$39.1 billion of that increase was computers and servers, phones and network gear, and computer parts, counting both directions.

Subscribe now

Take those three product lines out, and Laredo grew 1.6%. That’s before inflation.

The national picture is starker. US imports from Mexico reached $599.2 billion over the same 12 months, up $74.4 billion.

Imports of computers, phones, network gear and computer parts rose $77.2 billion. All other US imports from Mexico, added together, fell. That drop is autos: cars, trucks and auto parts fell $7.9 billion, and everything else rose $5.1 billion.

Every monthly release this year has carried the same headline: Mexico is the top trading partner, Laredo is the top port, another record. All three are true. They also describe one product cycle, the buildout of AI data centers, far more than they describe the corridor.

One line on the manifest

In June I wrote that the computer had passed the car at Laredo. It has kept going.

In August, Laredo imported $7.18 billion in computers and servers (HS 8471), the most in any month since at least 2019. It brought in $0.99 billion in passenger vehicles. In the 12 months through July, computers passed cars, trucks, and auto parts combined.

Most of that is one six-digit line, 8471.50, the “processing units” where servers sit: $48.3 billion of Laredo’s $53.8 billion in computer imports over the last 12 months, up from $19.9 billion a year earlier.

Here’s where the port’s growth came from, piece by piece.

Computer trade in both directions added $30.6 billion. Phones and network gear added $5.1 billion, and computer parts $3.4 billion.

Vehicles, engines and auto parts subtracted $3.3 billion. Every other product on the manifest, combined, added $8.3 billion.

Laredo has company. US computer imports from all countries went from $90 billion in 2019 to $382.6 billion in the 12 months through August.

Mexico supplied $146.2 billion of them, passing Taiwan ($132.4 billion). China, the largest supplier in 2019 at $44.7 billion, is down to $5.6 billion.

That last line is the nearshoring story everyone wanted, and it’s real. US computer buying left China. A growing share now gets assembled by Taiwanese contract manufacturers in Jalisco, Chihuahua, and Nuevo León and crosses the border by truck.

Core trade

Economists stopped reading headline inflation alone decades ago. Food and energy swing too hard, so the Fed watches core inflation, the index with those two set aside, to see what the rest of the economy is doing.

Cross-border trade needs the same thing now. I’m calling it core trade: trade excluding computers (HS 8471), phones and network gear (8517), and computer parts (8473). It’s the corridor without the data-center cycle on top of it.

Setting AI goods aside isn’t a new instinct. Minneapolis Fed research published in May sorted all US imports by AI relevance and found the non-AI side essentially flat. Core trade does that for the corridor, with a deliberately narrow basket, so nobody can say I stripped out too much.

Here’s what core trade says.

US core imports from Mexico peaked at $443.5 billion in the 12 months through January 2025. Through August 2026, they were $432.1 billion, 0.6% lower than a year earlier. The year before that, they fell 1.6%.

Laredo’s core trade is $315.4 billion, up 1.6%. That’s growth, barely, and slower than prices.

So the dollar value of everything outside electronics has been flat to shrinking since January 2025, through the tariff year and the USMCA review. The record covered that up.

The value inside the box

The second problem with the headline is what a server dollar carries.

Mexico’s deputy minister of foreign trade puts Mexican value added in the country’s electronics exports to the United States at 7% to 8% (Mexico News Daily, Sept. 24). An industry estimate cited by the Financial Times says Mexico could realistically reach 3% to 7% Mexican content per server, against about 39% for autos.

Both are estimates, one from a government official and one from industry. They point the same way.

Almost everything inside the box is imported. A UBS report citing INEGI and IMMEX data puts imported components above 95% in Mexico’s export-oriented computer and electronics manufacturing. Banxico data shows Mexico’s imports from Taiwan reached $43.9 billion in the first half of 2026, up 200.5%.

Run the rough math. It’s a content comparison, and I wouldn’t read it as GDP.

At 3% to 8%, the $77.2 billion gain in electronics carried about $2.3 billion to $6.2 billion of Mexican content. US imports of cars, trucks and auto parts from Mexico (HS 8703, 8704, 8708) fell $7.9 billion, and at 39% that took out about $3.1 billion.

Set side by side, the electronics gain and the auto loss net out to somewhere between slightly less Mexican content and about $3 billion more. That’s two sectors, not all of Mexico’s exports, but it’s a thin return on the biggest one-year jump in US imports from Mexico in at least seven years.

I’ll admit I got part of this wrong in June. I wrote that when compute is assembled in the region, “the value and the jobs settle on both sides of this border.” Some do. The content numbers say it’s a thin slice, much thinner than the auto trade it’s displacing on the manifest.

There’s a US side to it too. US exports of computer parts to Mexico (Schedule B 8473.30) rose from $17.2 billion to $31.0 billion in a year.

Boards and components go south, finished servers come north, and “two-way trade” counts both trips. It’s the loop that built the auto trade, running again with far less of the value staying in the loop.

What the trucks say

Laredo’s northbound truck count hasn’t moved: 2,968,087 crossings over the last 12 months, against 2,965,312 the year before.

The dollars per truck have. Divide the port’s imports by its northbound truck crossings, and you get an index of $87,292, up 17% in a year.

Take out computers, and it’s $69,166, up 3.8%. About four-fifths of the jump is computers.

The import side counts rail, air, and every other mode along with trucks, so read it as a measure of value density rather than one trailer’s load.

Trade value grew 12.6%. Northbound truck crossings grew 0.1%.

Bridges and yards carry trucks, not dollars. Over the past year, the truck count has looked far more like core trade than like the headline, so for anyone planning the corridor’s capacity, those are the two numbers to watch. I’ll take up the corridor’s real capacity limits in the next piece.

The review is negotiating last decade’s trade

Most of the public fight over the USMCA review, and the annual reviews that follow it, has been about autos and steel: regional value content, Section 232 relief. That’s where Mexico’s value added sits, so it makes sense.

It’s also the part of the trade that’s shrinking.

A server doesn’t need the review to cross. Computers and servers enter the United States duty-free at the general rate, the same zero that has applied since the US put the Information Technology Agreement into effect in 1997.

That leaves the USMCA rule of origin with almost nothing to do on the duty for a complete server. The rule is a simple tariff shift with no regional value content test, and a server that fails it still pays the same zero. The 10% Section 301 tariff on Mexico since July 24 exempts the main computer and computer-part lines, whether or not the importer claims USMCA.

The main new duty that can reach a server is the 25% Section 232 tariff on advanced chips and the products that contain them, in force since January 15.

It hits a server only if the chip inside falls in two narrow performance bands. It waives servers for US data centers that need more than 100 megawatts of new AI-dedicated load. USMCA status doesn’t change any of it.

So the rules that decide whether the corridor’s biggest line keeps growing are chip thresholds and data-center exemptions, written in Washington by proclamation. They sit outside the review.

The question that matters

Servers come because a handful of companies are spending on data centers at an unprecedented pace. Alphabet raised its 2026 capital budget to $195 billion to $205 billion in July and held it there on October 28.

Amazon is at about $220 billion, and Meta raised the bottom of its range to $135 billion to $145 billion. Microsoft trimmed its forecast to $170 billion from $175 billion, blaming delays in GPU capacity.

Wall Street has started asking when that peaks. When Alphabet and Microsoft reported in late October, neither raised its forecast. Wells Fargo’s analysts called 2027 a potential peak year (Fortune, Oct. 7), and UBS expects spending growth to slow from 76% this year to 25% in 2027 and 6% in 2028.

Slower growth doesn’t mean falling server imports. It means the headline climbs more slowly, and if spending actually peaks, it falls, with nothing about the corridor having changed. If the computer line simply went back to where it was a year ago, the port’s 12-month total would drop by about $30 billion, to roughly $363 billion.

That’s why the headline can’t be the scoreboard anymore. It measures Big Tech’s capital budget. Core trade measures the corridor.

The September data comes out November 4. I’ll be watching core trade.

Data: U.S. Census Bureau, general imports and total exports by port (Laredo, port code 2304) and by country, HS 4- and 6-digit, through August 2026; Bureau of Transportation Statistics border crossings (Laredo trucks). 12-month rolling totals, nominal dollars. Analysis by the Texas Center for Border Economic and Enterprise Development, TAMIU.

For about 25 years, the world had give. Demand went up, and somebody somewhere made more. Plants were interchangeable, chains were integrated, and the part you needed showed up the week you needed it. Economists call that elastic supply. In practice, it meant governments could push an economy without setting it on fire, because production always answered.

Subscribe now

That give is mostly gone now. COVID swapped efficiency for resilience on purpose. Energy dislocated and never fully came back. Fragmentation cut the chains, tighter immigration pulled workers back out of the system that globalization had put in, and uncertainty made companies stop building.

The instruments governments used to smooth all of it are spent too. US gross federal debt is past 120 percent of GDP, and a 6 percent deficit is unremarkable now in a year with no recession.

So when demand rises today, prices move before output does.

Here’s the part that reframes the China conversation.

At manufacturing scale, supply still has give in one country. China ran a merchandise trade surplus of about $1.19 trillion last year, the largest any country has ever posted, and July alone came in at $112.5 billion. That surplus is what a country posts when its factories can answer almost any order on earth and its own consumers can’t absorb what those factories make. Overcapacity is give with nowhere to go, and it leaves whether or not anyone wants it.

Which means the tariff fight is really North America choosing to buy less flexibility. That can be the right call. Adam Smith carved out a national-defense exception to free trade almost 250 years ago, on the logic that depending on a foreign supplier for something essential is worth paying to avoid. Strategic dependence is the modern version of that argument. But somebody should be honest about the cost.

Mexico had more exposure than most. China supplied under 2 percent of everything Mexico imported in 2000. By 2024 it supplied more than 20 percent, behind only the United States. The flood was real, and it took 25 years to arrive.

Mexico is where the bill comes due first, and the receipts landed two days ago.

On January 1, Mexico raised the tariff on cars from countries without a trade agreement from 20 percent to 50, and raised duties across 1,463 tariff lines spanning 17 industrial sectors. Last week the economy secretary presented the first results. Across those lines, imports from China fell 28.4 percent between January and May. Footwear fell 59 percent. Light vehicles fell 35. Home appliances fell 27.

The wall is working as a wall. Whether it builds competitive Mexican supply is the open question.

What’s showing up in place of the goods it blocked is the interesting part. Mexico’s auto dealers association expects Chinese brand prices to climb 6 to 8 percent in the second half of this year, as pre-tariff inventory sells through and gets replaced at the new rate. The first replacement to arrive is a price.

That’s the whole argument in a car payment. A tariff changes what an import costs. It doesn’t make the thing.

Making the thing is a different project on a different clock. A tariff is a signature and a date. Building the supplier that replaces the import takes land, power, capital, a trained workforce, and somebody willing to sign a 15-year lease in an environment where the rules changed twice this year. One of those can happen in an afternoon. The other is the hardest thing a government can attempt, because most of the work isn’t the government’s to do.

The honest scoreboard for Plan México is the 15 percent lift in national content it wants across strategic value chains, including automotive, aerospace, and electronics. That target lands in 2030. Five months to block the imports. Five years to build the replacement.

The first of 15 planned industrial hubs opened this year in Huamantla, Tlaxcala: 53 hectares, roughly $540 million, about 6,000 jobs counting indirect. That’s real, and I’d rather have it than not. It’s also worth doing the arithmetic out loud, because 14 more have to land, and they have to land in the sectors the imports were actually coming from.

The United States is running its own version of the same play under different names. The tariffs arrived on a policy timeline. The capacity they’re meant to protect arrives on an investment timeline, and those two clocks were never going to line up.

Which brings this back to the corridor, because a border crossing is capacity too. Every hour a truck spends in a queue is production that failed to answer demand, in the most literal form the economy offers. Lanes, staffing, booth hours, and the systems that clear a truck are flexibility this region can add faster than it can build a supplier base, even accounting for the money and the binational coordination each one takes. They’re also the part of every plan that gets deferred first.

North America already answered this in July, when the review turned into a year-to-year lease. Tariffs are a position. Capacity is a plan. The region has plenty of the first. China has spent two decades building the second.

Not in total trade, where Mexico is still the largest partner the United States has. In the goods trade deficit, the number that shows up in most headlines. Over the 12 months ending in April, the U.S. ran a bigger merchandise deficit with Vietnam than with Mexico, the first time that’s happened. By May it was $204 billion for Vietnam against $199 billion for Mexico, with Taiwan a few billion behind and climbing fast.

CSIS flagged the pattern back in April: the China deficit didn’t shrink; it moved to Asia. They were right. The question underneath is what’s actually inside the number that moved, because “the deficit went to Vietnam” and “China is still the deficit” can both be true at once.

Subscribe now

China left on paper

Start with China, because the official story looks like a win. The U.S. goods deficit with China fell from $418 billion in 2018 to $159 billion this May. China went from the top of the list to behind Vietnam, Mexico, and Taiwan. Imports from China were cut nearly in half, from more than $500 billion a year to $264 billion.

On the ledger, China walked off the field. Much of that deficit changed address while keeping its origin.

How much is really China

Here’s the tell that this isn’t a fringe read: U.S. customs keeps catching it. When investigators find that a shipment labeled “made in Vietnam” is actually Chinese, assembled with no meaningful changes along the way, they strip the label and apply China’s tariff rate instead. Solar cells and steel have both been caught getting routed through Vietnam to dodge the China duties.

How big is that flow? The most careful estimate comes from a Harvard Business School study covering 2017 to 2022. Raw country-level data makes about 28% of Vietnam’s export jump to the U.S. look like rerouted Chinese goods. Trace it to the actual firms and provinces, and the number falls to roughly 11%, or about $8 to $9 billion. That’s the most recent firm-level read available, and with Vietnam’s exports to the U.S. much larger since, the dollars have grown even if the share held. Pure transshipment is real, and smaller than the headline suggests.

The bigger channel is legal. China is now about 40% of everything Vietnam imports, the largest share of any supplier, with that share concentrated in chips, components, and textiles that go straight into what Vietnam ships to America. Even when the final assembly is genuinely Vietnamese, the inputs are often Chinese. Between the rerouting and the inputs, the China footprint is inside Vietnam’s number, whatever the country-of-origin stamp says.

Taiwan is the real thing

Taiwan is the opposite case, and worth keeping straight. Its deficit with the U.S. nearly doubled in 2025, and that one is genuine. It’s the AI-hardware boom: the servers and accelerator chips the buildout runs on, designed and made in and around Taiwan. Taiwan is climbing the deficit table because the world is buying what Taiwan actually makes.

Mexico is the number that comes home

Then there’s Mexico, the one the deficit statistic misreads worst.

Mexico’s deficit barely moved, up about 5% while the two Asian lines rocketed. And the trade underneath it looks nothing like the others. The U.S. bought $558 billion in goods from Mexico over the past year and sold $359 billion right back. Corn, machinery, fuel, plastics, auto parts that cross the river, get built into something, and cross back.

That flow shows up inside the goods. About 74 cents of every dollar of Mexico’s manufacturing exports to the U.S. is North American content: U.S., Canadian, and Mexican inputs built in together. A Vietnamese export leans on Chinese inputs. A Chinese one is Chinese. Mexico’s is the only one of the three substantially made in North America.

