A few days ago I shared my thoughts on the 9th IPO anniversary memo. If you haven’t read it, please do.

This article will focus on why Monee may need to IPO — or, in other words, why it might be reasonable to sell off part of the business for a few billion dollars.

Subscribe now

Monee IPO

Frankly, Monee doesn’t need to IPO.

Sea Limited generates plenty of cash through Shopee and Garena, and it could probably continue funding Monee’s growth for quite some time.

But I think an IPO might still be one of the best things management can do if Sea’s goal is to build an even larger company over the long term.

And it isn’t just about raising capital or unlocking Monee’s valuation.

I think there is a much bigger question here: How profitable is Monee really, once we account for everything Shopee and Sea Limited provide?

And conversely, how profitable would Shopee be if Monee had to operate as a completely independent business?

I think the answers will be quite interesting.

But first, we need to understand why Monee needs so much capital in the first place.

The Beauty of Banks and Fintech

Monee’s growth is extremely capital-intensive.

Currently, roughly 90% of its loan book is funded by Sea Limited’s own balance sheet. And the return on investment on that loan book is ultmost around 20% per year.

That is obviously not a great use of cash.

Lending money by itself is never a good business.

What makes banks great businesses is that they lend out other people’s money to make money.

And what makes fintech companies theoretically even more appealing is that they don’t even need to carry the lending risk.

They can simply build their own distribution systems, develop credit models, and make money by connecting borrowers with lenders.

That is basically Monee’s ultimate goal, which I discussed in my Monee piece earlier this year.

And Monee’s balance sheet is SUPER UGLY

If you think of Garena, Shopee, and Monee as three separate businesses, and imagine Monee operating independently with Shopee as its main source of funding, the picture becomes UGLY.

Monee would be way over-leveraged, with most of its funding effectively coming from Shopee.

That is nowhere near a healthy capital structure for a standalone business.

And I think there are three reasons why Monee has ended up in this position.

a) The local debt markets are not deep enough.

Since 2021, Sea’s management has repeatedly talked about diversifying Monee’s funding sources and reducing reliance on the group’s balance sheet.

But progress has been rather disappointing.

In Q4 2023, off-balance-sheet loans represented around 20% of Monee’s total loan book. That proportion has since fallen to roughly 10%.

To be fair, the absolute amount of off-balance-sheet loans has nearly doubled over the same period.

The problem is simply that Monee’s credit business is growing much faster.

And I don’t think this is entirely management’s fault.

Even U.S. fintech companies such as LendingClub and SoFi, which operate with the world’s most advanced capital markets behind them, have struggled to offload subprime loans and eventually ended up becoming banks themselves.

And Monee is basically playing this game on hard mode.

The small-dollar, short-term, poor-people BNPL loan market only makes it worse.

b) Deposits are growing, but not fast enough.

Monee’s banking business has actually been growing at a decent pace.

Deposits grew 40% in 2025, and its digital banking services now have more than 40 million members. Usage has also increased quite considerably.

But the deposit base still isn’t keeping up with Monee’s lending ambitions.

I don’t know exactly what happened, but I suspect timing played an important role.

Monee started building its banking business a few years later than its credit business.

On top of that, Sea cut Monee’s marketing budget by more than 85% during 2H22 and 1H23, which contributed to the relatively stagnant deposit growth during that period.

Then, around 24H2, the Off-Shopee SPayLater macro environment experienced a major breakthrough.

The rollout of national QR-code systems across several countries created new opportunities for off-platform payments and increased the urgency of acquiring active credit users.

So while Monee’s deposit-taking business was still catching up, its lending opportunities were expanding rapidly

c) Monee is racing against time.

As I mentioned in my earlier piece, Monee is essentially building two separate credit businesses at the same time.

The first is small-dollar cash lending.

This is fundamentally a user-acquisition and scale game. Monee wants to become one of the top three credit products that an unbanked or underbanked consumer turns to when they need money.

The second is off-Shopee SPayLater.

Off-Shopee SPayLater is a network-effect business.

Monee is basically trying to become the American Express of Southeast Asia.

It building a network connecting merchants, consumers, and lenders, eventually becoming a default choice for revolving credit.

It is competing for consumer and merchant relationships with payment networks such as Visa and Mastercard.

These two credit businesses require scale, and both needs to grow very fast to sustain the current advantage.

Now I’m not even sure Monee’s current pace of loan-book expansion is sufficient.

And even if deposits grow at the same speed as the loan book, and even with Garena and Shopee generating more than 1 billion in cash flows per quarter, Monee would eventually face a significant funding deficit if its credit business continues growing exponentially.

I fully expect Sea to follow a path similar to MercadoLibre in LATAM, increasingly relying on external borrowings to fund loan-book growth.

By the end of 2Q26, MercadoLibre’s external borrowings had already surpassed $10B, because most of its deposits are not legally allowed to be used to fund loans.

But I don’t think external borrowing can be a long-term solution.

At some point, even with Shopee’s weight behind it, Monee will still need fresh capital to continue expanding without becoming excessively leveraged.

Sea Limited’s Ambition

One thing I have always found interesting about Sea Limited is its approach to building businesses.

Sea has never been afraid to expand its circle of competence.

Throughout its history, management has pursued vertical integration, launched tons of smaller projects in parallel, and then put serious resources behind whichever ones showed the most promise.

Free Fire, Shopee, and Monee were all born against this backdrop.

And Sea also places enormous importance on flexibility and capital efficiency.

Its logistics strategy is a good example.

Sea has generally use leasing rather than committing large amounts of capital to ownership, even when leasing are obvious the more expensive over the long term.

Management has repeatedly shown that it is willing to sacrifice profitability in exchange for greater flexibility and capital efficiency.

So could Sea keep funding Monee with cash generated by Shopee and Garena? Definitely!

The only thing it needs to do is slow down Monee’s growth.

But I don’t think that is the optimal decision. There is reason why promising startups willingly give up equity to raise more capital.

Sometimes having enough funding to capture a much larger opportunity is worth far more than preserving 100% ownership. And Monee may be approaching exactly that point.

And there is another question worth asking in the spin-off process: If Monee cannot operate independently of Shopee, how valuable would Monee’s shares really be?

Monee’s Real Profitability

I have always considered Monee an extension of Shopee’s distribution network. And that raises an interesting question.

If Shopee is responsible for creating most of the value, shouldn’t Shopee also be the one capturing most of the profits?

And conversely, how profitable would Shopee actually be if Monee operated as a completely independent business? I think Sea’s current reporting are understating Shopee’s true profitability.

Back in 2020 and 2021, Affirm had to commit hundreds of millions of dollars annually to Shopify and Amazon just to gain access to their distribution networks.

And even today, Shopify and Amazon combined are estimated to account for “only” around 35~40% of Affirm’s total GMV.

Monee’s reliance on Shopee is way higher than Affirm’s reliance on Shopify or Amazon.

On-Shopee SPayLater still contributes around 80% of overall GMV, and most of its users or lenders were likely first acquired through Shopee rather than its standalone ShopeePay app.

I don’t think Monee currently pays Shopee any distribution fees, nor does it pay Sea, as its parent company, any interest for using its cash balance.

Essentially, Monee benefits from Shopee’s distribution and Sea’s balance sheet without necessarily bearing the full economic cost of either.

That brings us back to the real question:

What is Monee actually worth as a standalone business? And how much of its reported profitability would remain if it had to pay market rates for distribution and funding?

I think Sea’s current segment reporting are understating Shopee’s true economic contribution.

Of course, charging Monee distribution fees wouldn’t magically create additional consolidated profits for Sea Limited. Those fees would largely represent an internal transfer of economics.

But it could make the value of Shopee’s distribution much more visible to investors.

And I think that distinction matters enormously.

Because the question is not simply whether Monee deserves a premium fintech valuation.

It is whether the market is correctly valuing the economic contribution of both Monee and Shopee.

A Monee IPO could potentially help answer that question.

And perhaps discovering Shopee’s true profitability could be just as valuable as unlocking Monee’s own valuation.

Standing on Its Own Two Feet

In the recent memo, Forrest shared:

We believe Monee is approaching the point where it can stand on its own two feet and begin its own race. Going public could give it the resources and the room to go further and faster than it can today.

I think three projects show that Monee is preparing to stand on its own two feet:

a) The ShopeePay app
b) The digital banking business
c) Off-Shopee SPayLater

To me, these three are attempts to build customer relationships and distribution channels outside Shopee’s ecosystem.

And management is starting to spend aggressively to make that happen.

In the latest reported quarter discussed here, Monee’s sales and marketing expenses increased by more than 140% year-over-year.

Instead of relying almost entirely on Shopee to acquire users, Monee has started pouring more money into external marketing.

And the results look pretty good.

By the end of 2025, off-Shopee SPayLater loans had grown more than 300% year-over-year.

In 2Q26, more than 35% of SPayLater usage in Malaysia was already happening outside Shopee!

Meanwhile, judging by app downloads, user acquisition for both ShopeePay and Monee’s digital banking business also appears to be accelerating.

(Note: One reason Malaysia has the highest off-Shopee penetration rate is it relatively high credit card penetration, although it is still not as high as Singapore or Taiwan, where consumers are less inclined to use SPayLater on Shopee)

How Much Is Monee Actually Worth?

Although Monee still relies heavily on Shopee, its operations are already surprisingly huge.

It operates across several financial verticals, and some of its businesses are comparable in scale to leading fintech companies.

SeaBank was launched in 2021 (Monee’s digital banking business) has become Indonesia’s largest digital bank by deposits and users, and it is still growing rapidly.

It has a lot of similarities with Nubank in its early years.

SPayLater’s current GMV is also comparable to Affirm’s FY2025 GMV, assuming SPayLater’s penetration is around 18% and off-Shopee SPayLater accounts for around 25% of on-Shopee GMV.

But the crucial difference is that both Affirm and Nubank have always stood on their own two feet. Their ability to operate independently is a major reason they have commanded valuations of roughly $20B and $60B, respectively.

Monee scale is impressive, but scale alone doesn’t justify a premium valuation.

If Monee wants investors to value it like a major standalone fintech business, it needs to prove that it can be highly profitable without depending on Shopee for distribution and Sea Limited for funding.

And I think the next few years will be the real test.

Closing Thoughts

I personally think Monee has tons of potential. I think off-Shopee SPayLater will become much bigger than on-Shopee SPayLater in the next few years.

Its banking business could also become the equivalent of Nubank in the region, and I expect Monee to eventually export its banking and payment services into other regions where Shopee has no presence.

I would also like to add:

Perhaps Monee will eventually want to spin off SPayLater as a separate business.

After all, banking and BNPL are fundamentally different businesses. A traditional banking business is relatively asset-heavy and requires continuous capital to support loan-book growth.

A BNPL business, on the other hand, can theoretically operate as a much more asset-light, network-effect-driven platform, particularly if it can transfer the underlying credit risk to third-party lenders.

And businesses with those characteristics can potentially have a much higher valuation multiples.

As Monee continues expanding its banking operations, I think the differences between these two businesses will become increasingly apparent. A separate SPayLater business might eventually make sense.

All in all, Monee’s IPO could solve many of the capital issues it is facing. And, perhaps more importantly, it could provide much greater visibility into Shopee’s real profitability, which would be super good for Sea Limited’s valuation.

So while Monee doesn’t necessarily need an IPO to survive or continue growing, I think an IPO could make a lot of sense for Sea Limited as a whole.

Ultimately, this is not just about unlocking Monee’s value. It is also about giving Monee the freedom to stand on its own two feet, while allowing Shopee’s real economic value to become more visible.

I’m excited to see how management approaches this over the next few years.

Subscribe now

If you enjoy my writing and want to support my publication, you’re always welcome to buy me an (expensive) coffee here!

A few things have changed, and I don’t plan to launch a paid Substack anytime soon. But if you’d like to see more of my personal opinions, a pledge support on Substack will also show me your intention very well!

Thank you for reading!

Here’s the Sea Limited 2026.10.20 IPO anniversary memo if you haven’t read it.

This piece was originally published on my Twitter with a few links attached.

a) Monee IPO and the capital to grow:

As I said in my earlier Monee piece, Shopee and Monee have been competing for growth capital over the last few years. Monee’s current deposits are far from sufficient to fund its current growth.

I also didn’t expect Monee to be planning an IPO, but it does make sense for them to sell some Monee shares to get more funding.

I think listing Monee (with SE still holding the majority of the ownership) could also build more trust in the funding market, as it would reveal a lot more financial information.

Plus, standalone businesses focused on a single vertical usually get better valuations!

Don’t forget that Monee digital banking business has more than 40M clients and more than 40M active creditors!

And it’s still growing very fast.

I’ve always wondered whether Monee can enter new markets without Shopee’s presence. Perhaps now, as the banking business enters a more mature stage and Monee builds out more playbooks and products that don’t rely on Shopee’s distribution, it’s time for them to try new markets.

Source: NVIDIA AI Day Singapore 2026

b) Sea Cloud:

This is the main reason for yesterday’s crash. It basically signals huge CAPEX and a spending spree over the next few years.

As I shared a few days ago, Sea Limited will produce enormous profits in 2027 and 2028 if TikTok Shop doesn’t renew the pricing war.

And it’s basically impossible for a group of “young” founders who have successfully built a series of great products across different countries, with enormous capital from the capital markets, to suddenly become conservative nice boys within the AI trend.

That’s not the Sea Limited I know, and I’ve also expressed some frustration at seeing them not being aggressive enough.

Frankly, I like the idea of building a great new, big organization. SE has always had a series of experiments running in parallel. That’s basically what “other services revenue” is in the earnings release.

SE is always doing a ton of experiments, and that’s what makes them successful in the first place.

But I think they handled the announcement awfully, terribly. I think announcing a brand-new, capital-intensive project without a detailed explanation in the press release is simply awful. How could we, as investors, even think about SE‘s valuation as a group?

But to be fair, I do see the potential for SE to become a decent cloud provider. I believe they do have decent and differentiated know-how in underdeveloped nations and the Global South.

First, with eCommerce, live commerce, and video commerce, they’ve already built great infrastructure to support huge spikes in traffic.

With fintech, they’ve already built infrastructure with great cybersecurity and local compliance muscle.

With Garena, they specialize in real-time streaming on low-cost Android phones.

Now they’re partnering with dozens of public cloud providers around the world and also have quite a data center footprint globally. They now manage 40+ data centers around the world (a lot of them are colocation facilities, I believe).

They’ve already built out a distributed network of computing infrastructure that is also very cost-oriented, which I believe is quite differentiated from the U.S. giants, which are more centralized and performance-oriented.

I think they do have a decent chance of serving developing nations that are rather fragmented and cost-sensitive.

In the future, when hardware faces fewer constraints, I believe AI/computing will become less and less CAPEX-intensive, and the ones that capture more of the value will be in the application layer.

And the future won’t unfold equally everywhere. AI is obviously spreading everywhere in the U.S., but it will take quite some time for the trend to spread to less wealthy nations. And when that time comes, they’ll also need a lot of localization.

And localization has never been the strength of global giants, but it’s one of Sea Limited’s core specialties. They’re super good at building local solutions for every single small and mid-sized market.

And as AI becomes more and more regulated, I believe localization will become more and more important. Every government will want data to stay on its own soil.

I think Sea Limited has zero chance of winning if it competes head-on with the tech giants in developed nations. But if they play a cost-oriented and localization game, they have a decent chance of winning.

That doesn’t mean I think this new adventure has zero risk. I think it’s still something very concerning, especially since they’ve never been successful in B2B businesses. I don’t think they have any of the B2B muscle yet.

Sea Limited did try B2B businesses before with enterprise software (Sea Talk) and POS (Ocha). They simply weren’t successful.

It takes very different genes to serve consumers versus enterprises.

All in all, I’m decently optimistic about the business. I feel that SE is back! The aggressiveness is back!

But I’m also worried about the valuation prospects. As I shared earlier this year, I did buy back a small number of shares near the lows earlier this year and will buy more if it continues to trend down.

I’m planning to stick to the plan.

Subscribe now

If you enjoy my writing and want to support my publication, you’re always welcome to buy me an (expensive) coffee here!

A few things have changed, and I don’t plan to launch a paid Substack anytime soon. But if you’d like to see more of my personal opinions, a pledge support on Substack will also show me your intention very well!

Sea Limited delivered a strong quarter, with revenue growing 46.6%. In this article, I will mainly focus on what happened during the quarter. In the future, I will release a more in-depth article on Shopee, including some commentary on MercadoLibre and other e-commerce companies.

You can also check out my first two articles on Sea Limited: What’s Shopee Strategy in 2026 and Sea Limited’s Monee in 2026: What It’s Doing and What to Expect Going Forward. Both articles received very positive feedback. I think they are among the best pieces you can find on Sea Limited’s strategy. The things I discussed in those articles played out this quarter, and I believe many of those points will continue to matter in the coming quarters, and potentially for years.

Now, let’s focus on this quarter:

The things in this article

• Why did Shopee beat revenue? Was there a change in strategy?

• Shopee’s ads growth and the impact of Shopee GMV MAX

• The current stage and progress of Shopee VIP — I still expect it to surpass 25M users by year-end

• Why fulfillment matters for product selection in remote areas

• The difference between Grab and Shopee quick commerce strategies

• My concerns about the Monee acquisition and credit deterioration

• Why Brazil is now Shopee’s most important market and what it’s current growth

Besides, I think the management team is deliberately slowing the growth of both Shopee and Monee because of the fear of a drop in net income. I think the management team now does really care about “financial discipline” in fulfilling the guidance.

Shopee beat on revenue

This quarter, Shopee’s GMV growth was 30%, pretty much in line with my expectation. The real surprise was that marketplace revenue grew 44%, up from 36% last quarter. I think this was mostly due to the unexpected increase in take rates compared with the previous year.

The table shows that take rates increased by 0.26 percentage points QoQ, compared to a 0.26 point decline a year earlier. On a YoY basis, take rates were up 1.17 percentage points, which is a much larger improvement than in any quarter of 2025.

I think two factors drove the take-rate improvement. First, Shopee appears to have a much stronger ad business now. Second, management seems to have shifted more promotional discounts into logistics subsidies.

The improvement in Shopee’s ad take rate

Shopee started to focus more on ads in 2024Q2, as management mentioned in the earnings call:

One area we are placing greater focus on is improving our ad take rate. Currently, our ad take rate is lower than the industry average we observe in more mature e-commerce markets.

Over the quarter, we have made it easier and more attractive for sellers to join our ad platform. We also have a dedicated tech team working on improving our ad bidding algorithms to help sellers achieve higher returns from their advertising spend.

From then on, Shopee’s ads growth rate has been on an accelerated path.

2024Q2: ~ started to focus on ads
2024Q3: ~40%+ YoY (GMV MAX was officially launched)
2024Q4: ~50%
2025Q1 : ~56%
2025Q2: ~70% (ads take rate was around 2% GMV, + 70 basis point)
2025Q3: ~75% (+ 80 basis point)
2025Q4: ~75% (+ 80 basis point)
2026Q1: ~80% (+ 90 basis point)

In 2026Q1, ad spend increased by over 80% year-on-year, and the ad take rate also increased by more than 90 basis points. This represents another acceleration from the 75% growth last quarter.

The acceleration in growth was largely attributable to the new ad product “GMV MAX” that was launched in 2024Q3. GMV MAX has meaningfully shifted Shopee ads from a more traditional seller-controlled bidding, placement, and keyword-management system to a Shopee-controlled automated performance and ROAS-based ad system.

The shift to an automated performance system isn’t unique to Shopee. Both China’s eCommerce giant Pinduoduo and the world’s third-biggest ad network, AppLovin, were proponents of this method.

Shopee’s ad take rate should soon surpass or may have already surpassed 3% of GMV. The mid-term target would be 4–5%. I think in the long term, the take rate might be able to surpass 7% of GMV. There is still very sizable room for Shopee’s ads to grow.

The shift from discounts to logistics subsidies

If we look only at the take-rate increase in 2026Q1, it is easy to attribute most of the improvement to Shopee’s strong ad growth. But if we look closer, we see a different story driven by logistics subsidies.

The table shows that the 1.17 percentage-point increase in take rates was driven by a 1.95 percentage-point increase in the core marketplace take rate, partially offset by a 0.78 percentage-point decline in value-added services.

The latest quarter shows an interesting contrast. Both 2026Q1 and 2025Q1 had similar seasonality, with Ramadan and Chinese New Year both falling within the quarter. However, the QoQ change in Sales and Marketing expense was very different. In 2026Q1, Sales and Marketing expense grew 1.9% QoQ, compared to a 9% decline last year.

