefly

Source: Screener.in consolidated, 17-Sep-2026 close. Multiples computed by IPR on TTM to Q1FY27. Screener’s own displayed P/E of 139x uses a different EPS basis.

On trailing numbers, Paytm is not cheap by any measure. The only way to justify the price is forward. So here is some illustrative arithmetic, not a forecast. If revenue compounds at 25 per cent from FY26, it reaches roughly ₹16,500 crore in FY29. At 17.5 per cent, the middle of management’s margin target, that is about ₹2,900 crore of EBITDA. Add ₹440 to ₹1,400 crore of MDR-driven EBITDA and you land somewhere between ₹3,300 and ₹4,300 crore. Today’s market capitalisation is 26 to 34 times that FY29 figure.

That is the right way to read Paytm’s price: it assumes management hits its margin target, the MDR lands in the middle of the range, and nothing goes wrong with the regulator. None of those are unreasonable. All three together are a full price.

Part 5: Pine Labs, the Plumber of Indian Retail

Pine Labs Ltd (NSE: PINELABS, BSE: 544606)

A history in three acts

Act one, the petrol pump. Pine Labs began in 1998 running loyalty cards for Indian Oil’s fuel stations. The founders learned something that has defined the company since: whoever owns the device at the counter owns the relationship with the merchant, and the merchant will pay for software that makes the counter smarter.

Act two, the terminal and the instalment. Pine Labs became the dominant provider of card terminals to large-format retail in India, and then did something clever. It turned the terminal into a credit distribution point. When you walk into an electronics showroom and split a phone into six no-cost EMIs on your credit card, there is a very good chance a Pine Labs machine is orchestrating that transaction between you, your bank and the brand. Jefferies estimates Pine Labs’ share of store-based card EMI at 90 to 95 per cent.

It later bought Qwikcilver, which runs gift card and prepaid programmes for brands, and expanded into Malaysia, the Gulf, Singapore, Australia, the US and Africa. Its current CEO, Amrish Rau, joined in 2020.

Act three, the listing. The IPO in November 2025 was a lesson in humility. Pine Labs had been valued at about $5 billion privately in 2022. It went public at about $2.9 billion, in an issue cut from an originally reported $1 billion to about $440 million. The stock listed at ₹242, hit ₹284, and fell as low as ₹135. It trades around ₹185 today, down 22.5 per cent this calendar year.

The business today

Source: company Q1FY27 results, July 2026.

The digital infrastructure segment is the terminal business grown up: subscription fees from 2.17 million digital checkout points, transaction processing, online payments, and the affordability (EMI) franchise. The platform processed about ₹4.22 lakh crore of transaction value across 201 crore transactions in the quarter, and more than 70 per cent of those transactions already run on UPI.

The issuing and acquiring segment issues prepaid cards and gift programmes for brands and enterprises, processes issuing for banks, and earns interest on customer float and breakage on unused balances. It is growing faster, with international issuing up 47 per cent and new categories like gaming gift cards for Xbox, Roblox and Nintendo going live.

Growth triggers

• The MDR. Most of Pine Labs’ transactions already run on UPI, largely unmonetised. From 15 October, the acquirer leg begins applying to the eligible portion.

• New issuing categories that management expects to scale through the second half of FY27.

• International, now about 16 per cent of revenue.

• Operating discipline. Employee cost fell from 37 to 33 per cent of revenue year on year.

• Balance sheet repair. Borrowings fell from ₹1,046 crore in FY25 to ₹441 crore in FY26 after the IPO, which should keep bringing interest costs down.

• A lower tax line. The effective tax rate was 48 per cent in Q1FY27. Management expects it to taper toward 30 per cent for the full year.

How the MDR benefits Pine Labs

On direct revenue, less than Paytm. Pine Labs has no consumer app, so it touches one leg of the waterfall, and it shares that leg with its partner banks. Emkay assumes it retains about 6 bps of the blended pool, which yields roughly ₹155 crore of UPI MDR revenue in FY28 and about ₹43 per share of value.

But there are two second-order effects that I think the market underweighted on 16 September, and one that cuts the other way.

Its merchants have bigger tickets. Pine Labs serves large-format retail, brands and a growing mid-market. A far greater share of its UPI value is likely to sit above ₹2,000 than Paytm’s, which means a higher proportion of its volume is eligible. Per rupee of GMV, the acquirer economics may be better than the headline 6 bps suggests. I cannot size this without the company’s ticket-size distribution, and I will not invent one.

“Just take free UPI” is no longer free. For years, the easiest objection to paying for a smart checkout, for EMI orchestration, for a card acceptance setup, was that the merchant could simply put up a QR code and pay nothing. From 15 October, on large tickets, that merchant pays 0.4 per cent anyway. I want to be precise here: card acceptance still costs merchants several times more than 0.4 per cent, so this is not a price reversal. But the gap between “free” and “a fee” was always larger psychologically than arithmetically, and it just closed.

The cut that goes the other way. Pine Labs’ enterprise merchants are exactly the ones who now absorb the new fee. Large retailers negotiate. Some of that pressure will likely land on Pine Labs’ own subscription and processing pricing.

The risks

Operating leverage went backwards. This is the one I cannot argue around. Revenue grew 20 per cent in Q1FY27. Adjusted EBITDA grew 4 per cent. Margin fell 250 basis points. Contribution margin slipped a percentage point. Management attributes it to seasonal mix and deliberate investment, and that may be right. But the entire bull case for Pine Labs is margin expansion, and the most recent print did not show it. Exhibit 4 makes the point: on Screener’s reported basis, operating profit has declined for two consecutive quarters from its December 2025 peak.

Hardware is a structural tax. FY26 depreciation was ₹270 crore against ₹358 crore of operating profit. A business that deploys physical devices pays for them every year, whatever happens to UPI.

Debtor days of 140. Receivables have drifted up steadily, from 134 days in FY24 to 140 in FY26. Offset, to be fair, by large payables and customer float, which give the company negative working capital overall. But the trend deserves watching.

Credit on UPI. Everything I wrote in Part 3 about card EMI applies here with full force. The affordability franchise is Pine Labs’ crown jewel. A decade from now, it must either have evolved into credit-on-UPI orchestration, or it will have been disrupted by it.

No promoter, and a free float in transition. Pine Labs has no promoter group. Public shareholding fell from 83 to 66 per cent between March and June 2026 in Screener’s classification as institutions took larger stakes, and the stock has seen repeated large block trades. Which pre-IPO investors still hold material stakes, and when the remaining lock-ins expire, is worth knowing before anyone sizes a position in this one.

The profit is also mostly interest. Other income is 69 per cent of trailing pre-tax profit. Better than Paytm’s 89 per cent, but not yet a business funding its own earnings.

Valuation, briefly

On sales, book and EBITDA, Pine Labs trades at a steep discount to Paytm: roughly 60 per cent of its sales multiple, half its book multiple, and less than a third of its EBITDA multiple. On earnings the two are nearly identical, because both earnings lines are dominated by other income, which is precisely why P/E is the wrong lens for either.

Part 6: The Supporting Cast

Three more listed names complete the picture. Each tells you something about the industry the two main names do not.

MobiKwik: the consumer app without the scale

One MobiKwik Systems Ltd (NSE: MOBIKWIK)

MobiKwik began as a wallet and is now a UPI consumer app with a lending business, ZIP EMI, attached. It listed in December 2024. Its Q1FY27 was its third consecutive profitable quarter, with PAT of ₹7.6 crore against a ₹41.9 crore loss a year earlier, and platform GMV up 50 per cent to ₹58,700 crore. Its UPI transactions grew 130 per cent against about 24 per cent for the industry.

But revenue grew only 3.7 per cent, to ₹281 crore, and actually fell quarter on quarter. That is the consumer payments problem in miniature: volume is easy, monetisation is not. The profit turn came from cutting direct costs 21 per cent, not from revenue.

Prospects: The MDR helps at the margin through the consumer app leg, but with a small share of UPI, the dollars are modest. The real swing factor is lending, where the company is awaiting final RBI registration for its NBFC subsidiary to lend directly. At about ₹1,560 crore of market capitalisation, it is priced as an option on that. No brokerage target was readily available when the MDR news broke, which tells you how much institutional attention it currently commands.

Zaggle: the plumbing of corporate spend

Zaggle Prepaid Ocean Services Ltd (NSE: ZAGGLE)

Zaggle is a B2B spend management company: corporate expense cards, employee benefit programmes, channel incentive platforms, and the SaaS that ties them together. It issues prepaid instruments in partnership with banks, which puts it in the same issuing layer as Pine Labs’ Qwikcilver business, aimed at CFOs rather than consumers.

It grew revenue at a 51 per cent CAGR from FY23 to FY26, reaching ₹1,908 crore. Then Q1FY27 broke the spell. Revenue grew 28 per cent to ₹423 crore, but PAT fell 33 per cent to ₹17.5 crore and EBITDA margin compressed to 7.3 per cent, which the company attributed to acquisition integration costs and expenses previously capitalised now flowing through the P&L. The stock fell 20 per cent in a day to a 52-week low, and was down more than 55 per cent for the year at that point. Vijay Kedia’s firm then bought 1.48 per cent in a bulk deal at ₹164.72.

Prospects: Management guides to about 40 per cent consolidated revenue growth for FY27, with revenue heavily back-weighted into the second half. The MDR is largely irrelevant to this model. The questions are margin, working capital and cash conversion in a business whose revenue includes a lot of low-margin card and incentive flow. At roughly ₹2,500 crore of market capitalisation, it is a turnaround-of-trust story. The Q2 and Q3 prints will decide whether the market was right to punish it. I would personally exercise some caution here, as the management has had a consistent habit of ambitious guidances and regular tv appearances.

AvenuesAI, formerly Infibeam Avenues: the gateway that reinvented itself

AvenuesAI Ltd (formerly Infibeam Avenues Ltd; BSE: 539807)

A note on entity lock, because this one trips people up. The company you may know as Infibeam Avenues, owner of the CCAvenue payment gateway, now trades as AvenuesAI. It also owns Rediff, and an AI subsidiary. A share consolidation, from ₹1 to ₹10 face value, has been announced, which will break any per-share comparison across that date.

CCAvenue is one of India’s oldest online payment gateways, and the company has been pushing it into the UAE, Oman, Saudi Arabia and the US. Q1FY27 revenue from operations more than doubled to ₹2,680 crore, EBITDA excluding other income rose to ₹100 crore, and PAT rose 45 per cent to ₹85 crore. Transaction volume grew 74 per cent.

Read that revenue number carefully. Payment gateway gross revenue includes large pass-through flows, and the company itself flagged that net revenue dipped despite the gross figure. The ₹100 crore of EBITDA on ₹2,680 crore of revenue is the more honest measure of the economics: under 4 per cent.

Prospects: An online acquirer can benefit from the MDR’s acquirer leg on eligible e-commerce tickets, and the Gulf business is a genuine differentiator. At roughly ₹5,000 to ₹5,800 crore of market capitalisation through August, the question is whether the AI rebrand and the international push produce net revenue growth, not just gross volume.

Part 7: Head to Head, Paytm vs Pine Labs

Here is the comparison round by round. I have tried to make each round a genuine contest rather than a checklist, because these two companies are not really competitors. They serve different merchants, earn different fees, and carry different risks. Which is exactly what makes the comparison useful.

This is the chart almost nobody in this market publishes, because almost nobody wants to. The stock the market loves could not turn a reported ₹552 crore profit into operating cash last year. The stock the market is bored of generated ₹395 crore of it on ₹113 crore of profit, and posted its first positive free cash flow.

And the same test applied to earnings quality. Strip out treasury income and Paytm’s core business earned about ₹76 crore before tax over twelve months. Pine Labs is better on this measure, but neither company is yet funding its reported profit out of its own operations.

What the scoreboard says

Tally it and Pine Labs wins more rounds. But rounds are not equal, and I would be misleading you if I stopped there.

Paytm wins the rounds that matter most for the next two years: growth, operating leverage, and direct MDR exposure. It is a business in visible, accelerating repair, and the MDR lands on it at exactly the moment its cost base is flattening. That is why the market bought it by up to 7 per cent intraday on 16 September, and why the stock is up 37.5 per cent in a year in which the Nifty 50 is down 11.2 per cent.

Pine Labs wins the rounds that matter most for what you are paying and what you are really getting: cash conversion, earnings quality, regulatory cleanliness and price. It fell as much as 7.7 per cent on the same MDR news, and closed down about 4 per cent, on the same day Emkay raised its target to ₹230 and Jefferies to ₹235.

So the market’s direction on the day was defensible. Paytm does gain more directly. But I think the magnitude was wrong in both directions. Paytm was already priced for a large part of this, and Pine Labs was priced for none of it.

Part 8: The Verdict

Here is where I have landed, and I would rather be roughly right than precisely wrong.

Paytm is the better story, at a full price. The operating turnaround is real and the MDR is a genuine, two-legged tailwind. But on trailing numbers almost the entire profit is treasury income, FY26 operating cash flow was negative, and the valuation already assumes management hits its margin target and the MDR lands comfortably. What would change my mind: a clean reconciliation of the FY26 cash flow, and two consecutive quarters in which core pre-tax profit exceeds other income.

Pine Labs is the better-priced business with an unproven margin. It converts profit to cash, trades at a steep discount to Paytm on every measure that is not distorted by other income, and has second-order MDR benefits the market ignored. But its operating leverage stalled in the most recent quarter, and its crown jewel faces a credible ten-year disruption threat. What would change my mind: adjusted EBITDA margin recovering above the 19.6 per cent it printed a year ago, with MDR revenue visible in the Q3 or Q4FY27 numbers.

Neither passes IPR’s forensic gate for a full-sized position today. Both are on my watchlist with specific triggers attached, which, if you have read this publication for a while, you will know is not a polite way of saying no. It is a precise way of saying not yet.

Part 9: What I’m Watching

The retained basis points. Not the headline 40. The number each company actually reports keeping, starting with the Q3FY27 results. That single disclosure will collapse the ₹440 crore to ₹1,400 crore range for Paytm into something real.

Paytm’s FY26 cash flow note. Where the ₹743 crore went. I am going to do this reconciliation properly and publish it.

Pine Labs’ margin in Q2FY27. If adjusted EBITDA margin does not recover, the cheapness is not cheapness.

31 December 2026. Whether NPCI finally enforces the 30 per cent market share cap, defers it again, or changes it. Enforcement would reshape the consumer app leg overnight, and would be the first real opening Paytm has had in consumer UPI in years.

PhonePe’s IPO. Paused in March 2026. If it revives, Indian investors will finally have a listed benchmark for the consumer side of UPI, and both Paytm and MobiKwik will be repriced against it.

Credit on UPI adoption. The quiet variable behind Pine Labs’ long-term story.

Part 10: A Closing Thought

There is a lovely irony at the centre of this industry. India did something no other large economy managed: it built a public payments rail so good, and so cheap, that it made the private payments business almost impossible. The state gave away the road. And for a decade, the most sophisticated investors in the world kept trying to buy the road anyway.

The toll that arrives on 15 October does not change that. Most of it goes to the banks who maintain the lanes. What it does change is the conversation. It tells every company on the rail that the era of pure volume is over, and that the question from here is not how many trucks pass your dhaba, but what you sell to the ones that stop.

Paytm has learned to sell credit to the people who stop. Pine Labs has spent twenty-eight years owning the counter where they pay. Both have finally built something that can make money. Neither has yet proven it can make cash.

The road will never be the business. What you sell beside it always was.

See you at the next inflection point.

Karnik

Inflection Point Research

Portfolio Disclosure

I do not hold any positions in the companies discussed in this piece, directly or through connected persons: PAYTM, PINELABS, MOBIKWIK, ZAGGLE or AvenuesAI.

This disclosure is as of the date of publication.

SEBI Disclosure

This report is published by Inflection Point Research, a SEBI-registered Research Analyst (Registration No. INH000023807). The analyst may have personal investments in the securities discussed, as disclosed above. This content is for informational and educational purposes only and should not be construed as investment advice, an offer to buy or sell securities, or a recommendation of any specific investment action. Readers should conduct their own due diligence and consult a SEBI-registered financial adviser before making investment decisions. Past performance is not indicative of future results. Equity investments are subject to market risks. Brokerage estimates and target prices cited are those of the named brokerages as reported in the press, and are their views rather than mine.

GET TWO FREE IPR SAMPLE REPORTS

The toll on India’s free payments road is exactly the kind of structural shift Inflection Point Research narrows into specific, actionable company names for subscribers. Two free sample reports show that full framework applied to individual stocks: Gravita India (Future Titans tier, lead recycling with a regulatory tailwind, BUY framework) and Kalyani Cast Tech (Microcap Mavericks tier, container manufacturer with a founder moat, watchful-hold framework).

Get Free Samples

No payment required. SEBI Reg. INH000023807.

If this piece was useful, forward it to one person who still thinks UPI is a business. It is a road. The business is beside it.

Until the next one,

Karnik

In late July, with far less noise than it deserved, the Union Cabinet approved a programme called Samudra Manthan. Rs 84,084 crore, committed through 2031, aimed at one uncomfortable fact: India imports roughly 90% of the crude it burns, its producing fields decline 6 to 7% a year on their own, and its offshore basins, estimated to hold over 5,600 million tonnes of oil equivalent, have never been systematically explored. Earlier this year, when the Strait of Hormuz closed, the country found out in a fortnight what that dependence actually means.

The name is borrowed from the churning of the ocean, the myth in which the gods and the demons turn a mountain on the seabed until the treasures rise. It is a good name, and not only for the poetry. In the story, somebody has to hold the rope. In the programme, somebody has to shoot the seismic, operate the platforms, staff the rigs and build the plants. The money is spent in a specific order, and seismic comes first. Every rupee of it passes through a services contractor before it produces a single barrel.

This week’s video is about one of those contractors. Asian Energy Services was incorporated in 1992 to shoot seismic for ONGC and Oil India. Over thirty years it built genuine technical pedigree, and since 2016, under a new promoter, it has been rebuilt into something with four distinct engines: integrated oilfield services, where an 865 crore Vedanta mandate is the template; mechanised coal handling, its highest-margin business, underwritten by a regulatory requirement rather than a capex decision; a global offshore manpower arm bought last August that earns in dollars; and, arriving now through a reverse merger, the promoter’s own producing oil fields.

I have gone through each engine with its own revenue, its own margin and its own order book, rather than treating the company as a single line. I have spent time on management, because their record against specific guidance is the thing I weigh most heavily in a company this size. I have laid out the merger mechanics plainly, including the roughly 45% dilution that public shareholders take in exchange for the fields. And I have put all six of my open concerns into one honest minute near the end, rather than scattering them through the piece, because a list of concerns is the cost of holding something, not an argument against it.

Watch the full deep dive here:

Until next week,

Karnik

Inflection Point Research

Disclosures

Inflection Point Research is a SEBI-registered Research Analyst (Reg. No. INH000023807). I hold shares in Asian Energy Services Limited. Neither I nor the firm has any financial relationship with, or has received any compensation from, Asian Energy Services Limited or Oilmax Energy Private Limited.

Figures are drawn from BSE and NSE filings, the FY26 and Q1FY27 results and presentations, the merger scheme documents and the FY26 annual report. Oilmax Energy’s figures are management disclosure and are unaudited in public filings. This is written for information and education only. It is not investment advice, not a recommendation, and not an offer to buy or sell. Please do your own work or speak to a SEBI-registered adviser before acting.

Hello friends,

Since roughly May, a group of Indian metal companies has been bought as if it were one stock. Forgers, casters, bearing makers. The one-year moves run from +26% at Rolex Rings to +154% at Happy Forgings, and the sentence attached to all of them is the same: the cycle has turned; buy Indian metal conversion.

Before I get to whether that sentence is true, it is worth spending a few minutes on what these companies actually do, because “metal conversion” flattens three quite different trades into one word. Once you can see what each company makes and who buys it, the rally starts to look less like one wave and more like three tides in three harbours.

Part 1: Forging, casting, bearings. Three ways to make a metal part

Almost every load-bearing metal part in a truck, a tractor or an excavator starts life in one of two ways.

Forging is the older, harder way. You heat a steel billet until it glows, then shape it under a hammer or a press, sometimes with tens of thousands of tonnes of force. The metal is never melted. Because it is squeezed rather than poured, the grain of the steel flows along the shape of the part, and that is what makes a forging strong. Anything that has to take a repeated, punishing load is forged: crankshafts, axle beams, connecting rods, steering knuckles, the rings inside a bearing, railway wheels. Forging is expensive per part, which is why it is reserved for the parts that must not fail.

Casting is the pouring way. You melt the metal, pour it into a mould, let it cool, and break the mould away. Casting can make shapes forging cannot: hollow housings, intricate geometries, very large one-piece components, and it does so cheaply per kilogram once the pattern exists. The trade-off is that a cast part has no directional grain, so it is generally not as tough as a forging of the same size. Castings go where shape and wear resistance matter more than tensile strength: gearbox and axle housings, excavator undercarriage parts, bucket teeth, valve bodies, pump components.

The value in both businesses sits less in the raw shaping and more in what happens afterwards. A rough forging is a commodity. A forging that has been heat-treated, machined to tolerance and delivered ready to bolt onto an engine line is a component, and it earns a very different margin. When Happy Forgings says 90% of its output is machined, or MM Forgings says 67%, that is the number to read, because it tells you how far up the value ladder the company sits.

Bearings are the third trade, and a slightly different one. A bearing is an assembly: two rings, a set of rolling elements (balls, rollers, or needles), and a cage that keeps them apart. The rings are usually forged, then ground to micron tolerance. The finished bearing is sold both to vehicle makers and, crucially, to the replacement market, since bearings wear out and get changed. That replacement stream is why bearing makers tend to show a flatter margin cycle than pure forgers: some of their demand does not care whether anyone is building new trucks this year.

One name on the list belongs to none of the three. AIA Engineering makes high-chrome grinding media, the cast steel balls that mining companies feed into their grinding mills and wear away. It is a consumable, replaced continuously, and I keep it in the set only as a benchmark for what a genuinely non-cyclical metal business looks like on the same charts.

Part 2: Who makes what, and who buys it

Here are the nine, grouped by what they make. Mix figures are from Q1FY27 concall summaries where I have them, and from company disclosures otherwise.

Now put those customers into their cycles.

Read the two tables together, and the single-story version of this rally falls apart. Two forgers live almost entirely in the truck room, which is hot. One caster and half of one linkage maker live in the construction room, also hot. The other half of the linkage maker, and the farm third of the biggest forger, live in the large-ag room, where the lights are still off. The bearing makers sit in a domestic strip whose demand is only partly about new vehicles at all.

The market is reading nine names off one thermometer. They sit in three different rooms.

Part 3: The evidence in the margins

You can see the three rooms in the margin history without knowing a single product name.

Grey is each company’s worst operating margin in the last downcycle, FY19 to FY21. Teal is its worst in this one, FY24 to FY26. Where teal is taller than grey, the business came out of the bad years with a higher floor. Seven of nine did. Steelcast raised its trough from 19% to 27%, the largest gain, which is what an approved casting supplier into mining and earthmoving looks like once the approvals are in. NRB went from 11% to 16%, the replacement stream doing its work. Ramkrishna went from 17% to 14%, the only trough that fell, consistent with a forger carrying new plant and acquired capacity through a truck downturn. Menon was flat, 21% to 20%, on a standalone basis.

Two caveats. Happy’s first reported year is FY21, so its comparison is the weakest of the nine. And this downcycle was shallower for domestic names than the BS-VI and COVID trough, so part of the improvement is a milder exam rather than a stronger student. Tonnage and realisation per tonne would settle that. I have that series only for Happy, so margin percentage stands in as a proxy here, and I am naming the gap rather than hiding it.

The same history as lines makes the three trades visible. The step-change names (Steelcast, AIA, NRB, Uniparts) sit on a higher plateau after FY22. The flat names (MM Forgings, Menon, Happy) barely show a margin cycle at all, so their earnings swing with volume and interest rather than price. Ramkrishna is the one mean-reverting line, a round trip from 22% to 14% and a partial climb. Casters into mining, bearings into replacement, and forgers into trucks each leave a different fingerprint.

Part 4: Where the debt sits

Cyclicals die of debt, not of demand. So the one balance-sheet test I insist on divides net debt by the worst operating profit of the last five years, never the current one.

Seven of nine sit under a 3x ceiling, and four are net cash: Uniparts, Steelcast, Rolex and AIA. The two above it are the two purest truck-cycle forgers. MM Forgings carries 5.9x net debt to trough operating profit with interest cover of 2.3x. Ramkrishna carries 4.3x with cover of 1.4x. Both are mid-capex, MM Forgings with Rs 230 cr committed to its new press and Ramkrishna with Rs 337 cr of capital work in progress, so both rest on plant that is not yet earning.

That combination, the hottest end market and the heaviest balance sheets, is worth holding in your head. A forger’s revenue swings with truck builds. If the builds pause, the debt does not.

One footnote for anyone reading MM Forgings’ trailing P/E: Q1FY27 PAT of Rs 90 cr includes a Rs 56 cr land-sale gain.

Cash conversion tells a kinder story: seven of nine convert more than 70% of cumulative FY24 to FY26 operating profit to cash. The one that made me look twice is Uniparts at 94% while carrying 403 days of inventory. A system supplier to Deere-class OEMs holds a year of stock as the price of admission, and Uniparts converts anyway, which says the stock is contractual.

Part 5: What you are paying for

Because most of these names are still recovering, the multiple on five-year average EBITDA is higher than the multiple on trailing EBITDA, and the mid-cycle number is the one that tells you what you actually own. Three names trade under 20x mid-cycle: MM Forgings, Rolex Rings and Uniparts. Four trade above 30x. Happy Forgings at 59x has a compounder’s numbers, 21% revenue CAGR and a 27% to 31% margin band with no cycle in it, so a cyclical lens is the wrong instrument for it. Steelcast at 31x is the price of the best trough shift in the set, already paid.

Market cap from Screener at 03/04-Sep-2026. EV adds FY26 net debt as defined above. T3.

Part 6: Two headwinds already in the September quarter

Ocean freight is up 2.5 to 3x since mid-June. Rolex Rings has already guided FY27 operating margin down to 21% to 22% for it, against 23% in Q1FY27. Natural gas is up 50% to 55% since February; Steelcast runs its foundry on it and had not decided between gas and electric at the last call. Whoever cannot pass these through prints the first margin miss of the rally in Q2FY27. Watch gross margin, not revenue.

And one shape from history for 2028. In 2006, ahead of the 2007 EPA change, North American truck buyers pulled orders forward, and production then fell from 324,000 to 256,000 units over 2007 to 2008. EPA 2027 has the same shape. The two most levered forgers are the two most exposed to it.

What I am watching, in order. ACT Research’s September Class 8 order number. AEM tractor units for a real large-ag bottom. Q2FY27 gross margins across the set. Promoter disclosures at NRB (holding fell from 51.2% to 44.7% in the June quarter; 53 lakh shares still encumbered after an August release of 12 lakh) and Menon (promoter group sold 9.97 lakh shares between Sep-2024 and Aug-2026, holding now 66.7%).

See you at the next Inflection Point.

Karnik

Inflection Point Research

You can also watch a video we made on this topic here:

Disclosures and Disclaimer

Inflection Point Research is a Research Analyst registered with the Securities and Exchange Board of India (Registration No. INH000023807). The analyst and members of the analyst’s family hold positions in Ramkrishna Forgings (NSE: RKFORGE), Happy Forgings (NSE: HAPPYFORGE), NRB Bearings (NSE: NRBBEARING) and Menon Bearings (NSE: MENONBE). No view, verdict, target or entry level is expressed on any of these four companies in this piece; where they appear, the data is presented for completeness of the peer set only. The analyst and family may buy or sell any security discussed at any time, subject to the trading restrictions under the SEBI Research Analyst Regulations, and have received no compensation from any company discussed. Financial data in this piece is drawn from Screener.in, a third-party aggregator, and from press summaries of company earnings calls as of September 2026, and has not been reconciled to exchange filings; tonnage and realisation per tonne were not available for any company except Happy Forgings, so margin percentages are used as a proxy for unit economics throughout. This piece is for information and education only. It is not investment advice, an offer or solicitation, or a recommendation to buy or sell any security, and has been prepared without regard to the financial circumstances or objectives of any reader. Investments in securities are subject to market risk; small and micro cap securities carry additional risks including low liquidity and high volatility, and past performance is not indicative of future results. Registration granted by SEBI and certification from NISM in no way guarantee performance of the intermediary or provide any assurance of returns to investors. Please do your own research or consult a SEBI-registered adviser before making any investment decision.

I want to be clear at the top. This is a good business. I would like to own it. I am not applying, and this post is about why the price, not the company, fails my test.

The business in one paragraph

Rentomojo rents furniture and appliances on monthly subscription. It has 2.54 lakh live subscribers across 29 cities and 8.5 lakh items in circulation. It buys an asset once and rents it out across several households, refurbishing in between. Revenue grew 46% in FY26 to ₹387 Cr. It holds 42 to 47% of organised subscription revenue in its category, per the Redseer report commissioned for the issue.