What the deficit can’t see

Put them side by side, and the problem with the ranking is obvious. It counts a dollar of Vietnamese deficit, a dollar of Taiwanese deficit, and a dollar of Mexican deficit as the same dollar. They aren’t.

You can watch the difference in Laredo. About 40% of all U.S.-Mexico trade crosses the bridges here, and it runs both ways all day. Loaded trucks north, loaded trucks south. That’s a $917 billion two-way relationship, $359 billion of it American goods heading into Mexico. The deficit was built to track a gap between two numbers. It was never built to see a supply chain that runs in a loop, or a Chinese one hiding under a Vietnamese label.

What to watch

The USMCA joint review convened on July 1, and the deficit is one of the numbers cited most in that conversation. It’s shifting beneath the discussion in real time, and it points to something the ranking alone won’t show. The country that looks like it left is still in the number. The one that looks like the problem is the most North American of the three.

Taiwan crosses the line soon. When it does, watch which story gets told: that the deficit moved to Taiwan, or that a chip boom, a rerouting operation, and an integrated neighbor all got filed under the same number. Only one of those readings is true.

And whatever the ranking says, the trucks keep crossing both ways at Laredo.

That’s the horizon an assembly plant is built on: 30 to 40 years.

As of July 1, the trade agreement governing those plants’ supply chains runs on a different clock. The treaty’s legal term still stretches to 2036. Its planning horizon is now 12 months.

Subscribe now

What actually happened

The USMCA required its Free Trade Commission to conduct a joint review on July 1, 2026. The 3 governments met virtually. The United States declined to confirm renewal “in its current form,” and with that, the agreement rolled onto the sunset track: a yearly review until it’s either renewed or expires in 2036. USTR says the US and Mexico meet in Mexico City the week of July 20. USTR has announced no equivalent round with Canada.

The commentary since has converged on a reassuring point: the agreement remains in force, and the rules governing trade today are the same as they were June 30. True.

I read July 1 differently. The 3 governments made a decision, and it was the most consequential one available. They made review of the permanent condition of North American trade.

Last year I wrote about the Infinity Loop, the cycle of 90-day tariff extensions and exemptions that replaced coherent trade policy in 2025. The loop had one redeeming feature: it was improvised. A White House could end it with a signature. The annual review is of a different kind. It’s now written into the treaty’s own procedure, with a 10-year runway.

The Infinity Loop is the machine. The Precarity Premium is its invoice.

Pricing the loop

You can put a number on this. The Economic Policy Uncertainty Index for the United States has averaged 120 since 1985. It peaked at 460 in April 2025, when the tariff cycle hit full speed.

The peak passed. The baseline never came home. Through June 2026, the index hasn’t touched its long-run average in a single month since the cycle began, and it’s still running at more than 1.6 times the 40-year norm. Elevated uncertainty has quietly become the operating level.

The baseline shows up in the real economy on schedule. Mexico’s gross fixed investment fell 4.2% year over year in February, extending a slide that has run for more than a year, with machinery and equipment spending down 9.7%. On an investment base of roughly $450 billion a year, that’s about $19 billion in annual capital formation that stopped showing up.

Canada lost 111,000 full-time jobs in the first 4 months of 2026, the losses concentrated in its trade-exposed goods industries. The nearshoring wave that was supposed to remake the continent is sitting in committee.

Who pays first

The question the explainers skip: who carries this?

When the 2025 tariffs took effect on non-USMCA goods, businesses did exactly what the policy asked. They regionalized. They certified.

USMCA preference utilization for Mexican goods jumped from 49.5% in December 2024 to 76.1% by July 2025. Canadian goods went from 35.5% to 78.7%. Companies spent 2025 auditing supply chains, documenting rules of origin, and paying compliance teams to earn certificates under the agreement’s rules.

Those certificates are now the exposed flank. An annual review means the rules underneath them can be reopened every year for a decade. The tighter auto rules of origin Washington is expected to push (reports point to a demand that half a vehicle’s content be US-made) would re-grade the exact paperwork businesses just scrambled to earn. The companies that complied most aggressively now carry the greatest risk of rule changes. Good behavior bought exposure.

The view from the corridor

Here’s the twist in my own backyard. Trade through Laredo is at a record: $365 billion in the 12 months through April, with March ($34.4 billion) and April ($34.2 billion) the 2 biggest months in that window. I pulled those numbers the day of the review. Record flows and frozen investment, at the same time.

Both make sense together. Existing plants keep shipping. The annual review cycle doesn’t tax the goods moving today. It taxes the plant that doesn’t get built for 2029, the expansion that stays on the CFO’s desk, the supplier that keeps its Asian contract one more year. The Infinity Loop always collects in the future tense.

What July 20 can fix

The Mexico City round is the third US-Mexico bilateral this year, and the first of the annual-review era. One outcome ends the annual cycle: a joint confirmation of renewal, which resets the term for 16 years and hands capital back its horizon. Every other result, each partial deal, each extension of talks, keeps the lease year-to-year.

Toyota has built more than 10 million vehicles in Georgetown since 1988. Investments like that get made on 40-year horizons, and they’re the reason $365 billion crosses through 2 bridges in South Texas. The question for the negotiators in Mexico City: who builds the next Georgetown under an agreement that has to survive every year’s review?

Two years later, computer-related products are the single biggest category of products Laredo imports.

In April, Port Laredo was the number one trade gateway in the United States for the second month running, $34.16 billion across the bridges, ahead of Chicago’s O’Hare and New York’s JFK. That ranking made the news. What’s now sitting at the top of the manifest is the story.

Subscribe now

The computer passed the car

In April, Laredo imported $4.27 billion in computers and servers, the HS 8471 line, the data-processing hardware that fills data centers. In the same month, it brought in $2.27 billion in auto parts and $1.0 billion in passenger vehicles. Computers alone outweighed parts and cars combined.

The year tells the same story. For all of 2025, computer products were Laredo’s largest single import at $30.8 billion, ahead of auto parts at $25.9 billion. They passed passenger vehicles in September 2024 and auto parts a few months after that.

The speed is the part that’s hard to absorb. Laredo’s computer imports sat near $6 billion a year from 2021 through 2023. Then they jumped to $13.4 billion in 2024 and $30.8 billion in 2025, and they doubled again over the last 12 months, up 106% from April 2025 to April 2026, while auto parts grew 2%. Laredo now imports more computer products every six weeks than it did in all of 2021.

The flow that used to skip Laredo

Here’s what changed underneath the numbers. The first wave of AI hardware came by air, into hubs like O’Hare and JFK. Electronics already moved through this corridor: telecom base stations were a top Laredo import back in 2021. But the data-center buildout is on a different scale, and it’s arriving in a different way: nearshored to a Mexican assembly and trucked across the land border. Components cross south and return north as finished machines, an echo of the loop that built the auto trade, though for now it runs mostly one way.

The cargo that used to fly over Laredo now crosses through it, by truck instead of by plane. And when compute is assembled in the region instead of landing finished, the value and the jobs settle on both sides of this border.

A surge with an asterisk

More than one force is driving the jump. Some is genuine nearshoring of the hardware supply chain. Some is front-loading, with importers pulling shipments forward amid tariff uncertainty. And part of the gap is the other side falling: truck-tractor imports dropped from $11.6 billion in 2023 to $6.3 billion in 2025 as tariffs hit vehicles. Computers rose while a piece of the auto complex fell.

The direction is real. The slope is partly timing.

Why a commodity swap matters

You could file this as trivia. Computers passed cars, congratulations, Laredo. But composition is destiny for a trade corridor. The brokers, the warehouses, and the inspection lanes were built over decades around automotive freight. Now a second supply chain is layering on top of the first, with its own security exposure and its own geopolitics.

Last week I showed you that the cargo crossing Laredo keeps getting denser in value, more dollars per kilogram every year. This is the concrete version of that trend. The physical supply chain of artificial intelligence is rerouting onto four lanes of highway in South Texas.

And it arrives at the worst possible moment for ambiguity. The July 2026 USMCA Joint Review will signal whether the framework under this corridor is renewed, renegotiated, or left to run down toward its 2036 sunset.

This is a brand-new, strategically sensitive supply chain, deciding whether to commit to North American assembly while the rules remain unresolved. Capital choosing between Guadalajara and Penang reads that uncertainty as a price.

The auto industry built itself across this border over thirty years, under rules that held long enough to plan around. The compute industry is choosing where to build right now, on infrastructure designed for cars and under rules up for review in a month.

The data already shows which way it’s leaning. Policy decides whether it holds.

We noticed the convergence a few days ago and went looking for what’s behind it. The answer turned out to be 3 stories stacked on top of each other, and the most interesting one has nothing to do with trains.

Where the trains went

The cleanest signal in the data is cargo trucks: vehicles built to haul goods (HS 8704), a category that runs from pickups to heavy work trucks. Imports of them fell $3.07 billion at Laredo over the last 12 months and rose $2.59 billion at Eagle Pass. That’s close to one for one.

Subscribe now

It reads like what it is: railroads rebalancing their networks. CPKC runs the Laredo rail bridge; Ferromex and Union Pacific connect at Piedras Negras-Eagle Pass. When a manufacturer’s logistics team re-bids a lane, unit trains move from one gateway to the other in a single contract cycle.

The 6-digit codes say what actually moved. Gasoline light trucks, the Silverado class, shifted from one bridge to the other. Medium-duty diesel trucks went from almost nothing at Eagle Pass to $3.6 billion, a lane that didn’t exist a year ago. Diesel light trucks fell at both bridges at once, a pattern that repeats with cars below.

The geography backs the routing story. GM’s Silao plant loads Silverados onto Ferromex track, and that traffic moves via the Eagle Pass gateway. The commercial-truck plants sit on the same corridor: Daimler builds Freightliners in Saltillo, about 3 hours from Piedras Negras, and Navistar has built trucks outside Monterrey since 1998. Union Pacific, which owns a quarter of Ferromex, spent 2025 raising the gateway’s throughput by moving the crew swap off the international bridge to its yard 7 miles inside Texas. For a plant on Ferromex rails, Eagle Pass keeps the haul on allied track end to end.

Cars are a different story. Passenger vehicle imports (HS 8703) fell at both crossings, $2.14 billion at Laredo and $0.88 billion at Eagle Pass. A decline that hits every gateway at once points past routing to policy: the Section 232 auto tariff, in effect since April 2025, working through Mexico-built sedans sector-wide.

The third mover is grain. Corn exports to Mexico through Eagle Pass rose 36% while the McAllen-Hidalgo gateway lost more than half of its corn traffic. Grain matters for train counts far beyond its dollar value: a loaded corn train is worth roughly $2 to $4 million, compared with $25 million or more for a train of finished vehicles. Bulk moves the train counter; it barely moves the trade ledger.

The number that didn’t fit

Here’s where the story turned. If you only watched the rails, you’d expect Laredo’s trade totals to be shrinking. They’re doing the opposite: imports through Laredo rose $20.3 billion, up 9.4%, over the same 12 months.

The composition explains the apparent contradiction. What declined at Laredo is the rail basket: cargo trucks, cars, steel. What grew is machinery, up $22.7 billion, and electronics, up $5.3 billion, and those categories cross on trucks.

The port’s volume held. The mix moved from the rail yard to the highway.

The trailer test

That raised the question we found most interesting: what’s changing inside the trucks doing the hauling? (The tractor-trailers crossing the bridge, that is, not the cargo trucks riding the rails a section ago.) Laredo’s truck counts peaked in 2024 at about 252,000 a month and have eased since, while values keep climbing. Two explanations compete: inflation, or fuller trucks. We tested both against 8 years of data.

Inflation first. Deflated by the import price index, value per truck is up 48.5% in real terms since April 2018. Most of the gain survives the deflator.

If anything, that understates it. The all-commodity index rose 16.1% since 2018, driven up by fuels, which rose 44.7%. Prices for imported capital goods rose by 10.2% over the same 8 years, while consumer goods rose just 3.4%. The categories growing at Laredo are the ones whose prices barely moved; deflate the mix by its own categories and the real gain gets larger, not smaller.

Then weight, from the TransBorder freight files that record both value and shipping weight for every truck-borne import. The average trailer crossing at Laredo carried 9,738 kg of cargo in April 2018 and 9,415 kg this spring, 3.3% lighter.

That leaves one arithmetic possibility, and it’s the finding of the whole exercise: real value per kilogram is up 53.6% since 2018, from $4.55 to $6.99 in constant dollars. It has risen every single year since 2020.

The same trailer, at the same weight, crosses with cargo worth half again as much as it carried 8 years ago. That’s the nearshoring signature in physical form: the freight mix tilting toward machinery and electronics, the categories where a kilogram is worth the most, and away from freight that’s heavy and cheap.

It also says something about how we measure border trade. Crossing counts measure traffic. Value density measures what a corridor is becoming.

By the traffic measure, the border had a quiet year. By the density measure, it’s in the steepest 2-year climb the series has shown.

What to watch

Three things will tell us whether these trends hold. The monthly TransBorder releases will show if value density keeps climbing or 2026 turns out to be a tariff-distorted spike. The truck lanes will show whether the rail rebalancing is a contract cycle or a permanent network change. And the 232 auto review will determine how much of the car decline is reversed.

The bridges aren’t going anywhere, and neither is the freight. What’s changing is what’s inside the boxes, and that’s the number worth following.

Enterprises tried it. Then they did something the autonomy story didn’t predict.

According to KPMG’s AI Quarterly Pulse Survey for the first quarter of 2026, 63% of enterprise leaders now require a human to validate the outputs of their AI agents. A year earlier, that figure was 22%. The requirement nearly tripled in twelve months. KPMG surveyed hundreds of U.S. executives at billion-dollar companies, part of a global sample of more than 2,100 leaders across 20 countries.

These are the firms with the most capital and the most reason to automate. They had every incentive to remove the human. Instead, they put the human back, and they did it fast.

The lesson inside that number is one I keep returning to in my own work. The automation was the easy part. The structure you have to build around it is the hard part.

I call that structure the Accountability Surface: the layer of observability, audit trails, and human sign-off that surrounds an AI system, making its decisions explainable and defensible.

Subscribe now

The Accountability Surface is what the market has started to pay for. The feature that generates an output is now cheap. The apparatus that enables a company to stand behind its output is what carries value.

This is augmentation over automation, the position I’ve argued since 2023, showing up in enterprise budgets. Back then it was a claim about how people ought to work alongside these tools. In 2026 it’s a line item.

Governance turned out to be the job.

The path to 63% isn’t mysterious. Organizations moved fast. They shipped agentic workflows that booked, classified, drafted, and decided. Then a regulator, a client, or a board member asked a simple question: why did the system do that?

Too often, nobody could answer.

You can’t defend a decision you can’t explain. Not to an auditor reviewing a denied claim, not to a customer disputing a charge, not to your own leadership asking why the model flagged one account and cleared another. An AI agent that produces an answer with no trace behind it is a liability wearing the costume of efficiency.

So enterprises did what mature industries always do with powerful, opaque processes. They wrapped them in controls.