I personally think this is highly attributable to the change in promotion strategy. Shopee recognizes revenue on a net basis, so both platform discounts, e.g., 15% off coupons, and free shipping coupons reduce its revenue on the transaction. If the revenue isn’t enough to offset the subsidy, the excess amount becomes Sales and Marketing expense.

This is the reason take-rate improvements usually happen in non-promotion quarters, despite Shopee charging higher commissions and getting much higher usage of its ad offerings in promotion seasons.

What happened this quarter is interesting. Despite Shopee increasing its Sales and Marketing expense sequentially, its core marketplace revenue still grew sequentially by 0.4%, compared to a drop of 0.1% last year.

I personally think this increase in both core marketplace take rates and Sales and Marketing expense reflects the new strategy Shopee is pursuing — an even stronger emphasis on platform-wide free shipping standards.

The first country that went through this shift was Vietnam in 2025Q2. Before this shift, sellers could decide for themselves whether to pay a higher commission to join the shipping subsidy program. After the shift, nearly all sellers were forced to join the platform’s unified shipping subsidy program.

The unified shipping subsidy program seems to be a success. In 2026Q1, Shopee also forced all sellers in both Taiwan and Brazil to join a similar program. I expect the platform-wide unified free shipping push will continue to expand to other countries.

I will share deeper implications and more details in a future article.

Subscribe now

Shopee VIP surpass 10M subscribers

In my previous article What’s Shopee’s strategy for 2026? I compared Shopee VIP with two other types of eCommerce subscription models: free-shipping-led programs such as Amazon Prime and Coupang WOW, and loyalty- or discount-led programs such as Alibaba’s 88VIP and JD’s JD PLUS.

I also shared my model for Shopee VIP reaching 25M paid subscribers by the end of 2025, and eventually surpassing 100M paid subscribers over the longer term. I still hold that target, even though growth was “only” 40% in 26Q1. In my view, Shopee VIP’s growth rate remains on track for the 25M subscriber goal.

I think Shopee has just passed the stage of optimizing for retention in some countries, where retention is now averaging above 80% by the end of this quarter. Since April 2026, the annual subscription plan has started to roll out at full scale in Taiwan, while other countries have also shown clearer rollout progress. Indonesia and Brazil remain the only two countries without an annual subscription plan. In several other markets, the plan is still limited to selected users and has not yet reached a full-scale launch.

Some More Color on Shopee VIP

Shopee VIP members now contribute around 20% of GMV across Asia, and I believe they will eventually account for the majority of Shopee’s overall GMV.

I view the membership program less as a free-shipping program and more as a loyalty program designed to increase purchase frequency.

Over the past few years, Shopee has repeatedly used lower shipping thresholds to pull forward order volume and further reduce logistics costs. Lowering the cost to serve is Shopee’s way of strengthening its competitive moat.

The current investment cycle follows the same logic. Even without VIP, Shopee would still lower the shipping threshold eventually. Now, Shopee is just giving paid VIP members first access to these shipping benefits.

Shopee’s real goal is to build up a loyalty program in order to offer price discrimination between members and non-members. At this stage, the subscription fee itself is not important. It is mainly a filter to identify higher-intent customers.

For now, Shopee literally doesn’t care about the subscription fee. Take the market I’m from as an example. In Taiwan, Shopee not only gives a free trial to every user. It also gives out US$1.90 worth of cash back to every single subscriber and, at the same time, gives out tons of super generous subscriber-exclusive discount coupons. And most importantly, the subscription price is only US$1.90. This means that the cash back itself will immediately pay back the subscription on its own.

Singapore’s VIP program looks similar to Taiwan’s. Both directly offer very generous cashback. In other countries, the structure looks a bit different: it is more like giving users the option to buy a large bundle of discount and free-shipping coupons at a very low price.

My guess is that Singapore and Taiwan are wealthier markets with less geographic complexity, so Shopee does not need to customize the VIP benefits as much.

In other countries, the offerings should be more localized. For instance, in markets where ShopeeFood was operating, there seems to be a much stronger bundle of quick-commerce benefits. Urban users benefit more from instant-delivery coupons, while users in the countryside likely get more value from reduced shipping fees or faster delivery.

One more thing to note: the subscription also includes priority customer service and benefits from off-Shopee partners such as ChatGPT and Duolingo. In countries with longer-term subscription plans, the fee discounts can be very large. In Taiwan, the yearly plan is roughly 50% cheaper than paying monthly. Indonesia is even more aggressive: the 6-month plan is 66% cheaper than the monthly plan, bringing the total cost of six months of VIP to only about US$1.70.

The Road to Shifting VIP Costs to Sellers

Since the beginning of this investment cycle, Shopee’s marketplace gross margin has clearly come under pressure. In the latest quarter, marketplace gross margin dropped by around 4.5 percentage points year over year. Frankly, I expected an even bigger drop; I suspect the management team is implementing financial discipline in order to control the drop in Shopee EBITDA.

But, the good news is that Shopee seems to be shifting more of the VIP promotion costs to sellers.

In Taiwan, for example, the wording of Shopee VIP promotions has changed from broad category-wide discounts to more brand-specific discounts. This suggests that Shopee may be moving from platform-funded incentives toward more seller- or brand-funded promotions.

The pictures are from Shopee Taiwan’s VIP benefits. Currently, I think the promotion cost offloading is still in a very early stage. In Taiwan, small sellers haven’t gotten access to offer VIP-member-specific discounts. Shopee should still be the one carrying almost all of the cost.

Fulfillment, Selection, and Fast Shipping

Shopee’s ambition is to become the all-in-one shopping app. However, SPX is still mainly strong in C2C logistics. For Shopee to serve more customer needs, it needs to build capabilities beyond traditional C2C delivery.

Fulfillment and quick commerce are the two key pillars Shopee is pursuing to fulfill the goal.

Fulfillment and the selection problem in remote regions

In the previous quarter’s earnings call, management set a goal to double fulfillment order penetration by year-end. This quarter, Shopee’s fulfillment orders increased 25% sequentially. In Asia, more than one-third of parcels fulfilled by Shopee were delivered by the next day.

Interestingly, the ones that benefit the most from fulfillment aren’t the ones that live in the big cities. This is because most sellers already store their inventory near or inside major cities, and C2C logistics are already capable of sending packages between cities within 24 hours. So as a results, most people in the capital zones in Southeast Asia can already receive the packages within 48 hours, or even as short as 24 hours.

The bigger opportunity is in remote areas. Today, much of the platform’s inventory is still concentrated near major cities.

Because shipping to remote regions is costly and time-consuming, buyers in those areas often choose sellers within the same region to save time and money. That means their product selection is much more limited compared with buyers in cities.

With stronger fulfillment capability, Shopee can “forward deploy” products to warehouses closer to remote buyers, using cheaper and more efficient transportation methods.

“Forward deploy” is the key. Instead of relying on airplanes or small trucks to rush individual parcels, Shopee can move inventory in bulk ahead of demand. Once the products are closer to customers, delivery becomes both cheaper and faster.

In short, fulfillment allows Shopee to improve product selection for buyers in remote regions while also lowering delivery costs and shortening delivery times. That could unlock a lot more demand!

Quick Commerce — Shopee vs. Grab

Quick commerce is another focus that Shopee is pursuing. In the latest quarter, order volumes for instant delivery have increased by 35%, with costs reduced by around 20%. By the end of 26Q1, Shopee had around 7,000 offline stores available for instant delivery services.

This puts Shopee directly into one of Grab’s core territories. Grab, as a ride-hailing and food delivery giant, already has more than 10% of its orders coming from quick-commerce-related services.

What makes this competition interesting is that Shopee and Grab are approaching quick commerce from fundamentally different starting points.

For Shopee, quick commerce is mainly about improving the delivery experience for products that are already sold on its platform. For Grab, the focus is more on expanding SKUs and merchant coverage to better serve users using the same underlying on-demand delivery network. This leads to two very different ways of building the business.

Grab’s model is centered around fast delivery. Shopee’s model is more segmented. It basically divides products into three categories:

1. Urgent needs: delivered within 60 minutes

2. High-intent needs: fulfilled within 2~6 hours / same day

3. Other products: shipped through the traditional e-commerce logistics system

For urgent needs, orders are usually fulfilled by nearby local stores and delivered by ShopeeFood riders. The process is very similar to a normal food delivery order.

For high-intent needs, Shopee can send a ShopeeFood rider to the seller’s location, warehouse, or logistics network node, and then deliver the package directly to the buyer. This is still similar to food delivery, but the delivery distance is usually longer, and parcels are often sent in small batches.

For other products, Shopee can continue using its traditional logistics network. In reality, if buyers can wait more than 6 hours and the inventory is already stored within the same city, traditional e-commerce logistics should be able to deliver the package on time.

This is one of the main reasons same-day and next-day delivery services often have strict cut-off times. The quick commerce logistics layer that Shopee is building sits above its traditional logistics system. It gives users more flexibility and serves demand that falls outside normal cut-off times.

Shopee’s quick commerce push is obviously a huge threat to Grab. Shopee is giving out huge logistics subsidies through the VIP program. At the same time, it could also create an opportunity for Grab. If Shopee helps educate users and merchants around quick commerce, Grab may still be able to capture part of the ecosystem that Shopee is building.

In 2025, ShopeeFood holds about 20% of the weighted-average food delivery market in the four countries where it operates. I believe Shopee is currently more focused on strengthening the foundation of quick commerce delivery, especially in the high-intent needs segment.

The next logical step would be a stronger push into food delivery. Food delivery order density is very important to Shopee’s long-term instant delivery cost structure. However, Shopee may still need more time to prepare before it can fully capture this growth. Overall, I expect quick commerce order growth to continue accelerating.

First-Party Offerings

This quarter, first-party offerings have grown 50% year-over-year, significantly faster than GMV growth.

I don’t think this growth should be surprising. Fulfillment is a natural extension of Shopee’s C2C logistics network, and first-party offerings are especially well-suited to the mid-sized warehouses Shopee is now deploying.

At the same time, first-party offerings are generally a very good way to build up the initial supply of quick commerce, and they are also a critical element in increasing competitiveness at the service level.

I personally think Shopee will continue to deepen its first-party offerings, and this segment should keep growing much faster than overall GMV.

However, according to a friend who works in Shopee’s B2C unit, a large portion of Shopee’s first-party offerings operates under a consignment model. As a result, this shift may not be very obvious in the earnings numbers.

Monee

Monee had a decent quarter from a revenue perspective. Revenue growth accelerated to 57.8% year-over-year, slightly up from the 54.3% growth last quarter.

But this is the segments that give me the most concern. I can feel a strong “financial discipline” in the company’s earnings. It feels like management is deliberately limiting investments to sustain profit growth.

I personally don’t like companies controlling earnings deliberately. I know it is good for valuation since there will be more certainty. But I think it contradicts the natural rhythm of a business since each quarter should have different opportunities, and financial engineering might cause companies to lose some opportunities in the long term.

Growth & Credit Risk

Besides the understandable financial discipline, my major concern this quarter is the potential deterioration in both acquisition efficiency and credit performance. This quarter, active credit users were only slightly up, and credit allowance as percentage of receivable increased by 1% compared to last quarter.

The allowance amount is adjusted based on this loan performance during the quarter. Although 11.6% is still relatively low compared to the reported historical high of 13.8%, a 1% increase in a single quarter is still meaningful.

The concern is not just that the growth of loans outstanding has slowed to nearly a two-year low. Monee’s sales and marketing expense has also increased significantly over the past year. Even with sales and marketing expense more than doubling, the pace of acquisition still has not shown meaningful improvement.

The Likely Explanation

I think the likely explanation for Monee’s rising sales and marketing expense is the growth of off-Shopee loans. More specifically, it is being driven by the growth of Off-Shopee SPayLater loans.

In my Monee piece, I mentioned that Monee had hit an inflection point in 2024H2, and one of the major drivers was Off-Shopee SPayLater loans.

Off-Shopee SPayLater loans rely much less on Shopee distribution. They are basically another payment network. At this stage, Shopee likely needs to spend heavily to ramp up offline use cases and build the network effect of off-Shopee SPayLater.

At the end of this quarter, Off-Shopee SPayLater loans in Thailand and Indonesia exceeded 20% of the SPayLater portfolio. These are the two markets after Malaysia to cross this threshold. The overall loan portfolio of Off-Shopee SPayLater should surpass the one-billion mark fairly soon.

But this still does not fully explain the slowdown in active credit user growth. This quarter’s active user growth was the slowest in 10 quarters.

Credit Deterioration & the Slowdown

I think one explanation for Monee’s “potential” credit deterioration is the higher mix of Shopee Brazil’s loan portfolio and the longer loan tenures that Monee was pursuing.

We can see from the charts that long-tenure loans have grown rapidly. At the same time, loan outstanding in Brazil surpassed 1 billion this quarter. As we know, Monee needs a higher credit allowance for longer-term loans to reflect the additional risk, and Brazil’s default rates are multiple times higher than those in other Southeast Asian countries.

I think another reason for the slowdown is Monee’s funding constraints, as deposit growth and off-balance-sheet loans have not been able to catch up with loan book growth. The loan book funded from the balance sheet ballooned from 1.45B to 4.17B in 2025, while Sea Limited’s 2025 adjusted EBITDA was only 3.4B.

I think the need for cash to fund loan book growth is one of the main reasons why, despite announcing a buyback when the share price was $120, they still repurchased very few shares this quarter. Sea Limited has not yet reached the stage where it no longer needs cash.

But even with all those explanations, I still would not completely rule out some pressure in loan underwriting, since they had just rolled out an all-can-apply approach in 25H2. They might still need some time to adjust.

Shopee & Monee Brazil

Brazil has become increasingly important to Sea Limited over the years, and this is clear from the growing emphasis on Brazil in the company’s earnings calls.

Brazil is Sea Limited’s largest market by GDP and appears to be Shopee’s strongest growth markets. In 2026, Shopee Brazil will become Shopee’s second-largest market by GMV. It may even become the largest market by revenue, given its relatively high take rate.

From the picture, we can see that Latin America now accounts for 24% of Sea Limited’s overall revenue, up from 19% in 2024. In 2025, revenue from Latin America grew 69%, which is a very impressive number.

And while Latin America is a huge market for Free Fire, but Garena only contributed around 10.5% of Sea Limited’s overall revenue. Considering Monee still had limited operations in 2025 and Garena’s growth was much slower, I think Shopee Brazil may have grown around 80% in 2025.

The impressive growth did include a decent amount of currency tailwinds. However, I have also been told that GMV growth in Brazil was around 70% in local currency. I can’t verify that number, but it does seem reasonable based on the overall growth in the region.

Brazil Potential

Brazil’s eCommerce penetration rate is still low compared to Southeast Asia, and many of its competitors still operate in a rather old-school way. In my view, the main reason Shopee is not yet the market leader is simply that it entered Brazil quite late.

Shopee only started operating in Brazil in 2019 with a cross-border model. It began focusing more seriously on Brazil in 2020, when it launched the local-to-local model. By comparison, Shopee launched its local-to-local business in other markets back in 2015. This means Shopee Brazil is still at a relatively early stage of development.

In 2026Q1, management revealed that Shopee Mall, Shopee Brazil’s branded section, had reached 15% of overall GMV and had grown by triple digits year over year.

In comparison, Shopee Mall as a whole surpassed the 15% mark in 2022Q1. In that sense, Shopee Brazil is exactly four years behind other markets. Right now, Shopee Brazil is still seen as a more low-end and lower-quality platform, similar to how Shopee was perceived in other markets around five years ago.

However, Shopee has been constantly raising the bar for sellers. Over time, it has successfully climbed the learning curve and become a much better platform, with a large number of quality brand suppliers. Many of the very low-quality white-label products have also been eliminated through competition.

Now, at least in Taiwan, the white-label products mostly come with very decent quality. This is very different from five years ago. I think if Shopee can transform itself across seven markets, I believe it can do the same in Brazil.

You can expect a more detailed comparison of why I think Shopee’s main competitor is pretty old-school. There is a reason why Shopee was able to become the leader by order count and active users in five years of operation. I do expect Shopee to become the market leader by GMV within three years.

Monee Brazil

Monee Brazil is another business with significant potential. MercadoLibre and Nubank have already shown how attractive this space can be, with both companies delivering strong growth and impressive earnings. Overall, Brazil is a much more mature credit market than most Southeast Asian countries, with a relatively smaller underbanked population.

The most shocking thing when I learned about the banking world in Brazil is that it’s a market that has normalized triple-digit interest rates. Credit card revolving and installment interest rates average more than 400% and 190%, respectively. This creates a market where banks and fintech platforms can earn enormous spreads.

Another interesting factor is that Brazilian consumers are very used to paying in installments when shopping. Because of this, the quality of credit products is much more closely tied to conversion rates than in many other markets.

Monee has made significant progress in Brazil over the past few years. It only started its lending business there in late 2023, but Brazil has already become Monee’s fourth market to cross a $1 billion loan book, after Indonesia, Thailand, and Malaysia.

In 2026Q1, Monee Brazil was still growing over 250% year over year. Its on-Shopee SPayLater penetration rate is only around 10%, far below the 30%+ penetration seen in more mature markets. This suggests that Monee Brazil still has a long runway for growth.

The SCFI License

During the quarter, Monee obtained the SCFI license in Brazil. This license should broaden Monee’s foundation in financial services and allow it to build a more complete credit stack. This includes lending and financing, acquiring, selling, refinancing, and administering credit receivables, providing guarantees, and offering new post-paid payment instruments.

One of the most notable benefits is that Monee should be able to offer CDBs, or Certificates of Bank Deposit, to consumers. This could give Monee a new funding source for its credit business and help support further loan book growth.

MercadoLibre obtained the SCFI license at the end of 2020 and started to offer CDBs to consumers in April 2022. I guess we may see Monee Brazil start to roll out more financial services in the next few quarters. Hopefully, this new license can help ease Monee’s funding pressure and support its next stage of growth.

Garena

Overall, my view remains largely the same as last quarter, so I will simply quote what I wrote in my previous earnings call review.

I think compared to fintech, I have even less to say about gaming. I don’t play games, and I know very little about the sector. But despite that, I think Garena also had a wonderful year. Bookings were up 37%, and management is guiding for another double-digit growth year in 2026. At least in the short term, I don’t think gaming will be much of a drag.

My thoughts on gaming are that games are very niche and work in paradigms. The same genre will have a few winners, and the winner will last a very, very long time. If you look at the top games in the world, nearly all of them were first launched more than a decade ago. Good games can become evergreen franchises, and I believe Free Fire is already one of them. Because of that, I’m not too concerned about its long-term sustainability.

In the past few years, the management team seems to have cracked the code for driving growth through deep collaboration with other IPs, such as Squid Game and NARUTO. I believe in the future, the growth trajectory will also highly depend on these IP collaborations. So bookings and revenue would probably fluctuate more accordingly.

One thing I wanted to highlight is that, in gaming, revenue growth may not fully reflect earnings performance. In Garena’s case, game props have a weighted average life of 17 months, which means there can be a meaningful lag between bookings and revenue recognition.

This quarter, active users only grew 1%, while average bookings grew 20%. This also came on top of 36% average bookings growth a year ago, meaning average bookings were up 62.7% over two years!

Management is still guiding for double-digit bookings growth this year, despite the lack of solid active-user growth last year. This makes me wonder whether Garena is trending back toward a healthier user mix, with a greater share of users from wealthier countries, similar to what we saw in 2021.

Overall, I think Garena had a great quarter from a monetization standpoint, especially considering the tough comparison from last year’s NARUTO collaboration-driven quarter. I personally think Q2 year-over-year growth could be even better. Please let me know your thoughts!

Closing thoughts

I think there are still quite a lot of things to unpack from this quarter. I will leave the other stuff for the next few quarters. In my next Shopee article, I will probably focus more on the design philosophy and approach different platforms take. I think it will be a controversial but very interesting piece.

To address some of the questions from my readers on Shopee’s profitability, I think profitability this year remains a bit more uncertain because oil prices could have a meaningful impact on the overall platform economics and economies of developing countries. But the fortunate thing is that Shopee is the lowest-cost retailers. I think it will be less impacted, but that doesn’t mean it won’t suffer.

I think the best scenario is that if governments start encouraging people to drive less in order to save energy, Shopee might receive another Covid-like boost. But this is something I would not look forward to.

The management team is still guiding for Shopee EBITDA earnings growth this year. Since Shopee suffered a YoY decrease in the first quarter, I will assume that management is modeling stronger Shopee EBITDA earnings in the second half of the year.

About the long-term take rates, if we compare Shopee to offline retailers — many of which have gross take rates above 35% — it suggests that Shopee still has at least 50% to 100% room to grow! But it would likely take decades for Shopee to reach that level.