The engine works. Average gross rental block in FY26 was ₹411 Cr; it produced ₹163 Cr of EBITDA, with refurbishment cost already expensed. An asset earns back its purchase price in cash in about two and a half years. FY17 and FY18 asset cohorts still have 56 to 61% of items generating rent after nine years. Adjusted ROCE is 25%. Unrecognised contracted revenue at March 2026 was ₹293 Cr, which is 76% of FY26 revenue already signed.

You can view a video we made reviewing this IPO here:

Three-year snapshot (₹ Cr, RHP restated)

Two things stand out. CFO covers reported PAT 1.7 times, which is a clean forensic result on my usual gate. And free cash flow after rental capex is roughly zero in every year, by design: this is a leasing business that reinvests everything. That is not a flaw. It does mean depreciation is a real cost, not an accounting abstraction, and that matters for what follows.

From 40x to 62x

FY26 PAT of ₹104 Cr includes a ₹36.6 Cr deferred tax credit. That is the company recognising a tax asset on past losses now that it expects to stay profitable. It is legitimate. It is also not operating profit. Strip it out and PAT is ₹68 Cr. At ₹4,206 Cr that is 62x.

There is a second layer. The company pays no cash tax today because of those same past losses. The RHP shows about ₹36.5 Cr of deferred tax asset on the balance sheet. At current profit levels that shield is consumed by FY28, after which the effective tax rate goes from zero to about 25%. The P/E on today’s earnings therefore understates the P/E on the earnings you will actually own in two years.

The depreciation line

This is the finding I would ask you to sit with.

In FY24, management extended the estimated useful life of rented appliances from 3 to 5 years to 3 to 10 years. In FY25, it extended rented furniture from 8 years to 10. Both changes fall inside the three-year window the prospectus covers. The RHP discloses the effect: depreciation was ₹9.1 Cr lower in FY24 and ₹6.2 Cr lower in FY25 than it would otherwise have been. FY24 PBT was ₹22.4 Cr, so the change was 41% of that year’s pre-tax profit.

The FY26 effect is not disclosed. Rental assets are now depreciated at about 10.5% of gross block a year. On the earlier lives, my estimate is FY26 depreciation would have been roughly ₹22 Cr higher, taking PBT from ₹70 Cr to about ₹48 Cr.

I am not saying ten years is wrong. The cohort data supports long lives, and Schedule II of the Companies Act allows ten years for both asset classes. I am saying three things. First, it is the largest judgement in the accounts, at 18% of revenue. Second, the company moved it, in the direction that helps, right before coming to market. Third, the business does not need the ten-year assumption to be a good business. The IPO price does.

Here is what the multiple looks like on each set of assumptions:

The other things in the file

Eighty-eight percent is an offer for sale. Of the ₹1,256 Cr issue, ₹1,106 Cr goes to Accel, IDG, Chiratae, Edelweiss, ValueQuest, Madison, GMO and the promoter (₹34 Cr). The company receives ₹150 Cr, of which ₹70 Cr repays debt and ₹42.5 Cr pays warehouse rent. That is less than one year of rental-asset capex. This is a liquidity event for early backers first and a growth raise a distant second.

A co-founder dispute is pending. A former director sold his 9.41% stake to the company’s employee benefit trust in August 2023. In March 2026 he filed a petition at the NCLT seeking, among other things, to reverse that sale, remove the promoter from the board, and restrain the IPO. In July 2026 an FIR was registered against the promoter and CFO alleging a loss of ₹37.25 Cr; the Karnataka High Court stayed the investigation on an interim basis on 17 July. The company’s position, from the RHP, is that this is a civil dispute converted into a criminal one, that the petitioner sold voluntarily, and it has filed a perjury application against him. I take no view on the merits. I do note that the stake at issue is worth about ₹395 Cr at the cap price, and that the RHP discloses a ₹4.76 Cr loan from the company to the same employee trust in FY24.

What is clean. Related party transactions are about 1% of revenue and are all management remuneration. Auditor CARO observations are housekeeping: incomplete asset tags on ₹13 Cr of own-use assets, small statutory-dues delays, a one-day loan repayment slip. Supplier concentration is low. The promoter’s only group entity is a family trust. The Noida warehouse fire in June 2026 and an MCD stoppage notice on a Delhi store in August are small individually.

Return math from ₹404

Three-year hold, exit on FY29 earnings. Assumptions: rental capex at 45% of revenue, depreciation at 10.5% of gross block under current policy, cash tax from FY28, debt paid down.

The base case is the story. Revenue more than doubles by FY29 and PAT lands at ₹114 Cr, roughly where FY26 reported PAT is today, because the tax shield runs out and depreciation grows with the block. Three years of growth are already in the issue price. Even granting a 45x exit with no de-rating at all, base IRR is 7%.

For a 15% base-case IRR I need to pay about ₹290. For 12%, about ₹310. At the grey-market implied listing near ₹520, the bull case is +51% and the bear case is minus 71%. That is not a trade I want.

Verdict: WAIT FOR LISTING DIP

Rentomojo is a good business with a price problem and two open forensic items that a prospectus cannot resolve: whether ten-year asset lives survive public-market scrutiny, and how the co-founder litigation ends. I am not applying.

What would bring me in:

• Price in the ₹290 to ₹320 zone, or

• Two listed quarters with rental depreciation holding near 10.5% of gross block and provisions for asset loss not rising as a share of the block, and

• The NCLT waiver application decided, or the High Court quash petition disposed, and

• Revenue growth still above 35% with occupancy above 82%.

If the stock lists at a premium and holds it, I will have been wrong on price and I will say so. I would rather be wrong on price than wrong on the depreciation line.

Disclosures (SEBI Reg. INH000023807). Inflection Point Research is a SEBI-registered Research Analyst. This post analyses a publicly filed Red Herring Prospectus dated 3 September 2026 and is for information and education only. It is not investment advice, not a recommendation to apply for, subscribe to, or refrain from the issue, and not a solicitation. The analyst and connected persons hold no position in the issuer, which is unlisted at the time of writing, and do not intend to apply in the Offer. Litigation is summarised from the issuer’s own RHP disclosures; no view is expressed on the merits of any pending matter. Reg 16: issuer unlisted, 30-day pre-publication trading window not applicable. Financial figures are from the RHP restated financial information; scenario figures are IPR estimates; price band and dates are from public exchange notices as of 9 September 2026. Securities markets are subject to market risks. Verify every figure against the final Prospectus before acting. Investor grievances: research@inflectionpointresearch.in. SEBI SCORES: scores.sebi.gov.in.

GET TWO FREE IPR SAMPLE REPORTS

The method in this post, the depreciation-line check, the litigation read, the multiple ladder, the bull-base-bear IRR, is the foundation of every IPR report. Two free samples show the full framework applied to listed companies end-to-end: Gravita India (Future Titans tier, lead recycling with a regulatory tailwind, BUY framework) and Kalyani Cast Tech (Microcap Mavericks tier, container manufacturer with a founder moat, watchful hold framework).

[Get Free Samples] → https://www.inflectionpointresearch.in/free-sample-research-reports

No payment required. SEBI Reg. INH000023807.

Hello friends,

Short one this week, because the work is in the video.

Rishabh Instruments (NSE: RISHABH) reported FY26 profit after tax of ₹82.26 crore against ₹20.97 crore the year before, a 292% increase. Revenue over the same period went from ₹720.34 crore to ₹775.15 crore, up 7.6%. In the June 2026 quarter the pattern held and then some: revenue up 4.2%, EBITDA up 17.3%, and profit after tax actually down 1.4% as depreciation from the new Nashik capacity started landing. When profit quadruples and revenue barely moves, something structural has changed in the mix, and it is worth knowing exactly what.

The short answer is that Rishabh stopped doing something expensive rather than started doing something new. Its Polish die-casting subsidiary was selling into European automotive supply chains on contracts written before European energy and labour costs repriced. Management walked away from roughly ₹100 crore of that work. Casting revenue fell 41.2% year on year last quarter, and the segment went from an adjusted EBITDA loss of ₹15.0 crore in FY25 to a positive ₹3.3 crore in FY26. Underneath it the electrical instruments business is genuinely good, growing 34% last quarter at adjusted EBITDA margins of 24.8%. The two halves are almost exactly cancelling each other out at the consolidated line, which is why the group looks asleep while both are moving fast in opposite directions.

Two things in the video that I have not seen written up anywhere. First, this is an Indian-listed company that earned ₹526.84 crore of its ₹775.15 crore FY26 revenue in Europe. Sixty eight per cent. India is a quarter of the group. Second, and more important: across FY22 to FY26 Rishabh reported ₹243 crore of cumulative profit after tax and generated roughly minus ₹9 crore of cumulative free cash flow. Operating conversion is fine, around 105% of operating profit in FY26. The cash is simply going straight back out into capacity and working capital, with inventory days rising to 213 last year rather than falling. Whether that capex earns a return is the whole question, and it is a different question from the one the multiple appears to be answering.

One note on timing. I recorded this against the 31 August close of ₹705, a market capitalisation of about ₹2,724 crore and roughly 33 times FY26 earnings. The stock has run about 11% since, to around ₹784 and a market capitalisation near ₹3,030 crore, and it made a fresh all-time high this morning. That matters for one specific reason. On 20 August, CNBC-TV18 reported that the promoters were in talks to sell a controlling stake at a valuation above ₹3,000 crore, with a German entity among the parties, and the company filed a clarification under Regulation 30(11) saying no disclosable material event exists at this stage. In the video I point out that the reported control valuation sat above the market capitalisation. As of this morning it does not. I have nothing beyond the exchange record and I am not going to speculate past it, but if you are buying here, be clear about how much of the price is a view on a corporate event rather than on current transformers and panel meters.

The full picture, with the filings on screen, is here:

See you at the next inflection point.

Karnik

Inflection Point Research

Disclosure

I hold a position in Rishabh Instruments Limited (NSE: RISHABH). I may initiate, add to, reduce or exit this position at any time based on changes in fundamentals or portfolio rules.

SEBI Disclosure

This note is published by Inflection Point Research, a SEBI-registered Research Analyst (Registration No. INH000023807). It is for informational and educational purposes only and should not be construed as investment advice, an offer to buy or sell securities, or a recommendation of any specific investment action. Readers should conduct their own due diligence and consult a registered financial adviser before making investment decisions. Past performance is not indicative of future results. Equity investments are subject to market risks. Financial data is as of the FY26 audited results and the Q1FY27 disclosures. Price data is as of 7 September 2026 where stated and as of 31 August 2026 in the video.

On the morning of 11 June 2026, a Hyderabad precision engineering company lost about a tenth of its market value because a construction crew stopped work in Wyoming.

MTAR Technologies opened at ₹7,100 on the NSE and touched ₹6,330 intraday, down 11%. Nothing had happened in Hyderabad. Nothing had happened at any of MTAR’s three plants. No order had been cancelled, no customer had written in, no filing had been made. What had happened was that Bloomberg reported that Crusoe Energy Systems had paused work on a data centre project in Wyoming, that Bloom Energy had been contracted to supply fuel cells to that project, that Bloom’s stock had fallen about 10% overnight in New York, and that MTAR is the company which builds the hot box assemblies that go inside Bloom’s fuel cells.

MTAR’s management came on an investor call and said, in effect, that they had received no communication about any pause, that they work with vendors beyond Bloom, and that their order book had roughly doubled over the previous one to two months. The stock recovered 12% the next day.

Then it happened again. Bloom kept falling through late June and July on reports of regulatory hurdles delaying an Oracle-linked data centre deployment. Bloom lost roughly 38% in a month. MTAR touched a record ₹8,714.95 on 19 June, then fell for weeks, hitting the 5% lower circuit in six of seven sessions, down about 34% from that high by 20 July. From 25 June it was trading in the BSE’s ‘T’ group and had been pushed into Stage 4 of the long-term ASM framework.

Through all of it, MTAR’s own numbers were going the other way. Q1 FY27 revenue was ₹360.7 crore against ₹156.6 crore a year earlier, up 130%. Profit after tax was ₹50.2 crore against ₹10.8 crore, up 364%. Management had raised FY27 revenue growth guidance to around 80%.

I have been thinking about that gap for two months, and this piece is what came out of it.

Because MTAR is the clearest available specimen of something that has quietly become one of the most crowded trades in Indian small and mid caps: proxy investing. The logic runs like this. The biggest capital cycle of our lifetimes, the buildout of AI compute and the power infrastructure to feed it, is happening almost entirely outside India. You cannot buy Nvidia’s foundry, Bloom’s fuel cell, GE Vernova’s gas turbine, or Vishay’s current sensing franchise on the NSE. But you can buy the Indian company that supplies a part into each of those. So you buy the supplier, and you get the theme.

That logic holds up as far as it goes, and it goes about a third of the way. The incompleteness is where the money is lost.

This piece is an attempt to write down what is actually being bought when you buy a proxy. It uses three worked examples, all Indian, all listed, all currently priced as though the theme has already arrived on their income statements: MTAR Technologies, TD Power Systems, and Shivalik Bimetal Controls. It is going to be reasonably long, and it is going to be arithmetic-heavy in places, because the entire argument is arithmetic.

If you want the compressed version, here it is.

When you buy a proxy, you are buying a fraction of a fraction. The theme’s spending flows to a customer, and the customer’s spending flows to your company, and both of those transfers are far smaller than the narrative implies. In MTAR’s case, the second transfer works out to roughly two to three paise for every rupee Bloom Energy bills. In Azad Engineering’s case, against GE Vernova, it is a fraction of one paisa.

Worse, in any given year the transfer is governed by the customer’s inventory policy rather than by the theme’s demand at all. In FY24, Bloom Energy’s revenue grew about 11% and MTAR’s grew 1.5%, against management guidance of 45 to 50%. In FY25, at the absolute peak of global electrification narrative, Shivalik Bimetal’s standalone revenue fell 4.3% and its profit fell 10.5%, because one American customer was working through a stockpile.

And yet the equity prices as though the transmission were instant and complete. That mismatch, between an annual transmission mechanism and a daily correlation, is the whole trade and the whole risk.

The proxies that survive this are the ones increasing their content per unit faster than their customers can commoditise them, rather than the ones with the biggest customer. Content per unit is the metric. Volume is the story.

Pour something cold. Let’s begin.

Part 1: Everyone Says “Sell the Shovels”. Almost Nobody Checks Who They Sold To

The picks-and-shovels analogy is the most overused framing in Indian small-cap investing, and it is worth spending two minutes on it precisely because it is invoked so lazily.

The historical version is roughly accurate. During the California Gold Rush, Sam Brannan bought up every pick, shovel and pan he could find and sold them out of his store at Sutter’s Fort. Levi Strauss sold dry goods and, eventually, riveted trousers. Both made far more reliable fortunes than the prospectors did. The lesson everybody draws is: in a gold rush, sell shovels.

But notice the thing the analogy quietly assumes and almost nobody carries across. Brannan and Strauss sold to the entire field. Every prospector who walked into Sutter’s Fort was a customer. Neither of them signed an exclusive supply agreement with one miner, built a warehouse next to that miner’s claim, and then discovered eighteen months later that the miner had over-ordered shovels and would not be back until the pile ran down.

That second business is a completely different asset, with a completely different risk profile, and it is the business most Indian proxies are actually in.

So before anything else, it helps to be precise about the four different things people mean when they say “proxy”.

The category supplier. Sells a component or subsystem into a whole industry, across many customers, in many geographies. TD Power Systems has shipped 8,055 generators into 113 countries. It sells AC generators that get paired with steam turbines, gas turbines, gas engines and hydro units, and its customers are the OEMs who assemble those machines. When behind-the-meter gas generation for data centres booms, TD Power benefits regardless of which engine brand wins. This is the Sam Brannan position.

The single-customer supplier. Sells a critical, qualified, hard-to-replicate assembly to essentially one buyer. MTAR and Bloom, with more than 55% of MTAR’s revenue attributed to Bloom and a relationship going back to 2009. Azad Engineering, which has built four dedicated manufacturing facilities in the Tunikibollaram Industrial Park, each configured around a specific OEM: GE Vernova, Mitsubishi Heavy Industries, Siemens Energy, Baker Hughes. This is a genuinely deep moat and a genuinely deep dependency, and those are the same fact viewed from two ends.

The listed subsidiary. The Indian arm of the global technology owner. GE Vernova T&D India, Hitachi Energy India, ABB India, Siemens Energy India. You get the parent’s technology, the parent’s brand and the parent’s qualification base, applied to Indian revenue. Which means you get the technology moat of the global theme without the global theme’s revenue. I wrote about this cohort at length in the power infrastructure piece earlier this year, so I will leave it there.

The narrative proxy. The company whose name, history or product catalogue associates it with a theme that does not meaningfully appear in its revenue. The tell is always the same: a well-known brand, a real but small business line in the fashionable segment, and a segment disclosure showing that line contributing low single digits to the top line while the stock trades on the association. This is the easiest failure mode to spot, because the annual report settles it in one table, and it will not detain us further.

The three companies in this piece sit in the first two buckets. That distinction, category versus company, will turn out to matter more than anything else on the balance sheet.

Part 2: The Arithmetic of the Fraction

Here is the calculation almost nobody does before buying a proxy, and it takes about ninety seconds.

Bloom Energy reported revenue of $2.02 billion for calendar 2025, up 37.3% on 2024’s $1.47 billion. It has guided calendar 2026 to $3.9 to $4.2 billion, roughly a doubling, and reported Q2 2026 revenue of $1,065.4 million against $401.2 million a year earlier. Its total backlog is around $20 billion, with product backlog around $6 billion, up about 2.5 times year on year. This is, by any standard, one of the more spectacular industrial revenue ramps of the decade.

MTAR reported FY26 revenue of ₹876.2 crore, up 29%. Public reporting and credit rating commentary put Bloom at more than 55% of that. So MTAR’s Bloom-linked revenue in FY26 was somewhere around ₹480 to ₹570 crore.

Convert Bloom’s calendar 2025 revenue at the ₹95.50 per USD rate MTAR itself used in its May 2026 order disclosure, and you get roughly ₹19,300 crore.

₹480 crore into ₹19,300 crore is about 2.5%.

A caveat that matters, and I would rather state it than bury it: Bloom reports on a calendar year and MTAR on an Indian fiscal year, so this ratio spans a three-month offset, and the underlying deliveries lead the customer’s revenue recognition by an unknown lag. Treat it as an order of magnitude and not a share-of-wallet figure. The order of magnitude is the point. For every rupee of revenue Bloom Energy bills, roughly two to three paise reaches MTAR’s income statement.

Run the same sum on Azad Engineering, which is the cleanest gas turbine proxy on the Indian market. Azad’s entire consolidated FY26 revenue was ₹602.98 crore, up 31.8%, with exports at 93.3%. That is about $63 million. GE Vernova’s calendar 2025 revenue was $38 billion, with orders of $59 billion and a backlog of $150 billion. Azad’s total revenue, across GE Vernova, Mitsubishi, Siemens Energy, Baker Hughes and its aerospace and defence customers combined, comes to roughly 0.17% of GE Vernova’s revenue alone. Its GE Vernova slice is necessarily smaller than that.

None of this makes these bad businesses. Azad’s top three energy OEMs account for around 81.5% of FY26 revenue, and it holds an eight-year single-source contract with Mitsubishi for hot section nozzle vanes, a component reportedly made by only three companies globally. That is a real moat. MTAR’s hot box position with Bloom is a nine-year manufacturing relationship with genuine qualification barriers.

What the arithmetic does is calibrate the transmission. When a headline says Bloom’s backlog has reached $20 billion, the instinct is to multiply MTAR’s revenue by some large factor. The correct instinct is to ask what MTAR’s content per unit is, whether that content is rising or falling, and what fraction of Bloom’s build rate MTAR is qualified to serve. A $20 billion backlog at a 2.5% pass-through, spread over an execution period nobody has published, is a very different object from a $20 billion backlog.

You are buying a fraction of a fraction. The theme’s capex is a fraction of the narrative, the customer’s revenue is a fraction of the theme’s capex, and your company’s billing is a fraction of the customer’s revenue. Three multiplications, all of them less than one.

You can also watch a video we made on this topic, here:

Part 3: The Transmission Problem, or Whose Inventory Are You Actually Buying

This is the part that decides returns, and it is the part the market consistently gets wrong in both directions.

A proxy’s revenue in any given year is set by the customer’s ordering pattern rather than by end demand, and the customer’s ordering pattern is a working capital decision made by a finance function you will never meet, in a currency you do not hold, on a calendar that does not match yours.

We have two clean, independent, documented instances of this in exactly the companies under discussion.

MTAR, FY24. Management had guided the market to 45 to 50% revenue growth and a revenue target of ₹830 to ₹860 crore. The target was revised down repeatedly through the year. The company finished FY24 with ₹580.8 crore, growth of roughly 1.5%, and profit close to halved. The cause was not a collapse in fuel cell demand. Bloom Energy’s own calendar 2023 revenue grew about 11% and calendar 2024 grew about 11% again. The cause was inventory destocking at Bloom, combined with a product migration from the Yuma to the Santa Cruz hot box generation. Bloom was working down a pile MTAR had already built and shipped. Bloom’s business was fine. MTAR’s year was not.

The credibility damage was severe and lasting. Promoter holding had fallen from over 50% in FY21 to about 36.4% by early 2025 and stands at roughly 29.4% today, which management has attributed to non-core family members selling. Whatever the reason, an investor who bought MTAR in 2023 on the strength of the fuel cell theme spent two years watching the theme work and the stock not.

Shivalik Bimetal, FY25. Shivalik makes thermostatic bimetal strips and electron beam welded shunt resistors, the components that let a circuit measure its own current. In an EV battery pack, in a smart meter, in a battery management system, something has to sense how much current is flowing, and a low-ohmic shunt is usually that something. If any Indian company should have been geared to the global electrification narrative in FY25, this was it.

Standalone revenue in FY25 fell 4.3%, from ₹469.80 crore to ₹449.59 crore. Profit after tax fell 10.5%, from ₹81.13 crore to ₹72.60 crore.

Why? Because Vishay Intertechnology, historically around 40% of Shivalik’s shunt business, had built inventory aggressively through the post-pandemic EV boom, demand slowed, and Vishay entered a long correction. Shivalik’s Whole-time Director Sumer Ghumman put it plainly on the Q3 FY26 call, filed with the exchanges on 12 February 2026: the big US challenge was concentrated in one customer, and part of that customer’s design had been phased out on top of the destocking.

Two companies. Two different customers. Two different continents. Two different end markets. The same failure mode.

And the same recovery mechanism, which is worth noting because the cycle cuts both ways. Ghumman told the same call that Vishay had developed components for the top one and top two global EV manufacturers, that the business had already restarted, and that he expected Vishay volumes back to peak levels in FY27. MTAR, having gone through the FY24 destock, is now guiding to around 80% revenue growth in FY27 and printed 130% in Q1.

This is the central mechanical fact about proxies. Their revenue is a differentiated, lagged, amplified function of the customer’s inventory position, not of end demand. Which means the proxy’s income statement is more volatile than the theme in both directions, arrives later than the theme, and cannot be forecast from the theme’s own disclosures.

The market does not price it that way. The market prices it as a live correlation with a US ticker.

Part 4: The Disclosure Asymmetry

Here is the structural unfairness at the heart of the trade, and I think it is under-discussed to the point of being almost unspoken.

Bloom Energy files 10-Qs and 8-Ks with the SEC. It publishes quarterly supplemental decks with backlog, acceptances and guidance ranges. It runs earnings calls covered by a large sell-side following. Its own 8-K notes that its market capitalisation went from roughly $5 billion at the end of December 2024 to roughly $79 billion by 15 June 2026. Anyone in the world can read all of it, free, the moment it is filed.

An MTAR shareholder gets none of that about the thing that determines more than half of MTAR’s revenue. What they get is this.

On 14 May 2026, MTAR disclosed under Regulation 30 that it had received blanket purchase orders worth USD 238.76 million, or ₹2,278.96 crore at ₹95.50 per USD, from “an international entity”. The customer’s identity was withheld on grounds of confidentiality. The execution timeline was stated as remaining to be decided. The filing confirmed the orders were not related party transactions.

That single order is roughly 2.6 times MTAR’s entire FY26 revenue. It was subsequently amended upward by ₹819.94 crore, taking it to about ₹3,100 crore, again from an “undisclosed existing customer”.

I want to be careful here. There is nothing improper about this. Confidentiality clauses in long-term OEM supply agreements are completely normal, and the practice runs both ways. INNIO announced a 1.1 gigawatt gas engine order in July 2026 without naming the data centre developer, citing exactly the same customary confidentiality. MTAR has, at other times, named Bloom explicitly, for instance in the ₹386.06 crore fuel cell and electrolyser order disclosed in 2025.

But look at the position this leaves the Indian shareholder in. The single largest order in the company’s history, worth more than two and a half years of revenue, comes from a party you cannot name, on a timeline that has not been set. And the way the market resolved this information gap on 11 June was by watching a US share price overnight and inferring.

You bear the customer’s business risk without any right to the customer’s disclosure. And because the only real-time signal available is the customer’s stock price, the proxy imports the customer’s volatility at full strength while importing the customer’s fundamentals at a two to three paise pass-through and a multi-quarter lag.

That is a genuinely bad trade in information terms, and it is the price of admission for the whole category.

TD Power sits at the other end of this. It does not name its customers either, but it does not need to, because no single one of them decides its year. Order inflow in Q1 FY27 was ₹734.10 crore, up 87%, with exports at 93% or ₹683.60 crore, and the closing order book was ₹2,207.30 crore as of 30 June 2026. Management describes demand as broad-based across steam, gas engine, gas turbine and hydro, driven by AI data centres, grid stabilisation and renewables. When your generators go into 113 countries, no single project pause in Wyoming is a thesis event.

Part 5: The Scale Paradox

There is one more structural feature, and it is the one that separates temporary proxies from durable ones.

The qualification moat that protects a proxy is strongest when the customer is small and weakens as the customer gets large.

The logic is not complicated. When your customer builds 500 units a year, qualifying a second supplier costs more in engineering time, tooling, audit and validation than it saves in unit price, so you keep your position and your margin. When your customer builds 20,000 units a year, that same qualification cost is spread over a vastly larger volume, dual sourcing becomes cheap insurance rather than expensive luxury, and every basis point of unit cost is worth chasing. Your customer’s success is the mechanism by which your pricing power is eroded.

So the question that decides a proxy over five years is: as the customer scales, does your content per unit go up or does your price per unit come down?

This is where Shivalik becomes genuinely instructive, because it is the one of the three that has explicitly answered the question with capital.

On the Q3 FY26 call, management laid out a forward integration into automotive busbar and PCB assemblies, with a ₹20 crore facility in Pune, chosen for proximity to the customers rather than to Shivalik’s existing Solan base. The economics are worth reading closely. Ghumman said an assembly for a two-wheeler EV battery pack would carry a value of roughly ₹2,000 to ₹3,000 per vehicle with 60 to 65% of that being material cost. Supplying only the components into that same assembly, Shivalik would have billed ₹100 to ₹400. EBITDA margin on the assembly runs about 8 to 10 percentage points lower than the core business. Revenue guidance for the line is ₹70 to ₹75 crore in FY27, ₹150 to ₹200 crore in FY28 and ₹250 to ₹300 crore in FY29.

Read that trade carefully, because it is the correct trade and it looks wrong on a screener. Shivalik is deliberately accepting a lower margin percentage in exchange for a five to fifteen times increase in content per vehicle, on a product whose barrier to entry is the same electron beam welding capability it has had for thirty years. The same logic explains the strip-to-component shift forced by the US Section 232 surcharges on copper and stainless steel: management noted that roughly 36 to 37% of supply was still in strip form, that tariffs pushed Vishay and Shivalik to convert that into finished parts faster than planned, and that this raises both the top line and the value added per unit shipped.

The other two are doing versions of the same thing. MTAR has flagged a strategic shift to integrated systems, targeting 200 to 300 basis points of margin improvement. TD Power is moving up from its historical machine sizes into the 40 to 200 MW generator segment through a UK R&D centre, with management saying opportunities above 100 MW were close to being signed and expected to be announced in August 2026, with the caveat that an 18 to 20 month manufacturing cycle means no FY27 revenue from them.

Every serious proxy management team is running the same play, which is to climb the customer’s bill of materials faster than the customer can commoditise the current position. The ones that stop climbing get repriced as vendors.

Part 6: The Three, Scored

Here is where the three companies actually stand, on the numbers, as of late August 2026.

Order book and export figures from company results releases and investor presentations. Customer concentration figures are from management commentary and secondary reporting rather than a single audited disclosure, and should be treated accordingly.

The three are three different answers to the same question, rather than substitutes for one another or a portfolio.

MTAR is the pure single-customer proxy. The deepest embed, the highest torque to the theme, the highest torque to a customer’s news flow, and the widest gap between operating reality and equity behaviour. The FY27 setup is genuinely strong: record FY26 order inflows of ₹2,453.3 crore, an order book split 51.2% clean energy fuel cells, 26.3% civil nuclear and 14% aerospace and defence, Q1 FY27 revenue up 130%, and guidance around 80% for the year. The civil nuclear and defence legs are becoming independent anchors, which is exactly the right direction. The valuation, at a trailing PE in the 150s and roughly 26 times book, requires that direction to hold without interruption.