The KPMG data shows this is the mainstream now. The companies requiring human validation are the billion-dollar firms furthest along the adoption curve. They kept investing in AI. They added structure because they finally understood what it costs to scale it safely.

That cost has a shape. It looks like dashboards that surface why a model reached a conclusion, and logs that capture every input, output, and confidence score. It looks like review screens that let a qualified person approve or reject within seconds, so oversight is real work rather than a rubber stamp.

The model produces the output. Everything wrapped around it to make that output trustworthy is the Accountability Surface, and building it is where the durable jobs are.

I’ll admit this got concrete for me recently. I run a small set of AI agents that do research for me overnight, pulling data, tracking the developments I follow, and flagging changes while I sleep. It’s useful, and a little unnerving the first morning you wake up to a stack of findings you didn’t watch get assembled.

When I gave those agents the ability to spawn their own helper agents, my instinct surprised me. Instead of stepping back, I fenced them in.

I locked down what they were allowed to touch. I kept every irreversible action, anything that sends or changes a record, behind my own review the next morning. I added a separate pass whose only job is to check the first agents’ work before I rely on any of it.

I’d built an Accountability Surface for my own small operation before I let it scale. Nobody told me to. The work demanded it.

That instinct maps onto a framework I’ve used for years to describe who an organization needs in an AI economy. Three profiles do the work.

AI Producers build the systems. At the core are the engineers and developers who design the agents and the observability and audit layers beneath them. That circle is widening fast. No-code and AI-assisted tools now let analysts and business owners stand up working applications without writing much code, so producing is no longer the engineers’ monopoly.

AI Practitioners run the tools every day. The tariff analyst checking an LLM’s classification, the claims reviewer who can tell when a flagged file is wrong, the logistics planner validating a model’s route. They pair deep knowledge of how the business runs with enough understanding of the model to know when to trust it and when to overrule it.

AI-Aware Individuals are the executives and policymakers who set direction and sign off without operating the tools themselves.

Here’s what the KPMG number does to that picture. When 63% of firms require human validation, the demand for AI Practitioners stops being a forecast and becomes a hiring plan. Validation is their job. The Accountability Surface gets built and staffed by exactly the people the framework named.

That second profile is where the most valuable hire in a company now sits. For a decade, the prize was the person who could produce: build the model and ship the dashboard. AI made production cheap. Comprehension stayed expensive. The person who understands how the business actually runs and how the model actually works, both at once, is rare, and rare is what companies pay for.

That person is the spine of the Accountability Surface. Done well, validation is a senior judgment call: can this answer be defended, and would the reviewer catch the error and explain it? A company that treats the job as a clerical checkbox and staffs it with people who can’t tell when the model is wrong defeats the purpose. The firms that win at this pay for comprehension. The clicking was never the job.

It also raises the bar on the AI-Aware. Being aware is no longer enough. A leader signing off on an agent’s decision can’t vouch for what they don’t understand, and an executive running an AI overhaul she’s never tested with her own hands is making a bet she can’t see. “The system told me it was fine” is not a defense a regulator accepts. Awareness has to become readiness, the shift from knowing AI matters to knowing enough to govern it.

The World Economic Forum’s Future of Jobs report sharpens the stakes. It estimates 39% of workers’ core skills will need to change by 2030, expects AI and information processing to affect 86% of businesses, and projects 170 million new jobs over the same stretch. A real share of those jobs live on the Accountability Surface.

One clarification, because the headline number invites a misread. Requiring human validation isn’t the same as reviewing every keystroke. Most firms apply oversight where the risk is highest and let routine actions run. The Accountability Surface is structured oversight, calibrated to consequence, and companies are designing it on purpose.

If governance is the job, the question for everyone becomes where they fit and what to build. Each profile has a move.

For AI Producers and the firms that employ them: treat the Accountability Surface as a product. The observability layer, the audit trail, the review interface, these are features customers will pay for, because they’re what let customers sleep at night. A company that ships an agent with a clean explanation of every decision has a real edge over one that ships a faster black box.

For AI Practitioners: the skill that’s appreciating in value is judgment fused with fluency. Knowing a domain well enough to catch a confident, wrong answer, and knowing the tool well enough to see how it got there. That pairing is harder to automate than the underlying task, which is exactly why the stable work is heading toward it.

For AI-Aware leaders: make governance a requirement at the design stage. Ask how a system will be audited before you ask how fast it runs. Decide which calls an agent may never make on its own. The firms that wrote those rules early are the ones in the KPMG survey that are scaling with confidence today, rather than scrambling after a bad headline.

There’s a through-line under all three. The old survival line, the one quoted in business until it’s threadbare, still holds: the survivors won’t be the strongest or the smartest. They’ll be the ones that adapt best to change. In an AI economy, the best adapters are the best governors.

The autonomy story got the power of these tools right. It got the value's location wrong. The market has now settled that question with a number: 63%, up from 22%, in a single year.

Structure is how you scale trust. A company can buy the most capable model on the market, but with no way to explain and defend what it does, that capability stays stuck in a pilot, too risky to turn loose.

The next wave of advantage will go to whoever builds the Accountability Surface first and staffs it with people who can do the judging.

I built a modest one for a handful of overnight agents before I’d trust them with my own errands. The billion-dollar firms are doing the same thing, at scale. The data finally caught up to the argument.

On November 19, 2025, Mexico published a comprehensive reform to its Ley Aduanera. It took effect on January 1. On June 3, 2026, the White House signed an executive order restructuring customs enforcement on the U.S. side. Most of it takes effect within 180 days. Two governments, six months, one architecture.

Look at what each one actually does. Mexico’s reform makes customs brokers jointly liable for their clients’ unpaid duties, ends the old exemptions that protected agentes aduanales acting on importers' instructions, converts broker patentes from indefinite to fixed, renewable terms, and imposes formal know-your-customer obligations: supplier verification, valuation records, and five years of document retention.

Subscribe now

The American order works the same muscle from the other side. Every importer of record now needs to have minimum U.S. assets or bond coverage, beneficial ownership disclosure, and “good standing” with CBP. The registry gets purged and tiered by compliance risk. Foreign importers lose informal entry entirely and face new conditions on continuous bonds. Penalties get a floor of 50% of the assessed amount, with no mitigation for repeat offenders. And brokers face maximum penalties for weak due diligence or repeatedly representing noncompliant clients.

Strip away the legal vocabulary, and both governments did the same thing: they shifted customs enforcement from goods to actors. For two centuries, customs meant inspecting shipments. Both countries now vet the people and companies behind them. Banking went through this transformation a generation ago, when know-your-customer rules turned account-opening into a vetted privilege. That logic has now arrived at the border, in both directions, in the same fiscal year.

For Mexico, this is familiar territory, further formalized. Mexican importers have needed registration on the padrón de importadores for decades; standing in that registry has always been something you could lose. The American system, by contrast, has historically let almost anyone with a bond and a tax ID become an importer of record. The executive order ends that. In effect, the United States just built its own padrón.

In my opinion, the hardest part of the American order is the clock. The beneficial-ownership and U.S.-asset disclosures are not paperwork a foreign company assembles in a quarter. A Mexican manufacturer that has self-filed as its own importer for years now has 180 days to establish a qualifying U.S. presence, obtain CTPAT validation, or hand its filings to a validated broker. For many of them, this is the end of the non-resident importer as a working model.

That deadline lands hardest where the filings concentrate. Laredo’s brokerage ecosystem now sits between two newly hardened regimes. The same cross-border shipment answers to Mexico’s joint-liability rules southbound and the executive order’s importer requirements northbound. Brokers who are validated and staffed for the new vetting will absorb clients who previously filed on their own. Smaller firms have to build compliance intake on both legal systems simultaneously, on two governments’ timelines, with penalty floors waiting on each side. Many border businesses will struggle if they can’t react fast.

Here’s the part worth pausing on. USTR’s 2026 National Trade Estimate lists Mexican customs procedures among the frictions American exporters face, and Mexico’s new tariff schedule for goods from countries without trade agreements (1,463 tariff lines, average increases of 35%) has raised concerns among American business groups. Washington registered Mexico’s customs hardening as a trade barrier, then built a structurally similar regime six months later. I don’t read that as hypocrisy. I read it as two governments converging on the same enforcement philosophy without talking to each other about it.

And that’s the competitiveness problem. Each reform, taken alone, is defensible: both countries have real reasons to know who’s behind every entry, and historic underenforcement let non-North American goods slide through both systems. But convergence without coordination means one supply chain pays twice. The same Monterrey-to-San Antonio shipment now carries two sets of KYC obligations, two document-retention regimes, two penalty structures, and no shared compliance layer. Compliance spending that could have been built once, jointly, gets built twice, separately. Against a China that runs customs as one national system, North America just added friction to its own corridor and called it enforcement on both sides.

The USMCA Joint Review opens July 1. Customs cooperation is exactly the kind of unglamorous item that reviews are for, and for the first time, both governments arrive having just rebuilt their customs systems around the same idea. The architecture is converging on its own. The two systems already match. The review decides whether they ever meet.

At Laredo, truck crossings rose from 226,648 in January to 265,182 in March, a 17% increase in two months. That surge happened while national rates were climbing. Shippers moving goods through the busiest U.S.-Mexico land port are now paying premium linehaul rates to secure capacity in a tight market, then paying again for the operational realities of cross-border logistics.

The cost stack at a land port is different from the cost stack at a domestic distribution center. A shipper moving automotive parts from Monterrey to San Antonio pays a stack of charges, not a single rate. There’s a Mexican linehaul from Monterrey to the border. There’s drayage to shuttle the trailer across the bridge. There’s U.S. linehaul from Laredo onward, and that’s the segment where April’s rate cycle hit hardest. They pay for time: every hour a truck sits in a customs queue is an hour the driver isn’t moving freight, and carriers price that in. They pay compliance costs for USMCA documentation. They pay for the warehouse space to buffer against unpredictable crossing times. Southbound shippers face the same stack in reverse: U.S. linehaul to the border, drayage across, Mexican linehaul onward, and the April rate increase hits the U.S. leg of those moves, too.

When national linehaul rates spike 8% or 10% in a quarter, that increase sits on top of all those other costs. The percentage increase in total landed cost is smaller because the base is larger, but the dollar increase is real. For high-volume shippers moving thousands of loads per month through Laredo, April’s rate cycle added millions in annual freight spend. The shippers paying the steepest price are the ones still using Laredo as a transit point, crossing at the border, then trucking inland to Dallas or San Antonio for distribution. Those who’ve moved distribution operations to Laredo itself avoid the U.S. linehaul leg that just became more expensive. Every mile of inland trucking the rate cycle reprices is a mile a border-distributed shipper isn’t paying for.

Subscribe now

The driver math doesn’t help. FreightWaves' SONAR data shows van and flatbed rejection rates climbing through the spring. Carriers are turning down loads they don't want to run, and cross-border lanes price differently from domestic ones. The complexity of moving freight across an international border, with multiple regulatory regimes, multimodal handoffs, and binational scheduling, requires specialized carriers and specialized operations. That specialization costs more in a tight market, and in April, it cost meaningfully more.

The 265,182 trucks that crossed at Laredo in March moved roughly $30 billion in goods. That's close to a billion dollars a day across two bridges. The rate environment in which those trucks operate matters for U.S. manufacturers, Mexican suppliers, and every consumer who buys products assembled from cross-border supply chains.

The compounding is what matters now. In a stable policy environment, shippers can plan around seasonal rate cycles. They can lock in contract rates, adjust inventory buffers, and smooth out the peaks. In the current environment, where tariff policy shifts every 90 days and the July 2026 USMCA review deadline approaches, planning is harder. Shippers who can’t predict their tariff exposure can’t predict their total cost, so they pay premiums for flexibility. They book spot rates instead of contracts. They hold more inventory. They run smaller, more frequent shipments to reduce exposure on any single load.

All of that costs money. The Precarity Premium compounds with the rate cycle premium.

Infrastructure capacity lags trade growth. Laredo's two commercial international bridges are processing more trucks than ever, but the capacity expansions that would reduce wait times and increase throughput are multi-year projects. The last new commercial international bridge on the U.S.-Mexico border was the World Trade Bridge, which opened in 2000. The bridge is efficient by historical standards, but that isn't the same as being efficient enough for the corridor's volume growth.

The national trucking rate cycle will eventually turn. Capacity will loosen, rejection rates will fall, and linehaul rates will come down from cycle highs. That’s the pattern. But the cross-border cost stack doesn’t reset to zero. The structural premiums, the compliance burden, the uncertainty pricing: those persist.

Shippers who understand the full cost are already adapting. They're investing in visibility tools that track loads in real time. They're building relationships with carriers who specialize in the corridor. And they're moving warehouse capacity to the border itself. Locating distribution at the first U.S. touchpoint eliminates the inland linehaul leg that has just become more expensive. Goods cross at 8 AM, clear customs, undergo value-added operations, and enter distribution routes by noon. Every mile of inland trucking that the rate cycle reprices is a mile that a border-distributed shipper isn't paying for. Laredo's 50 million square feet of warehouse and distribution space exists because the strategic logic has been clear for years. The April rate cycle just made that logic more expensive to ignore.

A shipper moving 500 loads per month through Laredo at an average rate increase of $200 per load pays an extra $1.2 million per year. That’s before touching the drayage, scheduling premiums, compliance costs, and buffer inventory. The rate cycle will turn. The cost stack won’t.

Canada's $11.2 billion in covered US exports roughly matches the $19 billion bilateral metals trade. Trade lawyer logic. You hit my steel, I hit yours. Mexico's bilateral metals trade is about $5 billion and faces no equivalent counter-pressure.

Two theories of leverage. Both governments still pay 50% on the bulk of bilateral metals trade.

Subscribe now

What April actually did

The April 2, 2026, proclamation didn't lower the headline rate. It restructured the tier underneath. Annex I-A core metals stayed at 50%. Annex I-B derivatives dropped to 25%. Annex III industrial equipment got a temporary 15% cap through December 2027. Goods made abroad with at least 95% US-origin metal pay 10%. Anything with 15% or less metal content is entirely excluded from Section 232.

That’s the structure being sold as relief. Most of the press coverage stopped at the 25% Annex I-B number, which reads as a 50-to-25 cut.

Where bilateral trade actually lands

Mexico ships roughly $3.3 billion in steel and $1.85 billion in aluminum to the US each year. Canada ships $7.7 billion in steel and $11.5 billion in aluminum. The vast majority of both countries' shipments are HS Chapter 72 mill products and HS Chapter 76 aluminum. Annex I-A. The 50% tier.

Roughly 87% of US metals imports from Mexico fall in Annex I-A. About 82% from Canada. The relief tiers (I-B, III, US-origin carve-out, sub-threshold) account for the remaining 13-18%. The 10% US-content carve-out is effectively dead for USMCA partners because Mexican and Canadian producers smelt and pour their own metal, not US-origin metal.

USTR hasn’t published an official Annex-by-Annex split by country of origin. The shares above are modeled from Census product-mix data and the proclamation’s annex lists. Treat as an estimate. The methodology is in the chart footer.

The effective rate

Weigh the five rates by where bilateral trade actually sits, and the effective tariff lands at roughly 46%. Sensitivity testing across plausible Annex I-A coverage assumptions (75-90%) yields a range of 43-47%.