For the EBITDA margin, I think the long-term EBITDA/GMV take rate will be over 4%, or even 6%+. Even eCommerce players in China can monetize healthily despite having five or even six major competitors. Given that Shopee faces less competition and monetizes more of the value stack, I think Shopee could become much more profitable than the Chinese players on a GMV basis.

The top four Chinese players all enjoy a 2.5%+ EBITDA/GMV margin when using a similar GMV methodology. I will provide a much more detailed analysis of Shopee and Chinese eCommerce players in a future article.

Thank you for reading.

Subscribe now

As a reminder. My paid publication will officially launch in July. Early supporters who pledge now will lock in a significant lifetime discount.

The current annual price is $200/year and will increase to $220/year on May 30. It will eventually reach $350/year. After that, the price may continue to rise with each new article release. I believe you can also lock in a monthly option if that works better for you.

Subscriber benefits will include exclusive articles, AMAs, my portfolio, and a watchlist with detailed commentary. The founding tier will include private chats and personal assistance. ($600/year)

You can always expect high quality and highly differentiate content on my publication.

Share Contrarian Perspectives

This is my second piece on bubble tea, so I highly recommend checking out the first part. I promise my next article will be about e-commerce! Probably one on Shopee and one on Pinduoduo.

My paid publication will officially launch in July. Early supporters who pledge now will lock in a significant lifetime discount.

The current annual price is $150/year and will increase to $200/year in three days, on May 15. It will eventually reach $350/year. After that, the price may continue to rise with each new article release. I believe you can also lock in a monthly option if that works better for you.

Subscriber benefits will include exclusive articles, AMAs, my portfolio, and a watchlist with detailed commentary. The founding tier will include private chats and personal assistance. ($600/year)

The number of people who have already pledged has exceeded my expectations. I’m very grateful for your support and your kind words!

The Three Buckets and the Mainstream Market

To understand how bubble tea and coffee shops expand in China, we must first understand that China is a very unevenly distributed economy. Cities are divided into tiers, and these tiers basically reflect each city’s income and infrastructure level. Salaries in tier 1 cities are multiple times higher than those in lower-tier cities. For the 33% of the population in China who live in rural areas, the median disposable income was only USD 250 per month in 2024.

In China, markets are usually separated into three buckets: tier-one cities, mid-tier cities, and low-tier cities. There are only four cities in the tier-one bucket — Beijing, Shanghai, Guangzhou, and Shenzhen. This is the part of China that Western media often cites when talking about how “advanced” China is, but only 6% of the population lives in this bucket.

The second bucket includes the “new tier-one” cities and tier 2 cities. There are only a few dozen cities in this bucket, including many coastal cities and some important inland cities like Chengdu and Chongqing. The first and second buckets are basically the powerhouse that has driven China’s economic growth over the last 40 years. About 25% of the Chinese population lives in the second bucket.

The third bucket consists of low-tier cities. Around 70% of Chinese people are in this bucket, meaning about one billion people live in these areas. This is the market that had largely been ignored until the last ten years. The famous e-commerce company Pinduoduo, the owner of TEMU, is one company that grew out of this market.

But that’s not the whole story. In China, tiers do not only exist between cities; they also exist within cities. The income dispersion within a city can also be quite large.

Migrant Workers

Just like more advanced economies usually face labor shortages in entry-level service jobs or physically demanding jobs, China’s high-tier cities face a similar problem, but they solve it in a very different way.

In many developed countries, migrant workers are brought in from abroad to fill these jobs. The number of these workers is usually limited. But in China, high-tier cities can simply absorb labor from lower-tier cities. This situation is not always easy to notice because most of these workers are from the same ethnic group and speak the same language. The migration is internal, so it does not look as obvious as immigration in other countries.

This creates a huge gap in disposable income. Educated white-collar workers in Tier 1 cities may earn salaries that feel closer to those in middle-income countries (Taiwan, South Korea), while blue-collar workers and entry-level service workers may earn wages closer to those in low-income countries (Southeast Asia).

The city-tier system makes this gap even stronger. In China, higher-tier cities usually have better schools, better resources, and more favorable quotas for elite universities. As a result, people born in high-tier cities often have a much better chance of getting high-paying jobs.

That is why life can feel very comfortable for native residents of Tier 1 cities. They can enjoy middle-income-level salaries, excellent infrastructure, and cheap goods and services supported by migrant workers from lower-tier cities. This combination is hard to find in developed or middle-income countries.

City Size and Outside the Fifth Ring Road

City size also matters. Shanghai, for instance, is about 8.5 times the size of Singapore and more than 5 times the size of Hong Kong. So when people talk about China’s urban markets, they often separate cities into areas “inside the Fifth Ring Road” and areas “outside the Fifth Ring Road.”

This phrase comes from Beijing’s road structure. It is used to describe the difference between people who live in the real urban area and those who technically live in the city but are actually far from its center. Beijing has a population of around 22 million, but only half of that population lives inside the Fifth Ring Road. The same logic applies to other high-tier cities: not everyone counted as living in the city enjoys the same income level or lifestyle.

Of course, pay gaps exist in every major city. But the gap in China feels like another level because the supply of internal migrant workers is so large. It is almost like imagining a developed country with an unlimited supply of migrant workers from low-income countries.

I have been to the outskirts of Beijing, and some of those “county-level” areas look no different from lower-tier cities. But officially, they still count as Beijing. That is why looking only at a city’s total population or GDP can be misleading. In China, where you live inside the city matters almost as much as which city you live in.

Noted that: the area inside Beijing’s Fifth Ring Road is similar size to Singapore! So the area is quite huge.

The Target Markets They Serve

Starbucks China is one of the best examples of a company serving wealthy Chinese consumers who live within the “Fifth Ring Road”. It has around 8,000 stores in China, and most of them are located in higher-tier cities. Shanghai alone has more than 1,100 Starbucks. Outside of top-tier cities, Starbucks usually opens only a small number of stores, mostly in the city centers of lower-tier cities or counties.

Mixue and Guming are almost the exact opposite. They are built for the mass market. Take Guming as an example: its stores in Tier 1 cities account for only 3% of its total store count. In Zhejiang Province, among roughly 2,200 Guming stores, only about 150 are located in city-center areas.

So Mixue and Guming are playing a very different game from Starbucks. Starbucks mainly serves the higher-income urban consumer in high tier cities, while Mixue and Guming serve the much larger population outside the most comfortable and well-developed urban cores.

That is why these companies may be a better representation of China’s real mass market. They are serving the needs of the 70%+ of Chinese people who do not live in the best locations with the best infrastructure, but still make up the majority of the country’s consumer base.

Mixue and Guming’s different approaches of Scaling

The key difference lies in how Mixue and Guming built their respective supply-chain strengths. Mixue’s advantage comes from vertical integration and its factory cluster. Guming’s strength comes from its refrigerated trucks, storage capabilities, and specialized handling methods.

Guming’s scaling is much more constrained by the infrastructure that powers its growth. And because of the “mass market” it serves, a good and cost-effective cold chain was still not available. Guming needs to build everything from scratch. This severely limits Guming’s scaling.

This is one of the main reasons Guming, despite having more than 13,000 stores, still operates in only 17 of the 34 provinces in China, whereas Mixue started its national expansion in 2014 and its international expansion in 2018.

Guming also believes that store density is the key to drink freshness. With enough store density, Guming can deliver raw materials at a lower cost while keeping them fresher. Guming’s guideline was to reach a critical mass of stores in one market before entering the next market.

Mixue, on the other hand, doesn’t have as much of a cold-chain problem limiting its growth since most of Mixue’s ingredients are standardized by factories. It is also able to leverage the existing logistics network for delivery, which makes scaling Mixue much easier.

For Mixue, the bigger limitation is demand. Mixue is always actively expanding its product line and upgrading its products with better ingredients. For instance, Mixue used to only use jam and canned fruit in its beverages, then gradually upgraded parts of its beverages to frozen fruit, and now some products use fresh fruit. The same pattern can be seen with milk. Non-dairy creamer used to be the only option for beverages. Now room-temperature milk is used, and starting in 2024, some stores even began offering fresh milk.

With massive scale and supply-chain strength, Mixue will be able to gradually upgrade its beverage offerings. This is one of the main ways Mixue grows its buyer base.

The Chinese Coca Cola

Coca-Cola became ubiquitous largely because of bottling technology in the early 1900s, which allowed Coca-Cola to reach every retail store and restaurant. Since then, soda has become the most popular beverage for lunch and dinner, while coffee has dominated breakfast.

But in East Asia, things are different. While coffee and soda consumption remain low, coffee is growing at a 2X% CAGR, while soda consumption has been rather stagnant.

This challenges the common belief that tea is a very strong substitute for coffee. What bubble tea stores are really replacing is not coffee, but bottled beverages. These drinks are consumed during lunch, dinner, and afternoon snack occasions.

In East Asia, the upgrade in beverage consumption isn’t going from water to bottled beverages as it did in the U.S. Instead, it is moving from water directly to freshly brewed drinks. I believe nearly everyone would choose a freshly brewed drink over a bottled beverage, but freshly brewed drinks are usually too expensive or inaccessible. The real value that Chinese bubble tea shops provide is making freshly brewed beverages accessible at reasonable prices.

Food delivery platforms and the rise of bubble tea shops

The two biggest advantages of pre-made bottled drinks over freshly made drinks have been price and accessibility.

The rise of food delivery platforms has changed that. It gave bubble tea shops a powerful distribution channel, and when paired with their strong operating skills, fresh drinks became much easier to access. Now, in many parts of China, freshly brewed beverages are no longer a luxury. They are becoming an everyday option.

In the rest of the world, and even in many parts of China, most people still don’t have access to that level of convenience. But I think this will probably change over the next few decades, especially with autonomous vehicles. Once the marginal cost of delivery becomes much lower, bubble tea shops could really start to shine.

Note: As I mentioned in the first piece, in China today, more than 30% of orders from an average bubble tea or coffee shop come from food delivery platforms. This says a lot about how food delivery platforms and bubble tea shops work together to make freshly brewed drinks accessible. And if you’re interested in how Chinese food delivery platforms can offer such low delivery prices, sometimes even cheaper than in-store pricing, check out my piece on China’s quick commerce war.

What makes Chinese Bubble Tea Shops special?

When most people think of bubble tea, they usually think of the Taiwanese version: a simple drink made with tea, milk, and boba. But in China, the idea of a “bubble tea shop” has already evolved into something much broader and more complex. Today, the term is more like a general label for stores that sell freshly made beverages, as long as their main product is not coffee.

Coffee shops and bubble tea shops both sell freshly made beverages. The key difference between Chinese bubble tea shops and coffee shops is operational complexity. In general, coffee shops are much simpler to operate. A basic coffee shop only needs a machine, coffee beans, milk, and sugar. Most coffee shop ingredients, except milk, have a long shelf life, and the preparation of raw materials is usually handled by factories.

Bubble tea shops, on the other hand, usually require more space and place much greater emphasis on fresh ingredients. Staff also need to spend much more time preparing ingredients. For example, they may need to cut fresh fruit and cook pudding, boba, and other ingredients on-site. Many of these processes cannot be easily outsourced to factories. As a result, bubble tea shop products can become much more complex, with a wider range of flavors and textures.

This also makes their supply chains and product innovation cycles much more dynamic. These shops often need to handle 100+ new product launches and source different ingredients from many suppliers. Their products are also heavily driven by social media trends.

The picture above is from HeyTea Malaysia’s menu. HeyTea is a premium brand with drinks priced at ~ US$3. It is especially famous for its fruit-based drinks.

As you can see, each drink has many visible layers. The real products look very similar to the menu photos, so they not only taste good but are also highly suitable for social media.

Not all drinks in bubble tea shops have that much complexity. The ones from Mixue are rather simple. Ingredients are also from factories. Although they still have a lot more complexity than a typical coffee shop!

So in general, it’s much easier for bubble tea shops to sell coffee than for coffee shops to sell bubble tea! But it also means that the cost of operating a bubble tea shop is much higher than operating a coffee shop.

Closing thoughts

At first, I only expected this bubble tea article to be a short, five-minute section in my new series, which covers many different topics. But once I started writing, I realized I had much more to say about the industry than I originally planned.

Even after 27 minutes, I still think I had only touched a small part of the bubble tea story that I would like to share. For instance, I would like to discuss how Mixue built its IP business on top of its operational capability and supply chain strength; how different companies choose between company-owned and franchise models; what tradeoffs Starbucks China faces as it enters a new operating model; the expansion direction of each major player; the cost structures of different national brands; the operational differences between city tiers; and the challenges these companies face as they expand nationwide.

Another topic I wanted to discuss more is the endgame of the bubble tea industry. My current view is that there will probably be two to three long-term nationwide winners, each occupying a specific price range, along with several strong regional players in different markets.

Bubble tea is not like Coca-Cola or coffee, which have much longer life cycles. The industry carries more fashion risk, and each cycle will likely be influenced by new trends. I expect new chain stores to pop up from time to time and temporarily occupy the premium brand segment. I also think another bubble tea chain may eventually emerge with an extraordinary ability to manage indoor spaces, becoming something like the Starbucks of bubble tea. But I don’t think it will be Naìsnow. Its taste is just too bad.

That’s all for now. I will probably write some follow-up articles on this topic in the future.

Subscribe now

As a reminder. My paid publication will officially launch in July. Early supporters who pledge now will lock in a significant lifetime discount.

The current annual price is $150/year and will increase to $200/year in three days, on May 15. It will eventually reach $350/year. After that, the price may continue to rise with each new article release. I believe you can also lock in a monthly option if that works better for you.

Subscriber benefits will include exclusive articles, AMAs, my portfolio, and a watchlist with detailed commentary. The founding tier will include private chats and personal assistance. ($600/year)

The number of people who have already pledged has exceeded my expectations. I’m very grateful for your support and your kind words!

Share

My paid publication will launch in July. Members who pledge early will receive significant discounts.

The current price is $120/year and will increase to $150/year after three days (5/10). The final price will be $350/year. Pricing will increase with each article release.

Subscriber benefits will include exclusive articles, AMAs, my portfolio, and a watchlist with detailed commentary. The founding tier will include private chats and personal assistance. ($600 / year)

I’m very grateful for your support.

Subscribe now

Preface

Since launching this blog, my articles have focused mostly on eCommerce. Now, I want to try something different. In this series, I will write more about random stuff. I wish my audience could join me in the process of exploring business. This series won’t be as polished or thoughtful as my eCommerce takes, but I think the quality will still be rather good.

Today, I want to share some thoughts on coffee and bubble tea shops in China. I’m not deeply familiar with the industry, but as a business nerd and a hardcore bubble tea and coffee lover, let me share a few observations.

This will be a two part series, the second part will be release soon! ( Update: Click here for part two)

The Dynamic Pricing of Bubble Tea and Coffee Shops

The experience of buying bubble tea or coffee in China is unique — almost no one waits in line to order. Customers usually order through the shop’s WeChat mini-app, where everyone gets custom pricing.

You didn’t read that wrong. The pricing of beverages for each customer is different. This is caused by the highly digital environment and the coupon-driven discount model many companies adopt. In China, the listed price and the true price of products are two different things. For instance, Luckin Coffee’s menu prices usually range around 25~35 yuan (3.6~5.1 USD), but customers usually pay around 10~15 yuan (1.5~2.2 USD).

The “true” pricing will highly depend on regional competition, user habits, real-time foot traffic, and the promotions of different traffic platforms. So as I travel, I can see drastically different pricing between different stores, cities, and even different times.

Besides, companies tend to navigate the digital environment very well. Every company has teams dedicated to building viral content and staffs responsible for managing digital traffic. The reliance on traffic platforms is enormous. In a typical store, food delivery orders alone contribute 30%+ of the overall orders, and a very decent portion of the takeout orders are also driven by traffic platforms like Douyin and Meituan.

Sometimes the orders from traffic platforms can be out of control. For instance, for a few days in July and August 2025, in the midst of Alibaba Taobao and Meituan’s quick commerce war, many chain stores got the majority of their orders from the delivery platforms.

and I had written an article that dives into the quick / food delivery landscape in China. It goes through many of the most unique aspects of quick commerce in China and also touches on eCommerce. Feel free to check it out! The read time is 70 minutes.

The scale

There are now 500,000+ coffee / bubble tea shops in China, and seven brands already have more than 10,000 stores each. The biggest bubble tea chain is Mixue, with about 45,000 stores in China and 5,000 stores outside China. The biggest coffee chain is Luckin, with over 30,000 stores. Starbucks China now has about 8,000 stores.

Starbucks exit China? Will Luckin Coffee kill Starbucks in the U.S.?

By the end of 2025, Starbucks China had divested the majority of its shares to a local partner, and had also been cutting prices over time. One of the main drivers behind this shift was Luckin Coffee (aka the Starbucks Killer). Founded in 2017, Luckin Coffee aggressively undercut Starbucks on price from the start. But how did that happen?

What happened is fairly simple. Starbucks stretched its brand too much. In North America, Starbucks was an affordable premium option, but in China it was positioned as a high-end brand. Its products and experience simply were not good enough for the average consumer to pay 2~4 hours of salary for a simple drink.

Interestingly, now in China, Luckin Coffee has gradually raised its prices and has now become the new “affordable to affordable premium” option. Now brands like Lucky Cup, NOWWA, Cotti Coffee, and Guming are all the cheaper options.

Luckin entered the U.S. in June 2025, and many are expecting the “Starbucks Killer” to conquer Starbucks on its home turf. I personally don’t think Luckin’s expansion will be a smooth process. In the U.S., Luckin doesn’t have the open void to easily undercut Starbucks.

But I think what people should pay more attention to is Luckin’s speed of product iteration. The company launched 140+ new drinks in 2025 alone. That level of execution is very different from Starbucks’ much slower seasonal product cycle.

Besides, Luckin is more focused on being a broader beverage company than Starbucks. Compared with Starbucks, it has many more non-coffee-oriented products, and its best-selling coffee is in non-traditional coffee choices like lattes or Americanos, but more flavor-driven beverages such as Coconut Latte, Butter Latte, and Jasmine Tea Latte.

There are also some strengths that Starbucks has that Chinese brands are still not good at. The most outstanding one is how Starbucks manages its brand image through store design and its ability to maintain the “third place” experience.

For now, Chinese coffee or tea brands are mostly built around takeout, and frankly, even when they have seats, I still think none of those companies is anywhere near as good at creating a consistent space to work or chat. It feels like the sitting area is just an extension of the “little factory” that produces drinks.

For me personally, in China, I will still choose Starbucks as the go-to place to use my computer, especially after Starbucks cut prices by a meaningful amount (around 25%~30%) through lower list prices and discount coupons across different channels. In the past few years, the gap in pricing between Starbucks and Luckin has been basically cut in half.

I’m still expecting Starbucks China to further drop some prices, as Starbucks China now has a Chinese owner and competition in the coffee market has intensified, but I think Starbucks would still have a very decent chance of surviving profitably in China. They are still providing unique value, especially in the top-tier cities. (More on this in a moment, as there is a lot more nuance in the Chinese market.)

Bubble Tea Shops in China

The bubble tea industry in China was a cultural export from Taiwan. The first generation of shops mainly replicated Taiwanese flavors, but the industry later evolved into something more distinct.

Today in China, the four main categories of “bubble tea” are: 1) traditional Taiwanese bubble tea, 2) fruit-based drinks, 3) milk tea, and 4) dessert-oriented drinks.

(picture was generated by ChatGPT)

As consumer spending has upgraded and logistics networks have matured, China’s beverage chains are now in intense competition, with thousands of brands chasing the fast-growing market.

At the same time, many bubble tea brands are adding coffee, while coffee chains are expanding their non-coffee offerings, so the boundary between coffee and bubble tea in China is becoming increasingly blurred. This is the market context Starbucks China is operating in.

In the past few years, many brands have grown over tenfold. As the domestic market becomes increasingly saturated, a lot of the brands are expanding overseas at speed, and I think part of my Western audience will be able to see them in their daily life soon. Some brands, such as Mixue, are already well known in Southeast Asia and are now moving into middle-income countries and more Western markets.

Every successful company in this industry has either done something unique or has stepped onto the right trend to achieve current success. Let me share more of their strategies, and I think they show a lot of similarities with Western brands like Coca-Cola, McDonald’s, 7-11, Walmart, and others.

Different approaches and value propositions

Bubble tea and coffee shops in China can basically be divided into three price tiers:

Different price points map to different consumer segments, and each region and product category in China’s bubble tea market tends to have its own champion. The market is now in a kind of “Warring States Period.” This graph highlights only some of the more notable players, each with a clearly differentiated approach.