TD Power is the category supplier. It is exposed to the same behind-the-meter power buildout, but through the industry rather than through a company. Third consecutive record year in FY26, order inflow of ₹2,238 crore up 51% with 79% from exports, and a book-to-bill above 1.19, which is the specific signal that demand is running ahead of the ability to execute. FY27 guidance has been raised to ₹2,600 crore, with capacity roadmapped to ₹3,200 crore by FY28 and above ₹4,000 crore for FY29 and FY30. The watch item is on the balance sheet rather than the income statement: trade receivables of ₹785 crore, and a widening gap between profit growth and operating cash generation as receivables and inventory scale with the order book. Growth-stage capital goods exporters do this, and it needs tracking every quarter rather than annually.

Shivalik is the mid-cycle proxy with the honest management. It is the slowest of the three on the top line, at 12.3% consolidated revenue growth in FY26, with 9M volume growth of only 3%. It is also the only one of the three where management has told you, in public, that it cannot measure the thing you might be buying it for.

Asked directly on the Q3 FY26 call how much of the business is exposed to data centres, Ghumman said that for EVs and smart meters the parts are custom to the application and can be classified, but for data centres the switchgear and energy storage devices involved are general application products, so the company genuinely cannot specify the exposure. He then gave the mix that can be measured: thermostatic bimetal goes roughly 80% into switchgear; shunts go roughly 60% into automotive and 35% into energy meters.

I found that answer more useful than any bullish exposure estimate I have read this year. A company that will not manufacture a data centre revenue number when it does not have one is a company whose other numbers are worth more.

Part 7: The Proxy Ledger

If this piece is worth anything beyond three case studies, it is this. Five questions, in order, before buying any company because of a theme that lives somewhere else.

One. What fraction of the end product’s price do you bill? Not the customer’s growth rate, not the customer’s backlog. Take the customer’s revenue, convert it, and divide your company’s theme-linked revenue into it. If the answer is a low single digit percentage, then a doubling in the customer’s revenue does not double your company unless content per unit also rises. Do the sum before you read the broker note.

Two. Who pays you, and is that name in a filing? A customer relationship that appears in an investor deck, a media interview, or a WhatsApp forward is not a disclosed fact. A customer relationship that appears in a Regulation 30 filing, a concall transcript filed with the exchanges, or an annual report is. Shivalik’s management naming Vishay on a filed transcript is a stronger basis for analysis than any amount of inference about who MTAR’s “international entity” might be. Where the name is withheld, size the position for the fact that you cannot verify the counterparty’s health.

Three. What actually moved revenue last year, end demand or the customer’s warehouse? Go back four to eight quarters and find a year where the theme was working, and the proxy was not. Both MTAR’s FY24 and Shivalik’s FY25 are that year. If you cannot find one, you have not looked far enough back. The existence of that year tells you the amplitude of the destock risk you are underwriting.

Four. Is content per unit rising? This is the single most important question and the one least often asked. Track rupees of value added per unit shipped, not revenue. Shivalik going from a ₹100 to ₹400 component to a ₹2,000 to ₹3,000 assembly is content per unit rising, and it is worth accepting eight to ten percentage points of margin to get it. A proxy whose content per unit is flat while its customer’s volume triples is a proxy being set up for a price negotiation.

Five. Who is funding the growth? Proxies to fast-growing global customers almost always finance their customer’s ramp through working capital. TD Power’s ₹785 crore of receivables and Shivalik’s net working capital days running at 250 to 260 in Q3 FY26 are both examples. This is normal and it is survivable, and it is also the thing that turns a strong order book into a cash call at exactly the wrong moment. Read the cash flow statement before the order book press release.

Anything that clears all five is a proxy worth owning through a cycle. Anything that clears three is a trade. Anything that clears one is a correlation.

Part 8: What I’m Watching

For MTAR (MTARTECH): whether the ₹2,278.96 crore blanket order, as amended to roughly ₹3,100 crore, acquires a named counterparty and a stated execution schedule in any subsequent filing or concall. The FY27 order book target of ₹5,000 crore against ₹2,581.9 crore at March 2026. Whether civil nuclear and aerospace revenue grows fast enough in absolute terms to bring the Bloom share of revenue below 50%. And a governance item that is not about Bloom at all: promoter holding, which has moved from over 50% in FY21 to roughly 29.4%.

For TD Power (TDPOWERSYS): the large generator announcements above 100 MW that management said were close to signing in August 2026, and specifically whether they carry the 18 to 20 month cycle management flagged. Operating cash flow conversion against PAT, quarterly rather than annually. Receivable days. And whether the FY28 capacity roadmap to ₹3,200 crore stays on the ₹50 crore debottlenecking budget or requires a larger capex round.

For Shivalik (SBCL): whether Vishay volumes return to prior peak levels in FY27 as management has guided, which is the single cleanest walk-the-talk test available on this name. The Pune busbar and PCBA line against the ₹70 to ₹75 crore FY27 marker. The strip-to-component ratio moving down from 36 to 37%. And net working capital days returning to the prior year’s range, which management said would happen by March 2026 and which is worth verifying rather than assuming.

For the category: the gap between order inflow announcements and revenue recognition across all three. Blanket purchase orders with undecided timelines are not backlog in the way an EPC order book is backlog, and a market that has learned to reward inflow announcements will eventually have to learn the difference.

Part 9: A Closing Thought

There is a particular kind of investor error that only becomes visible in hindsight, which is mistaking proximity for participation.

Standing next to a gold rush is not the same as being in one. The Indian companies in this piece are standing very close to the largest industrial capital cycle in living memory, closer than almost any other Indian listed asset, and that proximity is real and it is worth something. MTAR’s hot boxes are physically inside machines that will power American AI training clusters. Shivalik’s shunts are measuring current in battery packs built by the two largest EV manufacturers on earth. TD Power’s generators are spinning at sites where the grid could not deliver a connection in under five years.

But the rupee that flows back from all of that is small, late, lumpy, and set by somebody else’s finance department. And the share price that responds to all of that is fast, daily, and set by a ticker in another time zone that most holders cannot read and none of them can influence.

The gap between those two sentences is the entire proxy trade. Everything useful in this piece is an attempt to measure it.

Which leaves a simple discipline. Before you buy the theme, find out what fraction of the theme you are actually buying. Then find out whether that fraction is growing. If it is, you own a business. If it is not, you own a correlation with a lag, and the lag will not be kind to you in the quarter you need it most.

Measure the fraction before you buy the story. That is the entire game.

See you at the next inflection point.

Karnik

Inflection Point Research

Analyst Certification

I, Karnik Shah, certify that the views expressed in this report accurately reflect my personal views about the subject companies and their securities. No part of my compensation was, is, or will be, directly or indirectly, related to the specific recommendations or views expressed in this report.

Portfolio Disclosure

Companies discussed in this report: MTAR Technologies (NSE: MTARTECH), TD Power Systems (NSE: TDPOWERSYS), Shivalik Bimetal Controls (NSE: SBCL) and Azad Engineering (NSE: AZAD).

I hold positions in MTAR Technologies (NSE: MTARTECH), TD Power Systems (NSE: TDPOWERSYS), Shivalik Bimetal Controls (NSE: SBCL).

I have no position in Azad Engineering, Bloom Energy, Vishay Intertechnology, INNIO Group, GE Vernova, Mitsubishi Heavy Industries, Siemens Energy or Baker Hughes, and no position in any other company referenced in this piece beyond what is stated above.

I may initiate, add to, reduce or exit any of these positions at any time based on changes in fundamentals or portfolio rules. This disclosure is as of the date of publication.

Regulatory Disclosures

In relation to the companies discussed in this report, as of the end of the month immediately preceding the date of publication:

• Neither I nor my relatives nor Inflection Point Research holds any actual or beneficial ownership of 1% or more of the securities of the subject companies, other than as stated in the Portfolio Disclosure above.

• Neither I nor my relatives nor Inflection Point Research has any other material conflict of interest.

• I have not received any compensation from the subject companies in the past twelve months.

• I have not managed or co-managed any public offering of securities for the subject companies.

• I have not received any compensation for investment banking, merchant banking or brokerage services from the subject companies, nor for products or services other than research.

• The subject companies have not been clients of mine or of Inflection Point Research at any time in the past twelve months.

• I have not served as an officer, director or employee of the subject companies.

• Neither I nor Inflection Point Research has been engaged in market making activity in the subject companies.

Disclaimer

This report is published by Inflection Point Research, a SEBI-registered Research Analyst (Registration No. INH000023807).

This content is for informational and educational purposes only. It does not constitute investment advice, an offer or solicitation to buy or sell any security, or a recommendation of any specific investment action. It has been prepared without regard to the individual financial circumstances, objectives or risk tolerance of any person who may receive it.

Readers should conduct their own due diligence and consult a SEBI-registered investment adviser before making any investment decision. Past performance is not indicative of future results. Equity investments are subject to market risks, including the possible loss of principal.

Financial figures in this report are drawn from company results releases, investor presentations, exchange filings and concall transcripts filed with the exchanges, and from SEC filings in the case of Bloom Energy and GE Vernova. Figures attributed to secondary reporting rather than a primary filing are flagged as such in the text. While the information is believed to be reliable, no representation is made as to its accuracy or completeness, and it is subject to change without notice.

Registration granted by SEBI, membership of a Research Analyst Administration and Supervisory Body, and certification from NISM in no way guarantee the performance of the intermediary or provide any assurance of returns to investors.

Hello friends,

Start with a question that sounds trivial and turns out to organise an entire industry.

What is the hardest thing about making something spin?

Not the spinning. Any motor spins. The hard part is that a spinning shaft has to be held by something, and the moment it is held by something, the two surfaces touch. Touching surfaces under load generate friction, friction generates heat, heat softens metal, softened metal deforms, and a deformed shaft destroys everything bolted to it. Every rotating machine ever built is a negotiation with that problem.

A bearing is the answer, and it is a strange kind of answer. It is the only component in a machine whose entire purpose is to let something else move. It generates no power. It transmits no torque of its own. It performs no function you could point to on a specification sheet for the finished product. Nobody has ever bought a tractor because of its connecting rod bearings. Its whole job is to sit between two surfaces and make sure they never actually touch, and then to keep doing that, roughly a hundred million times, for a decade, in a gearbox at 120 degrees Celsius filled with hot oil and metal particles.

And when it stops doing that, everything attached to it fails within seconds.

This piece is about the companies that make those parts in India. It opens with the physics, because in this industry the physics dictates the economics with unusual directness. It then covers where bearings are used and why the next ten years look structurally better than the last ten. It looks briefly at the multinationals, SKF, Schaeffler and Timken, because they set the ceiling that every Indian bearing company prices against. And then it spends most of its length on the two listed Indian small caps that make bearings: NRB Bearings and Menon Bearings.

Those two are the reason I wanted to write this. They are the cleanest natural experiment I have found in Indian small caps this year.

If you want the short version first, here it is.

Two Indian companies, both making bearings, both selling into the same commercial vehicle and tractor cycle. One is four and a half times the size of the other, with lower margins, lower returns on capital and half the growth rate. The other is smaller, faster, earns 43 per cent more on equity and runs 410 basis points more operating margin at a quarter of the scale. On forward earnings, the market prices the two of them within 60 basis points of each other.

Either the market has made a mistake, or the difference between them is sitting somewhere other than the profit and loss account. It is sitting in the cash flow statement, and in the year the fast one produced its best growth it converted 36 per cent of operating profit into cash while the slow one converted 104 per cent. That single divergence is the whole piece.

Pour something cold. This is a long one.

Part 1: Two Different Answers To The Same Physics Problem

There are two fundamentally different ways to stop two metal surfaces from touching, and the entire listed Indian bearing universe splits along that line.

The rolling answer

Put something round between the two surfaces. Instead of a shaft sliding against a housing, the shaft rests on a set of balls or rollers that turn as it turns. Sliding friction becomes rolling friction, which is roughly an order of magnitude lower, and the contact patch becomes a moving line rather than a static plane.

This is the rolling element bearing, and it is what most people picture when they hear the word. A ball bearing has spheres. A cylindrical roller bearing has short cylinders. A tapered roller bearing has cones, which lets it carry load along the shaft as well as across it. And a needle roller bearing, which is the product that matters for the first of our two companies, uses very long, very thin rollers, typically with a length several times their diameter.

The needle geometry buys one specific thing: load capacity in almost no radial space. A needle bearing carries what a ball bearing of the same load rating carries while occupying a fraction of the depth. That makes it the default choice anywhere a designer is fighting for millimetres, which in practice means gearboxes, transmissions, rocker arms, universal joints, planetary assemblies, two-wheeler engines and every place inside a vehicle where a component has to fit into a cavity that was designed around something else.

The engineering problem in a rolling element bearing is geometry held to microns. The rollers must be identical to within fractions of a thousandth of a millimetre, the raceways must be ground to a surface finish measured in nanometres, the hardness must be uniform through a case depth of under a millimetre, and the whole assembly must hold that geometry after millions of load cycles. Get one roller half a micron oversized, and it carries a disproportionate share of the load, heats, and initiates the fatigue crack that ends the part.

The sliding answer

Or, accept that the surfaces will slide and make one of them sacrificial.

A plain bearing has no moving parts at all. It is a shell, usually a semicircular half of a cylinder, made of steel for strength, with a thin layer of a much softer alloy bonded to the inside face. The shaft slides directly against that soft layer, separated in normal operation by a film of pressurised oil a few microns thick. When the oil film breaks down, which it does at every cold start and under every shock load, the soft layer is designed to give way rather than let the shaft gall.

This is the bearing inside an engine. Every connecting rod has one wrapped around the crank pin. Every crankshaft main journal sits in one. Thrust washers stop the crankshaft walking along its own axis. These parts are consumed by the engine’s most violent environment: combustion pressure arriving as a hammer blow several thousand times a minute, at temperatures that would anneal most alloys, contaminated by carbon and metal debris.

The engineering problem in a plain bearing has nothing to do with geometry held to microns. It is metallurgy. What alloy, at what composition, bonded to steel by what process, will be soft enough to embed a stray particle of grit rather than let it score the crankshaft, hard enough not to extrude out from under the load, and fatigue resistant enough to survive a hundred million hammer blows? For decades the answer involved lead. Emission and material regulations progressively removed it, which forced the entire industry to redevelop its alloy chemistry from scratch on lead-free grades.

Why the split matters more than it looks

Hold that distinction, because it explains almost everything downstream.

A rolling element bearing company is a precision machining company. Its capital goes into grinders, honing machines, heat treatment lines and metrology. Its competitive question is whether it can hold tolerance at volume and cost.

A plain bearing company is a metallurgy company. Its capital goes into powder atomisation, sintering furnaces, cladding lines and bonding processes. Its competitive question is whether it owns the alloy and the bond, or buys them.

One of our two companies buys steel and forms it with great precision. The other atomises its own alloy powder and sinters its own bimetal strip in-house. That is the same difference expressed in capital expenditure, and it turns up again as a difference in operating margin roughly four hundred basis points wide.

Part 2: Where They Actually Live

The reason this industry is durable is that the addressable surface is absurd. Anything that rotates has bearings in it, and the count is much higher than intuition suggests.

A modern passenger car carries somewhere between 100 and 150 bearings, depending on how you count. Wheel hubs, four of them, each a substantial precision unit. Gearbox shafts, typically a dozen or more. The clutch release mechanism. The alternator, the water pump, the air conditioning compressor, the power steering assembly, the turbocharger, the seat recliners, the sunroof rails, the window regulators. A two-wheeler is smaller but proportionately just as populated, and the engine bearings in a single-cylinder motorcycle are working harder per unit of mass than almost anything in a car.

Step outside vehicles and the surface expands again. A wind turbine has a main shaft bearing that is several metres across and costs as much as a small apartment. A railway wagon has axle box bearings that are inspected on a fixed schedule for the life of the wagon. Machine tool spindles run on bearings ground to a tolerance grade most industries never encounter. Electric motors, pumps, compressors, conveyors, agricultural implements, construction equipment, cement kilns, steel rolling mills. And, more recently, robotics, humanoid actuators and data centre cooling assemblies, all of which want small, quiet, low-friction rotation in very large quantities.

Two structural features of this demand are worth holding on to.

The aftermarket is an annuity that compounds independently of new sales. A bearing is a wear part. It has a design life, and the vehicle it sits in usually lives longer than that. Every vehicle sold in India in the last fifteen years is a future replacement bearing sale, and India’s vehicle parc has been compounding for a long time. This is why bearing companies do not experience the auto cycle at quite its full amplitude on the downside, and why aftermarket share is one of the more important numbers in the file.

Qualification is slow, and it is customer-specific. A bearing that goes into a specific gearbox is designed for that gearbox, tested on that platform, and approved for that part number. The approval takes months to years. Once it exists, the customer has an expensive, slow and risky path to replacing you, and the part typically stays yours for the life of the platform. This is the moat in this industry, and it is why the interesting question about any bearing company is never how good its product is in the abstract but how many designed-in positions it holds and how hard those are to dislodge.

If you would rather watch a video we made on this topic, you can do so here:

Part 3: Why The Next Ten Years Look Different

Five things are moving in the same direction at once, and one is moving hard against.

The import gap is the largest single opportunity

India does not make most of the bearings it uses. Industry estimates put the country’s bearing consumption met by domestic production at somewhere under half, with the remainder imported. In 2023 India imported over 1.2 billion dollars of bearings, and Chinese suppliers accounted for roughly 44.5 per cent of that by value. United Nations trade data puts India’s imports from China of ball and roller bearings and parts at approximately 648 million dollars in 2024.

The gap comes from range rather than from capability at the top end. India requires something on the order of 50,000 distinct bearing sizes and configurations across all its applications, and estimates of what the domestic industry actually produces run closer to 5,000. Every size not made here is imported by definition. Closing even a modest share of that spread is a structural volume opportunity that has nothing to do with the auto cycle.

(These are third party industry estimates rather than filings, and they are directional. They are useful for scale, not for modelling.)

Industrial and infrastructure capex is in a genuine upcycle

Railways, renewables, defence, cement, steel and general engineering are all spending. Every rotating asset installed in that build out carries bearings at the point of installation and then again on a replacement cycle for its whole life. The industrial bearing market is structurally higher margin than automotive because volumes are lower, specifications are more varied, and the buyer’s cost of failure is much higher than the cost of the part. This is why almost every bearing company in India, including both of ours, is trying to push its industrial share up.

The automotive cycle is turning up from a base

Commercial vehicles and tractors, which are the primary end markets for one of our two companies, have been through a full cycle and are recovering. The multinationals have been reporting double-digit growth through FY26. Cyclical recovery is not a thesis on its own, but it is the tide underneath everything else here.

Localisation is now the multinationals’ own strategy

Both SKF and Schaeffler have publicly committed capital to Indian manufacturing expansion. When a multinational localises, it does not only displace its own imports, it pulls a domestic supplier ecosystem up with it, because localisation targets are measured on value added inside the country. That is a tailwind for Indian component suppliers even when the multinational is the nominal competitor.

And then there is electrification, which does not point the same way for both products

This is the one thing in this piece I want to be completely unambiguous about, because it is the single largest difference in terminal risk between our two companies and neither of them is priced for it on a three-year view.

A battery electric vehicle has no connecting rod. It has no crankshaft. It has no crank pin, no main journal, and no thrust washer holding a crankshaft in position. The plain bearing content of a fully electric powertrain is zero. Not lower. Zero.

A battery electric vehicle does still have an e-axle reducer, and that reducer has gears, and those gears run on shafts, and those shafts run on rolling element bearings including needle bearings. The content per vehicle falls compared with a conventional powertrain because there is no multi-speed gearbox and no valvetrain, but it falls to a reduced number rather than to nothing.

The mitigation for the plain bearing side is real, and it is worth stating fairly: the end markets where these parts live, which are heavy commercial vehicles, tractors, gensets, marine engines, compressors and stationary diesel, are the slowest electrifying segments in India by a wide margin. A long haul truck and a tractor will run on diesel long after the passenger car market has moved. That pushes the risk out by a decade or more. It does not remove it, and on a fifteen-year view the two products face structurally different terminal outcomes.

Neither company’s three-year picture reaches far enough out for this to bite. It belongs in a comparison rather than in either standalone file, which is exactly why I am raising it here.

Part 4: The Ceiling, Or Why SKF, Schaeffler And Timken Matter Even If You Never Buy Them

You cannot understand Indian bearing economics without understanding who sets the price.

Three multinational subsidiaries dominate the organised Indian bearing market. SKF India (NSE: SKFINDIA), the Swedish group’s Indian arm, which has been here for over a century. Schaeffler India (NSE: SCHAEFFLER), the German group, which reports on a January to December financial year rather than the Indian April to March convention, a detail that trips up more peer comparisons than you would expect. And Timken India (NSE: TIMKEN), the American tapered roller specialist.

The structural development worth knowing is that SKF India completed the demerger of its industrial business effective 1 October 2025, with shareholders receiving one share of SKF India (Industrial) Limited for every share held, leaving the listed SKF India as the automotive entity. The industrial entity listed subsequently. Two focused companies now exist where there was one, each with its own capital allocation and its own investment programme, and the group has committed to several thousand crore of Indian capacity investment through 2030.

These are excellent businesses with genuine technology access from their parents, and they trade accordingly. As at 24 August 2026, the peer file carries Timken India at roughly 57 times earnings, Schaeffler India at roughly 51 times, and SKF India at roughly 38 times. Our two small caps trade at roughly 27 to 28 times forward earnings. Hold that spread in mind for Part 8.

But the multinationals matter to this piece for what their existence does to pricing rather than for their multiples.

Here is the uncomfortable fact about rolling element bearings as a product category. A crankshaft designed for a specific engine is that customer’s part and nobody else’s. A 25 millimetre needle roller bearing conforming to an ISO dimensional standard is, in principle, everybody’s part. The specification is set by an international standard rather than by a customer’s proprietary drawing. There is a global catalogue. There is a large, competent, low-cost Chinese supply base. And there are four multinationals with a century of brand equity, global distribution and application engineering support sitting on top of the market.

That is why the most finished, most precision-intensive, most technically demanding product in the Indian metal forming universe, a fully ground and boxed bearing, runs a lower operating margin than a forging company that ships something the customer still has to bolt into an assembly. Finishing the part gets you onto the field. It does not win the game. What wins is whether the specific position you hold is qualification gated and expensive to move.

Which brings us to the two companies.

Part 5: NRB Bearings, Or Four Thousand Ways To Be Designed In

NSE: NRBBEARING | BSE: 530367 | Market cap approximately ₹4,660 crore as at 24 August 2026

Read this before anything else. There is a separate listed company called NRB Industrial Bearings Limited (BSE: 535458). It is a different company, run by a different branch of the same family, and it is not the subject of this section and not a subsidiary of the company discussed here. NRB Bearings Limited is CIN L29130MH1965PLC013251 and consolidates the separately listed SNL Bearings (BSE: 505827). Confusing the two is the single most common error made about this name.

Where it came from

NRB was established in June 1965 as an Indo-French joint venture between Nadella of France and Trilochan Sahney, and it was the first company to manufacture needle roller bearings in India. The French partner has held a 26 per cent stake for most of that history. From 1997 the company made a deliberate and expensive decision to own its own technology rather than continue licensing it, which is the choice that separates a genuine engineering company from an assembler with a badge.

Sixty-one years later, the position is remarkable for a company this size. NRB’s own materials state an estimated 60 per cent share of the domestic needle roller bearing segment, and that over 90 per cent of vehicles running on Indian roads carry an NRB part. It holds a 60 to 70 per cent share of business at its lead customers and 100 per cent on some individual applications.

What it makes and who buys it

Needle roller bushes and cages are approximately 66 per cent of revenue. Ball and roller bearings are 23 per cent. Auto components are the remaining 11 per cent. Behind those three lines sit roughly 4,000 stock-keeping units, and the important characteristic of that catalogue is that most of it is designed to a customer platform rather than sold from a shelf.

The customer list is the kind of thing that should be read slowly. Mercedes-Benz, where NRB has content in every transmission from the A Class to the Maybach. BMW, from the 1 Series to the 7 Series. Mack, Freightliner, Scania and Volvo on the heavy truck side, largely through a patented combined cylindrical thrust bearing used in container trailer kingpins. General Motors, where a planet pin has been nominated for the Corvette through a tier one supplier. And Hindustan Aeronautics, for a plain spherical bearing used on the Sukhoi 30, which management states fewer than five firms worldwide are able to make.

Channel mix is roughly 65 to 70 per cent domestic original equipment and tier one, 12 to 15 per cent domestic aftermarket, and 20 to 25 per cent exports.

And then there is the number that I think is the most underrated disclosure in this entire pair. No single NRB customer, taken together with its tier one, exceeds approximately 7 per cent of revenue. For an Indian auto ancillary that is close to best in class, and it is stated in the annual report rather than asserted on a call. Losing the largest customer outright would hurt and would not be existential.

The second and third legs

Industrial bearings are approximately 14 per cent of revenue and growing at 34 per cent, targeted toward 20 to 25 per cent. This is the diversification that matters most, because industrial revenue carries better margins and has no powertrain exposure whatsoever.

Aerospace runs through Mahant Tool Room in Bengaluru, acquired as a business during 2026, doing precision machined components. The order book is approximately ₹30 crore, entirely with Hindustan Aeronautics, and AS 9100 certification was described in May 2026 as roughly six months away. Management’s stated aspiration is ₹300 crore of aerospace revenue by FY31. Treat that as an aspiration with a certification gate in front of it.

NRB Unitec Friction Solutions, a joint venture with Italy’s Mondial Group, is relocating to Aurangabad with commissioning targeted for April 2027.

And there is a list of genuinely open-ended vectors that management has described, covering data centre bearings, robotics, humanoids and urban air mobility, with initial orders described in the region of 250,000 units each. I mention these for completeness and assign them nothing. They are early enough that any number attached to them today would be invented.

The numbers, and a warning that has to come first

The definitional warning. NRB’s FY26 operating profit is ₹232 crore on the Screener computation and ₹267 crore on the company’s and Crisil’s, which is a margin of 17.4 per cent against 19.5 per cent. The gap is the treatment of other income and certain foreign exchange items. Menon has no equivalent divergence: ₹64.42 crore on ₹299.46 crore is 21.5 per cent on both bases. Every margin comparison in this piece is on the Screener basis for like for like consistency, with the company basis stated alongside where it matters. Mixing the two is the easiest way to draw a false conclusion about this pair, and I have done it myself in an earlier draft.

FY26 revenue was ₹1,335 crore, up 11.3 per cent. Operating profit ₹232 crore on the Screener basis, a 17.4 per cent margin, or ₹267 crore and 19.5 per cent on the company basis. Reported profit after tax was ₹146 crore. Return on equity was 15.6 per cent, and return on capital employed was approximately 17 per cent on the Screener basis.

Q1FY27, reported on 10 August 2026, was the strongest print in the series: revenue ₹370 crore, up 19.2 per cent, EBITDA ₹64 crore at a 17.3 per cent margin, profit after tax ₹38 crore.

Read that Q1 margin carefully, because it is the most important single number for this name. 17.3 per cent is below the FY26 average and below the 18 to 20 per cent band management guides to. Other expenses rose approximately ₹10 crore sequentially on electricity, logistics and petroleum inputs.

The ten-year growth rate is a trap

NRB’s ten-year revenue compounding reads about 7 per cent, which looks damning until you inspect the window. It contains a peak-to-trough from ₹965 crore in FY19 to ₹762 crore in FY21. FY16 to FY19 compounded at 12.8 per cent. FY21 to FY26 compounded at 11.9 per cent. Underlying earnings across FY24 to FY26, stripped of exceptional items, compounded at roughly 19.5 per cent.

This is a cyclical business that does low to mid teens in an upcycle and gives some back in a downcycle. Reading its long-window average as a structural growth rate is a measurement error rather than a finding.

What genuinely improved

Four years of working capital repair, and this is documented rather than asserted. Cash from operations against operating profit went 28 per cent, 56 per cent, 86 per cent, 61 per cent and then 104 per cent in FY26, which is ₹240.9 crore of operating cash against ₹232 crore of operating profit. Inventory days moved from 364 to 349 to 302 across FY24 to FY26. The cash conversion cycle came down from 338 days to 281. Debtor days sit at 67, with a more recent cut showing 57.4.

Inventory at 302 days remains the weakest line on the balance sheet in absolute terms, and against 4,000 stock-keeping units and a large aftermarket book, it is arguably structural. The direction of travel is what matters, and it has moved in the same direction for three consecutive years.

The ownership story, which reads badly and is not what happened

Promoter holding fell 6.47 percentage points in a single quarter. Here is what actually took place.