That’s a relief of 3 to 7 percentage points off the headline. Not the 25-point cut the coverage implies. Not the 35-point cut some readers assume.

There’s a second issue most coverage missed. Before April 6, 2026, Section 232 duties on derivative products applied only to the metal content. Since April 6, duties apply to the full customs value of the covered good. For a $100,000 piece of equipment with 30% steel content, the duty base went from $30,000 to $100,000. At the 25% Annex I-B rate, that’s a tax increase masquerading as a restructuring.

The Annex I-B “relief” is partly absorbed by the new full-customs-value calculation. The math gets worse, not better, for derivative manufacturers.

Drawback under 19 U.S.C. 1313 covers derivatives that are re-exported, which is real cash for maquila operators in Mexico who ship finished goods back across the border. It doesn’t help the metals consumed in US production, which is most of the bilateral flow.

The auto exception

The auto sector tells a different story, and it comes from a separate Section 232 action.

The March 2025 auto proclamation imposed a 25% Section 232 tariff on passenger vehicles and light trucks. USMCA-compliant vehicles pay that tariff only on the non-US content of the finished vehicle. A car built in Coahuila with 40% US content pays the 25% tariff on the remaining 60% of the vehicle value. Canadian content doesn't escape the tariff at the vehicle level. The carve-out is specifically for US-origin parts.

Auto and parts trade between the US and Mexico ran about $182 billion in 2024. The steel and aluminum content inside those vehicles is roughly 8 to 10% of the vehicle’s value (CAR and USGS estimates), which puts the embedded metals flow at $15 to $18 billion. That's larger than the combined bilateral trade in raw steel and aluminum under HS Chapters 72 and 76.

The auto proclamation adopted USMCA's US-content definition as the basis for partial relief. The April 2026 metals proclamation didn't replicate that approach. Two Section 232 actions, two different postures toward USMCA architecture. Where the administration chose to use USMCA mechanics, the relief is real. Where it didn't, the gap between the headline rate and the effective rate remains at 3 to 7 percentage points.

That asymmetry should shape how negotiators approach the July review. The lesson isn't that Section 232 needs more Annex categories. It's that applying USMCA's content-tracing logic to metals derivatives, the way the auto proclamation applies it to vehicles, would deliver the relief the April proclamation only gestured at.

Two bets on the same review

The July 2026 USMCA review is the inflection point. Mexico is betting that quiet negotiation through the formal review process will deliver more relief than retaliation. Canada is betting that economic pressure now creates leverage for the review later.

Both bets assume the review actually happens on schedule. Both assume the Trump administration treats USMCA partner status as meaningfully different from third-country status. The April proclamation only partly does. The drawback restoration under 19 U.S.C. 1313(a)-(b) is the clearest USMCA-specific benefit, but it covers only Annex I-B and Annex III articles, not the I-A core metals where 80%+ of bilateral trade sits.

Port Laredo handles close to $400 billion in annual two-way trade. Steel and auto components are a meaningful slice. Every percentage point of effective tariff on derivatives prices into Nuevo León plants and South Texas warehouses. The hemispheric supply chain is paying for the gap between headline policy and effective policy.

The April restructuring is progress. It’s not the relief package the coverage describes. Negotiators walking into the July review believing metals tariffs were meaningfully relaxed for North America will misread the room.

That’s the headline everyone has seen. Laredo also set a trade record the same year it lost the top spot. Imports at Laredo grew 7.8% year-over-year in 2025, according to U.S. Census Bureau data. Total trade through the port hit a record $354.65 billion.

Two airports passed it. Chicago O'Hare finished first at $423.22 billion in total trade, up 27.7% year-over-year. JFK finished second at $383.53 billion, up 53% year-over-year. The Port of Los Angeles fell to fourth at $301 billion, down 9.5%. Port Newark rounded out the top five at $234.71 billion.

Laredo kept moving what it has always moved: industrial machinery, vehicles, electronics. Three commodities that barely registered in the top-port conversation a year ago showed up in 2025 and reshuffled the geography of U.S. trade.

AI hardware. GLP-1 drugs. Gold.

Strip those three out, and the ranking looks roughly like it always has.

What actually moved through each port

The cleanest way to see 2025 is to look at what entered each top-five port by commodity chapter, using Census HS-2 import data.

Chicago O’Hare (imports: $329.11 billion, up 30.8%)

Three categories carried the bulk of the growth. Electrical machinery and electronics (HS 85) grew to $75.16 billion, up 30.8%. Nuclear reactors and industrial machinery (HS 84, which includes computers and AI servers) grew to $73.91 billion, up 57.1%. Pharmaceuticals (HS 30) came in at $62.41 billion, down 7.7% year-over-year. The real mover sits one chapter below in the Harmonized System. Organic chemicals (HS 29) tripled from $19.92 billion to $59.27 billion, up 197.6%. That's where the GLP-1 story lives. Finished drugs held roughly flat; the active pharmaceutical ingredients behind Ozempic, Wegovy, and Mounjaro are what surged through O'Hare.

That's where the GLP-1 story lives. Finished drugs grew modestly. Pharmaceutical precursors and active ingredients, mostly from Ireland and Denmark, surged.

Eli Lilly’s County Cork facility and Novo Nordisk’s Danish plants ship API rather than packaged Ozempic. The API flies into O’Hare. The finishing happens in the United States.

JFK (imports: $202.18 billion, up 42.9%)

JFK's 2025 is a gold story. Imports of natural or cultured pearls, precious and semi-precious stones, and precious metals (HS 71) nearly doubled from $60.51 billion to $119.98 billion, up 98.3%. That single chapter accounts for roughly 59% of JFK's total 2025 import value.

JFK handled 55% of U.S. unwrought gold imports and the vast majority of refined bar imports. On the export side, 82% of all U.S. gold exports departed from JFK, bound primarily for Switzerland and the United Kingdom.

Gold prices hit $5,589 per ounce in late January 2026, the fastest appreciation in four decades. U.S. gold ETF holdings reached a record 2,019 tonnes. JFK's rise to number two reflects how that dynamic moved through the physical trade system. Bullion in. Bullion out. Arbitrage between New York and London markets on a scale no U.S. port had seen before.

Subscribe now

Port Laredo (imports: $227.4 billion, up 7.8%)

The commodity mix looks different from the airports. Machinery (HS 84) at $67.67 billion, up 40.8%. Electronics (HS 85) at $40.25 billion, up 10.8%. Vehicles (HS 87) at $57.36 billion, down 10.9% — the 25% vehicle tariff announced in March 2025 couldn’t be front-loaded because it hit too late in the shipping calendar. Those three chapters still account for roughly two-thirds of everything Laredo imports. The industrial spine of U.S.-Mexico trade.

Inside HS 84, one sub-chapter tells the AI story from the Mexico side. HS 8471, which covers data processing units and servers, went from $6.06 billion at Laredo in 2023 to $13.44 billion in 2024 to $30.83 billion in 2025. A 409% increase over two years, and a 129% jump in the single year 2024 to 2025.

Foxconn, Inventec, Wiwynn, and Quanta are assembling Nvidia GB200 superchips in Chihuahua and Jalisco. The servers cross at Laredo. The same wave that lifted O’Hare lifts Laredo too, through a different geography.

Port of Los Angeles (imports: $268.53 billion, down 8.8%)

LA didn’t just fail to catch the three shocks. LA contracted. Imports fell from $294.51 billion to $268.53 billion, a drop of nearly $26 billion year-over-year. The decline was concentrated in exactly the categories you’d expect: toys down 22%, iron and steel articles down 24%, machinery down 14%, furniture down 13.5%, electronics down 9%. Nearly every major commodity chapter declined.

This is the China tariff story in one port. LA handles the bulk of Chinese container goods entering the United States. When importers shift sourcing to Mexico and Vietnam to dodge tariffs, LA bleeds. Port Executive Director Gene Seroka called 2025 “a roller coaster.” Container volume held steady above 10 million TEUs, but dollar value fell sharply. Lower-value goods per container, and a port caught between two trade realignments.

Port Newark (imports: $207.55 billion, down 5.8%)

Newark rounded out the top five, running the East Coast counterpart to LA's story. Like LA, Newark saw imports decline (down 5.8%), driven by the same tariff realignment. Unlike LA, Newark didn't catch any of the three commodity shocks. A big port having an unremarkable year while the action moved elsewhere.

The three shocks in one frame

Organize the 2025 data around the three commodities rather than the five ports and the story clarifies.

AI hardware. U.S. imports of digital processing units (HS 847150) exceeded $163 billion in 2025, per Census data analyzed by TradeInt. That single 4-digit sub-code was the largest import category in the country. The Federal Reserve Board’s February 2026 research note calls AI-related imports the single largest trade story of the year, noting that such imports “more than doubled” from 2023 to 2025. AI hardware entered through both air cargo (O’Hare, SFO, LAX) and land border (Laredo, Ysleta). Both routes grew triple-digits. One wave, two geographies.

GLP-1 pharmaceuticals. Ireland’s goods exports to the United States rose 52% to approximately $132 billion in 2025, per the Irish Central Statistics Office. Medical and pharmaceutical products made up 53% of that total. Denmark’s pharma exports to the U.S. rose 40% through July 2025. The flow concentrated at O’Hare because that’s where Eli Lilly and Novo Nordisk API enters the country before U.S. formulation plants turn it into finished drug.

Gold. U.S. gold exports rose from the fourteenth most valuable export category in 2024 to the fifth most valuable in 2025, per Census data analyzed by Ken Roberts. Non-monetary gold became the leading U.S. export in both October and November 2025, a first in the modern record. The flow ran almost entirely through JFK, driven by tariff-era safe-haven demand, Swiss refining arbitrage, and record ETF accumulation.

Front-loading, the invisible hand on 2025

One more variable distorted every number in this post. The Trump administration’s April 2, 2025 “Liberation Day” tariffs were pre-announced in late March, which gave importers a window to ship early. They used it.

March 2025 was the single largest month in U.S. import history. Laredo alone handled $31.35 billion in March, a port-level monthly record confirmed by Census data, up 12.4% from March 2024.

Then April collapsed. Consumer goods imports fell 32% month over month. Industrial supplies fell 31%. Container bookings at the Port of Los Angeles warned of 35% drops.

The Federal Reserve Board, in an August 2025 analysis, confirmed the pattern: “This shift was particularly concentrated in March, when U.S. import volumes from several major trading partners spiked, likely in expectations of the pre-announced date of April 2.”

Some portion of 2025’s headline trade numbers reflects anticipation rather than underlying demand. How much is an open question.

The AI import surge continued through the year, which suggests genuine data center buildout alongside front-loading. The Ireland pharma surge similarly persisted past Q1. Gold was driven by prices and ETF flows that kept moving regardless of tariff calendar. The durable-goods categories, vehicles, consumer electronics, and industrial supplies, showed the classic front-load signature: Q1 spike, Q2 crash, gradual normalization.

When the 2025 rankings get revised in mid-2026, expect some softening. Enough to know the underlying shifts were real.

Which of the three shocks lasts

This is the question that determines the 2026 and 2027 rankings.

AI hardware probably lasts. The CSIS analysis from March 2026 concluded that U.S. firms will depend on Taiwan for advanced GPU supply through at least 2028. Mexico’s assembly capacity keeps expanding, with Foxconn announcing additional capacity in Chihuahua and Nvidia’s GB200 line ramping through 2026. The flow through both O’Hare (air from Taiwan and Ireland) and Laredo (land from Mexico) is probably structural for at least three more years.

GLP-1s moderate but persist. Eli Lilly is building U.S. API capacity to reduce Irish dependence, which will reduce the O’Hare flow over three to five years. Demand, however, isn’t moderating. Novo Nordisk and Lilly combined for roughly $40 billion in global GLP-1 sales in 2025 and guidance suggests continued growth. The composition of the trade will shift from API imports to finished drug capacity. The overall flow stays elevated.

Gold is the one to watch. JFK's number two rank is the most concentrated of the top five. Take gold out, and JFK drops back to its historical range near $150 billion in total trade. The drivers, which include tariff uncertainty, ETF accumulation, and Swiss refining arbitrage, are all path-dependent on monetary and policy conditions that may look very different in 2027.

What happens to Laredo from here

Laredo finished third in 2025 after two years at number one. The composition data suggests the port is being lapped by commodity waves it was never going to carry.

The AI hardware flow through Laredo is still early. Mexico's HS 8471 exports to the U.S. hit $85.4 billion in 2025, up 145%, per Banco Base analysis. Chihuahua alone accounted for 46% of those exports, and Jalisco added another 23%. The two states that host Foxconn, Inventec, Wiwynn, and Quanta generated nearly 70% of Mexico's computer export boom. Taiwan’s ODMs, including Foxconn, Inventec, Quanta, and Wiwynn, keep expanding Mexican assembly capacity.

The gap between what O’Hare imports in computers ($51.5 billion in 2025) and what Laredo imports ($30.8 billion) could close. If Mexican assembly absorbs a larger share of North American AI hardware demand in 2026 and 2027, Laredo catches up on the same commodity, lifting O’Hare.

Meanwhile, the industrial spine that took Laredo to number one in the first place continues to grow. Machinery up 41%. Electronics up 11%. Vehicles down 11% as the March tariff announcement hit too late to front-load. Those numbers reflect a port absorbing a real tariff shock on one category while the AI wave drives another to record highs.

When gold reverts, JFK drops. When AI assembly keeps migrating south, Laredo rises. The 2025 ranking was a snapshot of three commodity shocks running through five ports. The ports that finished highest caught the biggest waves. The port that finished third caught a smaller wave and set another trade record doing it.

That’s a different story than “Laredo fell to number three.”

A bright-red humanoid named Lightning crossed the finish line of a half-marathon in Beijing on Sunday morning at 50 minutes, 26 seconds. It clipped a railing on its way in, got helped back up, and still finished about seven minutes faster than the human world record. Its time was faster than every one of the 12,000 human runners in the same event. Lightning’s two teammates took second and third in the robot category.

Lightning was built by Honor, a Chinese smartphone company. That detail usually becomes the story. It shouldn’t be.

Two weeks earlier, BMW announced it was expanding its pilot with Figure AI after the California company’s Figure 02 robots finished a ten-month run at the Spartanburg plant. In that window, the robots worked ten-hour shifts, five days a week. They loaded 90,000 sheet-metal parts onto welding fixtures within a 5-millimeter tolerance, on 84-second cycles. They contributed to the production of 30,000 BMW X3s. Hit rate on placement accuracy: 99%.

Spartanburg is the real number. The marathon was the ad.

The moment hiding inside the headline

The MHI and Deloitte 2026 Annual Industry Report, released this spring, surveyed more than 500 supply chain professionals around the world. The survey is in its thirteenth year. This year’s numbers broke from the pattern. The share of respondents who rated AI’s disruptive impact on supply chains as significant or transformational climbed 25 percentage points in a single year. Robotics and automation climbed 16.

Humanoid robotics shows up as its own category for the first time. Adoption today: 4%. Expected adoption within five years: 28%. That’s a seven-fold increase in five years.

