Juárez–El Paso, the USMCA review window, and the new binding constraint on North American integration
On a typical weekday morning, a truck loaded with AI-server components pulls onto the Ysleta-Zaragoza bridge and waits to cross from Ciudad Juárez into El Paso. The cargo has crossed the river before, as parts and subassemblies, and it will cross again. What determines its value at the line is no longer the tariff it owes. Under the USMCA the duty is mostly zero. What determines its value is whether its origin, its labor content, and the provenance of its suppliers can be proven to the satisfaction of two customs authorities and, increasingly, the corporate buyer at the other end. The binding constraint on North American integration has moved. It used to be the price of crossing. It is now the cost of proof.
That shift is one of the most consequential things happening on the U.S.-Mexico border, and it is almost entirely missing from a trade-policy conversation that still argues about tariffs, rules of origin, and geopolitics as if they were national questions settled in national capitals. They are not. Proof is produced where production happens: on factory floors, inside supplier relationships, at ports of entry, in the workforce that crosses every morning. For Juárez and El Paso, that place is a single metropolitan economy of roughly 2.4 million people, bound by three ports of entry, about 1.2 million truck movements a year, and between 15,000 and 30,000 workers who cross daily to their jobs. The El Paso customs district handled $197.3 billion in two-way trade in 2025. By any economic measure, Juárez and El Paso are one city. By any governmental measure, they are two. And the thing the USMCA now demands of them, verification, is precisely the thing no institution in the corridor is built to provide.
From tariffs to proof
For three decades, the organizing problem of North American trade was liberalization. NAFTA and its successor lowered tariffs toward zero, and the binding constraint on a cross-border supply chain was a price: the duty owed at the line, set in Washington, Mexico City, and Ottawa, uniform across every corridor. Capital cities set it; the border collected it.
The USMCA changed the problem. Tariffs were largely gone; what the agreement polices now is content. To claim preferential treatment, a producer must prove where things were made and by whom. Automobiles face the strictest regime, a regional value content requirement of 75 percent for finished vehicles, with tiered thresholds for parts and a labor value content rule that requires evidence of wages and hours. On July 1, 2026, the United States, Mexico, and Canada open the agreement’s joint review under Article 34.7. The review is not a renegotiation. It is the moment each party gets to challenge what the others have been certifying, and the cross-checks on Chinese inputs, transshipment, and labor compliance will tighten no matter who is asking.
The constraint that results is not a price. It is an evidentiary cost: the cost of proving origin, content, and provenance, of assembling and defending the documentation that converts a shipment into a qualifying one. Economists would call it a transaction cost, or a compliance cost. Whatever the label, it has a feature the tariff never had. It cannot be paid in the capital. It has to be produced in the corridor.
This is the part national trade debates miss. Origin and labor content cannot be certified from a capital, because the facts that have to be proven are generated elsewhere: which supplier made which subcomponent, in which plant, with what workforce, at what wage. Verification is a chain-of-custody problem, and chain of custody is geographic. The federal government can write the rule; only the corridor holds the evidence. The entity that could convene the plants, reconcile their data, and stand behind a compliance claim would have to be a metropolitan one, and in Juárez–El Paso there is no such institution. Hold that gap; the argument returns to it.
Verification is the first of three infrastructures that have quietly become metropolitan in exactly this way. The other two are logistics and energy. Each shares verification’s defining feature: it is a cross-border function the national government cannot perform locally, and increasingly it is a problem of proof.
The second infrastructure: a port you can rely on
A rule of origin is only as good as the port that enforces it. A wiring harness or a server board that crosses the river more than once during assembly cannot absorb an unpredictable customs delay; the reliability of the crossing becomes a competitiveness variable in its own right, a direct input into the cost of the verified supply chain. Port reliability is the operational substrate of verification, the place where the goods and their paperwork are physically checked.
How much does unreliability cost? On the western end of the border, SANDAG, the San Diego regional planning agency, modeled it: $3.4 billion a year and 88,000 jobs in border wait times across the California and Baja California ports of entry. That is a modeled figure, not a metered one, but it is the only hard price tag anyone has put on weak port governance on this border. Juárez–El Paso has no number of its own. Neither does Laredo. The metros that move the most North American trade cannot say what unreliable crossings cost them, which is the first thing any serious negotiation would need to know.
The reason no city has fixed this is structural. Customs is a federal monopoly on both sides of the line, and no city can run a port of entry. But measuring the problem, and petitioning jointly to fix it, are not federal.
The third infrastructure: power you can prove is clean
The third infrastructure is the one the trade conversation never reaches, and it may be the most consequential. The corridor is becoming a center of gravity for the most electricity-hungry industry on earth.
Chihuahua, the state Juárez anchors, has become Mexico’s computer-hardware export platform: it exported about $99 billion in goods in 2025, roughly 46 percent of all Mexican CPU exports, as national computer processors exports rose about 145 percent and overtook automobiles as the country’s top export category. The investment is landing locally, with Foxconn alone disclosing roughly $241 million to expand AI-server production in the state.