Harshbeena Sahney Zaveri holds 41.48 per cent and has not sold a single share. She went from 13.60 per cent to 41.48 per cent by inter se transfer from Trilochan Sahney Trust 1 in November 2025. The entire reduction is Trust 1 selling down from 7.00 per cent to 0.53 per cent between 11 May and 8 June 2026 at around ₹330, with proceeds used to prepay promoter loans at Tata Capital and Aditya Birla Capital.

The consequence is the part that matters. Encumbrance on the promoter stake collapsed from roughly 91 per cent in March 2025 to about 12.2 per cent, with further releases disclosed through August 2026. Institutions absorbed the entire sale: combined foreign and domestic institutional holding is 32.77 per cent, up 8.02 percentage points in one quarter.

Set against the Family Settlement of 14 February 2025, under which the Zaveri family took exclusive control of the NRB group at a cost of ₹55.12 crore paid to NRB Industrial Bearings, this is the end of a multi year family and leverage overhang rather than an operator losing conviction.

What I would want answered

Three chief financial officers inside FY26 and three company secretaries in eighteen months. That is churn, and on current evidence it is transitional rather than structural, but it is the kind of pattern that deserves a question on a call rather than a shrug.

Sahney Bearings Private Limited, a 99 per cent acquisition from a promoter, board-approved on 24 October 2025, absent from the FY26 annual report, consideration undisclosed. Unresolved for ten months and worth watching.

Standalone exports fell 2.3 per cent in FY26 to ₹248 crore. Consolidated international revenue grew a little over 4 per cent, which the managing director himself described as not satisfying. The FY27 guide of 10 to 14 per cent international growth is, on the record, the least proven line in the file.

The disclosure nobody models

NRB discloses lifetime nominated business, which is the total value of platform business it has been awarded and will supply over the life of those platforms. That number moved from ₹800 crore to ₹1,100 crore in a single quarter, up 37.5 per cent, largely on the Corvette planet pin nomination.

This is better forward information than an order book, because it leads revenue by two to three years and reflects design wins rather than purchase orders. As far as I can tell, almost nobody models it. If I were tracking one number on this company, it would be that one rather than the aerospace narrative.

Part 6: Menon Bearings, Or What Happens When You Own The Metallurgy

NSE: MENONBE | BSE: 523828 | Market cap approximately ₹1,539 crore as at 21 August 2026

Read this before anything else. Menon Bearings Limited, CIN L29130PN1991PLC062352, is a distinct listed entity from Menon Pistons Limited and from Menon and Menon Limited. The three share a family origin in Kolhapur and share nothing else that matters here: neither of the other two is a related party of Menon Bearings under Ind AS 24. Menon Bearings consolidates Menon Alkop Limited, Menon Brakes Limited and Menon Bearings New Ventures Limited. The stock trades in the NSE BE segment, which has implications for liquidity discussed below.

Where it came from

Menon Bearings was incorporated on 4 July 1991 and commenced business in September of that year, coming to market with a public issue in January 1993 to part-finance a project of ₹8.17 crore. Early technical know-how for design and manufacture came from King of Israel. It supplied Telco, Maruti, Mahindra, Kirloskar Cummins, Greaves and Eicher from the beginning.

Kolhapur was and remains a foundry and engineering cluster, which matters more than it sounds. A company that grows up inside a casting and machining ecosystem develops different instincts about what to make and what to buy than a company that grows up next to a customer.

Thirty-three years of listed history, and in that time the company has never had a qualified audit opinion and has never had a single share pledged.

What it makes

The core is bimetal engine bearings: connecting rod and crankshaft shells, camshaft and rocker arm bushes, thrust washers, flange bearings, PTFE-coated bushings. Roughly 71 per cent of revenue and around 80 per cent of profit. Everything in that list lives inside a high-stress diesel engine.

The signature technical position is lead-free. Menon runs a dedicated, physically isolated lead-free foundry and sintering line for BS IV and BS VI grades, and commissioned a thrust washer facility, Unit III, in 2025 using lead-free foundry technology. Annual bimetal capacity went from 486 lakh units to 580 lakh units in FY26.

Menon Alkop, the high-pressure aluminium die casting business separated into its own company in April 2024, did ₹72.9 crore of FY26 revenue at a 12.7 per cent profit before tax margin. It is the capital-heavy leg, with an asset turn of 0.93 times.

Menon Brakes, asbestos-free brake linings, did ₹9.1 crore of revenue at a 7.8 per cent margin with accumulated losses of ₹2.07 crore. It is sub-scale, and it absorbed the largest share of concall airtime for 3.1 per cent of revenue, which is worth noticing as a signal about where management attention sits.

Customers include John Deere, Tata Motors, Cummins, Mahindra, Escorts, International Tractors, Brakes India, Allison Transmission and Honeywell. Exports run to the United States, Japan, the United Kingdom, Italy, France, China, Mexico, Brazil, the Netherlands, the UAE and Belgium.

Customer concentration is not disclosed at customer level. Management states that bimetal relationships average over two decades and that Menon is the sole or major supplier for most of them. Sole supplier status is a moat and a concentration at the same time, and at ₹300 crore of revenue the arithmetic implies a small number of accounts carry a large share.

The control experiment

Every small cap manufacturer in India claims vertical integration. Menon is one of the very few where the claim can be tested against a control, and this is the passage I would keep if I had to throw the rest of the piece away.

Bimetal Bearings Limited (BSE: 505681), founded in 1961 in partnership with Clevite of the United States and Repco of Australia, sourcing technology from Daido Metal of Japan, makes the identical product list. Thin-walled bearings, bushings, thrust washers, bimetallic strip and alloy powder. Same products. Same market. Same raw materials. Overlapping customers. Same cycle. And, in FY26, almost the same revenue.

Two companies, the same products, the same revenue, and three and a third times the profit. That is a measurement rather than a narrative.

So what is Menon doing differently?

It atomises its own alloy powder by water jet, and sinters that powder into copper alloy bimetal strip in-house, then machines and plates on the same campus. The metallurgy, which is where the value in a plain bearing actually sits, is captured rather than bought.

It builds its own machines. An in-house tool room using SolidWorks and AutoCAD produces the company’s special-purpose machines, tooling, gauges and fixtures. The consequence shows up in the most striking capital number in this pair: Menon talks credibly about adding 25 to 30 per cent of bimetal capacity for ₹9 to 10 crore, and about reaching a consolidated asset turn of 2.0 on total capital expenditure of only ₹35 crore. For comparison, NRB’s approved programme is ₹270 crore across April 2025, February 2026 and May 2026, on a stated conversion of ₹100 crore of capital expenditure into ₹130 crore of sales, an asset turn of 1.30 times.

And it chooses mix over tonnage. The explicit stated choice is to sell high-value parts rather than chase volume, which is precisely the difference against the Bimetal control.

The year everything changed

For four years, Menon was a dull compounder that was not compounding. Revenue went from ₹195 crore in FY22 to ₹243 crore in FY25, about 11 per cent a year, and profit went nowhere at all: the FY23 profit after tax of ₹33 crore was not exceeded until FY26.

Then FY26 broke the pattern in one year. Revenue up 23.5 per cent to ₹299.46 crore. EBITDA up 42.6 per cent to ₹64.42 crore, a 21.5 per cent margin. Profit after tax up 53.4 per cent to ₹38.25 crore. Return on capital employed 25.5 per cent, return on equity 22.3 per cent.

Q1FY27, reported on 22 July 2026, accelerated again: revenue of ₹92 to 94 crore, up 37.2 per cent, EBITDA of ₹22.38 crore at a 23.8 per cent margin, profit after tax of ₹14.11 crore, up 67.3 per cent. That is the fourth consecutive quarter of acceleration and the second consecutive quarter above a 22 per cent margin, after Q4FY26 printed 25.3 per cent.

Where did it come from? Utilisation, mostly, on capacity already paid for. Bimetal utilisation moved from a 60 to 75 per cent range to 85 per cent. Alkop went from 45 per cent to 65 per cent against a guided 90 per cent. Consolidated asset turn is 1.5 against a 2.0 target. A meaningful part of the operating leverage runway is already sitting on the balance sheet, which is why the growth does not require a capital programme behind it.

The dilution inside that number is worth stating: Alkop at 12.7 per cent and Brakes at 7.8 per cent profit before tax margin are pulling down a bimetal core running at 18.9 per cent. Management presents segment EBITDA of 22, 21 and 12 per cent, which reads considerably more uniform than the audited AOC-1 profit line actually is.

And now the part that changes the whole comparison

Everything above favours Menon. This reverses it.

In FY26, the year the growth arrived, Menon converted 36 per cent of operating profit into cash from operations. ₹23 crore of operating cash against ₹64.42 crore of operating profit.

Here are the two series side by side, and they cross in FY24.

Two companies moving in opposite directions for four consecutive years, crossing in the middle.

The cause is named and it is not mysterious. Menon’s United States export terms run beyond 180 days, and the export ramp producing the revenue growth is the same thing producing the receivable build. Debtor days moved from 91 to 111 in the year revenue accelerated. On the Q1FY27 call management disclosed both a collections issue and an advance payment arrangement.

I want to be careful about what this is and is not. The receivables were explained on the Q4FY26 call, and Crisil confirmed the build was funded on 35 per cent utilisation of the working capital line. Nothing here reads as an accounting concern. What it is, is a structural constraint on how fast this company can grow without external funding.

Here is that constraint as arithmetic. Sustaining revenue in the region of ₹579 crore at 111 debtor days requires roughly ₹170 crore of receivables against about ₹91 crore today. That is ₹80 crore of incremental working capital, against FY26 operating cash of ₹23 crore. It either caps the achievable growth rate, consumes the 42 per cent dividend payout, or adds leverage. There is no fourth option.

(That revenue figure is an illustrative modelling input for the arithmetic, not a forecast.)

Ownership, and the buyer who matters more than the seller

Promoter holding has drifted down over a decade, from 74.67 per cent in FY17 to 66.67 per cent now. The most recent step was 9,96,909 shares sold at ₹240 on 14 August 2026, including ₹7.13 crore by Sucheta Nitin Menon. There has never been a pledge.

The buyer is the more interesting half. Persistence Capital Fund I took 8,75,000 of those shares for ₹21 crore. Set that against a five-year trend of institutional ownership shrinking, from 2.42 per cent foreign institutional holding in FY21 to 0.14 per cent now, with domestic institutions moving from effectively zero to 0.40 per cent in the June quarter. No broker covers the name. Combined institutional ownership is 0.54 per cent, against 32.77 per cent at NRB.

The company also carries one confirmed governance flag in my framework: Mani Auto Components, a promoter-connected partnership operating in the same industrial classification. It is unresolved.

And the liquidity picture needs saying plainly. Menon trades in the NSE BE segment, with a ₹1,539 crore capitalisation, 24,403 shareholders and effectively no institutional float. Exit is harder here than the market capitalisation alone suggests.

Part 7: The Head To Head

Read the governance row again, because the framework’s answer and the historical record point in different ways and both are true. NRB clears my governance module with zero confirmed fails, and Menon carries one. But Menon has never had a pledge and has never had a qualified opinion in thirty-three listed years, while NRB had 91 per cent of its promoter stake encumbered as recently as March 2025, three chief financial officers in eighteen months, and a family dispute that cost shareholders ₹55.12 crore to settle. NRB scores better because its problems are documented as resolved and Menon’s flag is unresolved. That is the correct treatment, and it should not be mistaken for NRB having the cleaner history. It has the cleaner forward position.

Read the moats correctly too. NRB’s moat is breadth: thousands of designed-in parts, no single customer above about 7 per cent, an installed base that renews itself with every platform refresh. Menon’s moat is depth: one product family made better and cheaper than anyone else in India can make it, proven against a listed control. Breadth protects against losing any single customer. Depth protects margin. Neither protects against the cycle, and both are in the same cycle.

And a word on correlation, since these two get filed in the same bucket. They are not substitutes, and they are not hedges. NRB skews to passenger vehicles, two-wheelers, transmissions and a deliberate industrial and aerospace push. Menon skews to heavy-duty diesel in commercial vehicles, tractors, gensets and marine. A commercial vehicle downturn hurts both and hurts Menon more. A passenger vehicle slowdown hurts NRB more. Treating the pair as diversification would be a mistake. They are two draws on the same cycle with different internal economics.

Part 8: What The Market Is Actually Paying

There it is. NRB is three times the market capitalisation and four and a half times the revenue. And on forward earnings, the market prices these two companies within 60 basis points of each other.

It does that despite Menon earning 43 per cent more on equity, running 410 basis points more operating margin on a like-for-like basis, and growing twice as fast.

There are only two readings available. Either the market is wrong about Menon, or the market is charging Menon exactly the right discount for a 36 per cent cash conversion, a ₹1,539 crore capitalisation with effectively no institutional ownership and no broker coverage, and precisely one year of proof.

So the honest statement of the difference between these two companies is this. The gap in modelled outcome is entirely a bet on the persistence of a growth differential, and none of it is a bet on which is the better business. Anyone who prefers Menon on the strength of its returns and its growth is, whether they have articulated it or not, underwriting three more years of roughly 25 per cent growth at an Indian micro cap with a 180-day export receivable. Anyone who prefers NRB on the strength of its cash conversion and its customer book is underwriting a margin reset that has exactly one year of evidence behind it and a first quarter that already printed below the guided band.

Both of those are defensible positions. Neither of them is a quality judgement.

Part 9: What I Am Watching

Both names have their primary test in the same six-week window, which is convenient.

Menon, H1FY27, late October 2026. The cash flow statement, and nothing else. Finished goods and work in progress, receivable days, and cash from operations against half-year operating profit. Cash conversion recovering from past 55 to 60 per cent validates the growth path as fundable. Another sub-40 per cent half tells you the growth is being bought with the balance sheet, and the whole case changes shape.

NRB, Q2FY27, early November 2026. The consolidated EBITDA margin on a company basis, and specifically the other expenses line, jumped about ₹10 crore sequentially in Q1FY27. A second consecutive quarter below 18 per cent is a serious problem for the margin reset story. At or above 19 per cent with industrial above 15 per cent of revenue, the reset looks real.

NRB, around November 2026. AS 9100 certification at Mahant Tool Room, stated in May 2026 as roughly six months away. Certification plus a non-Hindustan Aeronautics order is what moves aerospace from a narrative to a business.

NRB, quarterly. The lifetime nominated business figure. It moved 37.5 per cent in one quarter; it leads revenue by two to three years, and almost nobody looks at it.

NRB, any board outcome. Sahney Bearings Private Limited. Unresolved since 24 October 2025 and absent from the FY26 annual report. Consummation without an independent valuation opinion would be a meaningful governance event.

NRB, February and May 2027. Whether the FY27 international growth guide of 10 to 14 per cent is met, against standalone exports that fell 2.3 per cent in FY26.

Menon, ongoing. The institutional register. Domestic institutions moved from 0.03 per cent to 0.40 per cent in the June quarter and Persistence Capital Fund I took ₹21 crore at ₹240 on 14 August 2026. In a name with no coverage and no institutional float, discovery is the only re rating lever that has not already been pulled.

Menon, ongoing. Any disclosure that resolves or worsens the Mani Auto Components position.

Part 10: A Closing Thought

There is something worth sitting with in the fact that two companies making objects most people will never see, in a category most people cannot name, have arrived at the same market verdict by opposite routes.

One decided that the way to make money from friction was to know more designs than anyone else, four thousand of them, each shaped to a specific cavity in a specific gearbox, so that no single customer could ever hurt you and every platform refresh renewed the annuity. It took sixty-one years, and it has an installed base in more than nine out of ten vehicles on Indian roads.

The other decided that the way to make money from friction was to own the metal itself. Atomise your own powder, sinter your own strip, build your own machines, and refuse to chase tonnage. It took thirty-three years, and it earns three and a third times the profit of a company making the identical products at the identical revenue.

Both are right. Both work. And the market, looking at them, has decided they are worth the same multiple, because one of them cannot yet turn its profit into cash and the other cannot yet prove its margin will hold.

That is a more honest conclusion than the one I set out expecting to write, which was that the smaller and faster company was obviously the better one. The profit and loss account said so. The cash flow statement disagreed. In a manufacturing business running an 111-day receivable, those two documents do not carry equal weight, and finding that out was worth the several weeks this took.

Read the cash flow statement before you read the growth rate. In small-cap manufacturing, that is most of the game.

See you at the next inflection point.

Karnik

Inflection Point Research

Portfolio Disclosure

Inflection Point Research and connected persons hold no position in NRB Bearings (NSE: NRBBEARING) as at the date of this publication. The writer of this article, Karnik Shah, does have a position in Menon Bearings (NSE: MENONBE).

I hold no position in Bimetal Bearings, NRB Industrial Bearings, Menon Pistons, SKF India, SKF India (Industrial), Schaeffler India or Timken India. I may initiate, add to, reduce or exit any position at any time based on changes in fundamentals or portfolio rules.

SEBI Disclosure

This report is published by Inflection Point Research, a SEBI-registered Research Analyst, Registration No. INH000023807. This content is for informational and educational purposes only and does not constitute investment advice, an offer to buy or sell securities, or a recommendation of any specific investment action. Readers should conduct their own due diligence and consult their financial advisors before making investment decisions. Past performance is not indicative of future results. Equity investments are subject to market risks. All financial figures are stated with their source and as-of date; market prices and multiples move and should be re-verified before use.

If this piece was useful to you

Forward it to one person who has ever told you that a small cap with better margins and faster growth must be the better business. That is how this newsletter finds the right readers, one thoughtful forward at a time.

Until next week,

Karnik

MSTC reported FY26 profit down 46%, and then reported what it calls its best ever first quarter since listing. Both come from the same set of filings, and the gap between them is most of what is interesting about this company right now.

The short version is that a business most people still think of as a scrap trader stopped being one in the quarter just reported. Standalone marketing revenue for the June quarter was Rs 0.00 crore. What remains is a single-segment marketplace that auctions coal blocks, mineral blocks, iron ore, seized assets, state land and liquor licences on behalf of the Indian government, takes a fee, and never owns the asset. In FY26, goods worth Rs 79,737 crore crossed it. The fee on all of that came to Rs 329.72 crore, which is a blended take rate of 0.41%, earned at 59% operating margins and 30% return on capital employed. The reported profit decline was measuring the exit from the old business. The operating line was measuring the new one, and it grew 19%.

What I try to do in the video is separate two things that are easy to run together. MSTC has eight revenue engines. Four of them earn money today: the e-auction platform, e-procurement, the interest on bidder deposits it holds, and a vehicle scrappage business that has been much slower than anyone expected. The other four have been announced and earn nothing yet: an EPR certificate exchange built for the Central Pollution Control Board and awaiting the go-ahead to operate, a B2B travel portal, a TReDS invoice discounting platform awaiting RBI clearance, and coal and mineral exchanges still at the feasibility stage. No revenue guidance has been given on any of the four.

That distinction matters more than usual here, because effectively all of today’s profit comes from the first group, and management has been candid that the core has been flat for some time. There is also a threat worth understanding properly, which is that a rival exchange for coal has been announced by another operator, and coal is a significant part of MSTC’s revenue. I go through the evidence for all of it in the video, with the filings on screen.

No price target, no recommendation, and nothing here is investment advice. As always, the point is to give you the primary sources and the reasoning so you can disagree with me properly.

See you at the next inflection point,

Karnik

Disclosure: I hold a position in MSTC Limited (NSE: MSTCLTD), established prior to publication. I have not transacted in the security in the 30 days preceding publication.

SEBI Disclosure: This post is published by Inflection Point Research, a SEBI-registered Research Analyst (Registration No. INH000023807). This content is for information and education only and should not be construed as investment advice, an offer to buy or sell securities, or a recommendation of any specific investment action. Readers should conduct their own due diligence or consult a SEBI registered adviser before making investment decisions. Past performance is not indicative of future results. Securities investments are subject to market risk.

Hello friends.

Eternal’s B2B arm, Hyperpure, lost ₹1,261 Cr of revenue over the past year. It reported ₹2,295 Cr in the quarter to June 2025 and ₹1,034 Cr in the quarter to June 2026. Fifty-five per cent of the business was gone in four quarters.

That is the kind of collapse that ends careers. Except there were no layoffs. No profit warning. No downgrade. Nobody wrote about it at all.

Because it was not a collapse.

Two pages away in the same filing, Eternal’s quick commerce segment, which is Blinkit, reports revenue of ₹15,664 Cr against ₹2,400 Cr a year earlier. And Note 4 explains that these are one event: Blinkit’s move to an inventory-led model means its revenue now includes direct sales to customers rather than only marketplace commission, and hence there is a reduction in sales of Hyperpure supplies to the B2B business.

Blinkit used to buy its stock through Hyperpure. Now it buys direct from brands and sells it as its own. Nothing was created and nothing was destroyed. A single accounting decision moved roughly a thousand crore of quarterly revenue from one line to another, and in doing so made one segment look like a rocket and the other look like a company in trouble.

Nobody has done anything wrong here. The change is disclosed, audited, and explained in plain English in the filing. The problem is what happens downstream: a hundred WhatsApp groups and forty thumbnails take the Blinkit number and compare it to Swiggy, whose quick commerce segment reported ₹1,232 Cr in the same quarter, and conclude that one company is twelve times the size of the other in the same business.

It is not. Swiggy has not made the switch yet. Its own annual report says the IOCC structure “would enable the Company to evaluate and potentially adopt business models that are presently unavailable to it, including own-inventory arrangements within its quick commerce business”. Its board approved capping foreign ownership at 49.50% on 23 July 2026, subject to a members’ vote at the AGM on 18 August 2026. When that vote passes and the model flips, Swiggy’s revenue will jump by a multiple that has nothing to do with the business, and a fresh set of comparisons will break in the opposite direction.

That is the subject of this piece. Not “are platforms good businesses”, which people have argued about since 2021 and will keep arguing about long after both of us are gone. Something narrower and more useful: when a platform tells you it made money this quarter, what exactly is it telling you, and how would you check?

The elevator pitch, before we start.

India’s listed platform cohort is genuinely, structurally better than it was three years ago. Blinkit’s profitability is real. Paytm’s recovery is real. CarTrade’s margin structure is real. But across this cohort there are four specific, entirely legal mechanisms by which a reported result can flatter the underlying business, and in Q1FY27 at least one of them is operating in every single company in this piece. Learn the four and the cohort sorts itself into a much shorter list than the headlines suggest.

Nine companies: Eternal, Nykaa, Swiggy, Meesho, Urban Company, Paytm, BlackBuck, CarTrade and PhysicsWallah. Every figure below has been checked against the exchange filing, the shareholders’ letter or the annual report, and where a number in wide circulation could not be found in a primary document, I have said so rather than repeat it.

Part 1: What a Platform Actually Sells

The word “platform” now covers a grocery retailer, a payments processor, a coaching institute with an app, a truck telematics business and a classifieds site. Calling all of those the same thing is like calling a refinery, a petrol pump and a car hire firm all “energy companies”. Technically true, analytically worthless.

What distinguishes these businesses is not technology. Every one of them has an app; the app is table stakes. It is how close the business stands to the customer’s money, and how much of the transaction it has to carry on its own books. That determines the margin structure, the working capital, the regulatory exposure, and above all what the revenue line means.

There are four positions, and they are worth naming because the names do most of the work.

Attention. You sell eyeballs and leads. You own nothing and you touch no rupee of the underlying transaction. CarTrade lives here, through CarWale, BikeWale and OLX India. Cost of goods sold is near zero, so incremental revenue drops through hard. Growth is capped by the advertising budget of the industry you serve. No gig labour, therefore no gig labour regulation.

Rails. You take a fee on economic activity you do not perform. Paytm and BlackBuck. Contribution margins are extraordinary because the marginal transaction costs almost nothing to process: BlackBuck runs at 93% of net revenues, a number you would associate with software rather than trucking. The catch is that the fee is either regulated toward zero or competed toward zero.

Fulfilment. You do the work. You do not own the goods, but you own the network that moves them or the roster of people who perform the service. Swiggy, Meesho, Urban Company. The take rate looks like a margin and is not one, because fulfilment cost sits inside it. This is where labour codes bite and where switching costs for the customer are approximately zero.

Inventory. You buy the goods, own them, price them and sell them. A retailer with an app. Nykaa has always been here. Blinkit arrived on 1 September 2025. You capture the full product margin instead of a slice of it, which is why companies move. You also acquire inventory risk, working capital, and direct liability as merchant of record.

And the story of this year is not where each company sits. It is that they are moving between positions, and nobody restates history when they move.

Blinkit went from fulfilment to inventory. Swiggy has built every piece of legal plumbing to follow and is one shareholder vote away. Nykaa, an inventory business, is adding fulfilment through rapid stores. Meesho built its own logistics arm rather than move. Urban Company is trying to turn a marketplace for one-off services into something closer to a subscription.

Every one of those moves changes what the revenue line measures. Not one of them changes the prior-year comparative.

Part 2: The Four Flatterers

Here is the toolkit. Four mechanisms, all legal, all disclosed somewhere, all routinely missed. Each diagnostic takes under a minute.

One: The model change

Revenue recognition shifts from net to gross, and the growth rate becomes an artefact.

Diagnostic: look for a segment whose revenue moves by a multiple while a related segment moves the opposite way. That is the cold open of this piece. One segment exploding while an adjacent one collapses is almost always a reclassification rather than a business event.

Two: The adjustment

Every company here reports some adjusted profit measure. Almost none adjust for the same things, and several do not define the measure at all.

CarTrade is the cleanest specimen, and the reconciliation is exact. Reported EBITDA for Q1FY27 was ₹62.98 Cr at a 31% margin. Adjusted EBITDA was ₹97.50 Cr at 42%. The ₹34.53 Cr gap is precisely two disclosed lines: other income of ₹28.58 Cr plus share-based expenses of ₹5.95 Cr. Neither the results filing nor the investor presentation defines “Adjusted EBITDA”, so that reconciliation is arithmetic rather than a company statement, but it ties to the rupee, and FY26 behaves identically.

There is a second, subtler thing happening in the same table. The reported EBITDA margin of 31% is calculated on revenue from operations of ₹201.16 Cr. The adjusted EBITDA margin of 42% is calculated on total income of ₹229.74 Cr. Other income is added to the numerator and to the denominator. The adjusted margin is flattered twice by the same ₹28.58 Cr.

Diagnostic: find the reported figure, take the difference, and check the denominators match.

Three: The other income

The 2025 IPO class raised very large sums, parked what it had not yet spent, and at current Indian short rates that treasury throws off real cash. It lands below the operating line and flows straight to PAT.

Paytm reported PAT of ₹220 Cr for Q1FY27 with other income of ₹182 Cr. CarTrade sits on cash of ₹1,321 Cr with zero debt. PhysicsWallah held ₹2,273.98 Cr of unutilised net IPO proceeds in fixed deposits at 31 March 2026.

None of this is illegitimate. Interest on your own cash is real income. But it is not operating leverage, it does not scale with the business, and it disappears the moment the money is deployed, which is exactly what management has promised to do with it.

Diagnostic: PAT minus other income, year on year. In Paytm’s case this cuts the other way and I want to be scrupulous about it: other income actually fell from ₹241 Cr to ₹182 Cr, so the PAT improvement was achieved despite a smaller treasury contribution, not because of a larger one. That is a better picture than a naive reading gives.

Four: The line you stop at

The most common of the four and the least discussed. Between EBITDA and what a shareholder actually owns there are four separate stopping points, and companies quote whichever flatters.

BlackBuck reported PAT up 25% for Q1FY27. Profit before tax fell 8%, from ₹45.74 Cr to ₹42.12 Cr. The PAT growth is due to a tax credit.

PhysicsWallah reported profit before tax of ₹9.92 Cr for FY26, its first positive PBT. It reported a loss of ₹24.17 Cr, because the tax charge was ₹34.09 Cr on a ₹9.92 Cr pre-tax profit.

CarTrade’s Q1FY27 PAT was ₹56.75 Cr in total and ₹51.24 Cr attributable to equity holders of the parent, the ₹5.51 Cr difference being non-controlling interest in businesses shareholders do not wholly own.

Nykaa’s FY26 PAT was ₹203.94 Cr in total and ₹199.44 Cr attributable to the parent. The company’s own annual report uses ₹204 crores.

Diagnostic: always find the owners’ line. It is the only one you have a claim on.

Right. The companies.

Before we move on to the companies, if you would rather watch this as a video, you can do so here:

Part 3: Inventory, Where You Own the Goods

Eternal (NSE: ETERNAL)

Strip out the accounting and Eternal had a good quarter, and that needs saying before the caveats.

Quick commerce turned in a statutory segment result of ₹365 Cr for Q1FY27, against ₹265 Cr in Q4FY26 and a loss of ₹42 Cr in Q1FY26. For the full FY26 the segment result was ₹430 Cr. So Blinkit generated 85% of its entire prior-year segment profit in a single quarter. That is a genuine inflection and one of the more impressive operating achievements in recent Indian corporate history.

Consolidated PAT for the quarter was ₹92 Cr, against ₹174 Cr in Q4FY26 and ₹25 Cr in Q1FY26. So PAT is up 268% year on year and down 47% sequentially, and both statements are true. FY26 PAT was ₹366 Cr.