Supply chain professionals don’t tend to overreact. They manage variability for a living. When 90% of them report talent and workforce challenges and 58% call those challenges major (up 9 points from last year), and when 65% say they’ll invest in AI and 46% in robotics over the next two years, they’re not chasing a headline. They’re reading a labor curve they already feel.

In my view, this is where Logistechs stops being an idea about software and starts being an idea about bodies. I defined Logistechs a few years ago as the impact of exponential technologies on logistics: transport, handling, and customs clearance. Software has done most of the heavy lifting in that definition until now. Physical AI is the part that shows up in a warehouse, on a dock, in an assembly cell. It shows up with hands.

What deployment actually looks like

BMW is the clearest case because it’s been documented in public. Figure 02 at Spartanburg ran a classic pick-and-place task, lifting sheet-metal parts from racks and placing them on a welding fixture before a conventional six-axis robot welded them in. The targets were specific: 84 seconds per full cycle, 37 seconds for the load phase, and greater than 99% placement accuracy per shift. BMW is now extending humanoid pilots to its Leipzig plant in Germany, with high-voltage battery assembly and component manufacturing as the next use cases.

Tesla is converting assembly lines at its Fremont factory this year to produce the third generation of Optimus, with a target unit cost under $20,000 and a stated ambition of 10 million units per year from a new facility at Gigafactory Texas. Mercedes-Benz is testing Apollo, built by Austin-based Apptronik, at its Berlin plant on logistics and quality checks. Hyundai, which owns Boston Dynamics, has signaled plans to deploy the new Atlas humanoid across its manufacturing network. Apptronik raised $520 million this year with Google and Mercedes-Benz among its backers. Bank of America projects that humanoid unit costs will be below $17,000 by 2030.

Then there’s Réflex Robotics. Founded by MIT graduates. Announced in March by Nuevo León’s governor: it will build the first humanoid-robot manufacturing plant in Latin America. The location: Nuevo León, about 140 miles south of the World Trade Bridge.

The pattern here is a deployment curve already running.

Subscribe now

The marathon was a product demo

Lightning’s real competitor on Sunday was the assumption that we still have time to plan for this.

The actual argument in Beijing was an engineering argument. Balance, terrain recovery, battery endurance, and thermal management under 13 miles of stress. Those are the four problems that have kept bipedal robots out of warehouses and off factory floors for twenty years. When a machine manages all four of them in public, in front of cameras, at competitive speeds, the venture math changes. The purchase math changes. The deployment timeline compresses.

The MHI respondents forecasting 28% adoption within five years aren’t guessing. They’re telling you what their suppliers are telling them.

Why this lands on the border first

In 1956, a North Carolina trucker named Malcom McLean loaded 58 metal boxes onto a converted tanker in Newark and sailed them to Houston. Within a generation, his invention had re-priced port labor from Brooklyn to Rotterdam, rewired global trade, and made the ports that adapted rich and the ports that didn’t irrelevant. Physical AI is moving through manufacturing and logistics on a similar curve, and the corridor is where it first touches the ground.

Port Laredo processes over 6 million trucks and close to $400 billion in trade every year. Nearly a billion dollars crosses two bridges here every calendar day. What happens at this crossing previews what happens at every commercial port of entry from San Diego to Brownsville. This cross-border region’s comparative advantage was built, and is still defended, on a specific kind of labor equation: competitive manufacturing labor in Mexico meeting American capital, logistics infrastructure, and market access, with Laredo as the hinge where the two sides meet.

That equation is being re-priced in real time, exactly like port labor was in the container era. The labor didn’t disappear. The capital cost of replacing it just dropped 40% in two years, according to Goldman Sachs data cited in Deloitte’s 2026 tech trends work.

The 2025 Texas A&M International University Texas Center trilateral employment vulnerability analysis found 9.9 million jobs across the United States, Mexico, and Canada are exposed to trade disruption. Texas alone carries 805,000 of that U.S. exposure. Most of the Mexican share sits in the northern border states, in the maquiladora light assembly and automotive component work that feeds the corridor. Those are exactly the tasks that Figure 02, Optimus, Apollo, and whatever comes out of the Réflex plant in Nuevo León are being designed to do.

Deployment at scale is a decade-long process, not a quarterly one. Humanoid robots still fail more than traditional industrial arms. Maintenance costs are high. Supply chains for high-torque actuators and tactile sensors are thin. The Deloitte 2026 tech trends study found that only 11% of organizations are actively using agentic AI systems in production, even though 38% are piloting. Most pilots don’t scale. Most don’t scale soon.

But that’s the point. A decade-long process that’s already started is one we should be planning around. The labor shortage driving adoption is structural: 1.9 million U.S. manufacturing jobs are projected to be unfilled by 2033, according to Deloitte and the Manufacturing Institute, with industry analysts estimating that roughly a quarter of the current workforce is nearing retirement. Mexico has become the second-largest industrial robotics market in the Americas, importing essentially all of the units it installs. The money is already moving. The hardware is already shipping.

What the corridor has to figure out

Three things matter for Laredo and Nuevo Laredo in this window, and for every binational metro that lives off cross-border trade.

The first is workforce. When a warehouse reduces cost per box moved from 20 cents to 3 cents, as MHI documented in a real deployment this year with a major U.S. retailer, the people who used to move those boxes don’t disappear. They either get retrained into roles that manage, maintain, and improve the automation, or they end up in a worse version of the same economy. The supervision layer, the exception-handling layer, and the maintenance layer for physical AI all need people with skills that most regional training pipelines on both sides of the border are not yet producing at volume. The institutions that run those pipelines will have to develop curricula for roles that don’t yet have job descriptions.

The second is physical infrastructure. Humanoid robots running three shifts need 5G coverage, edge compute, redundant power, and clean data pipelines. Our commercial bridges are designed for trucks, not data. Public policy conversations on both sides of the border will have to start treating digital infrastructure for the corridor with the same seriousness as physical infrastructure, because you can’t run physical AI on fiber that wasn’t planned for it.

The third is USMCA itself. The 2026 Joint Review arrives just as humanoid deployment curves start to bend. The labor value content rules and the regional content requirements that sit at the heart of USMCA automotive provisions were written in 2018 under the assumption that a “worker” was a person. What happens to rules of origin when a Mexican maquiladora’s 2031 workforce is 30% humanoid, and those humanoids were designed in Cambridge, manufactured in Nuevo León, and programmed by a model trained in Austin? Is the car that crosses the bridge a North American car? Does the agreement’s labor enforcement machinery still mean what it meant seven years ago?

USMCA 2.0 requires institutional architecture, not just rule updates. The three proposals I’ve written about, the Binational Customs, the North American Industrial Coordination Council, and the North American Digital Infrastructure Coordination Initiative, were designed as the implementation layer for a continent that intended to be competitive. Physical AI is the case that proves why that architecture matters. Without NADICI-style coordination on AI-enabled trade interoperability, rules of origin in a humanoid-assisted economy become unmanageable. Without NAICC-style industrial policy coordination, the continent cedes humanoid manufacturing to companies and countries that built the institutional capacity first. Without a BCA, the customs system that has to classify, clear, and audit physical AI-produced goods remains two separate national systems talking past each other.

The race that actually matters

Borders, in my view, are laboratories. Societal petri dishes where the ability to coexist shapes economic and political prosperity. The Laredo-Nuevo Laredo corridor is one of those laboratories. NAFTA landed here first. USMCA’s enforcement mechanisms landed here first. The physical AI transition will land here first, because this is where the goods move.

The question for Texas, for Tamaulipas, for Washington, and for Mexico City is whether they shape that transition or whether they have it shaped for them. The companies that adapt fastest tend to win the next decade. The same is true of regions.

Lightning crossed the finish line in 50 minutes and 26 seconds on Sunday. The clock on our corridor is less precise than that, but it’s running at the same pace.

After I published “Mexico Passed China” on The Bridge, a few readers pointed out that the charts only covered goods trade. Fair point. The Census Bureau tracks goods by port, by commodity, by month. Services trade lives in a different dataset (BEA’s International Transactions Accounts), gets released on a different schedule, and requires a different kind of analysis.

So I ran the numbers. For both countries.

The full picture

When you add goods and services together, U.S.-Mexico trade in 2025 hit $969.3 billion. U.S.-Canada trade hit $869.7 billion. Combined: $1.84 trillion with just two countries.

Canada tells a different story.

Mexico moves more goods ($872.9B vs. $719.5B). Canada moves more services ($150.2B vs. $96.4B). Services represent 17.3% of the U.S.-Canada relationship but only 9.9% of the U.S.-Mexico relationship.

The U.S.-Mexico relationship is $30.7 billion away from $1 trillion. At current growth rates, it crosses that threshold in 2026.

Subscribe now

Two surpluses, opposite directions

The U.S. runs a services trade surplus with both neighbors. But the trajectories are moving in opposite directions.

With Mexico, the surplus hit $8.4 billion in 2025, up 59% from $5.3 billion in 2024. U.S. exports grew 3.9%. Imports from Mexico actually fell 2.5%. That’s the largest U.S. services surplus with Mexico in over a decade.

With Canada, the surplus fell to $28.9 billion, down 13% from $33.2 billion in 2024. U.S. exports to Canada declined slightly (-0.8%) while imports from Canada rose 6.4%. Q4 2025 posted the smallest quarterly surplus ($6.0B) since early 2024.

Two things stand out. With Mexico, 2022 was the inflection point: the surplus almost vanished (+$202 million) as post-COVID travel recovery was asymmetric. Since then, the surplus has rebounded sharply, consistent with strong growth in U.S. exports of IP licensing, financial services, and professional consulting globally.

With Canada, 2024 was the peak. The surplus began to compress in 2025 as U.S. services exports to Canada declined slightly while Canadian services exports to the U.S. grew steadily.

Honest about scale

The U.S. ran a combined services surplus of $37.3 billion with its two USMCA partners in 2025 ($8.4B with Mexico, $28.9B with Canada).

The goods deficits tell a different story. The U.S. ran a $196.9 billion goods deficit with Mexico and a $46.4 billion goods deficit with Canada. Combined: $243.3 billion.

The $37.3 billion services surplus offsets 15.3% of that combined goods deficit. Services trade doesn’t fix the deficit. But it does change the picture, especially with Canada, where the $28.9 billion services surplus cuts the $46.4 billion goods deficit by 62%.

With Mexico, the offset is smaller. The $8.4 billion services surplus covers 4.3% of the $196.9 billion goods deficit. Meaningful in dollar terms. Less so in the policy debate.

What’s in the $247 billion

Services trade between the U.S. and its USMCA partners includes categories that don’t fit on a truck. Travel is the largest component on both sides, followed by financial services, intellectual property licensing, professional and management consulting, telecommunications and IT services, transport, and insurance.

The U.S. runs its biggest bilateral advantages in intellectual property and financial services. Canada’s services exports to the U.S. are concentrated in professional services, insurance, and transport. Mexico’s services exports are concentrated in travel and, increasingly, professional services as Mexican tech and consulting firms grow.

The $247 billion in combined North American services trade exceeds the GDP of over 150 countries. It's roughly the size of New Zealand's economy.

These figures capture trade between U.S. residents and non-residents. They don't capture the broader circular economy along both borders: cross-border retail, medical tourism, real estate purchases, and education spending. That's a different dataset and a different analysis.

What this means for the USMCA review

The $243.3 billion combined goods deficit with Mexico and Canada is the number that drives headlines, tariff policy, and political rhetoric. It’s real. But it’s incomplete.

Adding services reduces the combined deficit by $37.3 billion, from $243.3 billion to $206 billion. That’s a meaningful correction, especially for Canada, where the services surplus absorbs nearly two-thirds of the goods deficit.

What’s more significant is the structural composition. The services categories driving U.S. surpluses (IP, financial services, professional consulting, digital services) are exactly the sectors where the U.S. has competitive advantages that aren’t going away. They’re also the sectors most affected by USMCA’s digital trade provisions, which are up for review.

The opposite trajectories matter too. The surplus with Mexico is growing. The surplus with Canada is shrinking. If the goal of the USMCA review is to strengthen North American competitiveness, understanding why these trends are diverging is as important as debating tariff rates on goods.

This week, I want to show you where that trade physically crosses.

$872.8 billion in U.S.-Mexico merchandise trade in 2025. That’s the number from the Census Bureau. But a number that large is hard to hold in your head. So I built three charts to make it concrete.

Where it Crosses

Hover over each bar for the full breakdown, including how much crosses per hour and per minute. → Interactive version

That compression is the story. An $873 billion trade relationship between two countries with a combined population of 470 million routes through a corridor in South Texas where 260,000 people live.

What’s on the Trucks

Every day, roughly 19,800 commercial trucks cross at Laredo. But what are they carrying?

The interactive version lets you toggle between total trade and the export/import split for each commodity. That's where the manufacturing integration story gets visible. → Interactive version

Three categories dominate: machinery ($88.3 billion), vehicles and auto parts ($74.1 billion), and electrical equipment ($54.6 billion). Combined, they account for 61% of everything that crosses.

Look at the import-to-export ratio within those categories. $64.9 billion in machinery crossed northbound from Mexico through Laredo, while $23.4 billion crossed south. Vehicles: $56.6 billion north, $17.4 billion south. Electrical equipment: $36.3 billion north, $18.3 billion south.

These aren’t finished products moving one direction. They’re components crossing back and forth in an integrated manufacturing system. Some automotive components cross the U.S.-Mexico border up to eight times before final assembly. An engine block might start as American steel, get machined in Monterrey, return to Texas for testing, go back south for assembly into a powertrain, and cross north again inside a finished vehicle.

Every one of those crossings passes through a port of entry. And 40% of them pass through Laredo.

If Laredo’s Port of Entry Were a Country

Hover over any country for population, context, and how Laredo's daily throughput compares. → Interactive version

$353.9 billion in annual trade passing through a single port of entry would place it between Austria and Hungary in the WTO’s global merchandise trade rankings. Roughly 31st in the world. Ahead of Norway, Denmark, and Israel.

Austria has 9 million people, a GDP of $550 billion, and centuries of industrial development behind its trade position. Laredo has 260,000 people and the most efficient logistics corridor in North America. The gap between them: $61 billion, and it’s closing.

The comparison that makes this tangible: Apple’s total revenue in fiscal year 2025 was $416.2 billion, about $1.14 billion per day. Laredo moves $970 million per calendar day. The entire consumer electronics output of the world’s most valuable company is in the same range as what passes through two bridges in South Texas.

$40.5 million per hour. $675,000 per minute.

Why This Matters

That concentration exists because the cross-border corridor works. Within a 40-mile radius spanning the border, over 500 customs brokerage firms, 500 freight carriers, and 300 warehousing facilities form a logistics ecosystem that supports more than 30,000 jobs. Add 61 million square feet of industrial space across 59 parks, 140 miles of toll highway to Monterrey's industrial base, and Nuevo León's 739 IMMEX manufacturing establishments with nearly 400,000 direct workers. KIA, Caterpillar, and General Motors all route major operations through this corridor. This infrastructure didn't appear overnight. It took decades to build.

But efficiency built on a single corridor means that every policy decision, every tariff change, every inspection delay at the World Trade Bridge ripples through supply chains across the country.