The affordable to affordable-premium range (roughly $1.50 to $2.60 USD) is where most bubble tea shops are positioned, and it is commonly regarded as the market’s “mainstream” price band.

Trends, product differentiation, and barriers to entry

The barrier to entry in the beverage industry is very low — a lot of people are capable of making something delicious. The hard part for huge chain stores is to build out a system that can deliver a consistent experience across locations. The game that most chain stores play is not about becoming the best location in town, but about becoming the best option for a consistent experience at a decent price point.

In every city, there will always be a standalone store that provides a better experience. But the problem for most of those standalone stores is that they can’t scale and maintain that same experience. Scaling with consistency is what makes chain stores unique.

Another interesting point is that it’s an industry that is heavily influenced by trends and offers limited technological differentiation. There are companies that catch the right trend or use better technology to deliver a strong product experience and suddenly become one of the most popular stores in the area. But those advantages usually won’t last long. Competitors will always step in and adopt similar methods and technologies.

In recent years, one of the best examples in China is Chagee. Founded in 2017, Chagee is a premium bubble tea brand that specializes in light milk tea and currently operates about 7,500 stores worldwide.

The Light Milk Tea Trend and Chagee’s Road to Thousands of Stores

At the end of 2020, Chagee was still a small chain store with about 200 stores, and it offered about 50 different drinks across various categories. From 2021 to 2022, with a fresh round of venture funding and “learnings” from ChaYanYueSe (茶顏悅色, a famous regional chain store), Chagee decided to focus on the “light milk tea” category and underwent several major branding upgrades.

At that time, there were no national chains focused on this product category. Chagee successfully capitalized on the light milk tea trend.

(Chagee’s brand upgrade - before and after - picture from Latepost)

Paired with new branding, a big-store strategy in prime areas, huge sales and marketing spending, trending products, and the introduction of automated tea-making machines that made light milk tea extremely fast with minimal training, Chagee has successfully garnered huge attention and built a franchise with high same-store sales, high margins, and low operational complexity.

The store count grew from around 400 at the end of 2021 to 6,440 by the end of 2024. That is nearly 15x growth in three years. The GMV grew more than 50x over the same period.

But the success attracted competition. Everyone in the industry wanted a piece of the good business. Chagee’s products were sold at a premium, and the company enjoyed a very fat margin. As a franchisor that sold raw materials to franchisees, Chagee’s gross margin was ~50%, whereas its peers’ gross margins were only around 30%. And many of those peers only reached 30% gross margin after spending years or even decades building factories, in-house logistics networks, and even farms to cut costs.

Before the end of 2024, most major brands had installed automated tea machines in their stores and rolled out similar products at lower prices. Some of them even claimed to use better milk and tea, which hinted at better flavors. I tried Guming and several alternatives multiple times; the tone was awfully alike, and they were just as good. After drinking the alternatives, as a light milk tea lover, I did reduce the frequency of drinking Chagee.

With competition intensifying and the trend fading over the past two years, the average same-store sales of Chagee stores have dropped by more than 40%. Some of the stores have begun losing money. In the latest quarter, the store count has started to shrink.

Subscribe now

The Distribution Problem

Chagee’s main problem is that its rapid expansion has outpaced its quality control, supply chain development, and product innovation. Before it built a strong enough foundation, competitors had already started catching up with its core product.

As a big fan of Chagee who has had more than 100 cups, I can clearly feel the inconsistency. The same drink often tastes slightly different from store to store, and my favorite drink is frequently out of stock. The in-store experience is also uneven. Some stores feel warm and welcoming, while others make me question why they offer seats at all.

Chagee is also not a company that is good at product innovation. It relies heavily on a single product category in a market where the barrier to entry is low. In some quarters, its top product contributed around 40% of overall sales, while its top three products contributed around 60%.

What makes this worse is the slow pace of new product launches. In 2025, there was a six-month period when Chagee didn’t launch any new products, and across the whole year, it only introduced eight new products, making its innovation speed roughly one-tenth of its peers.

As people in business often say:

The battle between every startup and incumbent comes down to whether the startup gets distribution before the incumbent gets innovation.

Chagee started with a good product and a wonderful brand. But when it scaled up the stores, it still relied on the same playbook of a small regional player. It did not build strong enough distribution capabilities (aka supply chain, in-store operations capability, and product innovation skills). As a result, when the incumbents caught up on innovation, such as light milk tea and automated tea machines, Chagee was crushed by the strong distribution capabilities the incumbents had built up through the years.

I still think Chagee has a decent chance of a comeback. Its brand is still very strong, but brand alone is not enough for a daily-consumption product. For daily brands, my view is that the brand is ultimately a reflection of the product itself. Brand image may fluctuate in the short term, but in the long term, a daily brand is basically equal to its consistent quality.

For high-priced goods like houses and luxury products, or for low-frequency purchases, brand can play a much bigger role because average consumers do not have enough repeated purchase cycles to fully understand the difference in quality.

Unfortunately, Chagee sells a simple premium beverage. If customers are under budget pressure, face stock shortages, see another trusted brand offering similar products, or simply get bored of the taste, they may churn quickly.

In the long term, I personally think the bubble tea industry will be won by the companies with the best distribution. Among all the bubble tea chains, Mixue and Guming look like the top two companies built to win this game. They may eventually achieve McDonald’s-like success, or in a certain sense, become the Chinese version of Coca-Cola.

Mixue and Guming(Goodme)

Mixue and Guming occupy very different positions in consumers’ minds. Mixue built its reputation on extreme affordability, becoming famous for its RMB 2 (US$0.30) ice cream and RMB 4 (US$0.60) lemonade, with the average item on its menu priced at around RMB 6 (US$0.90). For consumers, the appeal was straightforward: they could buy a freshly made drink for about the same price as a bottled beverage from a convenience store.

Guming, by comparison, was aimed at a different consumer need. It did not try to win purely by being the cheapest. Instead, it positioned itself as an affordable upgrade: still accessible to mainstream consumers, but with a more polished product and a stronger perception of quality. For customers, Guming was the brand they chose when they wanted something better than the lowest-priced option, but still well within their everyday spending range. The average price tag for Guming was RMB 12.5 ($1.80), roughly double the price of Mixue and about 25%~30% lower than premium alternatives.

Despite their very different brand images, both Mixue and Guming are relentlessly focused on the supply chain. They are companies that rely on distribution strength instead of purely relying on inventing trendy products.

To borrow the words of Guming founder and CEO Wang Yunan on product trend:

Each generation defines the iconic popular product of its time. In my view, for those born after 1985, it was Taiwan-style milk tea (COCO). For those born after 1990, it was Four Seasons Milk Green Tea (1 Point Tea). For those born after 1995, it was fruit tea (Heytea). And for those born after 2000, it is light milk tea (Chagee). Every five years, the trend changes. Behind the rise and fall of product categories is a generational pattern.

According to Wang, the way to success wasn’t to act as an innovator that leads the trend, and, frankly, big companies are also not the best innovators. From the same interview:

The most creative people in China are not the ones sitting in offices. They are people like Neo (the founder of Heytea) when he was twenty-seven or twenty-eight. They are grassroots entrepreneurs who spend every day in their stores, constantly thinking about how to make products better and personally crafting them by hand. They are the soul of innovation.

Guming’s current playbook is to make products that are proven by the market better and leverage its distribution to harvest popularity. Also from the same interview:

I don’t aim to set trends with my products. Whatever others are making, my job is to make it better. That’s all I focus on. I turn the brand itself into a channel.

Competitors introduce products like mango pomelo sago to stand out. I take the same quality, sell it across 10,000 stores at a lower price, and turn it into a mass-market product. Once it scales that fast, within a year, it becomes the tree my competitors planted, but I’m the one harvesting the fruit.

The McDonald brothers’ early success was built on the invention of the Speedee Service System, which allowed them to serve good hamburgers at exceptional speed. Today, McDonald’s competitive advantage is built less on product invention and more on distribution — a combination of brand power, supply-chain strength, and operational excellence. McDonald’s often lets smaller or newer players introduce “innovative” products first, then uses its unique distribution system to turn those ideas into the “best value at every tier of the menu.”

Mixue and Guming are both betting on distribution, and I think this is what matters in the industry over the long term. But while both companies share that same broad strategy, they focus on the supply chain in different ways.

Different approaches to supply chains

Guming’s price points are roughly double Mixue’s, and this basically anchors the raw materials used in the drinks. The two companies’ supply chain strengths and expansion strategies are largely shaped by this difference. Let me use milk and fruit as examples.

For Mixue, the milk it uses is mostly non-dairy creamer or milk powder. The fruit is usually jam, canned fruit, or frozen fruit. As a result, nearly all of Mixue’s raw materials can be transported at room temperature, with only a small portion requiring freezer trucks.

Guming is different. Its milk is mostly fresh milk, and its fruit is also freshly picked. This means a large portion of Guming’s raw materials can only be delivered through refrigerated logistics.

Overall, Mixue’s raw materials are mostly factory-produced, highly standardized, and have long shelf lives. Guming’s raw materials, by contrast, are mostly fresh goods with higher variation and shorter shelf lives. They require much more delicate handling.

This is the centerpiece of both companies’ supply chain strengths. Mixue’s strength comes from vertical integration and its factory cluster, which allows it to produce raw materials at a cost that almost no other player can match. Guming’s strength comes from its refrigerated trucks, storage capabilities, and specialized handling methods, which allow it to deliver fresh and non-standardized raw materials into lower-tier cities.

Next, I want to talk about how Mixue and Guming expanded in their own ways. But I’ll save that for Part 2. In the next part, I’m going to talk about the concept of the “mainstream” market they are really competing in and also why I think the drinks from bubble tea shops are the new Coca-Cola.

The next part will be released very soon! (Update: Click here for part two)

Subscribe now

As a reminder, my paid publication will launch in July. Members who pledge early will receive significant discounts.

The current price is $120/year and will increase to $150/year after three days (5/10). The final price will be $350/year. Pricing will increase with each article release.

Subscriber benefits will include exclusive articles, AMAs, my portfolio, and a watchlist with detailed commentary. The founding tier will include one-on-one consultations.

I’m very grateful for your support. See you in the next article!

The name Contrarian Perspectives comes from my desire to find non-consensus views that are actually right. Another reason is that I don’t like to repeat what people are already doing. I call my articles “non-deep-dive deep dives.” Usually in my posts, you would only see two things. One is very basic things that are the guiding direction of the company. Another is important details that only institutional investors or serious business nerds will care about.

I think most “deep dive” pieces out there are really just good introductions rather than true deep dives. There is still so much more! My publication is dedicated to what comes after the normal deep dive. So in most of my articles, I will assume that my audience already has a decent amount of understanding in the field. I hope you can still learn something even after spending hundreds of hours reading about the company.

Another thing that makes my publication special is how well the posts connect with each other. Usually, when I write about a specific company, I also have a solid understanding of all its main competitors. So concepts between articles can be easily linked together.

In my articles, I will clearly lay out the tradeoffs these companies make that make them truly unique. So there will be quite a bit of comparison between companies! Talking about the defects of the companies doesn’t diminish my admiration for the company and its past accomplishments. Often the company’s current defect is exactly what made it unique and successful in the first place.

I don’t consider myself an expert in any of the fields I write about — the executives are, not me. But I think what I can provide is a broader view and the ability to draw similarities across industries. I know quite a few of my regular readers are executives, and as a business nerd, I would be more than thrilled if you could point out where I am wrong.

I think the best way of staying right over the long term is to correct my mistakes as soon as possible. I think I have the humility to admit when I’m wrong. Since I’m Taiwanese and have been humiliated often throughout life, I don’t have too much of an ego.

In this publication, you don’t need to worry about wasting time. I won’t post much, but I will make sure every article I release is worth reading. Every long-form article I publish has been rewritten at least three times. I will make sure every article I release consists of only the absolutely crucial elements.

That’s all. Hope you all have a nice day!

Subscribe now

-

Oh, it’s not all. I’m planning to launch a paid publication in July. But I’m Taiwanese, and Stripe doesn’t work in Taiwan. I need to set up a U.S. company before I can enable paid subscriptions. For now, I want to test the waters.

If you are interested in supporting my publication, please pledge your support! The current pricing will be $15/month or $120/year. The price will rise gradually with every article release, and the official price will be $40/month or $350/year.

The subscriber benefits will include exclusive articles, AMA, my portfolio, and a watchlist. The founding tier will include one-on-one consultation. Now is the best price to get a subscription! I don’t think you would want to miss it!

As always, I very much appreciate your time.

Subscribe now

Preface

Just like Shopee, Monee also had a wonderful year. The credit business had just passed its critical inflection point and now is in a stage of enormous growth. And I believe the rapid growth phase had just started.

My thoughts on fintech in developing countries are that it’s basically the low-hanging fruit of any successful eCommerce business. It’s a business that could hugely boost the conversion rate of eCommerce transactions and could also help the eCommerce platform enjoy the lucrative profits of a banking institution.

Monee was born out of this backdrop. It serves three goals, which also map to the three stages of Monee’s evolution: 1) increase the conversion rate of Shopee 2) become the default choice of consumer credit 3) become the default app for digital life.

Basically, we are in the second stage of Monee’s transformation, which is increasingly focusing on credit. It doesn’t mean the first goal, increasing Shopee’s conversion, is no longer important. It only means that Monee is starting purse a broader vision, rather than simply acting as a complementary service to Shopee.

Increase the conversion rate

Monee is born to eliminate cash on delivery and to decrease the transaction cost in payment services. Now, with ShopeePay and national payment rails across the region, cash on delivery is no longer very popular in most Southeast Asian countries.

Monee’s goal has shifted from being a payment processor to being a broader payment solutions provider. Monee now provides not only basic payments, but also credit, installment, and monthly subscription services to users.

These services all boost Shopee conversions. One of the recent example is the VIP program in Indonesia. President Chris Feng mentioned this on the recent earnings call:

The renewal rate, it was one of the core challenges historically for a similar program in our region is that payment success rate. When when they roll from one month to another, many people drop off simply because there’s no credit card available for many of our users in our region versus if you look at the more credit card market.

I think we solved this by working closely between Shopee and Monee to enable the smooth payment process for our VIP program. And as a result, our kind of the subscription, re-subscription rate has grown from 40% to 70% in Indonesia over the past few quarters

I think Monee’s promotion to improve the conversion rate of Shopee VIP membership is one of several reasons why Monee’s sales and marketing expense has increased sharply in recent quarters. (I’ll talk more about the sales and marketing expense in a moment.)

Become The Default Choice Of Consumer Credit

After Shopee built up its payment business, it began layering in its credit business. In my opinion, in the long term, Shopee’s credit business will be worth a lot more than its eCommerce operation!

The four loan products

There are four main credit products in Monee’s loan portfolio, and they can be divided into two buckets: on-Shopee and off-Shopee.

1. On-Shopee loans: Working Capital loans (SME loans) and On-Shopee SPayLater (BNPL service that are use on Shopee)

2. Off-Shopee loans: Cash loans and Off-Shopee SPayLater (BNPL service that are use outside Shopee)

Although working capital loans were the first products launched, consumer loans now account for more than 95% of total loans. And since the end of 2024, more than half of the loans in Asia have been off-Shopee loans. Currently, off-Shopee loans are growing considerably faster than on-Shopee loans.

How Does Monee Scale Up Its Loan Book?

Given the enormous growth in both the loan book and sales expenses, many investors may assume that management is scaling the loan book recklessly.

In reality, the opposite is true. This rapid growth is not a sign of reckless expansion, but rather a result of different product cycles and launch timings across countries overlapping during their scaling stages.

Monee’s first consumer credit product, on-Shopee SPayLater, was launched in Indonesia in 2019, than in Thailand in 2021 and Brazil in 2022. A similar pattern applies to cash loans, which are generally launched 1–2 years after SPayLater. Meanwhile, over the past 18 months, the off-Shopee SPayLater distribution network has only recently begun to come online and scale.

I know this may sound a bit cumbersome, but I will break it down in detail in the following article. For now, let’s return to the question of how Monee scales up its loan book.

Management is scaling up the loan book by fully leveraging Shopee’s ecosystem advantage. As management explained on the 24Q4 earnings call:

In all our markets, on Shopee SPayLater purchases are the first and very natural touch point we have with most of our credit users. It allows them to build an initial credit track record and allows us to build out our credit model for the market.

Once we understand the users’ credit behavior, we give them access to other products with longer tenures and larger quantums. And when we have built a credit risk model for each market that we feel confident of, we then start to scale our loan book. Scaling our loan book includes diversifying into more off-Shopee scenarios, giving us access to a much larger pool of consumer spend.

In essence, management is using a land-and-expand strategy to grow the loan book:

1. Select a group of users based on shopping behavior.

2. Give them a low credit limit for SPayLater.

3. Scale up their SPayLater credit limits.

4. Give users the option to take out an off-Shopee cash loan.

5. Scale up their cash loan limits with longer tenures.

This land-and-expand approach is reflected in the disbursement of funds. Despite growing the loan book by 72% year-over-year and 14% quarter-over-quarter, loan disbursements to new users still accounted for less than 10% of total disbursements in the third quarter of 2025. And even more importantly, the loans that Shopee gives out are mostly small-dollar loans with tenures of 1~6 months. The default risk from the macro environment can be adjusted rather swiftly.

The inflection point in Monee’s credit business

There are three things that contributed to the inflection point in 2H24. 1) Shopee has stopped cash burn 2) the credit model is ready 3)the convergence of the product cycle

Shopee has stop cash burn

This is the most important reason of all: no matter what, Shopee is the group’s most important asset. Monee still relies heavily on Shopee for distribution. As soon as Shopee stops bleeding cash, Monee has the flexibility to deploy the cash on its balance sheet.

The Credit Model are Finally Ready

After years of testing, the credit model is ready for mass-market scaling. Non-performing loans over 90 days have declined from 2% to 1.1% in recent quarters.

Since the charge-off has a 4~6 month lag in the loan book, If we compare charge-offs against the loan-book figures from six months earlier, the numbers also appear broadly stable, if not slightly improving.

With better credit models to separate risk, the addressable market can expand into more riskier borrower segments while also allowing Monee to compete more effectively with prime-focused lenders for attractive prime borrowers.

Management has attributed the success in 2025 to AI-driven improvements in the credit model. From the 25Q4 Earnings Call, CEO Forrest Li:

Our credit business expansion in 2025 was made possible by improvements in our risk underwriting capabilities. These improvements tapped on our rich ecosystem data and advancements in AI.

Enhancing our models’ precision and performance enabled us to scale rapidly in 2025 while still maintaining a stable risk profile. Our 90-day NPL ratio held steady at 1.1% as of the end of the fourth quarter.

The Convergence of Product Cycle

In my opinion, Shopee stopping the cash burn frees up the cash to fund loans, while the fine-tuning of credit models makes the products ready for mainstream. And most importantly, the enormous growth is basically attributed to the convergence of product cycles in different countries.

The rollout timelines for the three main consumer lending products are shown below, with Indonesia generally launching first and Brazil latest:

1. On-Shopee SPayLater: launched in 2019~2022

2. Cash Loans: launched in 2020~2023

3. Off-Shopee SPayLater: launched in 2022, but the critical distribution route was launched in 2024Q3 and remains in rollout

1. On-Shopee SPayLater (launched in 2019~2022)

On-Shopee SPayLater is the most mature credit product, and it’s the one with the slowest growth rate. Despite that, we can still expect it to grow much faster than Shopee’s GMV growth over the next few years.

On-Shopee SPayLater is basically the low-hanging fruit. Even till now, a huge percentage of Southeast Asians are still underbanked, so for many users, SPayLater is their first formal credit product. In Indonesia, the credit card penetration rate is merely 7% of the population. Penetration is higher in Thailand and Malaysia, but still well below levels seen in middle and high-income markets such as Taiwan and the United States.

User penetration is also still in the early days. By the end of 2025, Monee had only 37M active credit users, implying that SPayLater had yet to reach even double-digit penetration across Shopee’s 400M+ active buyers.

On the GMV side, penetration also remains modest. In Q2 2025, On-Shopee SPayLater GMV penetration was in the mid-teens on a blended market basis. Meanwhile, at the end of 2024, roughly half of the $5.1B loan book consisted of On-Shopee loans.

The Different Cycles of On-Shopee SPayLater

Since PayLater was launched at very different times across markets, penetration rates also vary meaningfully by country. In Q3 2025, SPayLater GMV penetration ranged from single digits to over 30% in more mature ones.

Excluding Taiwan and Singapore, where underbanking is less of a structural driver, Brazil should be the earliest-stage market. In Q3 2025, penetration there was still in the single-digit to low-double-digit range. Brazil was a decent contributor to Monee’s loan book expansion in 2025, and its on-Shopee SPayLater business still appears to be in a phase of very rapid growth.