Now, the translations, of which there are three.

Consolidated revenue of ₹20,211 Cr is not comparable to anything. Neither is the Blinkit segment revenue. What is comparable is the segment result, because the profit line absorbs the model change on both sides.

The India food ordering and delivery segment is the quiet story. Revenue of ₹3,100 Cr against ₹2,261 Cr a year earlier, and ₹2,737 Cr the preceding quarter. A twelve-year-old business growing 37% year on year and 13% sequentially, while the entire market narrative is elsewhere, deserves more attention than it gets. I would note that the widely quoted net order value growth and take rate figures for this segment come from the quarterly shareholders’ letter, which sits outside the statutory filing; I have not reproduced them here because I have not seen them in a primary document.

Cost discipline is not the story either way. Advertisement and sales promotion was ₹945 Cr for the quarter and delivery and related charges ₹3,150 Cr, both consolidated. That is ₹4,095 Cr of variable spend against ₹92 Cr of PAT.

The governance item. Deepinder Goyal resigned as Director, Managing Director and CEO with effect from the close of business on 1 February 2026, after eighteen years, and was appointed Vice Chairman and Non-Executive Director from 13 March 2026. The annual report records that he had voluntarily waived salary from 1 April 2021 until resignation, and draws no remuneration or sitting fees in the new role. Albinder Dhindsa, who built Blinkit, is now Group CEO.

I do not read this as a red flag. The successor is internal and ran the division that now produces most of the profit. But a promoter stepping away from executive control of a company this size within four years of listing changes who is accountable for the numbers, and it belongs in any honest write-up.

Nykaa (NSE: NYKAA)

Nykaa’s Q1FY27 result landed on 4 August 2026, and it is the strongest quarter the company has produced in some time.

Consolidated revenue from operations of ₹2,782.00 Cr against ₹2,154.94 Cr a year earlier, up 29.1%. Profit before tax of ₹129.16 Cr. PAT of ₹79.76 Cr against ₹24.47 Cr, more than tripled, with ₹80.01 Cr attributable to equity shareholders of the parent. The same board meeting approved the acquisition of 51% of Aminu Wellness Private Limited for up to ₹32 Cr.

For FY26, PAT was ₹203.94 Cr in total and ₹199.44 Cr attributable to the parent, and consolidated gross profit grew 30% to ₹4,516 Cr, outpacing revenue growth. Nykaa Fashion, the segment everyone had written off, delivered GMV of ₹4,954 Cr and net sales value of ₹1,447 Cr, both up around 30%.

Two things to be careful about, one of them a correction to something I have said before.

The return metrics disagree with each other, sharply. The annual report gives consolidated return on net worth of 14.3% in FY26 against 5.6% in FY25, an improvement of 872 basis points. The same annual report gives a standalone return on equity of 5% in FY26 against 6% in FY25. Consolidated up strongly, standalone down slightly. Both are disclosed, in the same document, and they describe different entities. I have seen a three-year average ROE of 7.75% quoted for Nykaa in several places including my own earlier notes; it does not appear anywhere in the company’s filings and I cannot source it, so I am dropping it.

A causal claim I made and cannot support. I have previously written that Nykaa’s net revenue growth trails net sales value growth because House of Nykaa owned brands carry no marketing income component. The company discusses net sales value and marketing income separately and never links them. Marketing income is defined in the accounting policy only as visibility services provided to brands at the group’s retail outlets. The mechanism is plausible; it is not stated, and I should not have written it as though it were.

That leaves the interesting question about Nykaa unresolved but at least honestly framed: a business compounding gross profit at 30% and tripling quarterly PAT, whose consolidated and standalone return metrics point in opposite directions, and whose newest push, rapid delivery, sits exactly where everyone else in this piece loses money.

Part 4: Fulfilment, Where You Do the Work

Swiggy (NSE: SWIGGY)

Q1FY27 revenue from operations of ₹6,812 Cr against ₹4,961 Cr, up 37%. Loss for the period of ₹791 Cr, narrowed from ₹1,197 Cr, all attributable to owners of the Company. Sequentially the loss barely moved, from ₹800 Cr in Q4FY26. FY26 in full: revenue ₹23,053 Cr, loss ₹4,154 Cr.

The quick commerce segment reported revenue of ₹1,232 Cr. Advertising and sales promotion for the group was ₹1,160 Cr in the quarter, against ₹1,036 Cr a year earlier and ₹4,207 Cr for the full year.

Two things dominate everything else here.

The comparability cliff now has a date on it. The board approved capping foreign ownership at 49.50% on 23 July 2026, subject to members’ approval at the AGM on 18 August 2026. Aggregate foreign ownership as at the 1 July 2026 reference date was approximately 49.89% on a fully diluted basis. The annual report’s stated rationale is that IOCC status would let the company adopt models “presently unavailable to it”, naming own-inventory arrangements in quick commerce explicitly. After that vote, Swiggy’s revenue line becomes as uncomparable as Eternal’s is now.

On Instamart’s operating trajectory, the figures that circulate most widely, quarterly contribution margin as a percentage of gross order value and a stated breakeven month, come from the quarterly shareholders’ letter rather than the statutory filing. What the annual report does give, on an audited full-year basis, is an Instamart contribution margin of negative 2.8% of gross order value for FY26, improved from negative 4.0%. That is the number I am comfortable putting my name to.

The management churn. The Q1FY27 filing discloses that Amitesh Kumar Jha resigned as Chief Executive Officer of Instamart with effect from 28 July 2026, ceasing to be Senior Management Personnel. Instamart is the business on which the equity story rests, at the exact moment it is being restructured and readied for a new operating model. A monitorable, not a verdict.

Meesho (NSE: MEESHO)

FY26 consolidated revenue from operations of ₹12,626.35 Cr and a loss for the year of ₹1,357.74 Cr, all attributable to owners of the parent. Loss before tax was ₹1,209.27 Cr, so tax added roughly ₹148 Cr to the loss rather than reducing it.

The growth rate people quote for Meesho depends entirely on which prior-year line they use, and both circulating figures are real. FY25 restated revenue from operations was ₹9,389.90 Cr. FY25 restated total income was ₹9,900.90 Cr, the difference being other income of ₹511 Cr. Since FY26’s ₹12,626 Cr is revenue from operations, the correct comparison is to ₹9,390 Cr, which gives growth of 34.5%. Comparing against ₹9,901 Cr would understate it. This is Flatterer Four in a mild form and it is worth getting right.

On the loss. FY25’s headline loss was ₹3,941.71 Cr, and the prospectus is explicit that the increase over FY24 was driven by one-time exceptional expenses: tax incurred on the reorganisation of the company, additional costs from accelerated vesting of promoter stock options, and related perquisite taxes. Comparing FY26’s ₹1,358 Cr against that base and calling it a 66% improvement is comparing against a distorted denominator. The honest series is quarterly and forward from here.

On promoter holding. Promoter and promoter group holding stands at 16.41%, down from 18.50% on a fully diluted basis before the offer. The dilution has two sources: the IPO itself, and a further 44,605,655 shares allotted on stock option exercise during Q1FY27.

That is the lowest promoter holding in this cohort by a distance, and it is the arithmetic consequence of many funding rounds and a large offer for sale rather than of anything management did this year. It is not automatically a problem. It does change what “alignment” means, and it should be stated rather than glossed over.

Urban Company (NSE: URBANCO)

Urban Company is where Flatterer Two is most visible, and the company has generously provided three different versions of the same idea.

The core business is genuinely good. India Consumer Services excluding InstaHelp produced net transaction value of ₹1,056 Cr, up 29% year on year, with revenue of ₹356 Cr up 31%, and segment adjusted EBITDA of ₹73 Cr, equal to 6.9% of net transaction value against 5.2% a year earlier. Contribution profit was ₹226 Cr, or 21.4% of net transaction value. That is a business densifying and monetising better each quarter.

Now the three versions of “excluding InstaHelp”, all in documents dated 31 July 2026:

• ₹73 Cr, the India Consumer Services segment adjusted EBITDA

• ₹67 Cr, the consolidated adjusted EBITDA excluding InstaHelp, which is group-wide and includes Native at negative ₹9 Cr and International at ₹3 Cr

• ₹82.02 Cr, the statutory segment result for India consumer services excluding InstaHelp, on the CODM basis in Note 3

Three numbers, three bases, one quarter, and no reconciliation printed between them. Whichever gets quoted, the shareholder owns none of them, because the consolidated result for the quarter was a loss of ₹92.12 Cr, against a profit of ₹6.94 Cr in Q1FY26.

Then there is InstaHelp itself. Revenue of ₹11.22 Cr. Net transaction value of ₹53 Cr. Orders of 3.82 million, up 43% sequentially. Statutory segment result of negative ₹131.58 Cr, presented in the shareholders’ letter as an adjusted EBITDA loss of ₹132 Cr, or negative 250% of net transaction value.

The loss is two and a half times the transaction value that generated it. The company’s own figure for loss per order is ₹346, improved from ₹447 the preceding quarter. Dividing the statutory segment loss by the stated order count gives ₹344, and the letter’s rounded figures give ₹346, so the three tie. Average order value declined from ₹150 to ₹138 over the same period, which independently confirms the order count and net transaction value are on the same basis.

I want to correct myself here. I had previously used ₹495 for the Q4FY26 loss per order. The company says ₹447, in the same sentence where it gives ₹346, and the statutory numbers back-solve to ₹445. The improvement is therefore 23%, not the 30% I had. That is a less dramatic trajectory and the honest one.

Management targets consolidated adjusted EBITDA breakeven by Q3FY28 and roughly ₹1,000 Cr of adjusted EBITDA by FY31. The entire question is whether ₹346 keeps falling at 23% a quarter. There is no other question about this company worth asking right now.

Part 5: Rails, Where You Take a Fee

Paytm (NSE: PAYTM)

Q1FY27 PAT of ₹220 Cr, all attributable to owners of the parent, against ₹123 Cr in Q1FY26 and ₹183 Cr in Q4FY26. FY26 PAT was ₹552 Cr. For a company that lost a large part of its business to a regulatory intervention two years ago, this is a real recovery.

Three things sit underneath it.

Other income of ₹182 Cr against PAT of ₹220 Cr, which is 83% of profit arriving from below the operating line. But the direction matters and it favours Paytm: other income a year earlier was ₹241 Cr, so it fell 24.5% while PAT rose 79%. The operating business genuinely improved.

Payment processing charges of ₹794 Cr, up 36.7% from ₹581 Cr, growing faster than revenue. When you take a fee on someone else’s activity, the fee you keep is a spread, and the cost side of that spread is not under your control. This is the single line to watch each quarter.

The IPO proceeds. As at 20 July 2026, ₹1,686 Cr of the ₹2,000 Cr earmarked under Object 2 of the original offer remains unutilised, and the company proposes to extend the utilisation timeline to 31 March 2029, subject to a special resolution at the coming AGM. I emphasise “proposes” because it has not happened yet and I have previously written it as though it had. Read alongside the other income point: a meaningful part of current profit comes from money raised in 2021 and not yet spent, and the company is asking for another three years to spend it.

On EBITDA, a figure of ₹203 Cr circulates for the quarter. The word EBITDA does not appear anywhere in the eighteen-page filing. It is reproducible from disclosed lines, but that construction is mine and not the company’s, and Paytm’s own definition is not in front of me. I am leaving it out.

BlackBuck (NSE: BLACKBUCK, formerly Zinka Logistics Solutions)

The least fashionable company here and, on several measures, the most interesting. It sells software and services to India’s truck operators: tolling, telematics, load matching, vehicle finance. It moves no freight, which is exactly the point.

Q1FY27 revenue from operations of ₹204.17 Cr against ₹143.61 Cr, up 42%. Contribution margin of ₹152.74 Cr, equal to 93% of net revenues, unchanged from a year earlier. Core payments and telematics revenue of ₹145.18 Cr, up 21%. Growth businesses of ₹58.99 Cr, up 153% year on year and 44% sequentially.

Now the part the coverage skipped.

Revenue grew 42%. Adjusted EBITDA grew 16%, to ₹54.62 Cr. Plain EBITDA grew 23%, to ₹49.71 Cr, the ₹4.90 Cr difference being employee share-based payments. And profit before tax fell 8%, from ₹45.74 Cr to ₹42.12 Cr, while PAT rose 25% to ₹42.17 Cr on the back of a tax credit.

Revenue growing at nearly three times the rate of adjusted EBITDA, and PBT falling while PAT rises, is not operating leverage. It is the growth businesses being funded out of the core at lower margins, with tax doing the rest. That is a perfectly defensible strategy and it is the exact inverse of the story “revenue up 42%, profit up 25%” tells you.

One governance note. The board on 29 July 2026 classified BlackBuck Finserve Private Limited as a Material Subsidiary under Regulation 16(1)(c) read with Regulation 24, because its net worth exceeded 10% of consolidated net worth. BBFS is a wholly owned NBFC and the group’s lending segment. A lending book inside a platform company is a different risk animal, and it now carries enhanced disclosure obligations. Worth reading each quarter.

Part 6: Attention, Where You Own Nothing

CarTrade Tech (NSE: CARTRADE)

CarTrade is what the attention business looks like when it works, and it is the control case for this whole piece.

Q1FY27 revenue from operations of ₹201.16 Cr, up 16% year on year. EBITDA of ₹62.98 Cr, up 45%. PAT of ₹56.75 Cr in total, of which ₹51.24 Cr is attributable to equity holders of the parent. Cash of ₹1,321 Cr, zero debt.

And the line that makes the whole argument: marketing spend fell 9% to ₹7.92 Cr while revenue rose 16%. Compare that to Swiggy’s ₹1,160 Cr of advertising in the same quarter, or Eternal’s ₹945 Cr. A traffic habit built over a decade means you do not have to buy your customers again each quarter.

Three translations.

One, the adjustment. Covered in Part 2, but to restate: the ₹34.53 Cr gap between reported and adjusted EBITDA is other income of ₹28.58 Cr plus share based expenses of ₹5.95 Cr, and the two margin percentages use different denominators. Everything is disclosed. Only one of the two numbers will make it into a thread.

Two, the sequential picture. Revenue fell 0.97% sequentially, from ₹203.14 Cr in Q4FY26. And on the owners’ line, PAT of ₹51.24 Cr compares to ₹64.61 Cr the preceding quarter. “Record quarter” on a year-on-year basis and a sequential decline in owner earnings are both true.

Three, the cash. Return on equity is reported at 10%, footnoted as at year ended 31 March 2026, so it is an FY26 figure rather than a Q1FY27 one. A ten per cent return is what you get when a highly profitable operating business sits on top of ₹1,321 Cr earning deposit rates. The operating business is far better than 10%. The entity a shareholder owns is not. Whether that cash is an option or an anchor depends entirely on what management does with it.

Part 7: The One That Is Not a Platform

PhysicsWallah (NSE: PWL)

Every one of these pieces ends up with one company where the story people believe and the story in the filings have come apart. In the optical fibre piece, it was Finolex Cables. Here it is PhysicsWallah, and the gap is larger than I realised when I started.

PhysicsWallah is understood as an edtech platform: an app, digital distribution, software economics. FY26 revenue from operations was ₹3,899.54 Cr, up 35.1% from ₹2,886.64 Cr. Loss for the year of ₹24.17 Cr, sharply narrowed from ₹243.26 Cr, of which ₹22.49 Cr is attributable to owners of the parent.

Selected lines from the audited consolidated statement of profit and loss, FY26 results filing p.13. Operating expenses are not shown, so the rows do not sum to the profit before tax figure.

The word EBITDA appears nowhere in the eighteen-page filing, so there is no company measure to quote. Building one from the audited lines, profit before tax plus finance costs plus depreciation gives ₹549.20 Cr for FY26, and the same construction reproduces ₹193.19 Cr for FY25. Both tie to the rupee, which is what makes the basis trustworthy rather than convenient.

I should say plainly that I previously used a figure of ₹300 Cr for FY26 EBITDA. It is not reproducible from the audited filing on any basis I can construct, and I was wrong to use it.

Now the point. Depreciation of ₹437.39 Cr plus finance costs of ₹101.89 Cr equals ₹539.28 Cr, against EBITDA of ₹549.20 Cr. The business earns almost exactly what it spends on depreciation and interest. Then a tax charge of ₹34.09 Cr on a ₹9.92 Cr pre-tax profit tips it into a loss.

Why is depreciation so large for a company people describe as digital? Because it is not one. Right-of-use assets stood at ₹927.82 Cr at 31 March 2026, against property, plant and equipment of ₹445.01 Cr. Lease liabilities were ₹1,041.61 Cr. There is no rent expense line anywhere in the profit and loss statement. Under Ind AS 116, lease cost is not an operating expense; it becomes depreciation on right-of-use assets and interest on lease liabilities, both of which sit below EBITDA.

Which means that for this company, on any EBITDA definition built from this P&L, EBITDA is a measure that excludes rent. For a business whose principal fixed cost is rent on physical classrooms, that is close to the least informative number available. The relevant metrics are EBIT and free cash flow after lease payments.

And then there is Note 3, which is where this stops being a technical point about accounting standards.

Of the IPO net proceeds, ₹548.31 Cr was earmarked for “expenditure towards lease payments of existing identified offline and hybrid centers”, of which ₹491.92 Cr was still unutilised at 31 March 2026. Further amounts of ₹15.52 Cr and ₹28.00 Cr were earmarked for lease payments at Xylem and Utkarsh. Total unutilised net proceeds at year end were ₹2,273.98 Cr, held in fixed deposits and the public offer and monitoring accounts.

The company is explicitly funding the rent on its existing centres out of money raised at IPO. That is disclosed, it was in the offer objects, and investors who read the prospectus knew. It is also a materially different proposition from a business whose operations cover their own occupancy cost, and it does not appear in any EBITDA figure, adjusted or otherwise, because rent is not in the P&L at all.

One more asymmetry worth knowing before writing any sentence about this company: consolidated profit before tax was positive at ₹9.92 Cr, while standalone profit before tax was a loss of ₹15.77 Cr. The parent entity did not make money at the pre-tax line.

The exceptional items of ₹52.57 Cr break down as goodwill impairment of ₹29.01 Cr, IPO expenses of ₹8.29 Cr, and ₹15.27 Cr of remeasurement on employee benefit obligations following the labour code changes, which the group treated as non-recurring.

None of this makes PhysicsWallah a bad business. Offline coaching has real pricing power and the company’s insight, that Indian test preparation was priced far too high, has been commercially vindicated. It makes it a leased physical infrastructure business with a strong digital brand, and those are valued differently. Several claims that circulate about it, the share of offline centres that are profitable, the promoter holding, and a consumer protection penalty relating to checkout design, are not in the results filing and would need the annual report, which was not published at the time of writing.

Part 8: The Labour Codes, Honestly

Four labour codes came into force in India on 21 November 2025. Of the four, only the Code on Social Security recognises gig and platform workers. Central rules were notified on 8 May 2026.

Here is what the filings actually show, which is less dramatic than the commentary and more interesting than nothing.

Three companies in this piece have booked a charge. Swiggy took a ₹10 Cr one-time Labour Code charge in FY26. PhysicsWallah took ₹15.27 Cr, presented as an exceptional item covering increased gratuity liability from past service costs plus leave liability and overtime allowance. Meesho carries a note. Eternal’s Note 6 states that the provisions currently in force do not have a material impact, and separately discloses that the group is contesting the validity of the Karnataka Platform Based Gig Workers Act, 2025 before the appropriate forum, with rules notified on 19 November 2025 and no material impact from operationalised state legislation as at 30 June 2026.

So the position as at the June 2026 quarter is: charges in the low tens of crores, one company litigating a state statute, and the companies themselves saying the effect is immaterial so far.

I want to flag something I got wrong in my own thinking. I had wondered whether the widely quoted aggregator contribution of one to two per cent of turnover would land on reported turnover, which would mean that a company moving from a fulfilment model to an inventory model had multiplied its own levy base for identical economic activity. It is a neat idea. I could not find the definition of turnover for this purpose in any primary document, and I am not going to publish a mechanism I cannot source. It remains an open question and a good one, and if any reader can point me to the operative definition I would be grateful.

Alongside this sits the second regulatory axis, which is verifiable and does bind. The IOCC route that Eternal has taken and Swiggy is taking exists because foreign-owned e-commerce entities cannot hold inventory. Swiggy’s own annual report says IOCC status would enable models “presently unavailable to it”. Both companies are restructuring their shareholding to escape a foreign investment restriction, which permanently constrains how much foreign capital they can take in again without unwinding the operating model.

Owning your inventory buys you the full product margin. It costs you access to foreign capital. That is the trade, and it is rarely presented as one.

Part 9: Nine Companies, Four Flatterers

And there is a pattern in the position column that I did not expect when I started. The more of the transaction a company takes onto its own books, the more machinery sits between the customer’s rupee and the shareholder’s, and the more places a number can be presented favourably. An attention business has almost none of it. An inventory business has an entire revenue recognition model in the way.

Part 10: What I Am Watching

Eternal. The quick commerce segment result, quarter by quarter, ignoring revenue entirely until the model change anniversaries in Q2FY27. Whether food delivery holds 13% sequential growth. Whether the sequential PAT decline reverses. Albinder Dhindsa’s first full year against guidance.

Nykaa. Whether Q1FY27’s 29% revenue growth and tripled PAT sustain into Q2. The gap between consolidated return on net worth of 14.3% and standalone return on equity of 5%, and which one converges. Rapid store unit economics as the format expands.

Swiggy. The 18 August 2026 AGM vote, and whether the company bridges or restates prior periods when the inventory model goes live. Instamart contribution margin against the audited FY26 base of negative 2.8% of gross order value. Who runs Instamart, and for how long.

Meesho. Quarterly loss trend measured against Q4FY26, not against the distorted FY25 base. Revenue from operations growth against ₹9,390 Cr, not ₹9,901 Cr. Whether promoter holding stays at 16.41% as remaining lock-ins expire.

Urban Company. InstaHelp loss per order. Just that. ₹447, then ₹346. If the 23% quarterly improvement holds, the Q3FY28 breakeven target is credible. If it stalls above ₹200, the core business is being consumed to fund a category that does not work.

Paytm. PAT less other income, year on year. Payment processing charges as a share of revenue, currently growing faster than revenue. Whether the AGM approves the extension to 31 March 2029 and what the ₹1,686 Cr eventually funds.

BlackBuck. Whether adjusted EBITDA growth converges toward revenue growth as the growth businesses mature. Profit before tax rather than PAT, since the tax line is doing work. Disclosure quality at BlackBuck Finserve now it is a Material Subsidiary.

CarTrade. Sequential revenue and sequential owners’ PAT, both of which fell this quarter. What happens to ₹1,321 Cr of cash, since deploying it is the only route from a 10% return on equity to something that matches the multiple.

PhysicsWallah. EBIT, never EBITDA. The pace at which the ₹491.92 Cr of IPO proceeds earmarked for lease payments is drawn down, because that is the clock on the current cost structure. Whether standalone profit before tax turns positive. The FY26 annual report when it lands, for the lease note and the centre-level disclosure the results filing does not carry.

For the sector. Final rules and the operative definition of turnover under the Code on Social Security. The outcome of Eternal’s challenge to the Karnataka gig workers legislation. And the pace of new listings, since each one adds a fresh set of self-defined metrics to a cohort that already has plenty.

Part 11: A Closing Thought

There is a particular intellectual laziness this sector invites, and I have been guilty of it in this very piece.

It goes like this. A platform is complicated. Traditional metrics do not fit. Therefore the company should tell me which metrics do fit, and I should evaluate it on those. Adjusted EBITDA excluding the new vertical. Contribution margin as a share of gross order value. Net transaction value. Each is a genuinely useful construct invented for a genuinely good reason. Each is also a number the company chose, defined, and calculated, sometimes without printing the definition.

The discipline is not to reject them. Several are better than the statutory ones. The discipline is to insist they reconcile, and to notice when they cannot.

Writing this piece, I asserted a capture rate I could not source, an EBITDA figure that reconciled to nothing, a loss per order forty-nine rupees too high, a stock option arrangement that appears in no filing, and a causal mechanism at Nykaa the company never claimed. Every one of those came from reading commentary rather than filings. Every one is corrected above, and I would rather show you the corrections than pretend the first draft was clean.

The arithmetic that keeps you honest is not hard. Revenue against its own segment note. Reported minus adjusted, with the denominators checked. PAT minus other income. And always, always the owners’ line.

India’s consumption story is real. Blinkit generating 85% of a full year’s segment profit in one quarter is a genuine achievement. CarTrade compounding while cutting marketing spend is quietly excellent. BlackBuck’s 93% contribution margin is a beautiful business hiding in an ugly industry. Urban Company’s core is densifying properly. None of that needs the revenue line to mean something it does not.

Work out where the company is standing before you read its ratios. Then find the owners’ line.

See you at the next inflection point.

Karnik

Inflection Point Research

Portfolio Disclosure

I hold a position in Eternal. I do not currently hold positions in the other companies discussed in this piece. All positions were established prior to the publication of this report. I may initiate, add to, reduce or exit any position at any time based on changes in fundamentals or portfolio rules. This disclosure is as of 18th August, 2026.

SEBI Disclosure

This report is published by Inflection Point Research, a SEBI-registered Research Analyst (Registration No. INH000023807). The analyst may have personal investments in the securities discussed, as disclosed above. This content is for informational and educational purposes only and should not be construed as investment advice, an offer to buy or sell securities, or a recommendation of any specific investment action. Readers should conduct their own due diligence and consult a SEBI-registered financial adviser before making investment decisions. Past performance is not indicative of future results. Equity investments are subject to market risks.

Sources. All financial figures in this piece are taken from the companies’ Q1FY27 exchange filings, quarterly shareholders’ letters and investor presentations where cited, FY26 annual reports, and in Meesho’s case the Red Herring Prospectus dated 27 November 2025 and the current exchange shareholding pattern. Where a figure in wide circulation could not be located in a primary document, that is stated in the text rather than omitted.

If this piece was useful

Forward it to one person who quoted a revenue growth number from this cohort in the last month without checking what was in it.

Full research plans are at inflectionpointresearch.in/pricing-plans/list. If you would like to see the depth of the written work first, two complete sample reports, on Gravita India and Kalyani Cast Tech, are available at https://www.inflectionpointresearch.in/free-sample-research-reports

Until next time,

Karnik

Hello friends,

Start with a rock.

Acid-grade fluorspar is calcium fluoride, at least 97% pure, and India has almost none of it. The world’s reserves sit in China, Mexico and South Africa, with Mongolia and Vietnam supplying much of the balance. You cannot synthesise it, you cannot substitute it, and if you want to make any fluorinated molecule at all, you have to buy it from somebody else and ship it a long way.

Roughly a fifth of every modern medicine has a fluorine atom in it. So does the refrigerant in your air conditioner, the coating on your pan, the etchant that patterns a semiconductor wafer, and the salt that carries the charge inside a lithium-ion battery.

All of it starts with a rock India has to import.

That single fact does more to explain the last three years of Indian specialty chemical earnings than any amount of China plus one commentary, and it is why this piece is organised by element rather than by sector.

Because here is the puzzle. Four Indian specialty chemical companies closed FY26 within weeks of each other. All four sell molecules. All four export. All four have spent three years being described, by the same brokers in the same notes, as beneficiaries of the same China+1 tailwind.

Navin Fluorine grew revenue 41% and more than doubled EBITDA, ending the year at a 32.6% margin against 22.7%. Vinati Organics grew revenue by almost exactly nothing and still lifted PAT 9.5%. Neogen Chemicals grew revenue 11%, held EBITDA flat, and watched PAT fall 17%. Sudarshan Chemical tripled consolidated revenue to ₹9,787 crore and spent most of the year in a loss.

Same sector. Same macro. Same China. Same rupee. Same quarter-end.

If the sector thesis were doing the work, those four numbers would rhyme. They do not rhyme. They are not even in the same key.

What follows is the value chain, the chemistries that matter in India, and then six companies at close range: Tanfac, Navin Fluorine, Anupam Rasayan, Neogen, Sudarshan and Vinati Organics. Two of them sit on the same atom and are bound together by a shareholding that most coverage describes incorrectly. Two of them are racing toward the same molecule from opposite ends of the periodic table without, as far as I can find, ever being mentioned in the same note.

The diagnostic I used to read all six is at the end, as a field guide you can take to the next company that crosses your screen.

It is going to be long. Pour something cold.

You can also watch a video I made on this topic here:

Part 1: Three Families, and Where Each One Hides Its Scarcity

“Specialty chemicals” is a category defined by exclusion. A commodity petrochemical does not qualify. Nor does a bulk fertiliser, nor a formulated consumer good. Beyond those exclusions, the label carries almost no information.

Look at what actually sits under it in India. Refrigerant gases sold by the tonne into an HVAC supply chain. Multi-step intermediates sold by the kilogram into an agrochemical dossier that took a decade to register. Pigments sold into a coatings formulation by colour index number. Electrolyte salts sold into a battery cell qualification that runs for eighteen months. Surfactants sold into a shampoo. These have almost nothing in common. Their customers differ, their qualification clocks differ by an order of magnitude, and their price behaviour when China moves is close to opposite.