The I-35 highway that starts at the World Trade Bridge runs 1,568 miles to Duluth, Minnesota, through six states. TxDOT calculates that corridor supports $1.2 trillion in economic activity, 49% of Texas’s entire GDP. No single land border crossing in the world handles more commercial truck traffic. The Ambassador Bridge between Detroit and Windsor, often called the busiest U.S.-Canada crossing, handled 2.3 million trucks in 2024. Laredo handled nearly 6 million.

Trade policy conversations happen in Washington. Trade physically happens in places like Laredo. When policymakers discuss tariff rates, rules of origin thresholds, or customs modernization, they’re making decisions that affect not just this corridor but every community along the supply chains that depend on it.

The July 2026 USMCA Joint Review will determine whether the agreement that underpins this system gets renewed, renegotiated, or enters a 10-year sunset. That’s the policy decision that matters most for this corridor, and for the supply chains across the country that route through it.

The town is called Baarle. Half of it is Belgian (Baarle-Hertog), half is Dutch (Baarle-Nassau). It contains 30 enclaves: 22 Belgian parcels within the Netherlands and 8 Dutch parcels within Belgium, some of which are nested within one another like geographic Matryoshka dolls. The border runs through shops, restaurants, and living rooms. Where your front door sits determines your nationality, your tax code, and your electricity provider. One family famously moved their front door a few meters to switch countries and get a better deal. During COVID, Belgian lockdown rules shut down businesses on one side of a street while Dutch shops across the hallway stayed open.

I learned about Baarle over dinner at the Association for Borderlands Studies’ 50th-anniversary conference in Albuquerque last week. A European colleague told the story between courses in Old Town, and the table couldn’t stop asking questions. That’s what these conferences do: the panels are good, but the best material surfaces over meals. And it struck me: this is what border scholars actually spend their time thinking about. Not walls and checkpoints (though those came up too), but the infinitely strange, creative, sometimes absurd ways that human beings organize life around lines drawn on maps.

The conference brought together over 150 researchers from more than 20 countries across 40 sessions. The range tells you something: panels on border trauma and self-deportation sat alongside sessions on cross-border semiconductor talent pipelines in Sonora-Arizona, water diplomacy on the Rio Grande, the economic cost of non-cooperation at EU borders, and how the Cucapá indigenous community in the Colorado River Delta has navigated a line its members never drew. Someone mentioned that border studies, once considered a dead subfield, isn’t anymore.

Here’s why. Most people think of a border as a line. A wall, a river, a checkpoint. Something you cross or don’t. But spend a few days with border scholars and you realize that a border is actually several things stacked on top of each other. It’s a legal boundary (which laws apply to you). It’s an economic membrane (which goods move freely and which don’t). It’s a cultural seam (where languages, customs, and identities overlap or collide). And it’s a lived experience (how real people organize their mornings, their commutes, their kids’ schools around that line). Baarle isn’t strange because its border is complicated. Baarle is just honest about what every border really is: multiple systems layered on top of each other, often in contradiction.

Think about the Texas-Mexico border. A truck driver crosses the World Trade Bridge in Laredo, TX. carrying auto parts manufactured in Monterrey, bound for a plant in San Antonio. In the span of 20 minutes, that driver moves through Mexican commercial law, U.S. customs jurisdiction, USMCA rules of origin verification, Department of Transportation safety standards, and Texas state regulations. The cargo doesn’t change. The truck doesn’t change. But the legal, economic, and regulatory reality shifts with every step. That’s not one border. That’s multiple borders stacked on top of each other at the same GPS coordinate.

In Baarle, they solved this by building shared institutions. The two municipalities hold joint sessions. Their police share a station. The fire departments merged. The library has staff from both countries. The border still runs through every building, but the governance crosses it too. The lesson is almost too simple: the line stays, but the institutions learn to work across it.

I participated in a roundtable on the U.S.-Mexico border economy under neoprotectionism, alongside colleagues from El Colegio de la Frontera Norte, the University of California San Diego, and Arizona State University. The discussion centered on critical trends heading into the 2026 USMCA Joint Review. One colleague made a point that stuck with me: trade flows don’t equal integration. The fact that trucks cross a bridge doesn’t mean the economies on either side are actually integrated. Real integration requires institutional, environmental, and social infrastructure, not just commercial volume. Without it, you get what another panelist called “savage integration,” growth without the architecture to sustain it. That distinction matters. The manufacturing ladder that lifted previous generations of countries into high-income status is plateauing for middle-income economies like Mexico, just as AI begins restructuring the service sectors they’re supposed to transition into. The rules aren’t the problem. The institutional and educational infrastructure to manage that kind of structural shift doesn’t exist yet.

This was my last conference as the ABS Executive Secretary and Treasurer, a role I’ve held since 2022, but I’ll remain an active member. The ABS was founded in 1976 by Ellwyn Stoddard, Richard Bath, Oscar Martinez, and Anthony Kruszewski at a WSSA meeting in Tempe, Arizona. What started as a small group focused on the U.S.-Mexico border has grown into the leading international scholarly association on border issues, with members across 55 countries spanning the Americas, Europe, Asia, and Africa. The Journal of Borderlands Studies, first published in 1986, is now 40 years old and remains the primary forum for interdisciplinary border research worldwide.

There was also a tribute to honor Dr. Joan B. Anderson, who passed away in February. Joan was the first woman to serve as both ABS president and editor-in-chief of the Journal of Borderlands Studies. She once described the field perfectly: “A macroeconomist is working with sociologists, anthropologists, and geographers. That’s like the definition of border studies.”

She was right. And Baarle proves it. The town works because people stopped pretending the border wasn’t there and started building around it instead. That’s the shift border studies has been documenting for 50 years: from arguing about where the line should be, to figuring out how life works once it’s drawn.

That’s the work. Not erasing borders, but building the institutions that make them functional. Whether it’s a house on Loveren Street with two addresses or a port in Laredo processing $400 billion in trade, the question is the same: once you draw the line, how do you live with it intelligently?

The 150 researchers in Albuquerque last week are working on that answer. Fifty years in, the field has never been more relevant.

Two years later, the gap has exploded. I wanted to see what the full picture actually looks like.

Here's what I found.

Thirty years of U.S. trade with its three largest partners, from NAFTA’s launch to today. The story the data tells is more complicated than the headlines suggest.

Three decades, three acts

Act I: NAFTA’s promise (1994-2001). When the agreement took effect, U.S.-Mexico trade was $100 billion a year. Canada was the dominant partner at $243 billion. China was a minor player at $48 billion. Within seven years, Mexico doubled. China more than doubled. The trajectories were set.

Act II: China’s rise (2001-2018). China’s WTO accession in 2001 changed everything. By 2006, China had passed Mexico. By 2015, it had passed Canada. At its peak in 2022, U.S.-China trade hit $690 billion. For more than a decade, the economic center of gravity pulled toward the Pacific.

Act III: The reversal (2018-present). The 2018 tariff war broke the trend. In 2019, Mexico briefly reclaimed the #1 spot as U.S.-China trade contracted. COVID scrambled the picture in 2020 (China bounced back temporarily as its factories reopened first). But by 2023, Mexico had definitively pulled ahead. The gap in 2025: $458 billion.

That last number is worth pausing on. The distance between Mexico and China in 2025 is more than double the entire U.S.-Mexico trade relationship in 1999.

What’s actually crossing the border

Machinery and computers dominate, accounting for $216 billion in two-way trade. Vehicles and parts run second at $150 billion. Most people assume autos lead, but machinery and computers are $66 billion ahead. This is the integrated North American auto industry in physical form. An engine block crosses the border 8 times before it ends up in a finished vehicle.

Electrical machinery (appliances, semiconductors, wiring harnesses) comes in third at $148 billion. Mineral fuels, optical and medical instruments, and plastics fill out the top six.

The composition matters because it tells you this isn’t a low-cost labor story anymore. Mexico’s exports to the U.S. are concentrated in high-value manufactured goods, the same categories that made China’s trade relationship with the U.S. so consequential. The difference: Mexico’s trade is bidirectional. The U.S. exports $338 billion to Mexico. It exports $106 billion to China. That asymmetry is the structural story underneath the headline.

Where it crosses

Laredo, Texas, handles 39.5% of all U.S.-Mexico freight value. $344.6 billion moved through a single corridor in 2025, more than the next five ports combined. El Paso runs second at $139.3 billion, combining its two crossings (Ysleta and BOTA).

I work in Laredo. I run a research center here. And the concentration of trade through this one point is both an enormous asset and an enormous vulnerability. One bridge closure, one policy change, one infrastructure failure, and a significant share of North American commerce stops moving.

The port data reveals something else: this is overwhelmingly a truck story. Trucks carry 74% of all U.S.-Mexico freight value ($642 billion). Rail handles 10.9%. The physical infrastructure of U.S.-Mexico trade is four lanes of highway connecting two cities that most Americans can’t find on a map.

What the crossover actually means

The media framed Mexico’s rise as a China story: tariffs pushed trade away from Beijing and toward the Western Hemisphere. That’s partially right. U.S.-China trade dropped from $690 billion in 2022 to $415 billion in 2025, a 40% decline driven by tariffs, the IEEPA Supreme Court ruling, and Section 122 surcharges.

But Mexico’s growth isn’t just China’s loss redistributed. U.S.-Mexico trade grew from $100 billion in 1994 to $873 billion in 2025, an 8.7x increase over 30 years. That growth preceded the China tariff war. It preceded nearshoring. It preceded USMCA. The trade relationship was already enormous. The tariff environment accelerated a shift that geography and integration had been building for decades.

The real question isn’t whether Mexico will stay #1. It will. The question is whether the institutions, infrastructure, and policy frameworks can keep up with a trade relationship that’s grown 8x in 30 years while the governance architecture hasn’t fundamentally changed since NAFTA.

That’s the question I’ll be working on here, at The Bridge.

About this publication

I started The Bridge to do analysis like this: data you can verify, charts you can explore, and an honest interpretation from someone who works in the corridor the data describes.

Every couple of weeks, I’ll publish a data brief with live numbers from the U.S.-Mexico border. Once a month, I’ll go deeper on a policy or technology question that matters.

The interactive versions of these charts are at labs.drdanielcovarrubias.com.

She didn’t lose her job. She got more of it back.

That interaction, a human and a machine splitting the work along the line where judgment begins, represents one way AI is reshaping work. It’s not the only version. But according to the largest dataset we have on actual AI use, it’s the most common one.

What the Data Actually Says

Anthropic, the AI company behind Claude, released its fourth Economic Index report this month, built from analysis of 2 million real conversations. The headlines were predictable. One outlet warned of a possible ‘Great Recession for white-collar workers.’ Others ran the job exposure numbers. The usual cycle followed.

The finding that mattered most was quieter.

Of all the ways people use AI at work, 52% involve augmentation: humans working alongside the tool, making decisions together, with the person retaining final judgment. Automation, where the AI handles a task entirely, accounts for 45%. The remaining 3% is somewhere in between.

Augmentation is winning, at least for now. The most common use of AI isn’t replacing people. It’s a person and a machine collaborating on something neither could do as well alone.

The Gap That Matters

The report maps two things across 800 occupations. First, what AI could theoretically do. Second, what it’s actually doing. The distance between those two numbers is significant.

In the fields of computer science and mathematics, AI could theoretically handle 94% of tasks. It’s currently covering 33%. In business and finance, the theoretical number is also 94%. Actual adoption sits around 20%. Office and administrative roles: 90% theoretical, 25% real.

The gap is narrowing, but slowly. In January 2025, 36% of occupations had AI touching at least a quarter of their tasks. By November, that number reached 49%. That’s real movement. But when you measure what AI is actually completing successfully, the spread is still wide.

Researchers call this the adoption gap. I think about it differently. That gap is a strategy question.

Every organization sitting on a 60- to 70-percentage-point spread between what’s possible and what they’re doing isn’t facing a technology problem. The tools exist. The question is whether the organization has a plan for using them.

What This Looks Like in Practice

The customs broker example isn’t hypothetical. Variations of that scene play out across industries every day, in the organizations that have figured this out.

A regional healthcare system managing patient scheduling across 12 clinics. Staff spend hours on phone calls, insurance verification, and rescheduling. AI handles the scheduling logic, the verification steps, and the follow-up reminders. The staff redirects their time to patients who need human attention: the elderly person confused about medication changes, the new mother with questions that don’t fit a dropdown menu.

A mid-size accounting firm during tax season. Junior associates spend weeks pulling data from client documents, cross-referencing figures, and populating forms. AI handles extraction and cross-referencing. The associates review output, catch edge cases, and shift to advisory work: helping clients understand what the numbers mean and what decisions to make next. The firm bills more advisory hours. The associates develop judgment faster because they’re practicing it rather than copying data between spreadsheets.

A small law firm preparing for a commercial litigation case. Two paralegals spend a week reviewing 4,000 documents for relevance, flagging key clauses, and organizing exhibits. AI handles the initial review and sorting. The paralegals finish in two days, then spend the rest of the week on the work the attorneys actually need: building the timeline of events, identifying contradictions across depositions, and preparing the trial binder with the judgment calls that win cases.

In each case, the team doesn’t shrink. What the team does changes.

The Complexity Surprise

One finding in the report challenged my own assumptions.

AI doesn’t just speed up simple tasks. It speeds up complex ones faster.

Tasks requiring a college-level education were completed 12 times faster with AI assistance. Tasks requiring a high school education were completed 9 times faster. The more complex the work, the greater the time savings.

The popular version of the AI story assumes machines handle the simple stuff while humans handle the hard stuff. The data says something more interesting: AI’s biggest productivity impact is on the work that requires the most training and expertise to do manually.

There’s a deeper point here that goes beyond productivity metrics. For decades, organizations have treated expertise as scarce and rationed it accordingly. The senior analyst builds one financial model because that’s all they have time for. The logistics coordinator maps one set of routing scenarios because the alternatives take too long to evaluate. AI doesn’t replace the expertise. It removes the bottleneck that kept expertise from being fully deployed. The analyst builds five models and spends their remaining time interpreting them. The coordinator evaluates 30 routing options in the time it used to take to evaluate three.

The constraint on expert impact was never knowledge. It was time. AI compresses the time. The expert, finally, operates at full capacity.

The Deskilling Risk, and How to Think About It

The report flags a risk worth taking seriously. If AI handles the complex components of a job, the remaining work shifts toward lower-skill tasks. The researchers use travel agents as an example: if AI handles complex itinerary planning, the human role shrinks to routine ticket purchasing and payment processing. The job gets simpler. The person becomes less valuable.

That’s a real concern. But the deskilling scenario assumes a passive response, one where AI reshapes job content by default rather than by design.

The alternative is deliberate. If AI compresses the preparation phase of expert work, redirect that time toward the work that benefits most from a human in the room: building relationships, reading clients, and understanding context that doesn’t fit in a dataset.

The customs broker who no longer spends six hours on paperwork visits clients, maps their supply chains, and anticipates regulatory shifts before they arrive.

Deskilling is the default outcome. Reskilling is the intentional one.

The Productivity Signal

The Anthropic report estimates that widespread AI adoption could increase U.S. labor productivity growth by 1 to 1.8 percentage points per year over the next decade. Even the conservative end of that range would return productivity growth to rates the country hasn’t seen since the late 1990s.