Over the long term, I expect on-Shopee SPayLater GMV penetration to exceed 50% across all markets except Singapore and Taiwan. That would imply roughly 200% additional upside in GMV penetration for on-Shopee SPayLater. As a result, I think it’s reasonable to expect on-Shopee SPayLater to continue to grow faster than Shopee’s GMV in the coming years.

2. Cash Loans: launched in 2020~2023

Cash loans reflect Monee’s ambition to take a bigger slice of day-to-day credit. They are currently the dominant form of off-Shopee loans. By the end of 2025, cash loans accounted for roughly half of total loans outstanding. Growth is also materially faster than that of on-Shopee SPayLater: cash loans outstanding almost doubled year-on-year as of Q2 2025.

Monee’s cash loans can be seen as an extension of SPayLater, aimed at capturing more day-to-day borrowing demand outside the Shopee ecosystem. The nature of cash loans makes the average loan balance considerably larger than for SPayLater. Even so, Monee’s current loan portfolio is still largely made up of short-term products designed to meet immediate cash needs.

Compared with traditional bank loans, Monee’s cash loans are more like “small dollar loans,” with considerably smaller ticket sizes and very short payback periods, mostly around 3~12 months. The application process is also much easier and faster, with limited human involvement.

Monee is now actively scaling up borrowing limits and extending loan tenures as it moves further into segments centered on prime users.

From the Q3 2025 earnings call:

In Indonesia, we have been offering higher limits and longer tenures to attract more prime users who demonstrate strong repayment behavior. Loan sizes can typically range from a few hundred dollars to over a thousand dollars, allowing us to serve users with larger financial needs. Building on this success, we have similarly expanded access to prime users in Thailand and Malaysia, where user adoption is going up quickly.

Looking further ahead, I expect Monee to enter areas of secured loans like car loans and mortgages, but I guess it will take another few years. There is still a lot more to explore in the unsecured loans sector, especially in the more prime user case.

3. Off-Shopee SPayLater : Launched in 2022~

The business logic behind off-Shopee SPayLater is very different compared to on-Shopee SPayLater and cash loans. Off-Shopee SPayLater is basically a brand new category!

On-Shopee SPayLater and cash loans are both businesses that mostly rely on Shopee distribution. Shopee merchants in most countries are forced to accept SPayLater, so when scaling on-Shopee SPayLater and cash loans, Shopee does not need to worry about merchant acquisition.

Off-Shopee SPayLater, on the other hand, is built as a substitute for credit cards. It’s a product that involves merchants, consumers, and lenders. It’s a product that needs an enormous off-Shopee distribution to succeed. Monee needs to build the network from scratch.

The SPayLater Distribution Problem

Off-Shopee SPayLater was first introduced in Indonesia in 2022, but after years of recruitment, the number of off-Shopee merchants accepting SPayLater remained limited.

But things started to change in 24Q3 when SPayLater was integrated into Malaysia and Indonesia’s national QR code systems, DuitNow QR and QRIS respectively.

With this integration, the merchant onboarding process became a lot simpler; the merchants no longer need a full and direct integration with SPayLater. As a result, the places that could accept SPayLater have grown exponentially since then, and so have the off-Shopee SPayLater transactions.

By the end of 2025, off-Shopee SPayLater loans had grown over 300% year-on-year, accounting for over 15% of Monee’s total SPayLater portfolio. In Malaysia, close to 30% of SPayLater usage was already off-Shopee, up nearly 10 percentage points from just over 20% two quarters earlier.

Now SPayLater has been integrated with the national payment rails in Malaysia, Indonesia, Vietnam, Thailand, and the Philippines. I think we are safe to expect off-Shopee SPayLater to sustain triple-digit growth in the coming years.

Note: Every country’s national payment rails are different, but they generally drastically reduce the friction for merchants to accept credit and improve interoperability across banks and digital wallets.

The 10x growth ceiling of off-Shopee SPayLater

The off-Shopee scenario represents a significantly bigger market than the on-Shopee scenario. In China, more than 95% of Alipay’s transaction volume came from outside Alibaba’s own properties, and in the U.S., more than half of overall card-based transactions were made through credit cards.

Off-Shopee SPayLater is a three-side marketplace, and Monee is now competing to become one of the top consumer credit options. Monee is not only actively onboarding merchants, but also expanding credit use cases to extend its presence in higher-ticket categories and compete with more traditional forms of credit.

From the 24Q3 and 25Q4 earnings calls, respectively:

One recent example of an off-Shopee use case was to facilitate consumer’s large ticket purchases of mobile phones in offline retail stores.

We also expanded the use of SPayLater into higher-ticket offline categories, such as electronics and two-wheelers.

I think the recent ramp up of the offline scenario to build up the network effect of off-Shopee SPayLater is one of the main reasons Monee’s sales and market expense have been exploding.

Going to a broader market

Now Monee has successfully scaled its active credit user base to 37 million. Starting from 25Q3, the management has changed the approach from a “whitelist-based” approach to a broader “all-can-apply” approach and has been progressively rolling it out across markets for both SPayLater and cash loans.

According to management, the strategy has been a success! In 25Q4, Monee landed a record 5.8M new credit users, and the new cohort generally has positive unit economics. I think the shift to the all-can-apply approach is also another of the core reasons S&M spending increased so much.

Based on the 24Q3 earnings call, the management team said they generally target a payback period of less than 12 months. That suggests we should see the results of these initiatives fairly soon.

Subscribe now

The Next Stage of Monee: The Default App for Digital Life

Monee’s goal goes far beyond becoming a personal finance super app. The ShopeePay app has a much bigger ambition. It aims to become the default app for digital life, with financial services being just one part of that journey.

(See the screenshot of the Thailand ShopeePay App, translated by Gemini)

The Indonesia standalone ShopeePay App was launched in February 2024. By March 2025, it had surpassed 30 million downloads, and it has now reached at least 50 million downloads on Android alone. Today, more than 20% of ShopeePay’s monthly transacting users are using the standalone app.

The standalone ShopeePay app is basically an unbundled version of functions from the core Shopee app, and it signals Monee is entering the next stage of its expansion. The standalone app reduces payment friction in non-Shopee scenarios, and simplifies the UI design and frees up space for other lifestyle services.

From the screenshot, we can see users can book tickets, hotels, movies, order food, pay bills, and apply for credit products like SPayLater and cash loans. Everything can be done in one single app! In certain countries, users can also buy insurance, invest in mutual funds, and access other financial products.

For now, however, the services provided by Monee are still rather shallow, and the app’s capabilities remain limited. For instance, just like Grab, its hotel booking function is powered by Agoda. Its other services are also not very competitive compared with other standalone players.

But at this stage, that is not the most important point. Services outside of consumer credit are not Monee’s core focus. The standalone app is currently Monee’s way of improving conversion for its credit products and attracting more credit users from outside the Shopee ecosystem.

Monee is clearly thinking about a much broader picture than just credit. Once the business becomes more mature, it will likely make other services the next area of focus.

Note: The standalone ShopeePay app is currently available in only four countries: Indonesia, Thailand, Malaysia, and Vietnam. In other countries, ShopeePay and related services still remain inside the Shopee app and have not been separated. Interestingly, Indonesia was not the first market to launch the standalone ShopeePay app. There is also one exception: Taiwan is the only Shopee market without ShopeePay.

Another Marketplace

Shopee is a marketplace for physical goods, and Monee is Sea Limited’s ambition to become the marketplace for digital services.

In Southeast Asia, or in most developing countries, supply remains the biggest constraint in a marketplace business. Shopee first solved this supply problem by offering zero commission fees and free shipping to sellers to kick-off the marketplace.

In the beginning, sellers came to Shopee for free shipping to serve their existing customers. With more and more supply, the customers started to stay in the network. Interestingly, the sellers that first used Shopee as a tool for free shipping then got trapped by the consumer traffic and then eventually became sellers on the platform.

It’s similar to building a marketplace for digital services. Take credit services as an example: without existing credit history, traditional banks and credit funders have no idea who is creditworthy. Monee is now the intermediary between the two ends. On one end, Monee leverages its own balance sheet to fund the existing demand. On the other end, Monee trains models to build up financial records to attract future supply.

Now, only about one-tenth of the existing loans are funded by outside lenders. I believe the portion of loans funded by outside funds will grow substantially in the near future, especially when Monee enters more prime-oriented credit. Once that happens, it will become much clearer that Monee is building not just a lending business, but a marketplace for credit.

The Other Marketplaces Yet to Be Built

Among all the other marketplaces that could still be built, digital finance is the most obvious category. Mortgage lending platforms, digital banking platforms, financial advising platforms, insurance-matching platforms, and etc. are all huge opportunities that Monee may try to capture.

There are also a very decent number of O2O services that are essential to day-to-day life but remain highly fragmented in this market or haven’t even been properly developed yet. Monee will probably leverage its distribution to build an aggregation layer across many of these O2O use cases.

I don’t know how well Monee will perform in the future, but Alipay is a strong reference point for the ShopeePay app. It has successfully built marketplaces for many financial services and has also become an aggregation layer for a wide range of services that are relatively commoditized, yet still essential to everyday life.

Closing Thoughts

I think compared to eCommerce, I know a lot less about fintech in general; the whole thing isn’t as easy to write as the Shopee post. Even when I already know what I want to say, checking every detail takes much more time than I expected. But overall, I think I have successfully mapped out the long-term trajectory of Monee’s business and its path toward building another marketplace.

Another thing I want to point out is that I think Monee’s credit risk is a bit overstated. It’s fair to say that Monee has high credit risk, but compared to traditional lending, it is actually already pretty immune to the macro environment. Because its loans have very short tenures and small ticket sizes, management can adjust pricing rather quickly.

And besides, even if all the loans on the balance sheet suddenly default, it also won’t impact Monee’s long-term profitability!

From a macro perspective, the real thing to worry about is that when default rates rise, Monee’s ability to generate revenue will be severely impacted, and it could also undermine years of trust built with future loan funders.

Another thing I haven’t mentioned in the article is that Shopee‘s and Monee’s growth are a bit contradictory. Monee’s current model of funding credit through its own balance sheet is not very sustainable if Shopee does not generate enough profit.

Monee’s digital bank deposits are also not growing fast enough. In the past year, more than $2.7B of loans were funded purely by Sea Limited’s balance sheet. If Sea Limited doesn’t want to shrink its cash balance drastically, it will need to decelerate the speed of loan growth.

Garena

I think compared to fintech, I have even less to say about gaming. I don’t play games, and I know very little about the sector. But despite that, I think Garena also had a wonderful year. Bookings were up 37%, and management is guiding for another double-digit growth year in 2026. At least in the short term, I don’t think gaming will be much of a drag.

My thoughts on gaming are that games are very niche and work in paradigms. The same genre will have a few winners, and the winner will last a very, very long time. If you look at the top games in the world, nearly all of them were first launched more than a decade ago. Good games can become evergreen franchises, and I believe Free Fire is already one of them. Because of that, I’m not too concerned about its long-term sustainability.

In the past few years, the management team seems to have cracked the code for driving growth through deep collaboration with other IPs, such as Squid Game and NARUTO. I believe in the future, the growth trajectory will also highly depend on these IP collaborations. So bookings and revenue would probably fluctuate more accordingly.

Thanks for reading!

Another side note. I’m very grateful to have received quite a number of messages from fund managers and thoughtful individuals offering pushback and valuable insights. If you fall into this category, please feel free to shoot me a DM. I’d love to stay in touch.

If you are worried about taking up too much of my time, or if you simply want some help speeding up your understanding of eCommerce, or want to know more about my investing philosophy, you can now also reach out to request a paid expert call. I have done a few already, and the feedback has been very good.

I’m very grateful to have this opportunity.

In 2025, Shopee served more than 400M buyers. Conservatively speaking, Shopee now has over half of Southeast Asians and one-third of Brazilians as active buyers.

These numbers look even more impressive if you consider that ~18% of the Southeast Asian population still has no access to the internet and only ~70% of the population is over 18 years old.

I think it’s pretty safe to say Shopee has already penetrated 70%+ of the addressable population and has nearly hit the ceiling in terms of “traditional” active buyers. Now Shopee’s job in most markets has shifted from acquiring new users to boosting buying frequency.

This is exactly what Shopee is doing with the Shopee VIP membership program: buyers who join see their average spend rise by 30~40%. But this program does come with a cost.

This post won’t be a detailed financial analysis of the quarterly business. (I think there are already quite a lot of people who write good reviews from this angle.) Instead, I will focus more on the business logic itself.

There are still a lot more details behind Shopee’s management team’s thinking. Let’s start with some short bullet points on my thoughts on 25Q4 earnings.

TL;DR

• Shopee fundamentals are solid, but profitability is suffering from the Shopee VIP program’s success. This success is what hinders 2026 profitability.

• The 25% GMV guidance is good. We can now expect GMV growth to reach 30% in 2026. With take rate still improving, overall Shopee revenue growth should be sustained at 35%+.

• Monee’s credit business passed its inflection point in 24H2, and it should continue to grow rapidly in 2026.

• The increase in Monee’s Sales & Marketing spending is mainly tied to expansion into new categories. The revenue contribution will show up with a lag in later quarters, so I don’t think this is something to worry about.

• Garena didn’t perform well in 25Q4, but it’s also a symptom of the huge success in Q3 driven by Free Fire’s IP collaboration with Squid Game and NARUTO SHIPPUDEN Chapter 2.

Shopee & The Two Ways to Boost GMV

There are two ways for Shopee to boost GMV. First, acquire new users. Second, increase purchase frequency among existing buyers.

Acquiring New Users Through Discovery-Based Shopping

Shopee has become a household name in Southeast Asia a long time ago, and had already saturated most of its addressable population when its monthly active users surpassed 300M in 2023.

The marginal population that hasn’t tried Shopee is the ones that don’t have the strong intention of using search-based eCommerce in general. Pouring in more marketing dollars won’t do the work. What Shopee needs is to build a new set of products to satisfy these users.

Shopee is targeting people who are active on social media but not on eCommerce sites. They are the population that relies on passive recommendations instead of strong search intent. Lots of the older and younger population falls into this category.

Older users are accustomed to traditional shopping habits, while younger users may not yet have developed strong shopping habits at all. They will only try eCommerce when they see something very tempting but they can’t access it through traditional channels.

This is the population Shopee is capturing right now. Shopee began an aggressive push into content in 2023, and now more than 20% of orders are attributed to content. Shopee’s collaboration with YouTube in 2024 is another step further in this equation. Shopee is actively acquiring the users and monetizing the content that doesn’t even originate on Shopee.

The collaboration with YouTube was a huge success. Orders attributed to YouTube have more than tripled year-over-year. The new collaboration with Meta, which was launched last October, is also growing explosively.

Driving Buyer Frequency

Increasing buyer frequency is basically the most effective way to further boost Shopee’s GMV growth, and I believe it will basically be the main theme going forward.

E-commerce is a retail business. Just like physical stores specialize in different categories, e-commerce works the same way. Different products require vastly different service levels, sourcing capabilities, and cost structures.

Shopee still has meaningful gaps across several areas. For instance, most products on Shopee still have delivery times of over 24 hours, which means it cannot serve products with urgent needs. It also lacks the right sourcing and service levels in many verticals, including home appliances, furniture, electronics, motor parts, and fresh and frozen products. If Shopee can improve in these verticals, customers will use the platform much more frequently.

Cost to Serve

Another part of the equation is the cost-to-serve. Cost to serve is mainly made up of logistics costs and payment processing fees. Although shipping fees are usually quite affordable, most people are still unwilling to pay them for low-value items.

With lower cost to serve and faster shipping speed, Shopee can unlock a lot more use cases and compete more directly with supermarkets. Shopee’s VIP membership program and its new fulfillment push are both initiatives to address these issues.

Shopee VIP Program

The Shopee VIP program is Shopee’s flagship membership program. Members who pay a subscription fee of $1 to $3 USD, depending on the country, receive lower shipping thresholds, additional discounts, and some partner benefits from external providers (e.g. ChatGPT Go subscriptions, bike sharing subscriptions, etc.)

The program is designed to boost user loyalty and increase spending. So far, the program has been a success. Members who subscribe spend 30%~40% more on average, on an absolute GMV basis. The subscriber base is also growing exponentially, from 1M in 25Q1 to 7M in 25Q4, roughly doubling every quarter.

But the success comes at a cost. The lower free shipping thresholds and discounts that Shopee gives out materially hurt Shopee’s profitability. They contributed meaningfully to the 3% drop in gross margin and a spike in sales and marketing expense in 25Q4 compared to the last quarter.

Note: Sea Limited used to mix Shopee & Monee’s gross profits together and started to disclose them separately this quarter)

Note2: In my database, “EC & Others” includes the Provision of Credit Loss.

Is this the first time Shopee has done it? What’s the benefit?

This kind of investment cycle isn’t new; Shopee has done it multiple times before. The most notable example was 23Q2 when they re-started a huge cycle of investments. This time is similar but probably on a much smaller scale.

The main idea of another free shipping push is to pull forward order volume and further reduce logistics costs. We know that logistics is a volume game, and shipping costs still account for over 50% of Shopee’s marketplace revenue. The most effective way of reducing shipping costs is simply to increase order volume.

Building a durable moat

In my opinion, Shopee’s move is both offensive and defensive against TikTok Shop and other players such as Lazada, and in my view, Shopee’s move is correct. Now is a critical time for Shopee to build a durable moat.

In my post “The Biggest Myth of eCommerce Logistics” I have mentioned when logistics turns into a durable moat:

The durable moat basically boils down to the scale of the logistics network, but most importantly thing is than boil down to the eCommerce platform’s “parcel” share in the market.

Logistics incurs huge fixed cost, it’s one of the two main cost of operating an eCommerce platform. If the vertical logistics network can process a lot more parcels than the second biggest logistics player, than the eCommerce platform will enjoy a much lower cost to serve in the long term.

Now with TikTok Shop growing really fast in the region (I’m expecting mid-double-digit GMV growth), and with its main logistics partner J&T Express growing parcel volumes by 73%, Shopee is at a critical point in building a more differentiated logistics solution.

J&T Express now processes about 26.5M parcels per day in Southeast Asia. SPX, on the other hand, processes 30M+ parcels daily (including Taiwan and Brazil parcels).

The hidden benefits of subscriptions

The benefits of Shopee VIP go beyond logistics. It also serves as a tool for customer price discrimination and a less visible way to avoid direct price comparison across platforms. Once Shopee VIP reaches sufficient penetration, Shopee can begin off-load promotional discounts to qualified merchants in exchange for higher traffic exposure. I believe the off-load process will begin this year. This would slightly diminish the sales and marketing costs that Shopee is bearing.

What size can Shopee VIP become?

By the end of 2025, the program had reached 7M subscribers, which is less than 1.75% of the total 400M+ active buyers. The number of paying subscribers is now growing super fast. At the end of 25Q3, the subscriber count was merely 3.5M, meaning it roughly doubled in 25Q4.

In the markets where VIP services were launched earlier, VIP members now contribute more than 15% of market GMV. But what can Shopee VIP become? To answer that, let me lay out the differences between e-commerce subscription programs. These programs broadly fall into two buckets: one is centered on free shipping, while the other is built around extra perks.

Amazon Prime / Coupang WOW (Free Shipping oriented)

Both Coupang WOW and Amazon Prime are paid membership programs for reduced shipping fees. On these two platforms, the subscription works somewhat like a Costco membership: the paid membership is effectively required to shop on the platform smoothly.

In both Amazon U.S. and Coupang South Korea, paid subscribers contribute 90%+ of the overall GMV. In other words, paid subscription is an integral part of the experience. In the U.S., 80%+ of Amazon’s active buyers are paid subscribers. In South Korea, although only about 70% of Coupang buyers subscribe, a subscriber spends 800% more than a non-subscriber.

(I know both Amazon Prime and Coupang WOW provide services other than free shipping, but most people subscribe to Prime or WOW for free shipping, not for other perks. Amazon and Coupang bundle those extra services into the membership package mainly to reduce churn.)

Alibaba 88VIP / JD Plus (Extra perks oriented)

On the other hand, Alibaba 88VIP or JD Plus isn’t a service centered around reducing shipping fees; instead, it’s for extra perks. Customers subscribe for special discounts and extra services such as VIP customer service hotlines, electronics repair coverage, and streaming subscriptions.

This is because in China, free-shipping thresholds are usually easy to meet, or platforms already build free shipping into most products. Even when customers don’t subscribe, the experience is still really, really good. The membership functions more like a loyalty program than a core part of the shopping experience.