The label survives because it is useful to the people selling the index, not to the people reading the company.

A better first cut is the one the chemistry itself gives you. Every molecule in this industry is a carbon skeleton with something attached to it. What is attached, and how hard it is to attach, is where the economics live.

Family one: the halogens. Fluorine, chlorine, bromine, iodine. Here the scarcity sits at the very bottom, in a mineral or a brine. Fluorspar, as above. Bromine comes from brine, principally the Dead Sea, Arkansas, and in India the salt fields of the Rann of Kutch. Iodine comes from Chilean caliche and Japanese gas brine, and from almost nowhere else. For this family, the first question to ask is where the atom comes from, and who else can get it. What the company makes comes second.

Family two: the heteroatom insertion chemistries. Nitrogen, sulphur, phosphorus. Here the scarcity sits in the reaction rather than the input. Nitration, sulphonation, hydrogenation, phosgenation, cyanation, fluorination under pressure. These reactions are energetic, toxic, exothermic, or all three. Access to the atom is trivial, since nitrogen and sulphur are everywhere. The barrier is being permitted, insured, audited, and trusted to run the reaction at scale without killing anyone, and then being qualified by a customer who has to defend that choice to a regulator. Handling capability is the moat, and it is a permission rather than a technology.

Family three: the carbon skeletons. Aromatics from the petrochemical stream, oleochemicals from palm and coconut, the whole surfactant and additive world. Abundant inputs, well-understood reactions, many capable producers. Here the moat comes from scale, channel and brand, with the chemistry doing almost none of the work. This is the commodity floor of the industry, and where most of the disappointment of the last three years has been concentrated.

Part 2: The Value Chain, and Where the Money Actually Stops

Every specialty chemical value chain in India has the same four rungs, and value distributes across them in a consistent and lopsided way.

Rung one, the element or the mineral. Fluorspar, bromine brine, sulphur, elemental phosphorus, benzene, palm kernel oil. Nobody listed in India of any consequence sits purely here, and that itself is the important fact. The mineral moves along a route that runs, almost literally, along the old Silk Road, and India sits at the wrong end of it.

Rung two, the conversion. Fluorspar plus sulphuric acid becomes anhydrous hydrofluoric acid. Brine becomes elemental bromine. This rung is capital-heavy, hazardous, licence-constrained, and locally monopolistic: HF is not economically shipped over long distances in large volumes, so whoever has a permitted HF plant near a customer cluster holds something close to a geographic franchise. Margins here are real without being spectacular, and they compress hard when the mineral price moves, because this rung buys at the spot price of a mineral it does not control and sells to customers who negotiate annually.

Rung three, the derivative or intermediate. HF becomes R32 refrigerant, or aluminium fluoride, or a fluorinated building block. Bromine becomes a flame retardant or a pharmaceutical intermediate. This is where chemistry starts to be worth something, and where the qualification clock begins.

Rung four, the registered molecule. The multi-step custom-synthesised intermediate that appears by name in a customer’s regulatory dossier. Once your plant and your process are named in a crop protection registration or a drug master file, switching away from you becomes a regulatory event rather than a procurement decision, with a cost measured in years and a risk measured in the customer’s own product approval. This is where the durable economics of the industry live.

The industry’s own arithmetic reflects this. Navin Fluorine’s consolidated EBITDA margin in FY26 was 32.6%. Tanfac, which supplies the acid at rung two, ran roughly 16% across the first nine months of FY26 against 23% for FY25. Same atom. Same chain. Two rungs apart. Sixteen percentage points of margin.

The temptation is to conclude that the top rung is always the right place to be. The six companies below explain why that conclusion fails often enough to be dangerous.

Part 3: Tanfac, the Rung Everyone Skips

Tanfac was incorporated in December 1972 as a joint sector company between the Tamil Nadu Industrial Development Corporation and the Aditya Birla group, built for one national purpose: to make India self-sufficient in fluorine chemicals. Production began at Cuddalore in March 1985.

Almost everything about the business follows from what it handles. Anhydrous hydrofluoric acid dissolves glass, penetrates intact skin, and cannot economically be shipped in bulk over long distances. So you build one site, spend four decades learning to operate it safely, and sit next to a port. Tanfac’s plant occupies 60 acres in the SIPCOT industrial complex at Cuddalore, adjacent to the port, and it has been there the entire time.

The product list widened steadily rather than dramatically. Aluminium fluoride for the aluminium smelters, which is what the plant was originally built around, and which made Tanfac a supplier into India’s aluminium industry through its growth decades. Sulphuric acid and oleum. Cryolite and assorted fluoride salts. Organofluorine chemistry from the late 1990s. A captive power plant in 2005. A poly aluminium chloride line from April 2015.

The ownership change, and the thing most coverage gets wrong. In February 2022 Anupam Rasayan agreed to acquire 24.96% of Tanfac from Birla Group Holdings for ₹148.14 crore, replacing Birla as TIDCO’s joint venture partner under an agreement originally signed in October 1980, and launching an open offer for a further 26%. The transaction completed on 11 March 2022.

What Anupam ended up with deserves stating precisely, because it is routinely described as a majority stake and is not one. Anupam holds roughly 25.8% of Tanfac, rising to about 26.07% after a preferential allotment approved in July 2026. TIDCO remains a promoter and continues to nominate the chairperson: a serving IAS officer was appointed to that role in July 2026. What Anupam obtained was management control, conferred by the terms of the 1980 joint venture agreement rather than by owning most of the shares.

So Tanfac is a Tamil Nadu state joint venture with a listed specialty chemical company as its operating partner. A minority shareholder is buying into a structure where the operating partner is also a customer, the state is a co-promoter, and neither owns a majority. That is unusual, and it deserves to be understood rather than glossed. Read the related-party note in the FY26 annual report before sizing anything.

The capacity story, which is where a 1972 acid plant meets modern demand. The HF expansion was commissioned in October 2024, taking capacity to 29,700 TPA.

More interesting is what came next. In June 2025 Tanfac commissioned Phase 1 of a 10,000 TPA solar-grade dilute hydrofluoric acid unit. Solar grade. That acid textures and cleans silicon wafers, which puts this plant directly into India’s solar manufacturing build-out, while the same fluoride chemistry runs toward battery materials. A company built to supply aluminium smelters is now selling into two industries that did not exist when its plant was designed.

Further capex is committed: a new HF plant of 17,160 TPA at roughly ₹150 crore, a new sulphuric acid plant of 82,500 TPA at roughly ₹40 crore, and a separately announced programme of about ₹495 crore directed at HFC-32 refrigerant, with capacity indicated as moving to roughly 1.8 times current levels. There is also a share subdivision approved in February 2026, a ₹500 crore fundraising authorisation, and the ₹173.48 crore preferential issue in July 2026 at ₹2,341 per share.

FY26, and the number that matters. Revenue from operations ₹711 crore, up 27%, a record. EBITDA ₹112 crore. PAT ₹70 crore. Net debt ₹47 crore, ROE 19%, ROCE 20%.

Set that against FY25: revenue ₹556.98 crore, EBITDA ₹128.88 crore at a 23% margin, PAT ₹88.15 crore. Revenue up 27%, EBITDA down, PAT down. Margin fell to roughly 15% in Q3FY26 and 16% across the nine months, which management attributed to a spurt in key input material costs.

That is a rung two business answering for itself in public. Tanfac buys an imported mineral it does not control and converts it at a rung where the customer negotiates annually. When fluorspar moves, Tanfac absorbs it. Every converter at this rung absorbs it. The rung sets the margin here, not the management team.

The mix is what the capex is trying to change, and the utilisation disclosure in the FY26 filings is the most informative thing the company published all year: sulphuric acid ran at 101% of capacity, hydrofluoric acid at approximately 95%, and the specialty fluoride segment at approximately 41%. The two commodity plants are full. The value-added plant is running at four tenths.

What would change my mind, either way: specialty fluoride utilisation moving decisively off 41%, the solar-grade DHF ramp, and gross margin behaviour in the first two quarters of FY27 once fluorspar costs normalise. If revenue grows again and EBITDA does not, the pass-through does not exist.

Part 4: Navin Fluorine, What Four Rungs of Qualification Look Like

Navin Fluorine has been making fluorine chemistry at Bhestan in Surat since 1967, which makes it one of the oldest fluorochemical operations in the country. It belongs to the Padmanabh Mafatlal group, a textile lineage that became a chemistry lineage. That tenure matters more here than in most industries, because customer qualifications in fluorine chemistry are earned over decades and inherited rather than bought.

The company is best understood as three businesses sitting on one atom.

High performance products is the refrigerant and inorganic fluoride business, the volume end, now backward integrated following commissioning of its own anhydrous HF plant. R32 capacity is expanding from 9,000 TPA to 24,000 TPA, targeted for commissioning in Q3FY27, at a stated peak revenue potential of ₹600 to ₹825 crore per annum.

Specialty chemicals is the agrochemical and pharmaceutical intermediate business. This is rung four, where the molecules sit inside customer registrations and the switching cost is regulatory rather than commercial.

CDMO is the contract development arm, and the newest of the three. It grew 61% year on year in Q4FY26 to ₹186 crore on a mix of early stage, late stage and commercial molecules across oncology, respiratory, cardiovascular, neurology and animal health, with commercial supplies now running from the cGMP4 block against a European master supply agreement. A CDMO relationship is the purest form of the rung four moat: the customer’s own regulatory filing names your block.

FY26. Consolidated revenue ₹3,313.90 crore, up 41%. Operating EBITDA ₹1,081.68 crore, up 103%, margin expanding to 32.6% from 22.7%. PAT ₹663.55 crore, up 130% from ₹289 crore. Growth was broad rather than concentrated: high performance products up 34%, specialty chemicals up 44%, CDMO up 59%.

FY27 then opened with Q1 revenue of ₹1,044 crore, up 44%, an operating EBITDA margin of 34.2%, PAT of ₹243 crore, up 108%, and a move to a net debt free balance sheet during the quarter. Six consecutive quarters of growth in a business that was being written off as a fluorochemical cyclical eighteen months ago.

Now the honest part, because a 130% PAT increase invites the wrong conclusion. Management’s own attribution on the Q4FY26 call was that roughly 70% of EBITDA growth came from higher volumes and new capacity, and roughly 30% from pricing actions. That 30% is a refrigerant price cycle, elevated by a global demand and supply mismatch and by phase-down schedules that are regulatory artefacts with dates attached. Real, durable for now, and not permanent.

Management also described the agrochemical market as going through a slow reset, with volumes recovering and pricing still lagging. That is a considerably more sober read of the largest end market than the headline numbers imply, and it is the kind of remark worth listening for on a call where everything else is going right.

So the correct frame for Navin has little to do with the margin now reading 33%. Roughly two-thirds of the improvement is structural volume and mix, one-third is a price cycle in a regulated gas, and the question for FY27 and FY28 is whether the CDMO and specialty ramp arrives fast enough to replace the pricing component when it fades.

What would change my mind: the split of EBITDA growth between volume and price in the FY27 quarters. If the pricing share rises rather than falls, the quality of the earnings is deteriorating even as the number goes up.

Part 5: Anupam Rasayan, and the Stake It Has Just Pledged

Anupam Rasayan is a Surat company doing multi-step custom synthesis of life science and performance intermediates, largely for multinational agrochemical, pharmaceutical and personal care customers, with the molecules embedded in customer registrations. Agrochemicals are about 55% of revenue.

The strategic logic of the last four years is legible and coherent: own the fluorine chemistry, then own the acid that feeds it. Hence Tanfac in 2022, and hence the lithium hexafluorophosphate ambition discussed below.

FY26. Consolidated revenue ₹2,383.6 crore, up 65%, the best growth number of the six. Consolidated EBITDA ₹543 crore, up 32%, with the margin falling to 23% from 28%. PAT ₹222.2 crore, up 39%. Order book ₹14,646 crore. Gross debt around ₹1,500 crore.

Two observations, pointing in opposite directions.

The first is genuinely good and deserves saying first. Operating cash flow improved to ₹334 crore, against negative ₹30 crore in FY25, when a ₹391 crore adverse working capital movement swallowed the entire year’s operating cash. Working capital was the standing criticism of this company and it materially improved.

The second is that revenue grew 65% while EBITDA grew 32%, and a meaningful part of the revenue growth was purchased rather than qualified: Jayhawk Fine Chemicals in the United States for about $150 million, and Bliss GVS Pharma. Growth that arrives by purchase is a different asset from growth that arrives by qualification, and should not be valued as though it were the same.

The order book, held against its own clock. ₹14,646 crore against FY26 revenue of ₹2,384 crore, a ratio of roughly six times, reads spectacularly. Management’s own stated conversion arithmetic is that a letter of intent takes 18 to 24 months to reach commercialisation and a further two to three years to reach full volume. So a six-times order book is a portfolio of instruments at different stages of enforceability, spread across four to five years, some binding and some not.

The recent announcements make the point. A ten-year letter of intent with a Korean multinational valued at approximately $106 million, or about ₹922 crore. A letter of intent with E-Lyte Innovations and Fuchs Lubricants Germany for supply of up to 1,500 TPA of LiPF6, with commercial deliveries indicated as potentially starting in FY27. A contract and letter of intent with Elementium Materials for an advanced electrolyte chemical. Each is a real commercial event and a genuine signal about capability. None is revenue until it converts, and the word “potentially” is doing load-bearing work in at least one of them.

The item that went almost entirely unreported. In a disclosure dated 28 February 2026, made under Regulation 31(1) read with Regulation 28(3) of the SEBI Takeover Regulations, Anupam reported a non-disposal undertaking created on 20 February 2026 over its entire shareholding in Tanfac: all 25,73,081 equity shares, against a dollar borrowing.

An encumbrance is not a pledge invocation, nothing has gone wrong, and non-disposal undertakings are an ordinary feature of secured lending. But the backward integration that anchors this company’s strategic story now sits behind a lender’s undertaking, and gross debt is around ₹1,500 crore against ₹543 crore of EBITDA. Those two facts belong in the same sentence, and I have not found them put there anywhere else.

What would change my mind: disclosed conversion history on prior letters of intent, the FY27 margin against the 26% to 28% band management has historically guided to, and the direction of gross debt.

Part 6: The Convergence Nobody Has Drawn

Before leaving fluorine, draw the line that sector coverage does not.

Anupam Rasayan has signed a letter of intent to supply up to 1,500 TPA of lithium hexafluorophosphate, and has explicitly linked that ambition to its fluorine chemistry position and its integration through Tanfac. It is approaching LiPF6 down the fluorine chain, from the acid.

Neogen Chemicals is building the same molecule from the other side, through a bromine and lithium business and a Japanese technology partner.

LiPF6 is lithium, phosphorus and fluorine in one molecule, and it is where two of the three chemical families in Part 1 arrive at the same destination. It is the most commercially significant convergence in Indian specialty chemicals right now, and the two listed companies pursuing it are almost never discussed in the same note, because one is filed under agrochemical intermediates and the other under battery materials.

Follow the atom and they are competitors.

Part 7: Neogen, Where the Moat Is a Foreign Statute

Neogen was built on bromine. Bromine and lithium based specialty chemicals sold into pharmaceuticals, agrochemicals, polymers, water treatment, construction chemicals and fragrances: steady, unglamorous, single-digit to low-double-digit growth, and for most of the company’s life the whole of it.

Then it decided to build India’s lithium-ion electrolyte and electrolyte salt supply chain from scratch, through Neogen Ionics, and the reported financials have been unreadable ever since.

FY26. Consolidated revenue ₹862 crore, up 11%. EBITDA flat at ₹137 crore. PAT down 17% to ₹29 crore. Q4FY26 revenue ₹247 crore, up 22%, EBITDA ₹44 crore, up 21%, margin 17.8%. Consolidated total debt at year end ₹1,330 crore with net debt ₹1,295 crore; standalone total debt ₹671 crore and net debt ₹641 crore.

Read those numbers as a construction site rather than as deterioration. Revenue is growing, gross profit is expanding, and the profit line is being consumed by finance costs and expansion overheads at a subsidiary that has barely sold anything. Neogen Ionics contributed ₹13 crore of revenue in Q4FY26. That is the whole battery business, so far.

There is also a genuine one-off distorting the base: a fire at the Dahej plant. Neogen received ₹83.48 crore in insurance claims across the first nine months of FY26 with a net claim receivable of ₹251.12 crore outstanding, and the replacement plant was tracking for commissioning in Q1FY27. A ₹251 crore receivable against an insurer is a real asset and a recovery-risk item, and it is large relative to this company’s earnings base.

The joint venture, and the phrase that carries the case. Neogen Morita New Materials is structured with Neogen Ionics holding the majority and Japan’s Morita investing $20 million for the balance, to produce and sell solid LiPF6 globally using Japanese technology with a multi-decade track record.

Management describes the resulting facility as India’s only non-FEOC compliant electrolyte salt plant, and that phrase is the entire investment case despite not being a chemistry claim at all. FEOC stands for foreign entity of concern. Under the United States’ clean energy tax credit regime, battery materials sourced from entities linked to certain foreign jurisdictions disqualify the cell from credits under provisions such as section 45X. So an American cell maker cannot simply buy the cheapest LiPF6 in the world, because the cheapest LiPF6 in the world is Chinese and buying it destroys the credit. That creates a protected pool of demand which Chinese capacity, however cheap, is legally barred from serving.

A moat created by a foreign statute is powerful, real, and exactly as durable as the statute. This is the mirror image of the Chinese VAT circular in Part 11: policy in one direction created the opportunity, and policy in the other direction can remove it.

The milestones are dated, which I appreciate, because dated claims are falsifiable. A leading Indian gigascale customer completed the production part approval process at Dahej for long-term commercial electrolyte supply. Provisional approval for lithium electrolyte salt has been received from a key international customer. The Pakhajan greenfield project targets commercial electrolyte production in H1FY27 and electrolyte salt in H2FY27. Management has guided to battery chemicals revenue of ₹400 to ₹500 crore in FY27, against a FY26 contribution that rounds to nothing. Either it lands or it does not, and we will know inside four quarters.

One governance item stated plainly rather than skipped: the board approved a preferential allotment of 10 lakh equity shares to a promoter group entity aggregating ₹161 crore at ₹1,610 per share, described as around a 17% premium to the SEBI floor price. Promoter money arriving at a premium during a capital-hungry build is a better signal than most alternatives. It remains dilution, and it remains a related-party issuance.

What would change my mind: the FY27 battery chemicals number against guidance, the pace of insurance recovery against the ₹251 crore receivable, and whether net debt peaks in FY27 as the Pakhajan spend completes.

Part 8: Sudarshan, and the Difference Between Growth and Arrival

Pigments are a stranger business than the word suggests.

A pigment is a particle, not a dye. It does not dissolve. It is dispersed into a paint, a plastic or an ink, and what the customer buys is a specific particle size, a specific crystal form, and absolute batch-to-batch consistency, because a car manufacturer respraying a wing panel needs it to match a door made three years earlier. Chemically these are azo and high-performance organic pigments built through diazotisation and coupling, plus a range of inorganics, and the moat is a hybrid of formulation know-how, colour matching, regulatory registration across dozens of jurisdictions, and sheer catalogue breadth. This is a portfolio business rather than a molecule business.

Sudarshan has been doing it from Pune since 1951 and has been India’s largest pigment producer for years.

Then, in March 2025, it bought Germany’s Heubach Group out of insolvency and became one of the largest pigment companies in the world overnight. Heubach was itself a composite, having earlier absorbed Clariant’s pigments business. The combined group is being unified as “ONE Sudarshan”, with 19 manufacturing sites across 11 countries, customers in roughly 120 countries, over 1,600 pigment products, and combined turnover approaching €1 billion.

Buying a global top-three position out of distress is a legitimate way to create value, and the strategic logic is not in question. What FY26 shows is how far that is from being finished.

FY26 is the first full year of consolidation, and the numbers are a masterclass in why headline growth is not growth. Consolidated revenue ₹9,787 crore, against ₹3,346 crore in FY25. Consolidated business EBITDA excluding other income rose to ₹590 crore from ₹381 crore. Net debt ₹755 crore at 31 March 2026.

Now the line directly underneath, in the same filing. Standalone revenue from operations ₹2,397 crore, down from ₹2,534 crore in FY25.

Consolidated revenue up roughly 193%. Standalone revenue down 5.4%.

The legacy Indian business, the one with the seventy-year chemistry heritage, shrank. Everything else is the acquisition. And the acquisition was, for most of the year, loss-making: a ₹42 crore net loss for the nine months ended December 2025, an EBITDA loss of ₹38 crore in the acquired pigment business in Q3FY26 alone on soft demand and destocking, and combined adjusted EBITDA in Q3FY26 of ₹40 crore, a margin of 1.9%, from ₹116 crore the prior quarter.

There is a turn in the tail. Q4FY26 saw the acquired operations generate €11 million of EBITDA, ahead of internal expectations, against a revised full-year outlook of around €16 million, with global accounts resuming buying from January and destocking easing. The stated ambition is €90 million to €100 million of EBITDA from the Heubach operations within three to four years, to be achieved primarily through asset utilisation and pricing rather than large new capex.

Sixteen to ninety. That gap is the entire equity story, and net debt is ₹755 crore while they close it.

So underwrite Sudarshan as what it is: a leveraged turnaround of a formerly insolvent European industrial asset, executed by an Indian mid-cap, in a demand environment that was hostile for most of FY26 and has only just begun to normalise. It may well work. It should carry integration-bet risk pricing rather than specialty-chemical-compounder pricing.

A note on the trailing multiple, which is currently absurd and will be quoted at you: it is distorted by the restructuring-year profit collapse and tells you nothing. Forward estimates on FY28 look entirely different. Both are true and neither is analysis. The analysis is the €16 million to €90 million bridge, and the only honest position is that it is unproven with one good quarter of evidence.

What would change my mind: the FY27 quarterly EBITDA path of the acquired business against the €90 to €100 million target, net debt direction from ₹755 crore, and whether standalone Indian revenue resumes growth or continues to shrink.

Part 9: Vinati Organics, One Molecule, Done Better Than Anyone

Vinati Organics was founded in 1989 by Vinod Saraf and named after his daughter, who now runs the company. It manufactures at Mahad and Lote Parashuram in Maharashtra, and it has been debt free for years.

It makes one molecule better than anyone else in the world. ATBS, 2-acrylamido-2-methylpropane sulphonic acid, a sulphonated acrylamide monomer used in oilfield chemicals, water treatment polymers and acrylic fibre. Vinati is among the world’s largest producers. It is also a global leader in IBB, isobutyl benzene, the key precursor to ibuprofen, which means a meaningful share of the world’s ibuprofen begins in Maharashtra.

FY26 was, on the top line, a non-event. Consolidated revenue ₹2,226.89 crore, essentially flat and marginally down. Consolidated PAT ₹443.74 crore, up 9.5%, achieved through roughly 360 basis points of operating margin expansion rather than volume. Standalone revenue ₹2,280.79 crore against ₹2,292.39 crore, with standalone PAT of ₹487.78 crore, up 17.5%. Trailing ROCE 20.6%, ROE 15.8%, promoter holding 74.29%, standalone current borrowings nil, treasury around ₹190 crore.

Notice something in those two figures. Consolidated PAT is roughly ₹44 crore lower than standalone PAT. The US subsidiary contributed revenue of about ₹11.05 crore and net profit of about ₹1.48 crore, so it is not the explanation. The gap is essentially Veeral Organics, which generated a negligible ₹10 crore of revenue in FY26 while its plant underwent process reengineering after teething trouble with a new manufacturing process, with revenue contribution now expected only from Q3FY27. A subsidiary in the commissioning window is depressing consolidated earnings by roughly 9% while contributing nothing, and the standalone business is doing better than the headline suggests.

The mix, and the strategy underneath it. Revenue as disclosed on the Q4FY26 call runs roughly 35% ATBS, 15% to 20% antioxidants, 10% to 12% AB, 10% to 12% IBB, with the remainder customised and other products. Antioxidants grew 15% in FY26 despite aggressive price undercutting by Chinese producers; the company’s anti-dumping application was rejected and has been refiled, with a decision six to nine months away. Management targets combined antioxidants and butyl phenol revenue of ₹800 to ₹900 crore over the next two years, guides to 15% to 20% volume growth in FY27, expects ATBS volume growth around 15% annually for three years, and has two to three new specialty chemicals targeted for H2FY27 commercialisation with meaningful revenue only from FY28. FY26 capex was about ₹270 crore; FY27 is earmarked at ₹200 to ₹250 crore, all funded from internal accruals.

The honest reading, which is the most useful thing in this section. The moat is real. Being the dominant global producer of a specialty monomer, backward integrated, debt free, with 20%-plus ROCE through a brutal three-year sector downcycle, is a demonstration of quality almost nobody else in this piece can match. Vinati needed no rescue, no rights issue and no acquisition to get through FY23 to FY26.

But a monopoly on one molecule grows at the rate of that molecule’s end market and no faster. ATBS demand is a function of enhanced oil recovery, water treatment and acrylic fibre. Vinati can take the entire market and still grow mid-teens. The company clearly knows this, which is why the entire strategic effort has moved to new products and the antioxidants and butyl phenol chain. That effort has real momentum and is, by management’s own timeline, an FY28 revenue event.

So the question for Vinati is not whether the business is good. It plainly is. It is whether a debt free compounder growing at zero for two years while it builds the next leg is being valued as a compounder or as an option on FY28. That is a valuation question, and it is yours to answer.

What would change my mind: the anti-dumping decision on antioxidants, Veeral actually contributing from Q3FY27 as guided, and whether FY27 delivers the guided 15% to 20% volume growth or another flat year.

Part 10: Six Companies, Six Different Bets

Treat these six as six distinct positions on where in a chemical value chain a durable claim can be established, rather than as a portfolio or as substitutes for one another.

Part 11: The Field Guide

This is the diagnostic I ran on all six, compressed into something portable. When the next specialty chemical company crosses your screen, most of these can be answered from one annual report and one concall transcript, and the answers will tell you more than any peer table.

1. Where does the atom come from, and who else can get it? For any halogen chemistry, find the input mineral and its origin, then ask what happens to gross margin when that mineral moves 30%. If the answer is “we pass it through”, check whether the contracts say so or whether management merely hopes so.

2. How long is the qualification clock? Time from customer trial to first commercial supply. Under six months means a purchase order. Two to four years ending in a dossier that names your site means something close to an annuity. Hold every order book in this industry against this number.

3. Could the customer run this reaction themselves? For the heteroatom chemistries this is the whole moat. Multinationals outsource nitration and fluorination because they have decided not to own the hazard and the local liability. That decision is sticky in a way a cost decision is not. It also means the moat is a permission, and one serious incident can revoke it.

4. What does the contract actually bind? An order book, a long-term supply agreement, a letter of intent and a memorandum of understanding are four different things, and Indian disclosure routinely aggregates all four into one headline figure called visibility.

5. Where is the capex in its cycle? The dangerous window is commissioned plant, depreciation and interest running, customer approval not yet landed. Watch capital work in progress converting to gross block, and watch asset turns rather than revenue growth. A company in this window reports deteriorating returns whether it is executing well or badly, and only the approval milestones distinguish the two.

6. Did the growth come from molecules or from mergers? Split reported growth into organic volume, price and mix, and acquisition. Then look at the standalone entity. If consolidated tripled while standalone fell, underwrite it as an integration story.

7. What happens when policy changes abroad? In January 2026 China’s Ministry of Finance and State Taxation Administration announced withdrawal of export VAT rebates on 249 product categories from 1 April 2026, spanning chemicals, solar, ceramics and glass, with battery products cut from 9% to 6% and scheduled to zero from January 2027. That circular changed the landed economics of a long list of molecules overnight. A tailwind created by a foreign tax circular can be removed by another foreign tax circular, and the same is true of the American statute underwriting Neogen’s electrolyte case.

Part 12: What I’m Watching

For the sector. Whether the Chinese VAT rebate withdrawal shows up as sustained realisation improvement in the FY27 quarters or fades once pre-deadline inventory clears. Raw material availability, flagged as a Q1FY27 constraint. US tariff policy on Indian chemical exports, which several of these companies are now routing around by acquiring US manufacturing rather than exporting into it.

For fluorine. Fluorspar landed prices in India, because they set the floor under everything in Parts 3 to 5. India’s critical minerals policy and whether acid-grade fluorspar gets meaningful attention in it. Whether the coming Indian refrigerant capacity wave creates a domestic R32 surplus that compresses the pricing component of Navin’s earnings.

For LiPF6. Both routes together. Whether Neogen hits H1FY27 electrolyte and H2FY27 salt at Pakhajan, and whether Anupam’s E-Lyte letter of intent converts to a contract with volumes attached. Any change to US FEOC rules or section 45X.

Part 13: A Closing Thought

There is a reason this industry is so persistently mispriced in India, and complexity has little to do with it. The word “chemicals” invites people to treat a set of unrelated businesses as a single trade, and the market obligingly moves them together.