The U.S. has averaged about 1.4% annual productivity growth over the past two decades. Adding a full percentage point would be comparable to the surge that accompanied the personal computer and internet era.

But here’s the qualifier the researchers are careful to include: those gains depend on adoption. The technology has to be used. Right now, AI use remains concentrated in a narrow band of occupations and tasks. The top 10 tasks account for 24% of all AI interactions, up from 21% at the start of 2025. Usage is getting more concentrated, not less.

The organizations capturing productivity gains are those that have moved beyond experimentation and built AI into their workflows. The rest are running pilots, attending webinars, and waiting.

What the Gap Costs

In business and finance occupations, the gap between theoretical AI capability (94%) and actual adoption (20%) represents thousands of hours per year per organization spent on tasks that a machine could handle in minutes.

Those hours show up as slower response times, delayed reporting, and missed patterns in data. But the real cost is subtler. It’s talented people spending their days on work that doesn’t require their talent. It’s an organization that hired smart people and then buried them in process.

And yes, some organizations will look at that gap and see a headcount decision. That’s the blunt version of closing it. The data doesn’t pretend otherwise. But the organizations getting the most from AI aren’t the ones cutting teams. They’re the ones redeploying them.

The organizations that close the gap first won’t do it by buying more software. They’ll do it by rethinking how work gets done: which tasks belong to humans, which belong to machines, and how the two interact at every step.

The Real Question

Every technology transition eventually comes down to the same fork in the road. One path is reactive: wait for the technology to reshape your organization and deal with the consequences. The other is intentional: decide what you want your associates to do, and start building toward it.

The Anthropic data gives us something rare in the AI conversation: real evidence, drawn from millions of actual interactions, about how this technology is being used right now. The picture is complicated, but clearer than the headlines suggest. 52% augmentation. Complexity is being compressed, not eliminated. And a gap that tells us more about organizational readiness than about the technology itself.

But those findings describe the organizations that are paying attention. The gap between what’s possible and what’s happening is still enormous, even as it slowly narrows. Closing it isn’t a technology project. It’s an act of institutional imagination: looking at every workflow, every role, every process, and asking whether your people are spending their time on the work that actually requires them.

The tools are ready. The organizations that move first will be the ones that treat this as a design problem, not a technology problem.

Today, the market is projected to reach 61 million square feet by the end of 2026, spread across 59 active industrial parks, with an 11.3-million-square-foot construction pipeline representing a 247% increase from mid-2024. Trade through the Port of Laredo reached $339 billion in 2024 and is projected to exceed $350 billion in 2025. And the wave of industrial construction that began in 2020 has added significant new assessed value to the city’s tax rolls, with revenue implications that remain to be fully quantified.

What happened between 2018 and now is not a story about nearshoring, though nearshoring accelerated it. It is a story about what happens when market conditions, trade momentum, and private capital align simultaneously in the same place.

The Catalyst: When Conditions Aligned

For most of its modern history, Laredo’s industrial market has developed at a pace that has matched demand. Local developers built for local users: customs brokers, regional distributors, transportation companies, in facilities sized for the operations of the day. Large landholders maintained significant reserves. The market worked, and it worked well, but it worked at a scale calibrated to a different era of cross-border trade.

Three things shifted almost simultaneously. First, Mexico overtook China as the top exporter to the United States, and Laredo’s truck crossings began climbing toward 20,000 per day. Second, nearshoring discussions shifted from conference panels to corporate boardrooms, and site selectors began examining border markets with renewed urgency. Third, developers responded. Between 2018 and 2020, new industrial land came online across North Laredo and the Colombia corridor at a pace and scale the city had never seen, with full urbanization including paved roads, utilities, and drainage designed to meet institutional standards. Single-phase plats exceeded 600 acres.

The result was immediate. Many lots projected to sell over three years moved in eight months. National capital that had not previously considered Laredo began deploying. The absorption velocity surprised the market.

The Numbers Behind the Transformation

Consider the scale of what has been built. The Mile Marker 13 corridor alone, the industrial submarket stretching along FM 1472 in North Laredo, now accounts for nearly 22 million square feet across four major industrial parks, roughly 36 percent of Laredo’s total market. When you add the Colombia crossing submarket, the footprint grows further. By industry estimates, total investment across Laredo’s new industrial corridors, including land, infrastructure, and vertical construction, comfortably exceeds $1 billion, with significant capacity remaining in available lots.

The Mile Marker 13 corridor is the most concentrated example, but the transformation extends across the city. Of Laredo’s projected 61 million square feet, nearly two-thirds sit outside the North Laredo submarket, distributed across dozens of industrial parks serving the World Trade Bridge, the Colombia-Solidarity Bridge, and the I-35 corridor.

Marketwide, the signal from national capital is clear. Industrial investment firms managing billions in assets have broken ground on Class A facilities in North Laredo. One national developer alone committed $100 million to a three-building, 933,000-square-foot complex near the World Trade Bridge. Institutional joint ventures backed by global real estate platforms have launched multi-phase logistics parks. Private family office capital has acquired fully leased portfolios, citing nearshoring as a structural long-term play.

These investments signal that the market has graduated. Laredo is no longer a regional play. It is now attracting the same institutional capital that flows into the nation’s largest logistics markets, and this validation has implications for all stakeholders in the corridor.

The Honest Market Picture

Any serious analysis of Laredo’s industrial market must address the supply question. After years of record construction, vacancy has risen to roughly 11.5% as new large-format product enters the market, a natural normalization from the sub-2% rates that characterized the 2021–2023 boom. Millions of square feet of additional space remain under construction or in the permitting pipeline.

Much of the new inventory consists of buildings exceeding 300,000 or 400,000 square feet, designed for national and international tenants with large-scale logistics operations. The average local user operates at a fraction of that scale. Filling these buildings depends on attracting the same kind of institutional tenants whose investment decisions validated the market in the first place.

Every growth market normalizes. Laredo is experiencing the same adjustment cycle as peer logistics markets across the Sun Belt, and its fundamentals during this phase are stronger than at any point in its history. The conditions that drove the initial investment remain intact: the port now averages approximately 20,000 commercial truck crossings daily, trade volumes continue to grow, and Laredo sits at the convergence point for freight moving from Mexico’s major industrial centers, from Monterrey and Saltillo to the Bajio, Mexico City, and as far south as the Pacific port of Lázaro Cárdenas. That geographic reality is why 40 percent of all U.S.-Mexico trade flows through this single corridor.

Real estate markets are cyclical. Laredo has absorbed rapid growth before. The current adjustment period, while real, is happening in a market whose trade volumes and strategic positioning have never been stronger.

What Growth Built

Beyond the supply cycle, something more consequential is taking shape: Laredo’s industrial property tax base is undergoing a structural expansion, and the revenue implications are only beginning to come into focus.

Between 2013 and 2022, the City of Laredo’s property tax levy increased by 47%, from $67.6 million to $99.4 million, driven largely by residential growth. By 2025, the levy had reached over $137 million. That acceleration coincides directly with the industrial construction wave that began in 2020, which has added significant new assessed value, developed land, institutional-grade buildings, and infrastructure improvements, now flowing into tax receipts. The fiscal impact is substantial.

The city’s infrastructure needs are well documented, from roads to water systems to public safety facilities. The strategic opportunity is to build on existing planning efforts by more explicitly linking future infrastructure investment to the industrial growth reshaping the tax base, ensuring that new revenue reinforces the competitive position that generated it in the first place.

Other border communities offer instructive examples. Santa Teresa, New Mexico, is executing a coordinated public-private strategy that pairs state investment in highway and port-of-entry infrastructure with large-scale private logistics and residential development, including a $2 billion mixed-use project adjacent to the port of entry. Santa Teresa’s industrial footprint is a fraction of Laredo’s, but its coordinated approach illustrates how aligning public infrastructure with private investment can accelerate development in emerging logistics markets.

Laredo has the scale, the trade volumes, and the private capital. The opportunity is to match those assets with a coordinated public strategy that turns a construction boom into a durable competitive advantage.

What Comes Next

Laredo’s industrial transformation was driven by private capital. Local developers, regional landholders, and national institutional investors all played a role in building what exists today. That foundation is now in place: a market approaching 61 million square feet of logistics space, billions in investment, and a construction pipeline that signals sustained confidence in the corridor.
The question is what gets built on top of it.

In an earlier analysis, we introduced the concept of the First-Touch Advantage, the economic and logistical case for co-locating distribution operations at the point where goods first enter U.S. commerce, rather than hauling them 250 miles north to sort and redistribute. The infrastructure that private capital has built over the past five years makes that vision operationally possible in a way it wasn’t before. Laredo now has the building stock, the lot sizes, and the institutional tenant base to support a transition from gateway to gateway-plus-distribution hub.

That transition will not occur by default. It requires coordinated investment in workforce development, multimodal connectivity, and the permitting and regulatory environment that institutional tenants evaluate before committing.

The city that processed more than $350 billion in trade last year handles nearly 40 percent of all U.S.-Mexico trade. By container-equivalent throughput, it rivals the largest maritime gateways in the world. It sits at the northern end of Mexico’s industrial spine and anchors the most important commercial land corridor in North American trade. Laredo has both the assets and the opportunity to define this next chapter deliberately.

The private sector built the foundation. The opportunity now belongs to the entire community.

The plan gives both countries 60 days to develop coordinated trade policies, price floors, and joint investment frameworks for the minerals that power AI chips, electric vehicle batteries, semiconductor fabrication equipment, and advanced defense systems. On its face, it appears to be a mining agreement. It is not. It is a technology sovereignty play wrapped inside a trade negotiation. And how Mexico responds over the next two months will determine whether North America builds the physical foundation for its next technological era, or whether this becomes another nearshoring moment: enormous in promise, underwhelming in execution.

What the Action Plan Actually Says

The official text is one page. Its scope is not. The U.S. and Mexico will identify specific critical minerals of mutual interest, explore border-adjusted price floors for imports, and consult on embedding those mechanisms into a binding plurilateral agreement open to other allied nations. The plan extends to regulatory standards for mining and processing, geological mapping coordination between the U.S. Geological Survey and the Mexican Geological Service, coordinated stockpiling, investment screening, and joint rapid-response protocols for supply-chain disruptions.

Both governments framed this explicitly in the context of the USMCA Joint Review, due by July 1. The language from both sides treats critical minerals not as a side conversation but as a structural pillar of the North American trade relationship going forward, with Ebrard positioning mineral security as inseparable from industrial competitiveness and national strategy.

The timing reinforces that reading. Two days earlier, President Trump signed an executive order establishing Project Vault, a $12 billion critical-mineral stockpile structured as a public-private partnership, with initial participation from Boeing, GE Vernova, Western Digital, and Clarios. China’s restriction of rare-earth exports during the 2025 trade war demonstrated the very vulnerability this architecture is designed to address. As the USMCA review is underway, critical minerals could become part of the trilateral trade framework itself rather than an ancillary arrangement.

Mexico’s Mineral Endowment: What Exists and What Matters

A technology sovereignty strategy requires actual geology. Mexico has it, though the picture is more complex than headlines suggest.

Mexico ranks among the top ten countries globally in copper reserves, commonly placed around 7th, with approximately 53 million metric tons. Copper is used in nearly every clean energy technology, including solar panels, wind turbines, electric vehicle motors, battery storage systems, and the entire electricity grid infrastructure that connects them. Chile, Peru, and Mexico together account for roughly one-third of global copper production. Mexico is also the world’s largest silver producer, and silver appears on the USGS 2025 Critical Minerals List for the first time, recognized for its essential role in electronics, solar photovoltaic cells, and advanced conductors.

Beyond copper and silver, Mexico ranks fifth globally in molybdenum production, a mineral critical for high-strength steel alloys used in aerospace, defense, and energy infrastructure. It holds significant graphite resources and production capacity, and graphite is the dominant material in lithium-ion battery anodes. Mexico has an estimated 1.7 million tonnes of lithium resources spread across 82 known deposits in 18 states, with the largest concentrations in Sonora, Puebla, and Oaxaca. PEMEX has reported lithium concentrations in oilfield brines across several states and is evaluating extraction technologies, though commercial viability remains at an early stage.

The U.S. updated its Critical Minerals List in November 2025. It now includes 60 minerals. Mexico produces or has significant reserves of at least a dozen of them, including copper, silver, lithium, graphite, molybdenum, zinc, manganese, fluorspar, bismuth, antimony, barite, and lead. This is not a single-commodity story. Mexico’s geological endowment spans multiple critical mineral categories, making it a diversified partner rather than a single-resource dependency.

The Technology Stack Nobody Is Talking About

Most of the commentary on this deal has focused on geopolitics: countering China, securing defense supply chains, and building Western solidarity. That analysis is accurate but incomplete. Critical minerals are the physical substrate of every exponential technology curve currently reshaping the global economy, and treating them as strategic commodities alone misses the larger picture.

Consider the dependency chain. Semiconductors depend on silicon and increasingly on gallium and germanium, with rare earth elements critical in the manufacturing equipment ecosystem. AI infrastructure runs on data centers where servers consume large quantities of copper, use silver in circuit boards and connectors, and embed rare earths in storage and cooling systems. Leading quantum computing architectures rely on niobium and tantalum for superconducting circuits. Advanced robotics and autonomous systems use neodymium and dysprosium permanent magnets across high-performance motors. 5G and 6G telecommunications infrastructure depends on copper for cabling and power distribution, gallium compounds for RF amplifiers, and lithium for backup energy storage across the network stack. Space technology and advanced defense systems rely on titanium for airframes, cobalt for superalloys in turbine engines, and platinum-group metals in catalysts and sensors. Batteries across devices, vehicles, and grid storage systems use varying combinations of lithium, cobalt, nickel, graphite, and manganese, depending on the chemistry.

The country or region that controls critical mineral supply chains determines who participates in the next generation of technological development. This is what makes the U.S.-Mexico action plan fundamentally different from a trade deal about rocks. It is an infrastructure play for the AI, quantum, and clean energy eras simultaneously.

From a USMCA 2.0 perspective, critical minerals constitute a sector in which the trilateral framework can generate a genuine competitive advantage. The agreement already covers automotive rules of origin, digital trade provisions, and labor standards. Adding a critical minerals chapter, with shared processing standards, coordinated investment incentives, and joint stockpiling protocols, would create a North American technology supply chain that no other regional trading bloc can match. The raw materials are here. The manufacturing base is here. The question is whether the institutional architecture will connect them.

The Sovereignty Question: Mexico’s Crossroads

Here is where the opportunity meets friction. Mexico’s mining regulatory environment has been in flux since 2022, when the Lopez Obrador administration nationalized lithium, making exploration, exploitation, and use of the mineral exclusive to the Mexican state. The 2023 Mining Law reform introduced shorter concession terms, stricter environmental and water requirements, community benefit-sharing obligations, and new grounds for concession cancellation, which the mining industry has described as generating legal and regulatory uncertainty.