88VIP and JD Plus both have paid subscriber penetration of roughly 5–7% of the active user base, and they are estimated to contribute about 25% and 35% of Alibaba and JD’s GMV, respectively. A paid member spends 5x as much as an average non-paying user.

Shopee VIP

Shopee VIP isn’t like any of those subscriptions. It sits somewhere between the two categories. Shopee without membership is still really compelling; free-shipping thresholds aren’t as low as those of Chinese e-commerce platforms, but they are still rather easy to meet. It’s also not as unusable as Amazon or Coupang without a subscription.

From my perspective, the VIP subscription isn’t essential, but rather something nice to have. Take Shopee Taiwan as an example: with membership, the free-shipping threshold has dropped from $6.5 to $1.6. With the subscription, it does unlock access to a decent amount of low-value items.

Shopee VIP’s discounts are also quite generous, and the subscription can pay for itself after a few orders. I think most regular customers can get a decent amount of value out of the Shopee VIP program. And with lower friction and lower prices for products, customers will definitely spend a lot more money.

Since the membership value proposition sits between these two buckets, I believe the membership penetration rate should be at least 7% and have the potential to hit 25% of overall active users. I’m now modeling 28M paid subscribers by year-end 2026, which implies average quarterly growth of 40%–50%.

The bear case for Shopee Profitability in 2026

The bear case for Shopee profitability is basically that Shopee VIP is too big of a success. Shopee VIP may be so successful that it becomes a drag on profitability. Shopee added only 3.5M members in Q4 2025, and it might have contributed meaningfully to the 3% gross margin drop. If Shopee can’t reduce the cost-to-serve and offload the burden to merchants fast enough, Shopee profits might suffer from its own success in the VIP program.

Note: To be clear, I don’t think the entire 3% gross margin decline was caused by the VIP program. I believe the pursuit of faster shipping, which I discuss in the next section, also contributed to the gross margin drop. That said, I do think the VIP program accounted for a very meaningful portion of the decline.

Shopee Road to Increase Service Levels

Increasing service levels is what Shopee is now focusing on. Instead of being a pure marketplace that sells commodity products, Shopee is actively increasing service levels in order to differentiate itself from other players.

One of the more critical differentiators in the buying experience is shipping speed. This is what Shopee is actively pursuing.

Like what I mentioned in “Are Amazon, Coupang and MELI logistics the same? The 4 kind of eCommerce logistics” :

The most important factor in delivery speed is where the underlying inventory is stored. The fewer steps and the less that the parcel changes hands, the faster the delivery experience.

Besides routing parcels in small batches, Shopee is increasing delivery speed in three main ways. All of these strategies focus on removing unnecessary steps and bringing inventory closer to customers.

First, Shopee is making greater use of gig workers from its ShopeeFood operation. It has expanded its partnerships with supermarkets to fulfill groceries within one hour. For normal products on the platform, when the seller’s inventory is stored in urban locations or in suburban warehouses near customers, customers can now choose to let ShopeeFood workers pick up the products from the warehouses and deliver them directly.

The results have been very positive. In major metropolitan areas such as Bangkok and Jakarta, buyers using faster delivery services accounted for a double-digit share of order volume. By the end of 2025, buyers who adopted instant and same-day delivery were also spending around 15% more on average.

Second, Shopee is expanding its fulfillment network across all of its markets, especially in Brazil. As noted in the 25Q4 earnings call:

With our delivery capability well scaled, we started to roll out fulfillment service in various markets across 2025. We are seeing encouraging adoption trends from both buyers and sellers with double-digit order penetration in some markets. In 2026, we plan to expand fulfillment further across all our markets and aim to double our fulfillment order penetration by the end of the year.

Fulfillment brings tangible benefits when it comes to serving small sellers that do not have their own 24/7 facilities, and it is a very natural extension of Shopee’s C2C logistics network. Shopee is doing the equivalent of FBA (Fulfillment by Amazon).

As I mentioned in my earlier piece, one of the main advantages of Amazon’s fulfillment network is its built-in demand forecasting system. With Shopee managing the inventory, it can eliminate the need for merchants to send out packages themselves and move inventory closer to customers through cheaper delivery methods before an order is even placed. This reduces the need for last-minute expedited shipping and, in turn, lowers delivery costs.

Third, my suspicion is that Shopee is increasing its first-party offerings to gain more control over the product journey and to address the insufficient supply of low-value daily essentials in the marketplace. In Taiwan, we can see a very clear trend over the past eight months that Shopee is driving up the visibility of its own first-party offerings.

I think Shopee’s increase in the share of first-party offerings makes perfect sense. As I have mentioned in my old piece, the mid-sized warehouses that Shopee is now deploying work best with a combination of first-party offerings. In the China quick commerce war article, I had also mentioned that Meituan and other quick commerce operators are actively supplying first-party offerings in order to build up supply in the quick commerce marketplace.

I think the pursuit of first-party products has started to reflect in 25Q4 earnings. First-party sales grew 24% QoQ, and there was also an increase in the gross margin of first-party sales in 25Q4.

Note: Before the first-party daily essentials push, its first-party product mix likely consisted mostly of very low-margin branded consumer electronics. And for products like the iPhone, it was often used to attract more high-quality buyers, so the margin was extremely thin.

Note2: There is a major difference between Shopee’s fulfillment model and those of Amazon and MELI. The logic of Shopee’s marketplace is that each seller operates independently, and so Shopee fulfillment services work like an outsourcing facility that helps hold and send their own inventory. Products from different sellers are not mixed in the same box, whereas on Amazon and MELI they can be.

Note3: I don’t think the penetration of fulfillment will become too high in Southeast Asia, but it will be quite high in Brazil since its logistics ecosystem is still rather nascent.

Service level besides fast delivery

Faster delivery is one of the many components of a good service-level eCommerce platform. But there are still a lot more areas to improve. I believe this part is a secondary focus for Shopee but will probably become the first priority after it has done enough to build up a fast-delivery fleet.

These parts are specialized services for premium brands and other vertical categories. This is one of the key ways Tmall differentiates itself from Taobao. For instance, when shopping for medicine online, platforms should be equipped with pharmacists or doctors to help customers make the right decisions. When shopping for alcohol, the platform needs to verify product authenticity and enforce age restrictions for buyers. When buying electronics, platforms need to have trustworthy guarantees and also reliable trade-in policies, and when shopping for furniture and major home appliances, platforms need to provide seamless installation and maintenance.

Shopee currently offers only the most basic guarantees through Shopee Mall. There is still a lot more for Shopee to differentiate itself through high-quality services.

I believe those specialized and vertical services are the true long-term differentiator. I see fast delivery still more like a commodity; it’s something competitors can rather easily copy. Category-specific service quality is likely to become the more important long-term differentiator.

Closing thoughts on Shopee

I think Shopee executed well in 2025; things are basically all on the right track. Although I did expect 25Q4 GMV growth to be ~30%, 28.3% growth is still acceptable. Unfortunately, the market obviously hated Shopee doing the right thing with the idea of reinvesting amid uncertainty.

For the folks who ask about TikTok Shop, I personally believe TikTok Shop is a third-generation eCommerce platform that is more advanced than Shopee’s second-generation model.

I think TikTok might become the leader in some of the smaller Southeast Asian markets where Shopee is less aggressive. But I believe Shopee will be able to maintain its dominance in Indonesia, which accounts for about 40% of the SEA eCommerce market. I also believe Shopee will be able to maintain its overall leadership in SEA over the long term.

One of the critical differences between China and Southeast Asia is that in China, there are only a few traffic sites, and nearly all of them run eCommerce themselves. But in Southeast Asia, TikTok is just one of the five major traffic sites.

Now Shopee has become the default partner of Google and Meta. I believe the overall traffic of YouTube and Meta combined is larger than TikTok’s. But it will take some time for Shopee to ramp up and monetize its collaborative content business.

Social commerce isn’t just adding a link to content for audiences to shop. It’s about nurturing and creating engaging content designed for shopping, commonly known as Shoppertainment. This kind of content is still really scarce on both platforms.

Now the orders that are attributed to YouTube have more than tripled year-over-year, and the collaboration with Meta has also been ramping up. I believe social commerce / Shoppertainment will be a huge incremental market for Shopee in the future.

Overall, I think Shopee is on the right path to success.

Growth Rate & Agent Commerce

Some of my audience are also asking about my thoughts on Brazil and agent commerce.

My view on 2025 GMV growth is that Brazil is growing somewhere around 60%, Taiwan is growing in the low teens, and Southeast Asia is growing in the mid-to-low twenties.

I think this trend will continue in 2026, with Brazil’s growth slightly moderating and Southeast Asia’s growth slightly accelerating.

About agent commerce, I think it will be quite positive for Shopee, especially because TikTok already has its own chat product (Doubao), which is actively competing with Gemini and ChatGPT. Even setting aside that agent commerce benefits the long tail of products, this will still benefit Shopee in general when it comes to the agent-driven competitive landscape.

Also, I personally think the search intent of ChatGPT/agent commerce won’t be too disruptive to Shopee for the next few years. It will probably first disrupt the platforms with fewer SKUs and higher price points. The GPT shopping interface is still not too pleasant of an experience; the UI is simply not there yet.

That said, I do think it is already quite good at clarifying user intent when the intention is still vague. Because of that, I think it could be a bit more disruptive for OTAs and food delivery, where users often need to scroll through reviews and where AOVs are higher.

That’s all for Shopee. In the next post, I’ll share my thoughts on Monee and Garena.

Thank you for reading!

Share Contrarian Perspectives

In this article, we will explore the basic ideas of eCommerce logistics and how they have evolved, and the pros and cons of different logistics models and the limits of different logistics networks (e.g., Fulfillment by Amazon / Coupang / MercadoLibre).

First, let’s start with the most traditional form of an eCommerce platform with their four respective logistics systems that power their unit economics and user experience. Then in the bottom half, I will lay out what the traditional logistics have evolved into.

The Four Traditional eCommerce platform

In traditional eCommerce, there are four main platform types, each paired with a different delivery model that reflects its unique value proposition. Parcel volume in the graph is based on the current parcel volume in China.

Value Proposition of eCommerce platforms.

The main separation of different eCommerce platforms is the amount of SKUs they provide, whether the platform owns and holds its own inventory, and how time-sensitive the products are.

The most important factor in delivery speed is where the underlying inventory is stored. The fewer steps and the less that the parcel changes hands, the faster the delivery experience.

1) The Traditional C2C Logistics (Cross-Docking)

The main difference between traditional C2C logistics compared to B2C logistics is that fulfillment isn’t in the C2C logistics equation, and the packages and parcels are sent from places all around the country.

The biggest downside of C2C logistics is slower, less consistent delivery and packaging because orders are fulfilled by independent merchants. In practice, parcels are shipped from a city to a remote area and then shipped again to another city. This involves many more steps compared to B2C logistics.

The unpredictability is one of the main reasons consumers prefer B2C platforms, especially in the early days when logistics infrastructure was still developing.

2) The Traditional B2C eCommerce logistics

At this time, the main reason B2C platforms ship eCommerce parcels faster compared to C2C logistics is because the products are fulfilled by a warehouse that operates 24/7. Let’s use Amazon in 2010 as an example.

Amazon in 2010

In 2010, when Amazon was already 15 years old, Amazon had about 25 warehouses in the U.S. When an order came in, Amazon would identify the customer’s location and choose the most efficient shipping route.

The order would be fulfilled from one of those warehouses, often in remote areas where storage costs were lower. Then third-party carriers (e.g., UPS, USPS) would handle the parcel and deliver it to the customer.

Basic parcel flow: 1) Warehouse (remote regions) → 2) Logistics hubs (suburban areas) → 3) Customer

The good part of B2C logistics is that the delivery speed is much faster than C2C logistics since they skip the first two steps. But the downside is that the consumers using B2C platforms are usually a lot more time sensitive. To send the parcel within a 48-hour time frame, as the platform promises, airplanes are often used to expedite the shipping speed. Also, in the last mile, B2C logistics usually arrives in smaller batches compared to C2C logistics in order to increase the speed and delivery experience. The cost of expediting makes the delivery much more expensive.

3) Traditional Grocery eCommerce

Grocery eCommerce is mostly B2C; the main difference is the product selection and the experience they provide. Grocery platforms usually offers a much more limited selection. Unlike B2C platforms that have 2M+ SKUs and are limited to dry goods. Grocery-oriented eCommerce platforms usually have a built-in cold chain and will only provide daily essentials; the SKU count is usually limited to 30K SKUs. The SKU count they provide is similar to a huge grocery store (i.e., Walmart Neighborhood stores).

Grocery eCommerce’s value proposition is to provide users with an online grocery shopping experience; the products it sells are high-frequency purchase items.

Grocery eCommerce warehouses are usually located in suburbs to enable same-day or next-day delivery. However, suburban real estate costs and the need for a built-in cold chain make storage significantly more expensive than traditional B2C eCommerce logistics. At the same time, a smaller SKU range and tight delivery windows reduce order density. The delivery cost is much higher than tradition B2C eCommerce logistics.

4) Traditional Food Delivery Platforms

Food delivery platforms are like C2C marketplaces, business-wise, but there’s a catch: merchants (aka. restaurants & physical stores) on food delivery apps are region-based, and they can only serve customers within a 3-mile radius.

Food delivery platforms can offer ~30-minute delivery because products are fulfilled from a nearby physical location instead of a warehouse in the suburbs. They also run smaller batches per route, which helps keep delivery times down.

Because delivery is expensive, traditionally, these platforms focused on high-margin, service-heavy categories (mainly restaurants) where product gross margins are often 60%+.

Also, because the product needs to be stored near the customer, there are a very limited number of products they could inherently sell on the platform.

The complementarity of platform value propositions and shared logistics networks

The value proposition of each platform was drastically different during the first 15–20 years of eCommerce platform development. Different platforms basically served different profiles and different use cases, and they also used a good portion of the same underlying network.

For instance, the last-mile services of C2C & B2C eCommerce are mostly the same (e.g., all platforms rely on UPS). The only difference is that B2C packages will usually pay a premium for the package to be sent faster, and the C2C packages will be the supplementary packages that will only be sent when there is extra space in the delivery route.

Subscribe now

How does the logistics evolve? What has change?

As eCommerce platforms mature, the selection of all platforms has increased a lot, and with huge volume in the logistics network, the cost-to-serve of all platforms has drastically decreased.

The logistics system has evolved into infrastructure purpose-built for eCommerce parcels, and they found out that the best way to decrease cost is to shorten the last mile of delivery. Overall, eCommerce logistics evolution centers on two pieces: 1) Build an elastic logistics network only for eCommerce 2) Increase last-mile endpoints to decrease the last-mile distance.

1) The Modern C2C Logistics (Cross-Docking)

eCommerce’s small and dispersed packages make the last mile even more expensive. It’s the part that can’t be solved with bigger trucks. To this day, J&T Express, one of the leading C2C logistics companies in China & Southeast Asia, still spends 60%+ of the cost on first and last-mile delivery.

Modern C2C logistics uses a much denser logistics model by adding multiple small network branches (100m2~ 300m2) in cities to effectively decrease the last-mile and first-mile routes.

C2C logistics companies not only build out network branches, they also build a huge network of service endpoints by leveraging idle human resources to handle self-pickup or drop-off of packages.

Today, an average C2C logistics provider in China operates 20K+ network branches. Customers can choose from 400K+ service points nationwide. Alibaba’s Cainiao alone offers 170K service points—excluding smart lockers.

(There are multiple layers of sorting centers; the first level of sorting happens near the big cities.)

C2C logistics is the kind of logistics that improves the most with an increase in package volume. Not only does the cost per parcel drop significantly with increased volume, but speed has also improved with denser routes. This is paired with the professionalization of C2C marketplace sellers. C2C packages in China can now be delivered in 24 hours if the product is already stored in a warehouse near the city.

At the same time, a growing number of third-party providers offer fulfillment and inventory placement for C2C sellers, which makes B2C logistics less differentiated. ( More detail explanation in a moment)

Note: these C2C fulfillment providers are essentially third-party “FBA-like” services. Operationally, they look very similar to B2C fulfillment, but they’re often smaller, more flexible, and more customizable.

2) The Modern B2C Logistics

The biggest difference between the traditional and modern ones is the use of a multi-layer warehouse system. In the 2010s, Chinese eCommerce companies started to build out a massive fleet of mid-sized warehouses in the suburbs that only carry 100K~200K of the most frequently used SKUs.

Every huge regional warehouse is then paired with multiple mid-sized or small warehouses. Most packages are sent by the mid-sized warehouses; only long-tail products are sent by the huge regional warehouses.

The mid-sized warehouses that store inventory near the city effectively decrease the last-mile distance (cost) and drastically increase the speed of frequently used products. With a multi-layer warehouse system, same-day and next-day delivery become the standard for B2C logistics.

Amazon v. UPS.

Amazon started to build out its last-mile fleet in 2015 with the launch of Amazon Flex. There are three main things Amazon has improved, cost-wise, compared to UPS.

B2B v. B2C

UPS is a logistics network that is built for B2B services; the delivery process is optimized for rather fixed routes with similar big-box packages. eCommerce routes are inherently unstable, and package sizes are small and variable.

Full Time employees v. gig workers

UPS is centered around reliability; it utilizes full-time employees and “expensive” union employees to deliver important packages to business customers, and the process usually requires in-person signing and receipt. Amazon packages, on the other hand, don’t need that level of assurance and reliability; the only thing the delivery worker does is drop off the package.

eCommerce is also a business with huge spikes in volume depending on holidays or shopping festivals. Gig workers are very suitable for these kinds of jobs.

Hub-and-Spoke v. Forward Deployment

UPS is built to deliver packages from everyone to any place. Amazon is built on demand forecasting. Amazon will first move products to the nearby fulfillment center closest to the customer. In 2023, Amazon also separated the U.S. into 8 different regions, aiming to reduce the delivery miles needed to send packages.

More to note on B2C logistics:

Now it’s also common for B2C logistics companies to add another layer of small logistics hubs and service points inside cities, just like a standard C2C logistics network.

Even when packages are sent out by mid-sized warehouses, there is still a high chance that they might go through smaller logistics hubs first. It all depends on the cost and speed the platform promises.

3) The Modern Grocery eCommerce logistics

If you look closely at modern B2C logistics, you would notice that the buildout of mid-sized warehouses is like building out a traditional grocery warehouse, but with more SKUs. The warehouse location and the delivery speed are exactly the same.

In the U.S., the two biggest grocery e-commerce players are Amazon and Walmart. More broadly, many early standalone online grocery players have been acquired, outcompeted, or absorbed into larger B2C/C2C/Food Delivery platforms.

Given the tough value proposition today, independent e-commerce grocery players have now leaned toward a network of smaller warehouses with 3K~8K carefully selected items and try to differentiate themselves by providing instant delivery services, which overlaps heavily with quick-commerce platforms (covered in the next section).

4) The Modern Food Delivery platforms (Quick Commerce Platform)

Modern food delivery apps have evolved into quick-commerce platforms. In China, where quick commerce is a lot more mature than the rest of the world, instant retail has become 40%+ of the overall quick commerce market. (with the remaining ~60% coming from restaurant orders)

On quick commerce platforms, you can find not only restaurant meals but everything from clothing and electronics to groceries and everyday household items, effectively replicating the assortment of a local physical store, delivered within the hour.

The transformation is due to the massive scale the platform has accumulated over the years; it has become cost-effective to deliver more commoditized products with lower margins. The prevalence of micro-warehouses also contributes to the surge.

Still, there is a selection and pricing problem in modern food delivery apps since inventory is required to be stored near the customer and required to be sent at high speed.

and I are going to release an deep dive article on this matter within the next few days. ( Updates: now is out, click here)

The mixed use of Logistics

Just like traditional eCommerce platforms rely heavily on logistics that are not built solely for eCommerce parcels, different platforms rely on the same underlying last-mile delivery network.

Modern logistics are exactly the same in this part: with the maturity of C2C, more and more B2C packages have been sent through C2C last-mile routes, simply because it’s fast and already reliable enough.

Selection, Low Prices, Delivery speed and FBA

Selection, low prices, and delivery speed are basically three things that can’t coexist in eCommerce. Amazon tries to solve the selection problem by opening up its first-party platform to third-party sellers to solve the long-tail/selection problem.

Amazon third-party marketplace & FBA

The core idea of Amazon’s third-party marketplace is to make third-party products feel like Amazon’s own first-party offerings. Fulfillment by Amazon (FBA), launched in 2006, pushed that idea further by removing operational differences. With FBA, third-party sellers use Amazon’s warehousing, packaging, shipping, and returns just like first-party products.