They should not move together. One of them has a customer who can switch suppliers with a phone call. The other has a customer who would need to refile a dossier with a regulator in four jurisdictions. Those two businesses have no business sharing a multiple.

You will not find that distinction in a price chart, a sector screen, or a quarterly preview. It lives in the atom, the rung, and the qualification clock, and it takes an afternoon with an annual report to find.

Chemistry is the only industry I follow where the moat is quite literally an element of the periodic table, sitting in a mine several thousand kilometres away, in a country that has other ideas. That is a strange and rather wonderful thing to be able to underwrite.

Follow the atom, not the ticker.

See you at the next inflection point.

Karnik

Inflection Point Research

Disclosures

Analyst holdings. The author holds no positions in Tanfac Industries, Navin Fluorine International, Anupam Rasayan India, Neogen Chemicals, Sudarshan Chemical Industries or Vinati Organics, and has held no position in any of them at any time during the thirty days preceding publication of this report. The author may initiate, add to, reduce or exit a position in any of these securities at any time after publication, in accordance with applicable regulations.

Registration. This report is published by Inflection Point Research, a SEBI-registered Research Analyst. SEBI Registration No. INH000023807.

GET TWO FREE IPR SAMPLE REPORTS

The field guide in Part 11 is the same diagnostic IPR runs on every company before a single valuation number is written. Two free sample reports, on Gravita India and Kalyani Cast Tech, show that framework applied end to end on individual names, including the forensic checks and the exhibits.

https://www.inflectionpointresearch.in/free-sample-research-reports

No payment required. SEBI Reg. INH000023807.

If this piece was useful to you

Forward it to one person who holds a chemical stock and has never checked which rung of the value chain it sits on. That is how this newsletter finds the right readers, one thoughtful forward at a time.

Until next week,

Karnik

No long essay this week. I have put the whole thing on video instead, and I want to tell you why.

Rossell Techsys is an aerospace and defence supplier in Bengaluru. It makes the electrical nervous system of a military aircraft: wire harnesses, cockpit control panels, the looms that run through an Apache and a Chinook and a P-8. It was a division of a tea company until 2024. Boeing has twice named it a global Supplier of the Year, from a base of more than 12,000 suppliers.

Revenue grew 87% in FY26, to ₹485 Cr. The June quarter was the largest in its history.

Across FY24, FY25 and FY26, it reported ₹41 Cr of cumulative profit after tax and minus ₹124 Cr of cumulative cash from operations.

That is not weak conversion. That is negative, three years running, while the business doubled.

The reason I made this a video rather than an essay is that the explanation is a chain, and a chain is easier to watch than to read. It goes like this. In aerospace, when a process is qualified, the raw material is qualified with it. You cannot substitute the connector. You cannot re-source the wire because a lead time slipped. So the same qualification that builds the moat forces the company to hold more than a year of single-source, mil-spec inventory. That inventory is currently financed by ₹409 Cr of borrowings against ₹155 Cr of net worth, and the interest on it consumed 44% of operating profit in the June quarter.

The moat and the working capital are the same fact.

There is a second thread running through the video, and it is the one I found most interesting. Revenue has doubled twice. The confirmed order book has not moved. In January 2025 the company reported ₹900 Cr of confirmed orders against ₹228 Cr of trailing revenue, roughly 3.9 times cover. In June 2026 it reported ₹800 Cr against ₹552 Cr, roughly 1.45 times. Forward visibility has compressed by about 60% in seventeen months.

I put the honest counter-argument to that in the video too, because there is a real one, and I do not think the evidence closes the question yet.

The whole thing runs about twenty-two minutes. I keep a scoreboard on screen throughout: four questions, four verdicts, and the last one does not resolve until the end.

One note on method. Every load-bearing figure in this video appears on screen with its source and its as-of date, and where a number is a broker estimate rather than a company disclosure, I say so. I also flag one item I could not resolve at all. Unresolved things shown honestly are worth more than resolved things asserted confidently.

Watch it here.

Visual Research Report, free:
https://iprrosselltechsys.netlify.app/

Sample reports (Gravita India, Kalyani Cast Tech):
https://www.inflectionpointresearch.in/free-sample-research-reports

---

Disclosure: I hold Rossell Techsys (NSE and BSE: ROSSTECH). I have not transacted in this security in the 30 days preceding publication and will not transact in it for 5 days following publication. Karnik Shah, SEBI-Registered Research Analyst, Reg. No. INH000023807. This is not investment advice. Do your own due diligence or consult a SEBI-registered adviser.

See you at the next inflection point.

Karnik

IPR verdict: WAIT FOR LISTING DIP

Milky Mist is a better business than most things that list in India.

It is the largest private packaged paneer brand in the country, with roughly 19% of the organised market. It grew revenue 34% last year to ₹3,145 crore. It runs a 33.9% gross margin, which is the highest of any listed dairy company in India, because it sells only value-added products and no loose liquid milk. It owns its own cold chain, its own reefer fleet, its own chilling centres. It sources directly from 67,000-plus farmers.

I want to own this business.

I do not want to own it at ₹140.

That price is 85 times trailing FY26 earnings for a company that earned 9.5% on its capital last year, while paying 8.7% on its debt. Its cheapest listed rival earns 24% on capital and trades at 24 times earnings.

This note explains that gap, because that gap is the entire investment decision.

The three-year picture

All figures in ₹ crore. FY24 and FY25 from the restated DRHP accounts. FY26 from RHP-stage disclosure.

The top line is genuinely excellent. The bottom line just tripled. So what is my problem?

Everything above gross profit is best in class. Everything below it is the problem.

Here is Milky Mist next to its listed peers.

Read the first two columns together. Milky Mist has the best EBITDA margin in the set and the worst return on capital in the set.

That is not a contradiction. It is an explanation.

Milky Mist earns a superb margin on every rupee of sales, and then needs an enormous quantity of capital to produce those sales. Fixed asset turnover is 1.91 times. A single mega-plant at Perundurai. An owned truck fleet. Owned chilling centres. Owned solar capacity. Dodla runs a distributed, asset-light model and turns its capital nearly three times as hard.

The market is being asked to pay a branded-consumer multiple for a business whose returns are determined by the size of its asset base, not by its position on the shelf.

The cash flow nobody is quoting

This is the number that decided the verdict for me.

Over FY23 to FY25, Milky Mist:

Earned cumulative PAT of → ₹93 crore Spent capex of → ₹1,279 crore Added debt of → ₹578 crore Generated free cash flow of → minus ₹894 crore

Cash from operations, after interest paid, did not cover capex in any single one of the three disclosed years. Not once.

Interest cover last year was 1.68 times. Debt to net worth was 5.67 times.

This is not a company that has been converting its growth into cash. It has been buying its growth with borrowed money, and the return on that borrowed money has been running about 85 basis points above what the money costs.

To be fair to the bulls: the profit jump is real

Most pre-IPO profit spikes are accounting noise. I went looking for that here and did not find it.

FY26 EBITDA rose ₹125 crore. FY26 PAT rose ₹81 crore. That is a 65% drop-through, and the reason is honest: interest and depreciation were already sunk, so incremental gross profit fell almost straight to the bottom line. The incremental EBITDA margin on new revenue was 15.8%, against a blended 13.8%. Receivable days did not move. Related party transactions are about 1.1% of revenue. There is no auditor qualification.

This is operating leverage arriving. It is the real thing.

And it is precisely why 85 times is dangerous.

The same mechanic runs in reverse. PAT is a thin residual on a 4% margin sitting on top of a fixed cost base. A 100 basis point move in EBITDA margin is ₹24 crore post-tax, which is 19% of last year’s entire profit. Raw milk procurement costs are currently compressing every listed dairy company in India, and all four comparable listed names are down this calendar year on exactly that line.

Consensus is capitalising the most operationally geared number in the accounts as though it were the most stable one.

The governance file was clean only from six days before filing

Three findings from the DRHP that I have not seen discussed anywhere.

One. The brand was not owned by the company.

The trademarks under which Milky Mist sells were held personally by the promoter, Sathishkumar T. They were assigned to the company on 15 July 2025, six days before the DRHP was filed, for a consideration of ₹8.86 crore.

Shareholders funded ₹71 crore of advertising and brand promotion in FY25 alone. That spend was building equity in a mark sitting on the promoter’s personal balance sheet. The company then paid him for it.

It is now cured, and cured is what matters for a buyer today. But read it as a governance trajectory rather than a governance standard.

Two. A promoter-owned supplier, with no inventory records.

Asal Food Products Private Limited was owned by the promoters and sold roughly ₹20 crore a year of product into Milky Mist. It was bought by the company in March 2025 for ₹7.13 crore.

The auditors reported that Asal had no inventory records at all, so they were unable to comment on discrepancies between physical and book stock, and that its accounting software had no audit trail facility whatsoever.

Three. Promoter remuneration against profit.

Over FY23 to FY25, the two promoters drew ₹50 crore in remuneration. Cumulative PAT over the same three years was ₹93 crore.

In FY24, promoter remuneration was 88% of the company’s entire profit for the year.

The control-side findings

Not accrual engineering, but not nothing either. From the CARO annexures:

→ The audit trail (edit log) was not operating on the customer master, vendor master and price master until September and December 2024. In FY24, it was absent from all non-accounting modules entirely, and the auditor stated they were unable to comment on whether it had been tampered with.

→ Inventory reported to lenders in the quarterly stock statements diverged from the books in three of four quarters in FY23, once by ₹24 crore on a book figure of ₹96 crore. It diverged again at Q4FY25, by ₹22.5 crore, attributed to a revenue reversal at audit.

→ Coolers and freezers held at distributor premises have not been reconciled to the books. The auditor said they were unable to comment on material discrepancies. Note that ₹155 crore of the issue proceeds is earmarked to buy more of that same asset class.

What is genuinely good about this issue

I want to be straight about this, because the structure is better than most.

→ 92% of the issue is fresh capital going into the company. The OFS is only ₹125 crore. → The promoters halved their own cash-out, from ₹250 crore in the DRHP to ₹125 crore. → The objects are specific: ₹497 crore to debt repayment, ₹469 crore to the Perundurai expansion, ₹155 crore to cold chain. Very little is parked in general corporate purposes. → Receivables are 16 days. There is no debtor stretch and no channel stuffing. → Temasek, through Jongsong Investments, came in at ₹139.76 in April 2026 for 5.16%. The IPO is priced flat to that. No markup on the last round. → The ask has already been cut. The promoters were reportedly seeking ₹20,000 crore in 2024. They are asking ₹10,778 crore now, a 46% reduction.

The reservation on the use of proceeds is arithmetic, not intent. The company repays ₹497 crore of debt and commits ₹624 crore straight back out to new capex. The balance sheet gets one year of relief, not a structural reset.

Valuation

At the ₹140 cap price:

P/E on FY26 earnings → 84.9x EV/EBITDA on FY26 → 26.8x EV/Sales on FY26 → 3.71x Dairy peer median P/E → roughly 25x Most expensive listed dairy comparable (Hatsun) → 63.2x

On my own FY27 estimate, assuming 25% revenue growth to about ₹3,930 crore at a 14.5% EBITDA margin, with interest falling on the reduced debt, I get roughly ₹219 crore of PAT. At the cap price, that is still about 49 times forward earnings.

I cannot construct a measure on which this is a cheap issue.

Verdict: WAIT FOR LISTING DIP

The thesis is sound. The entry is not.

This is a quality watch, not an apply-at-any-price. The business is real, the brand is real, the growth is real, and the operating leverage is real. But 85 times trailing earnings and 26.8 times EBITDA, for a company earning 9.5% on capital with the highest leverage in its own peer set and no free cash flow in any disclosed year, is a price that requires everything to go right.

Temasek is in at ₹139.76. The anchor is at my price, not below it. The issue offers no discount for the risk being taken.

And the dip I am waiting for is a real one, not 10%.

My re-entry checklist. Meet four of these five and the verdict changes.

→ Price. ₹110 to ₹115 or better, roughly 40 times my FY27 estimate. → Two quarters of evidence. Gross margin held above 32% through the current milk cost cycle, EBITDA margin sustained above 13.5%. → Capex discipline. FY27 capex guided under ₹500 crore, with the first sign of positive free cash flow. → Balance sheet. Net debt to EBITDA at or below 2x a full year after listing, not just on day one. → Governance. Promoter remuneration disclosed against a stated policy and falling as a share of PAT, and completion of the fixed asset reconciliation the auditors flagged.

Open item, stated honestly

My promoter entity footprint check is not fully closed. Two of the five checks failed historically and were cured before filing: the trademark assignment and the Asal acquisition. Still outstanding is an independent MCA and ZaubaCorp sweep of the full promoter and promoter group entity list against NIC codes, the basis of any combined-group credit rating treatment, and an e-commerce seller GSTIN check.

Nothing in the DRHP suggests a further problem. I am stating it as an unverified gap rather than claiming a clean pass I have not earned.

The full four-page forensic note is attached as a PDF below. As is a video I made on this topic.

Ipr Milkymist Ipo Review

227KB ∙ PDF file

Download

Download

Data sources and caveats

My primary read is the Draft Red Herring Prospectus dated 21 July 2025, covering FY23 to FY25. FY26 figures, the final price band and the final issue terms are drawn from RHP-stage public disclosure and press reporting, and have not been independently verified against a primary read of the Red Herring Prospectus. Peer market multiples move daily and should be treated as an order-of-magnitude anchor, not a precise screen. Verify every figure against the final RHP before acting.

Disclosures

Analyst position: I hold no position in Milky Mist Dairy Food Limited, which is unlisted at the time of writing, and I will not be subscribing to this issue.

This report is published by a SEBI-registered Research Analyst (Registration No. INH000023807). This note is for information and education only. It is not investment advice, not a recommendation to apply for or subscribe to the issue, and not a solicitation. Securities markets are subject to market risks; read all related documents carefully before investing. Past performance is not indicative of future results. Readers should conduct their own due diligence before making investment decisions.

Investor grievances: research@inflectionpointresearch.in | Compliance / SEBI SCORES: scores.sebi.gov.in

Inflection Point Research. Forensic-first analysis of Indian micro and small-cap equities.

Want to see the method applied to two full company deep dives? Get two free sample reports: https://www.inflectionpointresearch.in/free-sample-research-reports

Shiprocket’s Red Herring Prospectus landed on 5 August. The issue opens on 12 August at a price band of ₹92 to ₹97, raising ₹1,617.5 crore and valuing the company at roughly ₹7,000 crore.

The headlines have settled on two facts. Losses down 88%. First positive Adjusted EBITDA. Both are true. Neither means what it appears to mean.

I want to start with the second one, because it is the whole note.

Adjusted EBITDA is a definition here, not a result

Shiprocket reports FY26 Adjusted EBITDA of ₹17.6 crore. Positive for the first time. The company defines it, and the definition is in footnote 11 on page 169 of the RHP, as earnings before interest, tax, depreciation and amortisation, arrived at after deducting cost of merchant solutions, traded goods, employee benefits expense excluding share based payment expense, other expenses and rent.

That excluded share-based payment expense was ₹112.3 crore in FY26.

It is 6.4 times the Adjusted EBITDA the company reports. It is 5.6% of revenue. It has been present every single year in the disclosure period, at ₹192.6 crore, ₹91.3 crore and ₹112.3 crore. It is not a one-off. It is compensation, paid in stock instead of cash, and the bill arrives as dilution rather than as an outflow.

Treat it as the recurring cost it is and the bridge looks like this:

Adjusted EBITDA as the company defines it, FY26 → ₹17.6 Cr Less: share-based payment expense excluded from that definition → ₹(112.3) Cr EBITDA after treating ESOP as a real cost → ₹(94.7) Cr

The FY25 comparative on the same basis is ₹(84.2) crore. So on a post-ESOP view, FY26 was worse than FY25, not better.

And the “88% loss reduction” is a FY24 to FY25 comparison against a year that carried ₹244.4 crore of one-time goodwill and intangible impairment. From FY25 to FY26, the reported loss actually widened, from ₹74.4 crore to ₹79.2 crore.

The receivable book and the provision that shrank

The second finding is the one I would want any subscriber to check for themselves, because it is the classic pattern in the year before a filing.

Trade receivables → FY24: ₹117.1 Cr → FY25: ₹147.0 Cr → FY26: ₹236.5 Cr Revenue growth → FY25: +24.0% → FY26: +24.0% Receivable growth → FY25: +25.6% → FY26: +60.9% Debtor days → FY24: 32.5 → FY25: 32.9 → FY26: 42.6 Receivable plus unbilled days → FY24: 58.0 → FY25: 59.2 → FY26: 69.7

Receivables ran two and a half times faster than revenue in FY26. Now look at what happened to the provision against them in the same year.

Expected credit loss charge → FY24: ₹19.8 Cr → FY25: ₹10.1 Cr → FY26: ₹5.4 Cr As a percentage of closing receivables → FY24: 16.9% → FY25: 6.8% → FY26: 2.3% Provision coverage on gross receivables → FY24: 28.1% → FY25: 27.3% → FY26: 17.8%

The receivable book itself ages cleanly. All the good debt is under six months, which the RHP ageing schedule confirms. So this is not a collection crisis and I am not alleging one.

But run FY26 provisions at even the FY25 intensity and roughly ₹10.7 crore comes off the profit and loss account. Against a reported Adjusted EBITDA of ₹17.6 crore, that is over half of it. A provisioning judgement moved in the company’s favour in the one year that was going to be sold to public investors.

The Core is good. The Emerging Business is eating it.

Split the segments apart and the picture becomes much clearer than the consolidated numbers allow.

Core Business (domestic shipping and shipping apps) Revenue → FY24: ₹1,084.7 Cr → FY25: ₹1,305.9 Cr → FY26: ₹1,485.4 Cr Adjusted EBITDA margin → FY24: 6.65% → FY25: 12.02% → FY26: 12.56% Customer acquisition cost → FY24: ₹4,101 → FY25: ₹3,361 → FY26: ₹2,829

That is a real asset-light platform. Improving margin, falling acquisition cost, ₹186.6 crore of segment profit.

Emerging Business (cross-border, checkout, marketing, hyperlocal, merchant credit) Revenue → FY24: ₹231.3 Cr → FY25: ₹326.1 Cr → FY26: ₹538.7 Cr Adjusted EBITDA → FY24: ₹(200.1) Cr → FY25: ₹(149.9) Cr → FY26: ₹(169.0) Cr

Read that last line again. The margin percentage improved, from negative 46.0% to negative 31.4%, and every deck will lead with that. But the absolute burn grew by ₹19 crore in FY26. The loss is not narrowing. It is getting bigger while getting more efficient per rupee of revenue, which is only good news if you believe the revenue keeps compounding at 65%.

And ₹205.8 crore of the fresh issue is earmarked for marketing “primarily for our Emerging Business.”

The number almost nobody is looking at

Power Merchants, the merchants doing more than 100 transactions a month, are where Core revenue actually comes from.

Power Merchants → FY24: 9,020 → FY25: 10,005 → FY26: 10,090 Growth → FY24: +10.13% → FY25: +10.92% → FY26: +0.85% Power Merchant ARPU → FY24: ₹12.8 L → FY25: ₹14.4 L → FY26: ₹17.8 L, up 23.6%

The Core added 85 net Power Merchants in FY26. All of the Core’s 13.7% revenue growth came from monetising the same base harder.

ARPU-led growth is not illegitimate. Cross-sell is the stated strategy and it is working. But it has a ceiling, and a flat merchant count on a platform whose entire pitch is network effects is the single fact I would want management to address on the first earnings call.

What is genuinely clean, and I want to be fair about this

I run every filing through the same forensic checklist and Shiprocket passes more of it than most.

Auditor: S.R. Batliboi & Associates LLP, a Big Four network firm. Unmodified opinion for FY26. Contingent liabilities as at 31 March 2026: Nil. Not low. Nil. Related-party transactions: managerial remuneration and intra-group eliminations only. No promoter-entity purchases, sales or vendor routing. Customer concentration: top 10 merchants at 16.23% of revenue in FY24, falling to 12.19% in FY26. Largest single merchant 2.83%. Regulatory enforcement: no SEBI, ED or CBI action disclosed anywhere in 516 pages. Promoter entity footprint: not applicable. The company is professionally managed with no identifiable promoter, no promoter group, and no group company.

Those are honest clears and they stay in the scorecard.

The offer for sale reads worse than it is

Forty-five percent of the issue is an offer for sale, and my checklist flags anything above 40%. Read it shareholder by shareholder, though, and the story changes.

Lightrock (LR India Fund) → selling 100% of its 28.0 million share position Tribe Capital III Series 1 → selling 30.4% of its stake Saahil Goel, MD and CEO → selling 20.4% Gautam Kapoor, ED and COO → selling 20.4% Bertelsmann Nederland, 21.32% holder → selling nothing Tribe Capital III Series 5, 7.75% → selling nothing Eternal Limited (Zomato), 6.85% → selling nothing KDT Ventures, 5.49% → selling nothing MacRitchie (Temasek), 5.29% → selling nothing

Lightrock’s full exit is fund-life mechanics, not a verdict. The four largest holders are not selling a share. The founders are trimming a fifth each at a price 41% below their own last mark. That is not a cash-out.

The supply problem is somewhere else, and it is much bigger

Because there is no identifiable promoter, Regulation 14(1) of the ICDR Regulations does not apply. There is no minimum promoter contribution and no eighteen-month lock-in on a single share. Everything gets six months under Regulation 17(1).

Free float at listing → 166.8 million shares, 22.9% of post-offer capital Locked for six months → 560.8 million shares, 77.1% of post-offer capital Potential increase in tradable stock at month six → 4.4 times

That cliff lands around mid-February 2027, and there is no promoter block standing behind it. This, more than anything in the financials, is why my call is about timing.

Valuation

At ₹97, post-issue market capitalisation is roughly ₹7,057 crore.

EV to gross revenue → 3.13x EV to contribution margin → 17.0x EV to Adjusted EBITDA → 359x EV to post-ESOP EBITDA → not meaningful, negative Price to last primary round of ₹163.14 → 0.59x Price to last secondary of ₹116.38 → 0.83x

The RHP offers exactly one listed peer, Unicommerce eSolutions, which is a pure software order-management business at ₹204 crore of revenue with no freight gross-up. That is one-tenth Shiprocket’s size and a completely different revenue model. It cannot carry a valuation.

Anchor to Delhivery instead. As at 6 August 2026, Delhivery trades at roughly 3.2x EV to sales on ₹10,508 crore of FY26 revenue, with a 6.1% operating margin, positive PAT and ₹911 crore of operating cash flow.

So Shiprocket is asking for the same EV to sales as a peer five times its size that is profitable and cash generative. The asset-light model argues it should get more. The negative post-ESOP EBITDA and the contribution margin that fell in FY26 argue it is already getting the benefit of the doubt.

The genuine bull point, and it is a strong one: at ₹92 to ₹97 this is a down round. The last primary was struck at ₹163.14, there has been no primary or secondary transaction in the eighteen months before filing, and the issue size was cut 31% from the ₹2,342.3 crore proposed in the updated DRHP. That is a reset. It is also, read less charitably, the book telling the company what it is worth.

The verdict

WAIT FOR LISTING DIP.

This is not an AVOID. The forensic file is cleaner than most and the Core Business is genuinely good. If Shiprocket were listing at ₹65 I would be interested.

But I cannot get to subscribe. Reported Adjusted EBITDA is positive only because a ₹112.3 crore recurring cost is defined out of it. Receivables ran 61% on revenue of 24% while provisioning intensity fell by two thirds. The Core’s merchant base has flatlined. Fresh capital is directed at a segment whose absolute losses are widening. And 77% of the capital unlocks at month six into a 23% float with no promoter lock behind it.

The entry is the problem here, not the business.

What would change my mind, three of five:

1. Power Merchant count growing above 5% year on year, not 0.85%

2. Emerging Business Adjusted EBITDA loss narrowing in rupee terms, not just as a percentage

3. Debtor days back below 40, with receivable growth no faster than revenue growth

4. Expected credit loss charge normalising toward 6% to 7% of closing receivables

5. First post-listing goodwill impairment test passing clean, with the Omuni and Pickrr cash-generating units disclosed separately

The full four-page forensic note is attached as a PDF below.

Ipr Shiprocket Rhp Review

252KB ∙ PDF file

Download

Download

Get two free IPR sample reports. The forensic-first method you just read, applied to two full company deep dives. No payment required: https://www.inflectionpointresearch.in/free-sample-research-reports

Disclosures (SEBI Reg. INH000023807). Inflection Point Research is a SEBI-registered Research Analyst. This is an analysis of a publicly filed Red Herring Prospectus and is for information and education only. It is not investment advice, not a recommendation to apply for or subscribe to the issue, and not a solicitation. The analyst and the analyst’s relatives hold no position in the issuer, which is unlisted at the time of writing, have no financial interest in it, have received no compensation from it, and have not acted as a market maker or in any advisory capacity for it in the preceding twelve months. Securities markets are subject to market risks. Read all related documents carefully before investing. Past performance is not indicative of future returns. Investors should consult their own financial advisers before acting. Investor grievances: karnik@inflectionpointresearch.in. Compliance and SEBI SCORES: scores.sebi.gov.in.

Data note. All financial, operational, litigation, related-party, shareholding, lock-in and ESOP figures are drawn from the Shiprocket Limited Red Herring Prospectus dated 5 August 2026. Peer market data for Delhivery is as at 6 August 2026. The price band of ₹92 to ₹97, the bid dates and the lot size are sourced from public reporting dated 6 August 2026 and have not been verified against the price band advertisement. Every valuation figure moves if the final band moves. Verify against the Prospectus before acting.

Indo-MIM Limited’s ₹3,811 Cr mainboard IPO closed bidding today, subscribed north of 7.6x, GMP running ~38-39% above the cap price. It’s the world’s largest manufacturer of precision components using Metal Injection Moulding (MIM), a category most investors have never had reason to think about until this week.

The business

Five segments feed off one manufacturing platform: Automotive, Defence, Medical, Consumer and Aerospace. Fifteen plants across India, the US, UK and Mexico. The moat isn’t a slogan, it’s a mechanical fact of the industry: OEM supplier qualification runs 2-3 years of audits and validation, and dual-sourcing a precision component is usually uneconomical once tooling costs are sunk. That shows up in the numbers: repeat customers contributed 91.6% of FY26 revenue.

One detail worth dwelling on: Indo-MIM makes its own feedstock “the elemental way”, mixing raw metal powder and binder in-house rather than buying premix compounds like most competitors do. The industry report the company commissioned calls this out as a genuine design-freedom advantage, not just a cost play. It already makes its own stainless-steel powder, and has an iron-powder facility under construction, targeted for completion by the end of FY27.

Financial snapshot, FY24 -> FY26 (Rs Cr)

Revenue -> 2,870 -> 3,330 -> 4,193
EBITDA margin -> 25.90% -> 28.01% -> 25.54%
Reported PAT -> 284 -> 424 -> 534
CFO -> 458 -> 506 -> 1,077

CFO exceeded PAT in every single year, the single highest-signal earnings-quality check in our framework. Clean on that front.

The valuation question

Here’s where this note earns its keep. The RHP itself states plainly that no listed Indian company does what Indo-MIM does. Its own disclosed comparable is a Chinese peer, Jiangsu Gian, trading at roughly 148x P/E on a RoNW of just 3.1%. That’s not a useful reference point for anybody.

So we built our own basket, the Indian precision-engineering and defence-component names investors actually rotate within:

Company -> P/E (current)
Indo-MIM (at Rs485 cap) -> ~45x
Azad Engineering -> ~116x
Craftsman Automation -> ~58-111x
Paras Defence -> ~62x
MTAR Technologies -> ~189-260x

Against that set, not the Chinese comp the RHP hands you, Indo-MIM is priced as the cheapest name in the room, and on RoNW of 21.3% (versus Jiangsu Gian’s 3.1%), plausibly the best-run one too. That’s a relative-value argument, not a claim that 45x is cheap in any absolute sense. Some of those peers may themselves be priced on policy sentiment rather than current earnings.

Where we’d slow down

Two things temper the enthusiasm. First, capacity utilisation across the core MIM plants sits at just 30.6%, and it actually fell from 35.8% a year ago, even as revenue grew 25.9%. The growth came from a shift in mix (a tripling of the lower-margin “Others” line: powder, tooling, trading) rather than higher throughput, and EBITDA margin compressed from 28.0% to 25.5% as a result. The operating-leverage re-rating case is real and well evidenced (utilisation this low, with no fresh capex needed since the IPO’s fresh issue funds debt paydown, not expansion) but it is a forward trigger to watch over the next two to three quarters, not something already validated in the FY26 numbers.

Second, three straight years of “exceptional” goodwill and PP&E impairment charges (Rs76.5cr, Rs101cr, Rs78cr in FY24, FY25 and FY26) tied to the UK and US acquisitions deserve a fourth-year check before being waved through as one-off. And defence-segment (DPG) revenue, the segment most tied to the Make-in-India growth story, actually fell 12.2% this year, a fact the RHP discloses without explaining.