Under President Sheinbaum, the signals have been mixed but generally more encouraging. The administration reconsidered the proposed ban on open-pit mining after recognizing it would paralyze sand extraction for construction and lithium mining for the energy transition. Economy Minister Ebrard has consistently framed mineral access as a national security priority, and senior officials have signaled the government’s intent to resume large-scale exploration in 2026. The Mexican Geological Service received a modest budget increase for lithium-specific work, and PEMEX is evaluating lithium extraction pathways from oilfield brines in several states.

But the structural challenges remain real. The regulations needed to implement the 2023 Mining Law reform have not been published. For lithium and other strategic minerals, exploration authority has been centralized within state entities, with the Mexican Geological Service given a leading role; however, it lacks the resources to match the scale of private-sector investment it displaced. Mining concessions have been frozen since 2018, and Sheinbaum has confirmed the moratorium will continue. Exploration investment fell from $543 million in 2023 to an estimated $400 million in 2025, which CAMIMEX warns would represent a decade low. LitioMX, the state entity created to develop lithium, operates on a budget in the low six figures in U.S. dollar terms, with only a handful of permanent staff, enough to cover salaries and administration but not serious exploration.

This is the tension at the heart of the deal. The Action Plan calls for investment promotion, screening, and identification of projects that comply with internationally recognized responsible business conduct standards. That language assumes a regulatory environment that attracts private capital at scale. Mexico’s current framework, despite its promising rhetorical direction, has not yet delivered that environment.

The Nearshoring Parallel

For those of us who have tracked North American trade integration for years, this moment carries a familiar feeling. The nearshoring narrative, which accelerated after 2020, promised a fundamental restructuring of global supply chains toward Mexico. There was real movement: FDI reached record levels, with annual inflows climbing to nearly $41 billion by 2025, and goods exports grew at high-single-digit rates annually. But a persistent gap remained between announced nearshoring projects and the pace and sectoral composition of executed investment. The narrative preceded institutional delivery.

Critical minerals could follow the same trajectory. The geological endowment exists. The geopolitical incentive is overwhelming. The trade architecture, through USMCA, provides a ready-made framework for integration. But none of that matters if the regulatory environment discourages the private investment needed to build mines, processing facilities, and refining capacity. S&P Global data indicate that mines that began production between 2020 and 2023 averaged nearly 18 years from discovery to first production. Mexico does not have 18 years. The 60-day action plan window, while obviously not meant to resolve these structural issues, signals the urgency of the moment.

Four Requirements for Success

The data points are in different directions, and it would be irresponsible to predict outcomes. But the conditions for success are identifiable.

First, the USMCA Joint Review should incorporate critical minerals into the trilateral framework, not as an appendix but as a substantive chapter, with rules of origin, processing standards, and investment facilitation mechanisms comparable to those for automotive and digital trade. This would give the critical minerals sector the same institutional weight that transformed North American automotive manufacturing over three decades.

Second, Mexico would need to publish the implementing regulations for the 2023 Mining Law reform and establish a concession framework that balances environmental and community protections with the investment certainty required by mining capital. Ebrard’s rhetoric has been right. The regulatory architecture needs to match it.

Third, the focus should extend beyond extraction. The real value in critical minerals is not in digging them out of the ground. It is in processing, refining, and integrating them into manufactured goods. China dominates not because it has the most minerals, but because it built the processing infrastructure that converts raw material into technological capability. North America’s critical minerals strategy will succeed or fail based on whether it builds that processing capacity on this continent.

Fourth, the technological research component of the Action Plan deserves serious attention. Joint R&D between U.S. and Mexican institutions on extraction technologies, recycling processes, and alternative materials could accelerate timelines and reduce environmental impacts simultaneously. This is where universities, national laboratories, and the private sector intersect, and where frameworks such as NADICI (the North American Digital Integration and Competitiveness Initiative) provide the connective tissue between cross-border research collaboration and industrial application.

The Stakes

China mines 60% of the world’s rare earths and processes 90%. During the 2025 trade war, Beijing demonstrated its willingness to weaponize that dominance. The October 2025 truce included a one-year reprieve on export curbs, but restrictions remain tighter than before. Every month that passes without alternative supply chains is another month of strategic vulnerability.

North America has the geology, the trade architecture, and now, as of February 4, the bilateral framework to build something different. Mexico sits at the center of that possibility. Not as a commodity exporter shipping raw materials north, but as an integrated partner in a technology supply chain that stretches from mine to semiconductor fab to data center.

Whether that possibility becomes reality depends on what happens in the next 60 days and in the years that follow. The minerals are in the ground. The question is whether the institutions, regulations, and investment frameworks will rise to meet the challenge. North America has been here before. The outcome is not predetermined. But the clock started on February 4.

How does AI change work? What decisions require human judgment? What processes benefit from machine speed? When does autonomy create value, and when does it create vulnerability? These are the strategic questions that will determine North American competitiveness for the next decade.

2025 has seen me traveling to give keynotes and participate in panels, all the while discussing cross-border trade, digital transformation, and economic integration. What struck me across these conversations, whether in Mexico City, San Antonio, or policy roundtables in Washington, was how the threads kept interweaving. Trade policy. AI governance. Workforce transformation. Energy infrastructure. Continental security. These stopped being separate conversations this year. They became one conversation with different entry points.

As 2025 draws to a close, I want to share what this convergence has taught me and what I believe it will demand of us in 2026.

The Year Uncertainty Became the Product

If I had to characterize 2025 in a single word, it would be uncertainty. Tariff volatility became the defining economic feature of the year. Rules changed quarterly. Exemptions appeared and disappeared. Industries whipsawed between relief and panic depending on which announcement landed that week.

The economic damage isn’t just from tariffs themselves. It’s from the inability to plan. I called this the Precarity Premium: the hidden tax that businesses pay when they cannot forecast what regulatory environment they’ll face six months from now. Investment decisions get delayed. Expansion plans get shelved. Capital is tempted to flow to regions with more predictable policy environments. The uncertainty itself extracts economic cost, separate from any specific tariff rate.

My conversations with business leaders shifted noticeably this year. The question used to be “How do we grow?” Now it’s “How do we survive the unpredictability?” That’s not a sustainable posture for the most integrated manufacturing region on the planet. Companies cannot build supply chains, train workforces, or commit capital when the ground keeps shifting beneath them.

This cannot continue. 2026 will force a resolution one way or another, whether through the USMCA review, market pressure, or competitive losses to regions that offer stability. The Precarity Premium will eventually be paid, either by North American businesses adapting to permanent uncertainty or by policymakers recognizing that predictability itself has economic value.

Shared Security, Shared Prosperity: No Going Back

The framework that crystallized for me this year is simple to state but profound in implications: security and prosperity can no longer be treated as separate policy tracks. For decades, trade negotiators worked in one building while defense and security officials worked in another. That separation made sense when economic interdependence and national security occupied different spheres. It makes no sense today.

Consider what 2025 revealed. Tariffs justified on security grounds caused damage to the prosperity of integrated supply chains. Supply chain security, meanwhile, requires the very production integration that protectionist policies threaten. You cannot secure what you’ve dismantled. The automotive sector alone demonstrates this: parts can cross the border as many as 7 or 8 times before final assembly, meaning security and prosperity are literally welded together in every vehicle rolling off North American assembly lines.

The AI developments of November brought this convergence into sharp relief. In the same month, an AI-generated country track hit No. 1 on a Billboard country sales chart. At the same time, Anthropic, creator of Claude, disclosed disruption of what it described as the first reported AI-orchestrated cyber-espionage campaign. Creation and destruction. Productivity and vulnerability. Same underlying technology, radically different outcomes. The difference wasn’t the AI. It was the human governance framework surrounding it.

This is Shared Security, Shared Prosperity in action. The technology that drives economic growth is also the one that threatens critical infrastructure. The supply chains that create jobs are the same supply chains that require protection. The energy grid that powers AI development is the same grid that adversaries target. North America either coordinates these dimensions together or fragments and loses to competitors who integrate them.

This past week, Mexico’s Congress approved tariff increases of up to 50% on more than 1,400 products, largely affecting imports from countries without free-trade agreements, with China expected to be among the most impacted, effective January 1, 2026. The policy is being read as both industrial defense and strategic positioning ahead of the 2026 USMCA review: tightening exposure to non-regional content while reinforcing the logic of North American rules of origin. It’s a real-time example of the argument in this essay: trade policy is now inseparable from security, competitiveness, and the credibility of regional integration. Security and prosperity, moving together.

The AI Imperative: Why the Continent Must Move Together

I’ve been developing a framework I’m calling the North American Digital Infrastructure Coordination Initiative, or NADICI, which I’ll publish in full before the year ends. The core insight driving this work is straightforward: the United States can win the AI race, but it cannot afford to leave Mexico and Canada behind. This isn’t charity. It’s a strategic necessity. This work builds on the USMCA 2.0 framework I’ve proposed for modernizing North America’s trade architecture.

AI requires energy. Massive amounts of it. North America possesses an integrated energy grid with diverse generation capacity: Canadian hydroelectric, American natural gas, and Mexican solar potential. No single country has all the pieces. Together, the continent has abundant clean power to fuel AI infrastructure that would make competitors envious.

AI requires semiconductors, which require rare earths and critical minerals. Canada has deposits. Mexico has processing potential and manufacturing capacity. The United States has design expertise and fabrication facilities. The EV battery supply chain already demonstrates this trilateral complementarity: Canadian nickel and cobalt, American cell manufacturing and R&D, Mexican pack assembly and vehicle integration. AI chip supply chains will follow similar patterns.

AI requires talent. Not just engineers who build systems, but professionals across every sector who can work alongside AI effectively. I’ve written about the evolution from AI-aware to AI-fluent professionals: people who use these tools reflexively in daily work, who understand both when AI accelerates outcomes and when human judgment must remain in control. The continent that builds AI-fluent workforces across industries wins. The continent that hoards talent in one country while neighbors fall behind loses collective capacity.

AI transforms logistics. This is Logistechs in action: the convergence of logistics operations with exponential technologies. Autonomous vehicles. Predictive customs processing. AI-optimized routing. Smart border infrastructure. These capabilities are emerging now, but they require coordinated standards, shared data frameworks, and harmonized regulations. Fragmented national approaches mean fragmented benefits. Continental coordination means continental advantage.

The alternative is three separate national AI strategies competing against China’s unified approach. Beijing isn’t coordinating with neighbors out of goodwill. They’re doing it because scale wins in AI development. Larger markets for data. Larger pools of talent. Larger energy capacity. Larger manufacturing bases for hardware. China understands that national AI strategies compete against continental ones, and continental ones win. North America has the assets to compete at that scale. The question is whether it organizes to use them.

What I’m Watching in 2026

The USMCA joint review will be the defining North American trade event of 2026. With public consultations and testimony having just passed, the mandatory six-year review triggers a fundamental question: does the trilateral framework that has governed North American commerce since 2020 continue, evolve, or fragment? Our research across North America identified 9.9 million jobs exposed to trade disruptions. Some voices are calling for bilateral deals to replace the trilateral structure, arguing that one-on-one negotiations give the United States more leverage. I have strong views on this question that I’ll share in detail in the new year.

For now, I’ll simply note that what’s at stake is whether thirty years of integration deepens into genuine continental competitiveness or fragments into competing national interests that benefit no one except our global rivals. Thenearly $56 billionMexico–Canadatwo-way merchandise trade in 2024 alone demonstrates why trilateral thinking matters: bilateral U.S. deals would orphan supply chains that currently flow seamlessly across all three borders.

There’s an irony worth noting: 2026 is also the year the United States, Mexico, and Canada co-host the FIFA World Cup, the first men’s World Cup hosted by three nations in history. These three countries will need to coordinate security for millions of visitors, harmonize visa procedures, synchronize transportation across dozens of venues, and present a unified face to the world. If North America can organize that, but can’t figure out how to extend a trade agreement that’s been functioning for thirty years, something has gone badly wrong with its priorities.

AI governance in trade frameworks deserves close attention. USMCA’s digital trade provisions were forward-looking when negotiated, but predate the explosion of generative and agentic AI. How do rules of origin work when AI designs products across borders? How do labor provisions apply when AI performs tasks previously done by workers? How do we prevent AI-enabled circumvention of trade rules while encouraging AI-enabled productivity gains? These questions don’t have obvious answers, but 2026 will force us to start developing them.

Coordination on critical minerals and energy will accelerate. The EV transition is the test case, but the implications extend far beyond vehicles. AI data centers consume enormous power. Semiconductor fabrication requires reliable clean energy. The countries that coordinate energy infrastructure development will lead; those that don’t will depend on those who do. North America has natural advantages here. The question is whether policy enables those advantages or squanders them.

Logistics transformation will continue to reshape cross-border operations. The border itself is becoming a laboratory for AI-human collaboration. Customs brokers, freight forwarders, trade compliance specialists: these professionals are either becoming AI-fluent or falling behind. The firms that figure out human-AI collaboration in logistics will set patterns that spread across every sector. I’ll be watching closely to see which models emerge and which scale.

The Stakes

What I tell my students: you’re not preparing for a future where AI exists. You’re preparing for a future where AI fluency is table stakes. The jobs that remain purely human will be fewer than you expect. The jobs that combine human judgment with AI capability will be more numerous and more valuable than current career guides suggest. Build the skills to thrive in that collaboration, not to compete against it. Learn to ask both questions: “Why not use AI for this?” when seeking productivity, and “Why should AI have autonomy here?” when security matters. Those who master both questions will lead. Those who ask only one will struggle.

What I tell policymakers: you’re not choosing between security and prosperity. That framing is obsolete. You’re choosing whether North America architects both together or lets competitors do it first. The Shared Security, Shared Prosperity framework isn’t a slogan or an aspiration. It’s the operating system the continent needs to compete in an era where economic strength and national security are inextricably linked. Policies that sacrifice one for the other will fail on both dimensions. The November paradox of AI creation and AI destruction should have made this clear to anyone paying attention.

What I tell business leaders: the uncertainty of 2025 cannot become the new normal. Investment requires predictability. Growth requires stability. Workforce development requires knowing what skills will matter in five years. If 2026 doesn’t deliver clarity on trade frameworks, energy policy, and AI governance, capital will flow to regions that offer what North America won’t. That’s not a threat or a prediction. It’s already happening. I’ve watched it happen in conversations across the continent this year. Leaders want to invest here, build here, and hire here. But they need to know the rules won’t change before their facilities open or their training programs are complete.

The year taught me that these threads are inseparable. Trade policy shapes AI deployment. AI capability shapes workforce requirements. Workforce readiness shapes continental competitiveness. Energy infrastructure enables all of the above. Security concerns constrain and channel all of the above. Pull one thread and the whole fabric moves. Policymakers who treat these as separate domains will produce incoherent results. Leaders who recognize the convergence will shape what comes next.

2026 will test whether North American leadership recognizes this convergence. The USMCA review will reveal whether we can think trilaterally about trade. AI governance debates will reveal whether we can coordinate digital infrastructure. Energy transitions will reveal whether we can build continental capacity. Each of these tests arrives in the same year, and each will be shaped by how we approach the others.

I’ll continue working at this intersection: developing policy frameworks, teaching the next generation of professionals, and contributing to public conversations that shape how we move forward. The stakes are too high for anyone who cares about this continent’s future to sit on the sidelines.

The paths are crossing. The question is whether North America crosses them together.