Amazon didn’t stop there. Soon after launching FBA Amazon started to pool the “exact same products” (same UPC/EAN) together across different third-party sellers in order to save space and speed up shipping.

Pooling third-party inventory together obviously creates problems when some third-party sellers sell fake products. After near 20 years of operation, Amazon is planning to stop this absurd policy this year in order to fight counterfeit products.

Amazon’s third-party marketplace, at its core, operates like a consignment model; third-party sellers are just another type of supplier that is responsible for its own inventory.

FBA - Amazon, MercadoLibre and Coupang

FBA was built to improve delivery speed and consistency in the fulfillment of long-tail products. First-party platforms don’t need third-party sellers for the best-selling 1M SKUs; they can source and sell those items themselves, leveraging scale and sourcing power.

But since FBA mainly serves the long tail, it inherits the long tail’s core problem: demand is unpredictable.

Modern B2C logistics is built on demand forecasting and pre-positioning. To reduce costs, companies move items that might sell in the near future using lower-cost methods (e.g., trucks rather than airplanes) from a huge regional warehouse in a remote area to a mid-sized warehouse in the suburbs to reduce the need for costly expedited last-mile delivery.

Long-tail products are extremely hard to forecast; this is the most important limiting factor, so most long-tail SKUs will always be stored in large regional warehouses.

MercadoLibre (MELI)

MELI is the leading C2C eCommerce player in LATAM; 90% of its GMV is from third-party sellers. It’s known for an Amazon-like fulfillment and delivery network that gets products to customers quickly.

As of the end of 2024, MELI only had ~30 warehouses/fulfillment centers in all of LATAM. That footprint suggests the network still leans heavily on large regional warehouse rather than a dense, “modern” multi-tier warehouse system closer to end customers.

To this day, MELI’s average delivery time is still close to three days, and its average cost per parcel (including fulfillment) has hovered around $3.5 for years, despite massive growth in parcel volume.

The reason is structural and highly attributable to the limitations of the fulfillment model that MELI uses. To deliver fast from a regional warehouse that is far away from most customers, the need for expensive delivery methods becomes inevitable.

MELI is now changing into a more modern delivery network by building mid-sized warehouses in secondary cities and the suburbs of dense cities. But there’s a limit when you’re serving the long tail: you can’t place every SKU close to every customer without blowing up costs and complexity. (More on this in a moment)

At the end of the day, the number-one factor impacting delivery speed and cost of delivery is where the inventory is located.

Coupang

Coupang is the opposite case; most products sold on Coupang come from its first-party operations and are fulfilled through a modern B2C logistics model. In Q1 2023, FLC (Fulfillment and Logistics by Coupang) accounted for only 4% of total items sold.

By the time Coupang IPOed in 2021, it had already built 100+ warehouses, putting about 70% of the population within 7 miles of a Coupang logistics hub. That coverage translated into speed: 99%+ of Coupang’s first-party/FBA orders were shipped the same day or the next day.

Now 90%+ of South Korea’s population is eligible for same or next-day delivery, up from 70% in 2021.

Coupang’s first-party & third-party offerings

Coupang is famous for its fast delivery, such as Rocket Delivery and Dawn Delivery. It’s impressively fast, but there are trade-offs. Rocket Delivery covers an eligible assortment of about 6 million SKUs, while Dawn Delivery supports significantly fewer SKUs and is highly dependent on location.

Rocket Delivery covers most of Coupang’s first-party SKUs, plus third-party sellers that use Fulfillment by Coupang. However, across the remaining 100M+ third-party SKUs, delivery is typically standard, and customers usually receive orders within 2–3 days.

Amazon

Amazon is somewhere between MELI and Coupang. Amazon’s first-party business still contributes about 40% of its overall GMV. And its logistics model is also in a transitional stage; Amazon started transitioning to a more “modern” delivery network about a decade later than major Chinese eCommerce firms.

Amazon's push to build out mid-sized warehouses to support same-day delivery started in 2019, but it was delayed until 2023 due to COVID disruption. Now in the U.S., 60%+ of the orders in metro areas can be delivered within one day. However, same-day and next-day eligibility is still constrained by inventory placement. These services mostly apply to items already positioned in warehouses near customers.

Only 10%–15% of Amazon’s “Prime” assortment (FBA SKUs) are qualified for the services, and the majority of same-day and next-day deliveries are from Amazon’s first-party offerings. Even with Amazon’s top-tier demand forecasting system, it’s still not doable for all products to be delivered within the next day.

The push into a more modern system is a huge success; it has drastically reduced cost and improved delivery time. As Andy said in the Q4 2023 earnings call, “In 2023, for the first time since 2018, we reduced our cost to serve on a per unit basis globally. In the U.S. alone, cost to serve was down by more than $0.45 per unit compared to the prior year.” The expansion of these facilities also drove nearly 70% year-over-year increase in items delivered the same day or next day overnight, which he later reiterated in the shareholder letter.

Now, Amazon is still aggressively increasing same-day and sub-same-day facilities.

How to think of FBA?

FBA, at its core, is a fulfillment outsourcing service for merchants’ products, bundled with Amazon’s platform data for demand forecasting. Since it’s an outsourcing facility for third-party merchants, it’s actually a part of a C2C logistics network, and it inherits many of the same drawbacks of an outsourcing facility.

When talking about FBA, there are two main issues: a) the huge complexity of SKUs that drives up the cost; b) merchants lose direct control over its inventory.

I’ll save the deeper breakdown for a separate piece. This article is already a bit too long.

JD & C2C Platforms

JD, China’s leading B2C eCommerce platform holds a similar first-party SKU count as Amazon. They currently have more than 3,600 warehouses and 18,000 logistics hubs in operation. Of the 3,600 warehouses, 1,600 are dedicated for first-party products. Now, 95%+ of its first-party products are delivered within 24 hours.

The most scary thing is JD’s inventory turnover rate. JD has more than 10M SKUs on their first-party platform. Their inventory turnover rate is similar to top notch wholesale clubs with limited SKUs (i.e. Costco), and is much much better than Walmart or Amazon.

JD success is highly attributed to their mid-sized warehouse network.

China C2C logistics players

In China, there are now different tiers of C2C logistics networks that serve different needs, and many providers offer FBA-like services. Shipping costs vary dramatically depending on merchant preference; merchants typically pick a provider based on what they’re shipping and the service level they need.

S.F. Express

S.F. Express is a high-end C2C logistics player that focuses on premium services with ultra-high speed.

It also offers dedicated handling for different parcel types. Its logistics product line includes: Cold Chain & Fresh, Pharmaceutical & Life Sciences, 3C Electronics & High-Tech, Apparel & FMCG, Luxury & High-Value Goods, Industrial Manufacturing & Auto Parts, and other vertical niche services.

Similar to UPS relies on full-time unionized employees for a high quality, consistent experience, S.F. Express operates a largely full-time courier workforce and provides signature-on-delivery (in-person) service.

J&T Express & Other low-cost players.

There are mainly five low-cost C2C logistics providers in China with parcel market share ranging from 10%~20%. J&T Express is the fifth-largest logistics player by parcel volume.

Unlike high quality C2C logistics players (i.e. S.F. Express & JD Logistics) that utilized full time employees for the service, low-cost C2C players operate under a franchise model. The franchisor owns the sorting centers and handles package transportation, while franchisees own local delivery hubs and are responsible only for the first and last mile.

Overall, franchisors are not very involved in day-to-day delivery execution. Instead, they provide tools and simple operating metrics that are centered around delivery speed and volume for the franchisees to fulfill. Many of the parcel end up routed through service points, which often “forces” customers to self-pick up. These service points are ubiquitous and are usually within a 5-minute walk.

As a result, delivery costs for low-cost players are often 70%+ lower than those of high-quality C2C players.

The impact of C2C to B2C logistics network

As C2C marketplaces have grown, sellers have professionalized alongside the maturation of C2C logistics networks. C2C logistics are now so fast that they erode one of B2C’s classic advantages: fast delivery.

When I was in tier-one cities in China, I would usually receive C2C packages within 48 hours, and sometimes within 24. The speed of delivery has become fast enough to make B2C platforms’ speed as a value proposition less compelling.

Why did it get this fast? As I mentioned in the article, delivery speed is largely a function of where inventory is stored. Now, with the professionalization of sellers, a huge number of C2C merchants also operate 24/7 warehouses or use FBA-like fulfillment services. The parcels are often fulfilled 24/7 and are stored near or within major cities.

Combined with very low delivery costs, this speed has helped make C2C marketplaces the dominant e-commerce model in China. In fact, JD is the only major B2C left, and about half of its GMV now comes from third-party sellers.

Closing Thoughts

It takes 20+ years of continuous improvement for C2C logistics to be good enough to challenge B2C logistics. C2C marketplaces and C2C logistics are now mainstream because they offer the highest ceiling on the cost-to-experience tradeoff.

To be more specific, in some places C2C delivery is now nearly as fast as B2C, but at a fraction of the cost.

Meanwhile, food delivery platforms are cutting delivery costs and expanding SKUs, effectively evolving from meal delivery apps into quick-commerce platforms. That shift puts even more pressure on B2C eCommerce’s traditional value proposition, and it’s one of the main reasons JD kicked off the eCommerce war in early 2025.

and I are releasing our “China Quick Commerce War” piece in the next few days. It’s a deep dive that will cover a wide range of eCommerce topics. Stay tuned!

Thank you for reading!

Subscribe now

Share

-

Pinduoduo call itself “Costco + Disney” but what does the “Costco” part refer to?

The Core idea of Pinduoduo

The core idea of Pinduoduo’s low-price model is based on a trade-off: less variety in products, but much more volume per item.

For sellers on PDD, that means one good product is enough, but it needs to be really good quality for its price. It’s the Costco playbook — fewer SKUs, higher throughput per SKU, with lower prices.

But when it comes to branded products, it’s a bit different. Before PDD invented the 10 Billion Subsidy section, its platform filled with white-label and fake goods.

The 10 Billion subsidy section. - a huge seller bidding network with selective product.

The “10 Billion Subsidy” program is PDD’s push to offer high-quality, branded products and earn consumer trust.

As part of the initiative, Pinduoduo guarantees product authenticity — and promises to pay ten times the price if any counterfeit is found.

-

For this section, PDD built a catalog with only a few thousand products — all of them are the most popular branded items. The items are carefully chosen, often from the top-branded items on JD and Tmall.

However, there’s a problem: brand owners don’t want to be associated with PDD and are unwilling to work with the platform (e.g., Apple). As a result, PDD is forced to source these products from resellers.

The reseller network

Take Apple in China as an example: there are 10+ authorized resellers and 10,000+ small brick-and-mortar shops selling iPhones. Early on, if any reseller is willing to sell, PDD will buy out inventory in cash and resell it to consumers — most at a loss.

Because the platform can move volume so quickly, those resellers often open their own storefronts on PDD and become part of its seller network.

Over time, this builds a long-tail network of very small suppliers. That’s one reason PDD-sold branded goods but struggle to qualify for the national subsidy program: the supply comes from many tiny resellers rather than a centralized, clearly authorized channel.

So Pinduoduo sell things at a loss?

The answer is yes — that’s why it’s called the 10 Billion subsidy program. But things have changed through the years. It started to become a network that makes money.

It helps to split products into two buckets: premium flagships (e.g., iPhone) and mid-tier brands. $PDD often treats iPhones as a loss leader, but it can earn solid margins on the mid-tier mix.

-

When you buy premium brands on the 10B section, the platform doesn’t immediately reveal the seller. Instead, you typically wait a few hours to several days while $PDD confirms the source.

During this time, PDD runs a bidding process among its network of sellers and assigns the order to the lowest bidder. If the winning seller’s price is higher than what PDD charged you, PDD takes a loss. If it’s lower, PDD keeps the difference.

When a new iPhone launches, supply is limited, so PDD often has to offer heavy subsidies to stay the lowest-priced option. As inventory improves and scarcity fades, PDD begins to break even or turn a small profit.

For smaller, high-quality brands, $PDD typically profits from each sale. These sellers cover both the platform’s discounts and commission fees, but in return, $PDD drives significant traffic to their products, making the exposure worth the cost.

In conclusion, Why PDD can sell for less than JD and Alibaba?

1. $PDD prioritizes volume over variety, focusing on fewer products but selling them in large quantities.

2. For premium brands that don’t allow trade-offs, $PDD covers the price gap through subsidies to stay competitive.

3. Unlike its competitors, $PDD often doesn’t have official partnerships with major brands. Instead, it sources products from resellers—some of whom are offloading excess inventory, even at a loss.

-

JD and Aliababa have strong brand partnerships to protect, which puts them in an innovator’s dilemma. Unlike PDD, they can’t easily adopt a broad reseller model.

That’s why their seller channels remain more reliable—products typically come directly from the brands, not third-party resellers like on $PDD.

While $PDD quality control is now decent, fake products still occasionally slip through.

-

$PDD handpicks products for its 10B section, manages its own inventory, and oversees the entire sourcing process. Sounds a lot like JD, doesn’t it?

And that’s just one of many similarities. Stay tuned—follow me for deeper eCommerce insights.

Also, I’d be really happy if you subscribed to my Substack! and follow me on twitter @the_zack_zhu

If you have any questions, feel free to comment. A like or repost would also be super helpful. Thank you so much!

Subscribe now

This was a old twitter post. But I think it’s worthwhile to stay in my blog.

Pinduoduo sells more iPhones in China than JD

shocking, right? But this is the reality in China.

But are people stupid to buy on $PDD instead of $JD or $BABA Tmall?

Sub-Platforms

In China, every eCommerce company has sub-platforms inside the main platform that operate under different policies and service levels.

The people that buy from $PDD aren’t stupid or don’t care about the services. It's because there is a separate policy for their "branded products."

For $JD, $BABA, and $PDD, they all have two "main" sub-platforms — one for serving the high-end and one for serving the lower-end market.

The credibility of Sub-Platforms

In terms of credibility among the main sub-platforms, here’s how they rank:

Tier 1 : $JD First-party platform
Tier 2 : $BABA Tmall / $JD Third-party platform
Tier 2.5 : $PDD Ten Billion Subsidy Section
---(Big Gap) ---
Tier 3 : $BABA Taobao
Tier 4 : $PDD other section

Many brands have Official Stores across all three major platforms. However, high-end or premium brands often choose to only be present on $JD and $BABA Tmall.

When a brand operates Official Stores on all three, most consumers simply buy from whichever platform offers the lowest price.

Pricing between $BABA Tmall Official Stores and $JD Official Stores is usually similar, though Tmall can sometimes be slightly cheaper.

$PDD Tmall Official Stores tends to have the lowest prices, but its Official Stores often offer a much smaller product selection.

There are tons of Sub-Platforms

Each platform includes a range of smaller sub-platforms. For example, $BABA runs both a first-party channel and a section that functions more like $PDD. $JD offers similar structures.

In every single platform, there are all a lot of different small sub-platforms. For instance, $BABA also got there first-party platform and the platform that operates like $PDD. $JD also have things that are similar.

The ranking is just for the "main platforms" that creates the majority of the sales.

-
Interestingly, a single company may operate multiple stores on Taobao, each serving a different purpose. It's also common for brands to have official stores on both Tmall and Taobao.

This is mainly due to differences in traffic and commission structures across the sub-platforms.

Also, $PDD’s "A Billion Yuan Subsidy" section operates in a way that’s quite similar to $JD’s self-operated business model.

-
Thank you for reading, Welcome to read my article on “How does Pinduoduo sell branded products at a lower prices”.

Feel free to leave a comments or question! Welcome to follow me on twitter.

Grab–GoTo Merger: A Brewing Quick Commerce Battle with Shopee Food

One of the biggest developments in Southeast Asian tech is the potential merger between Grab and GOTO, the parent company of Go-Jek. Indonesian government officials are directly involved, and GOTO recently appointed Hans Patuwo as CEO to help fast-track the deal.

The integration of two long-time rivals will likely create significant chaos for at least two years after the merger closes.

Ironically, this turbulence could give competitors like Shopee Food an opportunity to gain ground. If the merger is officially confirmed, I expect Shopee Food to ramp up its investment and take advantage of the short-term chaos. The quick commerce race in Southeast Asia may accelerate as a result — though I don’t believe Shopee Food is fully prepared for that fight just yet.

In the markets where it operates, Shopee Food held an average market share of 12.4% in 2023 and 16.43% in 2024. Grab faced significantly more competitive pressure in those same countries.

Shopee Teams Up with Meta and YouTube to Boost Social Commerce

Shopee has announced multiple deals with Meta and YouTube, and the results of the collaboration seem to be overwhelmingly positive. In Indonesia, YouTube-attributed orders grew over sixfold after the partnership launch, showing strong synergy.

I believe Shopee has significant disadvantages when fighting TikTok Shop. TikTok’s reach in the countryside and its traffic advantage can’t be underemphasized. I believe TikTok Shop is still the most undervalued eCommerce player in the world, despite its massive success in China and Southeast Asia.

In my opinion, the collaboration with Meta and YouTube is Shopee’s top emergency to defend against TikTok Shop.

SPX Express, Shopee’s logistics arm, opens up to non-platform sellers.

In The Biggest Myth of eCommerce Logistics, I argued that a lasting logistics moat comes down to network scale — but, even more critically, to a platform’s share of parcel volume in the market.

Today, Shopee routes over 50% of its eCommerce packages through its own logistics arm, SPX Express. It now holds nearly half of the eCommerce market in Southeast Asia (52% when considering major platforms only).

Logistics is a high fixed-cost business — the more volume, the lower the cost per delivery. SPX Express is now the second-largest player in the region, holding a 27.7% market share, behind only J&T Express’s 32.8% in 25H1. While Shopee has steadily brought more parcel volume in-house, J&T continues to grow — largely fueled by the rapid rise of TikTok Shop.

To build a defensible logistics moat and become a true cost leader in logistics, Shopee must expand its logistics services beyond its own ecosystem. Opening SPX Express to more third-party sellers will be key to increasing volume and lowering unit costs.

Shopee has already started pilot collaborations with selected sellers in a few countries. But to compete effectively with TikTok Shop over the long term, and build a durable moat, scaling its logistics operations would be the top priority. It will be interesting to see how Shopee executes on the plan.

TikTok Shop Expands In-Store Business—What Does It Mean for Grab?

In my Top 6 things I’m wataching in China’s eCommerce, I mentioned the rapid rise of Douyin’s in-store business. As TikTok often follows Douyin’s playbook, it’s no surprise to see TikTok Shop now expanding similar services in Southeast Asia—starting with pilot programs in Thailand and Indonesia in 2024.

But what exactly is the “in-store business”?

Simply put, it’s about using online traffic to drive customers into physical stores. Think restaurants, salons, or even hospital offering digital discount coupons to attract customers, with platforms like TikTok Shop taking a cut from the transaction or advertising spend. Influencers can promote these venues, making it easy for viewers to redeem deals and visit the store—creating a seamless loop between digital content and physical commerce.

In China, this model has become a $280 billion USD industry. It’s enabled businesses to thrive even in areas with little foot traffic, relying purely on internet-driven demand. Entire industries have emerged around this local services model.

If TikTok can replicate this in Southeast Asia, it could pose a real challenge to Grab. TikTok may lure restaurants away from spending their marketing budgets on Grab, weakening Grab’s efforts to own a bigger pie in the physical in-store experience.

That said, TikTok faces real hurdles. Unlike China, Southeast Asia remains supply-constrained, making rapid scaling harder.

In Conclusion

Southeast Asia’s eCommerce landscape has so much more to explore—it's one of the fastest-growing markets in the world. We're also seeing rapid growth in related sectors like Fintech and the distribution of high-quality branded goods from China. I'll dive deeper into these in an upcoming episode!

Thank you for reading and welcome to follow me on twitter @the_zack_zhu and read my previous post on the Top 6 things I’m wataching in China’s eCommerce.

Subscribe now

Alibaba and Meituan quick commerce war.

Before the quick commerce battle began, Meituan and Alibaba held over 70% and 20% of GMV market share, respectively. Despite Alibaba spending ¥50 billion (~$7.91 billion) in this fight, Meituan still retains more than half the market — though Alibaba has caught up in order volumes. As of October, both companies were handling around 75 million daily orders.

Alibaba is now shifting focus to supply chain upgrades and cost reductions, with media reports hinting at a renewed push this spring.

Meituan continues to benefit from strong ties with restaurants, physical retailers, a well-managed rider network and most importantly a much affluent user base. However, Alibaba’s deep pockets—fueled by Taobao’s $25 billion in annual profits—mean the competition is far from over.

( Despite reaching similar order volumes, Alibaba and Meituan still differ by 2 yuan in unit economics per order.)