The read

SUBSCRIBE, built on relative valuation and moat quality rather than on operating leverage already showing up. Watch utilisation, the defence mix, and the FY27 impairment line before extending conviction from “cheap relative to peers” into “genuinely re-rating.”

Full four-page PDF, including the complete forensic scorecard and the 3-year bull/base/bear scenario table, attached below.

Indo Mim Ipo Review

232KB ∙ PDF file

Download

Download

Disclosures (SEBI Reg. INH000023807): Inflection Point Research is a SEBI-registered Research Analyst. This note analyses a publicly filed Red Herring Prospectus and prevailing market data, for information and education only. It is not investment advice, not a recommendation to apply for, subscribe to, or trade the Equity Shares, and not a solicitation. The analyst holds no position in the issuer. Read all offer documents carefully before investing.

Inflection Point Research | Karnik Shah | July 2026

Hello friends,

Let me start with something that should not make sense.

In December 2025, Pernod Ricard sold Imperial Blue. Not a struggling brand. India’s third-largest whisky by volume, roughly 22 million cases a year, sold to Tilaknagar Industries for about ₹4,150 crore. A few months earlier, Diageo had done something similar with a clutch of its own popular, mass-market labels. Within the space of a couple of years, the two largest global spirits companies operating in India had voluntarily walked away from the very brands that sold the most bottles.

Read that again, because it is genuinely strange. The biggest, most sophisticated consumer companies in one of the fastest-growing markets on earth looked at their best-selling products, the ones that took decades to build, and decided they would rather not sell them any more.

They are not confused. They can see something in the data that the rest of us feel but rarely name. The money in India is not spreading out. It is climbing up. And the smart operators have decided that the profitable place to stand is no longer in the middle of the market, serving everyone. It is at the top, serving the few who are pulling away from everyone else.

This is the first of a two-part series on what I am calling the new Indian consumption story. This part is about premiumisation: the rise of a thin, fast-growing band of affluent Indians, the businesses repositioning themselves to capture that band, and how an investor should think about the whole thing without getting swept up in the narrative. The second part, which I will tease at the end, is about the new-age internet platforms feeding the same customer through a different pipe. We will come to that another day.

This one runs long. If you want the single sentence to carry in your pocket, here it is.

India is not one consumer market that is getting richer. It is two markets pulling apart, and almost all the interesting money is being made by the businesses that understood this early, repositioned deliberately, and can prove the shift is already showing up in their numbers rather than merely in their investor decks.

If you would rather watch a video we made on this topic, you can do so here:

Part 1: The Number That Reframes Everything

Start with per-capita income, because it is the engine under all of this.

India’s per-capita income crossed roughly $2,600 in FY25 and is, on most credible projections, on track to reach something in the region of $5,000 by FY30. Private final consumption expenditure now runs at roughly 61% of GDP, the highest share since FY12. India is the sort of economy where domestic demand, not exports, is the spine.

That is the headline every consumer-story deck opens with, and it is true, but on its own it is almost useless. An average is a lie we tell ourselves to feel comfortable. The interesting question is never what the average Indian earns. It is how the earning is distributed, and which direction the distribution is moving.

Here the picture sharpens dramatically. Goldman Sachs, in its now widely-cited “Rise of Affluent India” work, defined an affluent cohort as those earning more than roughly $10,000 a year. That cohort numbered about 24 million people in 2015. By 2023 it was around 60 million, roughly 4% of the population. On Goldman’s projection it reaches about 100 million by 2027.

Now hold the two growth rates next to each other, because this is the whole thesis in one comparison. That affluent cohort has been compounding at something like 12 to 13% a year. India’s overall population grows at barely more than 1%. So the group of Indians who can comfortably buy a premium product is not just large. It is growing more than ten times faster than the country as a whole.

This is what people are gesturing at when they say India’s consumption has gone “K-shaped.” The upper line of the K keeps climbing. The lower line flattens. And the businesses that sit on the upper line, selling to the people whose incomes and, crucially, whose asset values are compounding, have been growing revenue and profit at rates that mass-market companies serving the broad base simply cannot match.

The word to sit with is “asset values,” because it is the part most people miss.

The wealth-effect flywheel

Income is only half the story. The affluent Indian is not just earning more. She is watching her existing wealth compound, and that changes behaviour more than a salary rise ever does.

Consider what the top of the pyramid actually owns. One government survey suggests roughly 90% of India’s gold is held by the top 10% of earners. The same cohort owns the overwhelming majority of directly held equities and property. Over the last several years, all three of those asset classes, Indian equities, gold, and prime residential property, have risen substantially and, for stretches, together.

When your salary rises, you feel a little richer. When your portfolio, your gold, and your flat all rise for the third year running, you feel structurally, permanently wealthier, and you spend accordingly. That is the wealth effect, and it compounds on itself: rising asset prices make the affluent feel richer, so they spend more on premium goods and invest more, which supports asset prices, which makes them feel richer still. It is a flywheel, and for the last few years it has been spinning in the right direction.

This is why the premiumisation you can see on the ground feels early, and slightly improbable. India is buying single malts, ₹73 lakh cars, and South Mumbai flats at a national per-capita income at which most economies barely had a middle class at all. The reconciliation is that “national per-capita income” is the wrong number to look at. The top slice behaves like a rich country bolted onto a poor one, and it is spending not just from income but from a rising stock of assets. That is what makes the theme bigger than the income data alone would suggest. It is also, for the same reason, more fragile than it looks, because a market powered by asset prices is only as durable as the asset prices.

I flag this now, at the top, because it is also the single biggest risk to everything that follows, and I would rather you hold it in your mind the whole way through than have it sprung on you at the end. Every business in this piece is, to some degree, a leveraged bet on that flywheel continuing to spin. A sustained fall in equities, gold, and property together would hit all of them at once, because it would hit the confidence of the exact customer they all share. Keep that thought. We will return to it.

Part 2: Two Kinds of Premiumisation (Do Not Confuse Them)

Before we look at any company, we need one distinction, because getting it wrong is how investors overpay.

“Premiumisation” is used to describe two quite different things, with different margins, different durability, and different valuations. Confusing them is the most common mistake in this theme.

The first kind is pure luxury: capturing the top few per cent. This is the watch retailer, the luxury-car dealer, the ultra-luxury developer, the wealth manager for the very rich. The customer base is small, the ticket size is enormous, and demand is powered directly by the affluent cohort and the wealth-effect flywheel. Growth can be spectacular. But the addressable market is genuinely narrow, the businesses are often working-capital heavy, and they are the most exposed to any reversal in asset prices, because their customer is the wealth effect.

The second kind is mass-premium trade-up: mix-shift within an everyday category. This is the dairy company moving from loose milk to cheese and whey protein, the spirits company shifting its mix from cheap whisky to “prestige and above,” the FMCG major selling premium detergent and hair care. The customer here is not the top 5%. It is the far larger aspiring middle, choosing to “drink better, not more,” or to buy the ₹1,000 bottle instead of the ₹500 one. The volumes are bigger, the demand is more resilient in a downturn (people do not stop buying milk), but the pricing power is lower and the competition fiercer.

Most of the excitement, and most of the eye-watering valuations, sits in the first column. Some of the more durable, if less thrilling, compounding sits in the second. A serious investor in this theme should always know which column a given name belongs to before forming a view on the price.

With that in hand, let us walk the categories. I am not trying to be exhaustive. These are illustrative windows onto the theme, chosen because each shows the mechanics clearly. The named companies are examples, not recommendations, and every figure below is flagged for verification against primary sources.

Part 3: The Categories, and What the Operators Are Actually Doing

Hospitality: the capital-light way to sell an experience

If you want the cleanest example of premiumisation done well, look at Indian Hotels (NSE: INDHOTEL).

The obvious version of the story is that rich Indians are travelling more, luxury hotel rooms are scarce, room rates are rising, and IHCL owns the Taj. All true. In FY26, IHCL delivered consolidated revenue of about ₹9,971 crore, up 16%, with EBITDA of roughly ₹3,477 crore at a margin near 35%, and its best-ever PAT of about ₹2,084 crore. Same-store revenue per available room grew around 9%, and the Taj brand commanded a same-store RevPAR premium of over 70% to the industry. Sixteen consecutive record quarters.

But the interesting part is not the room rates. It is what management is doing with them. Under its “Accelerate 2030” plan, IHCL is deliberately shifting from owning hotels to managing them: signing a record 250 hotels in FY26 to reach a portfolio of about 630, with management-fee income growing roughly 22%. That fee income is nearly pure margin and requires almost no capital. So IHCL is using its brand, the Taj name that took a century to build, to earn a high-margin annuity on other people’s real estate, while keeping a handful of iconic owned assets for prestige and pricing power. The credit rating was upgraded to the highest domestic notch during the year.

That is the template for the best premiumisation businesses: a brand with genuine pricing power, monetised in a capital-light way so that rising demand converts into rising return on capital rather than just rising capital expenditure. The honest caveat is that this is a discretionary, cyclical business exposed to shocks it does not control; the West Asia conflict dented inbound and Dubai occupancy in Q4FY26. And the market knows all of this, so the stock is not cheap. But as an illustration of how to capture the affluent traveller, it is close to textbook. Peers worth watching in the same current include EIH, Chalet Hotels, and the recently-listed luxury names, each a different bet on the same rising tide.

Watches: the retailer as a curator of scarcity

Ethos (NSE: ETHOSLTD) is India’s largest luxury and premium watch retailer, and a near-perfect specimen of the first kind of premiumisation.

An entity-lock note first, because it matters: Ethos and its listed promoter KDDL (NSE: KDDL) are two different companies. Ethos is the retailer; KDDL holds a large stake in Ethos and separately manufactures watch components. If you are buying “the Ethos story,” know which ticker you actually hold.

Ethos does not make watches. It curates access to them. It holds relationships with dozens of global luxury brands, Rolex, Omega, and the rest, and sells through boutiques in premium locations. In FY26 it grew revenue strongly, with Q4FY26 sales up about 33% year on year, roughly 94 boutiques by year-end against about 79 a year earlier, same-store sales growth in the mid-teens, and a mix that continued shifting towards luxury and high-luxury watches at around 71% of sales. Average selling prices sit around ₹2 lakh per watch.

What is management doing to capture the tailwind? Three things. Adding exclusive brand partnerships that competitors cannot easily replicate. Expanding into new affluent pockets beyond the top metros. And positioning for the India-EFTA trade agreement, which over time should ease the duty on Swiss watch imports and improve economics.

And here is the discipline the theme demands: even a clean premiumisation story has an unglamorous underside. Ethos is intensely working-capital heavy; it must hold enormous, expensive inventory to offer the in-store experience, and its cash conversion cycle runs to several months. Q4FY26 profit actually fell year on year despite the revenue surge, as expansion costs and currency movements compressed margins. And the stock trades at a rich earnings multiple against a currently modest return on equity. Wonderful category, genuine leader, but the price already assumes years of flawless execution. That combination, great business, demanding price, is the recurring signature of this entire theme. Titan (NSE: TITAN), through its own watches and its far larger jewellery franchise, is the large-cap comparator to keep in the same field of view.

Cars: premiumisation is real, but mind where the value sits

Everyone can see the premiumisation in Indian cars. Sport-utility vehicles have overtaken small cars. Mercedes and BMW keep posting records. The question for an investor is not whether the trend is real. It is where in the chain the profit actually lands.

Landmark Cars (NSE: LANDMARK) is the listed way to play luxury-car retail, a dealer for Mercedes-Benz, BYD, Mahindra, and others. In FY26 it posted record revenue of about ₹6,713 crore, up 19%, with the average Mercedes selling price rising to roughly ₹73 lakh and electric vehicles crossing a fifth of sales. Management has finished a multi-year network build-out and is now pivoting to sweating those assets, growing the higher-margin after-sales business, and preparing for a BYD volume ramp.

But look at the shape of it. Against ₹6,713 crore of revenue, Landmark’s market capitalisation is only around ₹1,800 crore, and its return on equity has run in the high single digits. That is the tell of a distributor, not a brand owner. Dealerships are thin-margin, working-capital-intensive businesses tethered to the demand cycles and pricing policies of the manufacturers they represent. This is the picks-and-shovels version of the luxury-car boom: real exposure to the trend, but the structural pricing power sits upstream with Mercedes, not with the showroom.

There is a subtler, and arguably better, way to play the same trend at the component layer. As cars premiumise, the “aesthetic content” per vehicle, the badges, dials, chrome, decorative surfaces, and increasingly the display covers, rises in value. SJS Enterprises (NSE: SJS) supplies exactly this kind of decorative and functional content to automakers and consumer-durable brands, and its revenue per vehicle should structurally rise as vehicles move upmarket. And in two-wheelers, the same trade-up shows in Eicher Motors (NSE: EICHERMOT), whose Royal Enfield has made premium the entire point. The lesson across all three: the trend is one thing, the layer that captures the trend’s economics is another. Never buy the theme without asking where the pricing power actually lives.

Alcohol: “drink better, not more”

We opened with the spirits giants selling their mass brands. Now the data behind that decision.

Total Indian spirits volumes have been broadly flat to slightly down, falling around 2% in FY25 after a similar drop the year before, squeezed by high state taxes and inflation at the bottom end. And yet the “prestige and above” segment has been growing at a healthy clip, with premium and super-premium products (priced above roughly ₹1,000 for a 750ml bottle) growing around 15% and on track to make up something like 38 to 40% of spirits revenue. The phrase the industry uses is “drink better, not more,” and the data supports it: this is genuine trading up, not consumers being priced out.

Radico Khaitan (NSE: RADICO) is the cleanest domestic expression. FY26 revenue of roughly ₹6,050 crore, up about 25%, with “prestige and above” now more than 70% of its Indian-made-foreign-liquor value and margins expanding as the mix improves. What is it doing to capture the shift? Launching genuinely premium products, Rampur and Jaisalmer single malts and gins, that carry luxury margins, backed by disciplined debt reduction so the profit of the mix-shift drops to the bottom line rather than to interest.

This is the second kind of premiumisation, mix-shift, and the peer set, United Spirits (NSE: UNITDSPR), United Breweries (NSE: UBL), Allied Blenders (NSE: ABDL), and Tilaknagar (NSE: TI), is all playing versions of the same game: retreat from where margins are unviable, push where the affluent are trading up. The regulatory point cuts both ways, though. State-level excise and pricing controls are the sword hanging over the whole sector; a favourable move (Karnataka’s recent pricing deregulation) can re-rate the group, and an unfavourable one can gut a state overnight. And these stocks are not shy: several trade at 70 to 85 times earnings. The margin expansion story is real; a lot of it is already in the price.

Wealth management: the annuity on the rising pool

Here is the most elegant way to play the whole theme, and the one most retail investors overlook. If the affluent cohort is compounding at 12% a year and its assets are compounding on top, somebody has to manage all that money. The wealth manager does not need to pick the winning consumer brand. It earns a recurring fee on the entire, rising pool.

360 ONE WAM (NSE: 360ONE), formerly IIFL Wealth, is the pure-play leader for the very rich. In FY26 it reported PAT of about ₹1,225 crore, up roughly 21%, on revenue near ₹3,144 crore, managing total assets of around ₹6.75 lakh crore, of which the recurring “annual recurring revenue” pool grew strongly, serving over 8,500 families. Its recurring-revenue base has compounded at roughly 32% a year over five years. What is it doing to capture the tailwind? Deepening into the ultra-rich through its partnership with UBS, adding capital-markets muscle via its B&K acquisition, and scaling alternatives, private equity, private credit, real assets, which is exactly where the very wealthy are shifting allocation.

Nuvama Wealth Management (NSE: NUVAMA) is the broader-based cousin, with a mix that spans wealth, asset management, and capital markets, and it typically trades at a lower earnings multiple than 360 ONE. The distinction matters: 360 ONE is the concentrated bet on the UHNI wealth annuity; Nuvama is the more diversified, more capital-markets-sensitive version.

I want to be clear-eyed here too. The wealth-management model is beautiful on the way up and cyclical on the way down, because the fee pool moves with markets. This is, more than any other name in the piece, a direct, undiluted bet on the wealth-effect flywheel I flagged at the start. When it spins, few businesses compound as cleanly. When it reverses, the annuity shrinks with the assets.

Real estate: fewer flats, more money

The single most illustrative statistic in this entire piece belongs to Oberoi Realty (NSE: OBEROIRLTY).

In FY26, Oberoi sold roughly 698 homes, down about 24% from the 929 it sold the year before. And yet its sales bookings rose, to about ₹5,447 crore. Fewer units, more money. That single line is premiumisation distilled: the deliberate choice to sell fewer, far more expensive homes in the best locations to a narrower, wealthier buyer, and to let pricing power do the work that volume used to.

Zoom out and the sector shows two structural shifts at once. First, value is migrating to the top: across 28 listed developers, FY26 pre-sales reached nearly ₹1.95 lakh crore, with growth driven overwhelmingly by luxury launches and price appreciation rather than volume. DLF’s (NSE: DLF) ultra-luxury project The Dahlias reportedly sold over ₹11,500 crore in nine weeks. Prestige Estates (NSE: PRESTIGE) grew FY26 pre-sales about 76% to roughly ₹30,024 crore. Second, share is migrating to the branded: buyers increasingly prefer large, financially strong, RERA-compliant developers who can actually deliver, so listed players are taking share from the unorganised long tail. That is a double tailwind, premiumisation plus formalisation, and it is why the branded developers have compounded.

Dairy: premiumisation of the everyday

Dairy is the corrective to the whole glossy narrative, and worth including precisely because it is not glamorous.

This is premiumisation of the second kind, mix-shift, and it happens inside the humblest category imaginable. Loose liquid milk earns a gross margin of maybe 4 to 8%. Value-added products, cheese, ghee, curd, Greek yoghurt, whey protein, earn 20 to 35%. So the entire game for a listed dairy is to move its mix up that ladder. Heritage Foods (NSE: HERITGFOOD) grew its value-added share to over 38% of revenue and is buying into premium formats like high-protein ice cream; Hatsun Agro (NSE: HATSUN) leans on scale and brand and is pushing into protein products; Dodla Dairy (NSE: DODLA) and Parag Milk Foods (NSE: PARAGMILK) are each climbing the same ladder through cheese, whey, and functional products.

But dairy also teaches the limits of the theme. These businesses remain hostage to the milk procurement cycle; when input prices spiked through late 2024 and 2025, margins compressed hard, because you cannot always pass a cost rise on to a customer buying a daily essential. Premiumisation is a genuine margin pathway here, but it is a slow, capital-intensive climb up a value chain, not a luxury pricing-power windfall. That is the honest difference between trading up milk and selling a Rolex.

Part 4: The Comparison

These are not a portfolio, and they are certainly not substitutes. They are windows onto one force, arranged by which kind of premiumisation each represents and where in the chain it captures value.

Notice the pattern. The pure-luxury names offer the steepest growth and the greatest exposure to the wealth effect, at the highest prices. The mass-premium names offer more durable, if slower, compounding at the mercy of input cycles. And within pure luxury, the businesses that convert the trend into return on capital, IHCL’s fee engine, 360 ONE’s annuity, are structurally better positioned than the ones that convert it into working capital and thin distributor margins.

Part 5: A Framework, So You Are Not Just Buying a Story

Here is the part that matters more than any single name, because names change and frameworks last. If you take one thing from this piece, take these five questions. Run any “premiumisation” idea through them before you form a view.

One. Is the affluent cohort’s spend in this category structurally rising, or are you assuming it? Travel, watches, prime homes, private wealth, premium spirits: yes, the tailwind is documented. Be suspicious of any pitch that asserts the tailwind rather than showing it in category data.

Two. Which kind of premiumisation is this, and have you priced it accordingly? Pure luxury (narrow, high-margin, wealth-effect-linked) or mass-premium trade-up (broad, mix-shift, input-cycle-linked)? The two deserve different multiples and different downside assumptions. Most mistakes are made by paying luxury prices for mix-shift durability, or by treating a fragile luxury TAM as if it were a broad staple.

Three. Where does the value accrue: brand, distributor, or enabler? The brand owner with pricing power (IHCL, Radico’s premium labels, an ultra-luxury developer) beats the distributor with thin margins (Landmark) beats nothing. The enabler that earns an annuity on the whole pool (the wealth manager) is a distinct and often superior third model. Always ask who actually keeps the incremental rupee.

Four. Is the mix-shift showing up in the reported numbers yet, or only in the strategy slides? This is the walk-the-talk test, and it is where rigour lives. Understand first what you are testing for. Premiumisation is really a margin story wearing a growth costume: the topline can grow for all sorts of dull reasons, but the thing that re-rates a stock is the richer mix dropping through to margin. So the tell is specific. Real premiumisation shows up as two lines moving together in the actual segment table: the premium share rising and the margin rising with it. Radico’s “prestige and above” share of IMFL value, Heritage’s value-added percentage, Oberoi’s booking value climbing while units fall, Ethos’s 71% luxury mix, IHCL’s growing fee income. You can check every one of these in a filing. A premiumisation story that lives only in a management quote, with no fingerprint in the mix and no lift in the margin, is a story, not a thesis. And beware the reverse trap that Ethos illustrates: a genuinely rising premium mix whose profit is being swallowed by working capital and expansion before it reaches the shareholder. Rising mix is necessary. It is not sufficient.

Five. What does the valuation assume, and can the wealth-effect flywheel keep spinning? This is the one that governs all the others. Almost every name here trades richly, from Ethos in the mid-sixties to spirits names in the eighties on earnings. That is tolerable only if growth persists, and growth persists only while the affluent keep feeling wealthy, which depends on equities, gold, and property staying broadly buoyant together. A sustained, simultaneous reversal in those asset classes is the systematic risk that would hit every business in this piece at once, because it strikes the confidence of the single customer they all share. Size positions as though that reversal is possible, because it is.

That fifth point is not a footnote, and it deserves a name, because it is the most expensive mistake available in this theme. Call it the one-bet-six-times problem. An investor who owns the luxury hotel, the watch retailer, the premium spirit, the wealth manager, and the luxury developer feels diversified: five sectors, five businesses. They are not diversified. They own the same bet five times, because all five sit downstream of one factor, the affluent Indian continuing to feel wealthy, which is to say asset prices staying high. On the way up, that correlation is invisible and feels like skill. On the way down, it is the whole story, and it arrives everywhere at once. Premiumisation is a genuine, multi-year, structural force. It is also, at today’s prices, a consensus one, and consensus stacked on a single shared risk factor is exactly the configuration that punishes the investor who mistook a strong narrative for a margin of safety. The partial antidote is the unglamorous second column, the mass-premium names selling essentials, whose demand does not evaporate when the market falls.

Part 6: What I’m Watching

On the macro: the trajectory of Indian equities, gold, and prime property together, because that trio is the fuel for the wealth effect. Any sign of a broad, sustained reversal changes the risk on every name here at the same time. Also, the pace of the affluent cohort’s growth as fresh income-tax and consumption data land through FY27.

On the operators: whether the mix-shift keeps showing up in the reported segment data, IHCL’s management-fee income share, Ethos’s luxury mix and, critically, its working-capital cycle, Radico’s P&A salience, Oberoi’s value-per-unit, 360 ONE’s recurring-revenue growth and net flows, Heritage’s VAP percentage against milk-procurement costs. The slides are cheap; the segment tables are the truth.

On regulation and trade: state excise moves in alcohol (a Karnataka-style deregulation can re-rate the group; a punitive state move can gut it), the phased benefits of the India-EFTA and India-EU trade agreements for luxury imports and watch retail, and any change to import duties on luxury goods.

On valuation: simply whether growth keeps pace with the multiples. At these prices, a single quarter of decelerating premium mix, in any of these names, tends to be punished disproportionately, because so much of the future is already paid for.

The Bottom Line

The strangeness we opened with, global giants selling their best-selling brands, is not strange at all once you see the shape of the market. India is becoming two economies. One is broad, price-sensitive, and growing slowly. The other is thin, roughly the population of a mid-sized country, compounding at more than ten times the pace, and getting wealthier not just through income but through the assets it already owns. The businesses in this piece are, each in their own way, deliberate bets on that second economy: the brand that earns a fee on it, the retailer that curates scarcity for it, the developer that sells it fewer, dearer homes, the wealth manager that earns an annuity on its rising pile.

The opportunity is real and structural. So is the fact that most of it is now well understood and richly priced, and that nearly all of it leans on the same wealth-effect flywheel. The investor’s job in this theme is not to discover that premiumisation exists; everyone knows that now. It is to tell the businesses converting the trend into durable return on capital from the ones merely standing in its path, and to pay a price that survives the flywheel slowing down. The framework above is how I try to do exactly that.

Read the mix-shift before you read the multiple. That is the whole game.

Coming in Part Two

There is a second pipe feeding this same affluent customer, and it is the one that has quietly rewired how India consumes: the new-age internet platforms. Quick commerce delivering to the top of the pyramid in ten minutes. Food delivery, online travel, digital brokerages, the platforms that sit between the affluent Indian and almost everything she now buys. They benefit from the identical income tailwind, but they play by entirely different rules, cohort economics, take rates, and the ferocious question of who actually captures the profit in a platform war. Part Two is about how to think about them without getting burned by the hype. Subscribe so you do not miss it.

See you at the next inflection point.

Karnik

Inflection Point Research

Disclosures

Disclosures and Disclaimer

This article is published by Inflection Point Research, a SEBI-registered Research Analyst (Registration No. INH000023807).

Analyst holding. The analyst (Karnik Shah) does not hold any position in any of the securities discussed in this article.

SEBI Disclosure: This report is published by Inflection Point Research, a SEBI-registered Research Analyst (Registration No. INH000023807). The analyst may have personal investments in the securities discussed, as disclosed above. This content is for informational and educational purposes only and does not constitute investment advice, an offer to buy or sell securities, or a recommendation of any specific investment action. There are no price targets or buy/sell calls in this piece by design. Readers should conduct their own due diligence and consult a SEBI-registered adviser before making investment decisions. Past performance is not indicative of future results. Equity investments are subject to market risks. All financial data in this piece requires independent verification from primary sources (BSE/NSE filings, company results, Screener.in) prior to use in any investment decision.

GET TWO FREE IPR SAMPLE REPORTS

The kind of structural thesis you just read is where Inflection Point Research starts. Where it goes next, for subscribers, is into specific, forensically-screened names with the full framework applied company by company. Two free sample reports, on Gravita India and Kalyani Cast Tech, show exactly what that finished research looks like.

Get Free Samples: https://www.inflectionpointresearch.in/free-sample-research-reports

No payment required. SEBI Reg. INH000023807.

Hello friends,

A quick one this week; the analysis lives in the video below rather than on the page.

Venus Pipes (NSE: VENUSPIPES) is a stainless steel pipe maker from Dhaneti in Kutch that has grown revenue from 119 crore to 1,167 crore in 7 years, nearly 10 times. But the number that made me stop was a split inside the factory: last year the seamless line ran at over 95 per cent of capacity, effectively sold out, while the welded line ran at 60 to 65. Same company, same steel, broadly the same customers. That gap turns out to explain most of how money is made in the pipe industry, so the first third of the video is the industry map itself: how pipes are made, which welding families exist, who the listed players are in each lane, and why margins widen as you climb the materials ladder from carbon steel to titanium.

Then the company. Four first-generation promoters, 2 of them traders who crossed the counter after 2 decades of selling other people’s pipes. An IPO that was entirely fresh issue, no offer for sale. A piercing line in 2023 that took quarterly operating margins from 12 per cent to 20 in a year, which is what backward integration looks like when it works. And a climb up the value ladder that is still underway: fittings live this May, a titanium and hygienic tubes line live last year, and now a ₹70 crore entry into spooling, backed by a ₹185 crore letter of intent from a data centre customer.

And the honest tension, which I spend real time on: operating margin has fallen 2 years running, 18 to 17 to 16.3 per cent, while revenue set records. In the video I take that number apart properly. It is struck before interest and depreciation, so the pressure sits in the operating line itself: export mix, under-absorbed new capacity, and raw material with roughly 6 months of inventory on the shelf. Management has guided it back towards 17 per cent this year and 18 by FY28. Whether you believe that guidance is, in my view, the entire question the current price is asking.

Watch the full breakdown below

Read the full depth from Inflection Point Research

If you want the complete research, initiation reports, models and position-level detail, the best place to start is a research plan. Subscribe here: https://www.inflectionpointresearch.in/pricing-plans/list

Or, if you would like to see the quality of the work first, you can read two full sample reports (Gravita India and Kalyani Cast Tech), free and with no payment: https://www.inflectionpointresearch.in/free-sample-research-reports

Disclaimer

This article and the accompanying video are published by Inflection Point Research, a SEBI-registered Research Analyst (Registration No. INH000023807). The analyst holds a position in Venus Pipes & Tubes Ltd, the subject company; the last transaction was more than 30 days prior to publication, in compliance with Regulation 16 of the SEBI (Research Analysts) Regulations. This content is for information and education only and is not investment advice, an offer to buy or sell securities, or a recommendation to replicate the analyst’s position. Registration granted by SEBI and certification from NISM in no way guarantee performance of the intermediary or provide any assurance of returns to investors. Investment in securities market are subject to market risks. Read all the related documents carefully before investing.