JD’s Push into On-Demand Logistics

Many may have forgotten, but JD was the one that kicked off the quick commerce war—driven by the need to protect one of its core value propositions : fast delivery. That value proposition has been increasingly challenged by Meituan’s 30-minute on-demand delivery network.

In China, there are four major types of eCommerce logistics models. JD has traditionally excelled in B2C and C2C logistics but lacks strength in the on-demand space. This strategic push into on-demand logistics is JD’s attempt to defend its brand promise.

As the smallest player in this battle, JD’s evolution will be interesting to watch. I will co-write an article on this topic with Gab—stay tuned.

National subsidy Program

China’s national subsidy program, launched in July 2024, covers large-scale equipment upgrades and trade-in incentives for consumer goods. It has significantly influenced eCommerce platforms — JD has benefited the most, as reflected in the earnings started from Q3 2024 earnings, while PDD has seen a negative impact.

The program’s total subsidy in 2025 was around 300 billion yuan and could exceed 500 billion yuan this year.

Pinduoduo supply chain transformation

Since Q2 2024, Pinduoduo has been upgrading its supply chain to help factory-based sellers meet more diverse consumer needs. This initiative is also part of a broader effort to attract branded merchants, who have traditionally been hesitant to sell on the $PDD platform.

On April 25, the company launched a 100 billion yuan merchant subsidy program. It will be interesting to see how this shapes the platform’s next stage of growth.

Welcome to read my article - “A guide of Pinduoduo bad growth and earnings in 2025”

Douyin continues its expansion into the in-store business.

One of the hottest topics in the eCommerce space in 2023/2024 is the Meituan war with Douyin in the in-store business, related to group-buy coupons and hotel bookings in lower-tier cities.

Chinese media reports project that Douyin’s Local Life unit will hit 800 billion RMB in GMV by 2025 a , growing over 50% year-on-year, with a 3% share in the OTA hotel market. I recently stayed at a four-star hotel that was livestreaming room bookings on Douyin. According to the staff, it's been surprisingly effective — just two months in, and it already accounts for over 10% of their total bookings.

Meituan in-store business GMV was a little bit more than 1 trillion in 2024, the growth was not too good partially because of the delivery war.

The potential rising OTA war with Alibaba Fliggy, Meituan, Trip.com and Douyin

In mid-2025, Alibaba merged its OTA business into its eCommerce division and integrated the Fliggy user interface into the Taobao app (Before the integration, it was in the ecosystem but with separate user experience). From my personal experience, it’s been a real game changer.

Trip.com now commands over half of China’s OTA hotel booking market, with high take rates and the operating margins is above 25%. It’s a rapidly growing and highly profitable space.

Alibaba has already ramped up its investment in this space, and Douyin is clearly moving in the same direction. China’s hotel industry still feels young and full of potential. It’s definitely a space worth keeping an eye on.

-

A write-up on SEA and LATAM is coming soon—stay tuned!

The good and bad of Taobao & JD.com & Pinduoduo from a user perspective.

$JD biggest problem is the product pricing isn't optimal, and the variety of product is often times disappointing. But the good part is that it still has the best trust when it comes to electronics and top tier logistics.

For instance, when I try to buy my camera from the DJI store official store, the same official store on Tmall has a lot more product set. And $JD happens to lack the set I want to buy. Pricing and variability is $JD's weak spot.

But interestingly, DJI Official Store on Taobao still used $JD logistics, the user experience of $JD Logistics just feel super reliable.

Alibaba - Taobao / Tmall

$BABA biggest problem is it used to be too dominant so other newcomer started to take away it's market share. The biggest and true advantages of $BABA is that it's product variety is still the best. It's basically the default choice when I search products.

The down side of $BABA is that it's usually doesn't have the best pricing when it comes to branded product. I usually start my journey on Taobao and than when I find the "branded" product I want, I than turn to $PDD. And sadly, 70% of the time, I would find the same "branded" product on $PDD Official stores or the 10B subsidy selection with cheaper price. But interestingly, even when $PDD got the same branded official store, they might lack the SKU in the official store. So I needed to go back to Taobao.

(I will still used $BABA to buy non-branded / small branded product, especially clothes, I don't trust $PDD enough 🤣🤣🤣)

Pinduoduo

For $PDD, it's generally pretty wonderful when I buy branded products. I haven't got any issue when buying product from it's official stores or 10B selection. But I'm not satisfy with it's white label, (aka small branded product).

I need to say, I have returned a decent amount of things, and now, I simply only buy branded product that I know from $PDD official store or the 10B selection. For me, the pricing difference isn't good enough for me to justify the time spend waiting for the logistic and then spend time return the product.

Another big problem of $PDD is it's gaming function is quite annoying, but interestingly, if you used the "functions" correct. Users can easily got more discounts. It's a platform that needs a decent amount of experience to get the optimal pricing but even without the extra discount, it already the cheapest price of all the platforms. I love $PDD simply because it's pricing is the best.

In the future, I will give out a much more detail version of the reasons that cause the difference in value proposition. Thank you for reading.

-
And if you are interested, welcome to read my old and short tweet about China's eCommerce platform key value proposition.

-

If look at history, Amazon first started their own self-operated last mile service Amazon Flex in 2015 after 20 years of founding. Does Amazon don’t have durable growth and not having a moat in the first 20 years of founding?

( data organized by gemini, not 100% correct, but can get the idea)

But when will an integrated logistics service works better, why will eCommerce platform build their own fleet and when will logistics become a durable moat?

There are a lot of nuance. First a quick concept of modular and vertical integrated product.

Vertical Integrated products

Vertical integration is only important when a brand new product hit the market that needs fast iteration, or when the product ecosystem isn’t well developed enough.

When a product or a product ecosystem become mature, all supply that surrounded the core product that control the user will started to be modularized and commoditized.

When a brand new innovative product was first developed, since the whole industry is “invented” by the company. It will be force to be an integrated player since there are literary no other suppliers except themself.

But when the product become much well developed, the qualified suppliers will increase and the components will become more and more modularized. The modularized supplier often times started to compete with each other, than the cost of the component dropped.

This happens in the PC / Smart Phone / Auto and a lot of more other industries. eCommerce is also one of them.

Don’t get me wrong. It doesn’t mean the one with integrated product will lose, they sure didn’t. In reality, there will usually be both integrated players and more modularized players, but often times in the end, the modular player will win the biggest market share simply because lower cost.

Although the modularized players usually got the most market share, but it doesn’t mean the integrated player won’t get most of the profits. e.g. Apple with iPhone. But in PC and Auto or most other industry, usually the modular players got the most market share and the most profits.

Modular or integrated products are just depends on the perspective. Apple can get 50%+ gross margin on iPhone is also because they modularized most of the components that are inside the iPhone.

Inclusion, the vertical integration of products often comes from necessary, and the when the industry got more mature, it will become more and more modularized.

This topic is really interesting, but lets us get back to eCommerce logistics.

The modularized logistics, and why it’s not that important in the long term.

The reason why Amazon doesn’t build it’s last mile fleet until 20 years of it’s founding is because U.S. or other Amazon operating countries already got a decent, although not optimal, logistics network. The last mile fleet is really costly, and with a more “mature” 3PLs, the delivery is already good enough.

But it’s another story for other developing nations. Logistics are usually not something have been well developed, it’s a core part of the experience.

The two reasons of building self-operated logistics fleet.

For eCommerce companies that own logistics, there are two basic reasons of why they do so.

1) To Cut Cost

This is very intuitive, logistics is part of the eCommerce value chain, owning another piece of the value chain means another layer of margin especially when they can always have their logistic network fully utilized and than outsource the “hard”, “costly” and “spike volume” part to the 3PLs. This is the main reasons why Amazon started to build their fleet.

2) To better serve users

The main reasons of why eCommerce platforms in developing nations started to buildout logistics network is simply because the 3PLs is still too bad, it’s usually slow, too costly and also the service areas are limited.

There are a lot of example. Coupang in Korea because the local players are slow and costly, $SE Shopee in South East Asia because the country doesn’t even have a good map and no logistics services in more rural areas, JD.com $JD in early China because no one else can send package fast without harming the electronic products. Similar reasons with $MELI in LATAM.

Vertical integration logistics is very important to serve users when the eco-system isn’t mature enough, and 3PLs lack the incentives or don’t know how to do the job correctly. A vertical logistics that operating in non-mature regions is usually losing money on it’s shipping in order to better serve customer. MercadoLibre loss about 20% of it’s Commerce revenue on shipping subsidy, same with Shopee, Shopee loss about 30% of it’s marketplace revenue on shipping subsidies.

Logistics in China

In China, it’s logistics network are already really well developed, platforms no longer subsidies on logistics, in fact, only the one of the top five players operates it’s own logistic network.

Now in China, all the things B2C eCommerce logistics can do in other countries can be done with third party logistics. This including tracking the package, send the package under 6 or 24 hours, integrated customer services, return the package before received it, you name it, they can all achieved it, and it’s probably a lot better than the ones outside China.

That’s why $BABA $PDD TikTik Shop in China can still strive without 3PLs. The important thing isn’t logistics itself, is the control and the software integration with the logistics network that really matters.

How China’s logistics work is another post, China ships nearly three times the parcels of the US–EU–SEA combined., it’s operated in a very diverse and very interesting way. It solved a lot of defects of $MELI or $AMZN logistics, $AMZN logistics is also moving into the same direction.

( Welcome to read my old post on the biggest problem of $MELI B2C logistics - fulfillment by Mercado Libre or the replicate of FBA , spoil alert, the B2C logistics isn’t good for scaling, and it isn’t suitable for lots of product categories.)

-

When does a vertical logistics really convert into a durable moat?

The durable moat basically boils down to the scale of the logistics network, but most importantly thing is than boil down to the eCommerce platform’s “parcel” share in the market.

Logistics incurs huge fix cost, it’s one of the two main cost of operating an eCommerce platform. If the vertical logistics network can process a lot more parcel than the second biggest logistics player, than the eCommerce platform will enjoy a much lower cost to serve in the long term.

-

I think a lot of people will argue, isn’t vertical logistics better when it comes to delivery of products, especially electronics? Isn’t that a huge moat? I would say that’s pretty much true before the ecosystem achieved maturity, for most developing countries, that’s probably a decade later.

In China, the best logistics and most reliable logistic player is actually not JD logistics, it’s SF Express although JD is also consider one of the top tier companies. When customer buy products that are more expensive from Taobao or Pinduoduo, the seller will just used SF Express instead, and shockingly the experience will be just as good.

-

Thank you for reading. Welcome to follow me at @the_zack_zhu on twitter. Feel free to comments, push likes or share to your friend that might be interested.

Really appreciate!

-

Shopee $SE EBITDA are mainly impact by 1) Logistics 2) Sales & Marketing Spend. In the last post, I have already share that Shopee is still losing 30%+ of it’s marketplace revenue on logistics subsidies. This post, let me share eCommerce acquisition model and why Shopee GMV might continue to accelerate.

Let me elaborate!

-

I think the amount of logistics cost as a percentage of revenue doesn’t improve much compare to last year. But the marketing cost did, it’s drop from 27.6% to 23.69% in the last twelve months.

The nearly 4% drop of the S&M is about the operating margin improvement of Shopee’s business.

-

I personally consider logistic subsidies a kind of Sales & Marketing spend. It’s the cost of improving the platform experience, which decrease the cost the marketing spend needed for the platform.

The SaaS acquisition model

eCommerce acquisition model is like a subscription SaaS business. It’s separate in two parts:

1) Acquisition of User

Acquisition part is straight forward, it takes a lot of ads and discount to attract new customer, so eCommerce platform will first book a lot of loss when acquire a user and than get the revenue back through the customer LTV (Life Time Value).

2) Land and Expand of Existing User

The second part is less obvious. But the Land and Expand part of the eCommerce business is actually huge, the same customer can easily see a 500% boost in sales(ARPU, Average Revenue Per User) after a few years on the platform, eCommerce platforms do still see multiple years of user spend expansion long after hitting the ceiling of active users.

The reason is that eCommerce users usually started from buying from a certain product vertical and than gradually expand into other more expensive product vertical. Customer usually isn’t too comfortable buying from an unfamiliar vertical. It’s usually takes multiple years of using the platform to expand to every single vertical.

A huge part of the eCommerce marketing expense is not for acquiring new users, it’s actually use to boost the learning curve of the eCommerce user into other verticals.

Using marketing expense to pull forward the user learning curve of buying from new vertical not only let a user spend more. It will also boost the stickiness of the user. It will vastly increase the Life Time Value of a user.

-

We can see the profit expansion pattern happens in the last cycle of investments started in 23Q3, the GMV growth accelerate after Shopee started to tamper it’s S&M spend and logistics subsidies after two quarters of aggressive investments.

All in all, I won’t be surprise to see Shopee GMV numbers continue to accelerate! I believe the new investment cycle will be really fruitful to Shopee.

-

I think now is a critical time for Shopee to step up it’s investment to defend it’s market share from TikTok Shop. I think market share is the real and most important moat of Shopee.

But the market share theme is for another post. Welcome to follow me on Twitter @the_zack_zhu for more. I still got a lot of fundamental ideas of the eCommerce business that haven’t shared!

Thank you for reading, and welcome to subscribe to my Substack!

If we map J&T Express (South East Aisa largest logistics provider, often consider the cheapest) cost of deliver per package with Shopee orders, we would get a scary number, before 2022, delivery cost is over 100% of Shopee’s revenue.

Now with dramatic increase in unit economics, it’s still in the range of 50%+ of revenue. The loss on VAS (Value add services, mainly logistics) is still in the range of 30%+ of marketplace revenue.

Shopee huge operating leverage!

Shopee’s operating margin is still really low, but delivery cost per order is shrinking fast and the take rate is still going higher. The Shopee platform’s operating leverage is still crazy high. (click here to know about the potential take rate ceiling)

There are a lot of things to note:

factors that decrease the cost on calculation

• The way of Shopee counts orders will overestimate the real logistics volume, since even when customer cancel the order before the package is send, it will still counts as orders. This activity alone may decrease the real order volume by 10%.

• Shopee when sending orders that are intra the city, the cost will probably be lower than the cost of J&T express. Since Shopee will give the better orders to SPX to send.

• In 25Q3, I use the 2025H1 J&T express cost, 25Q3 cost will be lower than 2025H1

factors that increase the cost on calculation

• Shopee SEA cost of delivery is relative cheap compare to Taiwan or Brazil. Brazil & Taiwan probably accounts near 20% of the overall order volume. The delivery cost of these two areas is probably 100% higher than the average of SEA.

• Shopee still rely heavily on 3PL, probably 40% of the orders are still send by 3PL, the cost orders can’t be as low as J&T express cost, it’s more near J&T revenue per package (which is an 20% markup) + J&T express is actually a relative cheaper alternative.

• SPX send more and more package with fast delivery in recent years, fast delivery cost is significant higher than standard delivery. As a results, in the past year, the average delivery cost in Asia only decrease mid-single digits compare to J&T express’s double digit. Using J&T delivery cost is too optimistic.

• SPX doesn’t get an cost advantages when sending orders in rural areas, the estimate cost will be higher compare to J&T express

  • The reason is two folds, one the logistics hub of SPX is build base on existing order data from 3PLs, but in rural areas, SPX doesn’t got enough data to gain much advantages.

  • The volume is still much lower than J&T express in rural areas.

Thank you for reading, welcome to follow me on twitter, if you have any questions, welcome to shoot me a message.

Time is your best friend when a company got a structural edge over it’s competitors. When times goes, the company with structural advantages will only eaten more market share from it’s competitors.

But how can we know a company got structural edge or it’s the one will be soon be eaten?

An easy way, although not an accurate way, is to look at the year of the company founding. Usually a younger company will got a better edge.

The reason is simple, young companies is build on the best technology on the moments, and will utilitized the newest technics at that time, they won’t be path dependency like the old companies.

-

The older companies will always have a more compelling offering on paper. It’s the conventional wisdom to like them.

Startups will never win if the incumbent never ignore the market. And interestingly, most of the time incumbent don’t, so little startups become generational companies.

But if a young company become really successful in a very short period of time. Then it have a really high chance it’s has already garner huge edge over the incumbents.

-

“The battle between every startup and incumbent comes down to whether the startup gets distribution before the incumbent gets innovation.”

A startup usually always got a better product, but 99% of the time incumbent got a better distribution. Often times the incumbent can just buy out the startups and then plug the new product into the distribution and repeat all the profits of the startups.

This is why big tech each acquires hundreds of startup in the past decades. The thing they do is to acquired the tech talents and put products into their existing distribution offerings.

This doesn’t only happened at tech. It’s also happened in traditional industries. L’Oréal, Coca Cola, LVMH all did a decent amount of acquisition in the past few years.

Acquire a product and than leverage existing distribution is one of easiest way an incumbent can grow it’s revenue.

-

But when will acquisition don’t do the trick? It’s when their is a fundamental platform shifts in the business that requires a new distribution strategy.

There are basically two platform shift in the past 35 years. The first shift is from PC to Web, and another one is from Web to Mobile Apps. Now the third is emerging with LLMs, but still not as obvious.

-

The platform shifts is often when the incumbent started to get the incentives problem. The startup product doesn’t fit the distribution pipeline and not only that, the new product might even cannibalized the old product line. This is basically what disruptive innovation like.

People usually only focus on the big big players like smart phones versus the PC. For now Google versus OpenAI.

There are still so many companies that are hidden, and it takes a lot of time for them to emerge. There will also be second order effect on huge platforms, and in the being often the new platform will look like a sustainable innovation that the old incumbent can benefit from.

But in reality, nearly without exceptions. The “native” platform application will emerge, and it doesn’t mean the old platform application won’t continue to grow. But the new platform application will grow even faster, and often times even more stronger.

There are so many examples, Netflix vs Blockbuster. TikTok vs YouTube, Shein vs Zara, and so much more. In my opinion it’s relatively hard to grab the big platform layer shifts when it comes to investing, but it’s rather easier to catch the application layer one that is the secondary effect the platform.

Most people just never notice a platform’s secondary effect, or a native platform application is such a big deal until it’s too late.

-

Platforms shift is often when a company got a structural edge over it’s competitors, and time will become your friend in the process.

-

$PDD are facing two business transforming, both in and outside China.

To make matters worst, there is also policy headwinds in and outside China, makes things even more difficult.

Let me elaborate.
-
In China, $PDD find out serving low quality product will only take them so far. High quality sellers prefer $BABA and $JD a lot more. It’s nearly impossible to get the same supply without doing any transformation.

The transformation started in 2024H2, they decided to treat their seller that sell high quality goods better, and reduce the reliance on low quality product merchants.

These cause two problems. First, the low quality sellers is the life blood of $PDD, it’s the most monetizable sellers on $PDD. The same traffic monetize more on those sellers, which decrease marketing revenue and operating income.

The second is they give out huge subsidies for high quality merchants and increase R&D to help build software for the merchant and sometimes even provide R&D budget and guidance to upgrade the seller. (BTW this is the Shein model) This cost tons of money.

-
Outside China, cross border eCommerce is slow, it numbers of SKU is limited and the logistics is subject to congestion. The most important part is the market size is super small.

They decided to go from a cross border model into a local to local platform model, and started to onboard more foreigner sellers instead of Chinese sellers.

They are building local teams around the world and constant expanding into more countries. The complexity is extremely high. So goes up 40% YoY on the R&D cost.

The cross-border business is not a new thing in China, there are tons of talent related to it, but the local to local is. The local business is also not as monetizable as the cross border business for now.

This is the two main transformations.

-
To make matters worst are policies in and outside China.

In China, government had started a electronic subsidy program. $JD and $BABA become the beneficiary of it.

Why $PDD is benefit from it? It’s because $PDD branded electronic mainly rely on reseller and dealers. They are hard to comply with subsidy program. (details here) So $PDD not only can’t sell more. They also need to spend the money to make up the difference of the subsidy themselves in order to maintain the competitive of products.

This increase a ton of $PDD marketing budget, but now the subsidy is about to end or already run out of budget. (Depends on the province) $PDD will be fine soon.

That’s the main reason $PDD got better than expected operating income and marketing budget. They don’t need to pocket the difference that much anymore.

Outside China, it’s rather obvious. Trump creates chaos on tariffs, it’s direct impact TEMU pricing, TEMU’s cross border business also hugely impact the local business. No government like cross border eCommerce. They don’t create jobs and they don’t pay taxes. So more regulations are hindering $PDD growth.

-
So all in all, with these headwinds on earnings. $PDD can still maintain 4B of net income is very shocking to me. I think they are doing a better job than I expect.

I will continue follow-up on these transformation! Stay Tuned!