The hardware Juárez assembles is what makes downstream computing enormous. An AI server rack draws 40- to 60 kilowatts against a handful for a conventional one, and the largest AI campuses now plan for continuous loads of 650 megawatts to a gigawatt, more than the entire installed capacity of many Mexican states. A corridor that makes this hardware is, before long, exposed to the same power constraint that will govern any computing it later hosts.
And the requirement has changed in a way that mirrors the USMCA. The buyers driving the boom, led by the major cloud providers, no longer accept clean electricity as an annual average; they have committed to matching their consumption with carbon-free power on an hourly basis, at the place of use, by 2030, and European rules now require data centers to report hourly matching. Clean power is becoming something you have to prove, at the time and place you used it, exactly as origin is something you have to prove at the border. It is a verification problem wearing a different uniform.
And here is the corridor’s bind. Only about 19 percent of Chihuahua’s installed generating capacity is clean, against a national average near 37 percent, on a gas-heavy fleet that added no new operating capacity between 2024 and early 2026. The federal answer, a national transmission plan, moves power around; it does not make Chihuahua’s mix cleaner. The state winning Mexico’s AI-hardware race is, today, among the least able to give its new customers the clean power they increasingly require. And the seam runs straight through the metro: El Paso draws on the U.S. grid, Juárez on the Mexican grid, joined by a single asynchronous tie, with no binational body planning the energy future the corridor’s own production is making urgent.
Figure 1. The corridor’s bind. Chihuahua has become Mexico’s AI computer-hardware export platform even as its clean share of installed capacity sits at roughly half the national level, on a gas-heavy fleet that added no new operating capacity in 2024–2026. Sources: SENER PRODESEN 2024 (state capacity, Dec 2023) and PLADESE 2025 (national, Dec 2024); OEC, Mexican customs 2025; CNE permit registry (Feb 2026).
A transition mistaken for a contraction
All of this rests on a claim that deserves scrutiny, because the surface data look like decline. Between August 2023 and August 2025, Juárez lost 57,500 maquiladora jobs, an 18 percent fall, from 317,000 to 259,500, according to the Federal Reserve Bank of Dallas. The reflex reading is that the corridor is unwinding. It is not. Plant counts held roughly flat, near 330, and output rose as employment fell: in March 2025 the city’s plants exported about $9 billion to the United States in a single month, up roughly a third from the same month a year earlier. Fewer workers were producing more, and three independent pieces of evidence say the cause is upgrading, not merely the automation of the same low-wage work.
The first is wages. If plants were simply automating unchanged assembly, pay would not rise broadly. It did. Average maquiladora wages in Juárez rose about 10 percent in nominal terms in the year to October 2025, to roughly 23,300 pesos a month, well above inflation of 4-5 percent during the period, and they have roughly doubled over five years (INEGI data, reported by El Diario de Juárez). Rising wages alongside falling headcount is the labor-market signature of a workforce moving up, not out.
The second, and most direct, is the composition of the jobs themselves. Between 2019 and 2024, according to research from the University of Texas at El Paso’s Hunt Institute, computer-manufacturing employment in Juárez rose by about 9,500 jobs, a 238 percent increase, and medical-equipment employment by about 12,500, up 41 percent, while textile employment fell by 8,000 and automotive electrical-systems employment by 5,000. The basket of jobs rotated toward higher-complexity work, with employers reporting growing demand for bilingual engineers and technicians.
The third is the export basket, which rotated the same way. Chihuahua’s exports of processing units, the chips at the heart of AI and data-center hardware, rose from about $14 billion in 2023 to about $42 billion in 2025, nearly tripling. Part of that gross figure is re-exported imported components, so the net gain is smaller, but the rotation toward a more complex product is real and backed by real plant investment. On the U.S. side the same gradient shows up in pay: in El Paso County, computer and electronics manufacturing and computer-systems-design work pay 55-70 percent above the all-industry average wage.
Figure 2. Fewer workers, more output. Ciudad Juárez maquiladora employment fell 18 percent from mid-2023 to 2025 while single-month exports to the US rose about a third and the plant count held near 330. Source: Dallas Fed, Southwest Economy (April 2026).
Figure 3. What crosses the river. Chihuahua’s top exports, 2025 (gross, state-level proxy for Juárez): processing units for AI and data-center hardware nearly tripled since 2023. Source: OEC, Mexican customs 2025 (HS6).
The corridor, in other words, is shedding lower-margin assembly roles while adding more technical manufacturing work. That is good news, and it is also what makes verification both harder and more valuable: technical, supplier-deep production is exactly the kind that generates a heavy burden of proof, and exactly the kind whose margins can bear the cost of carrying it.
The governance gap, and why it persists
Set the three infrastructures side by side. Verification, logistics, and energy are all metropolitan, all cross-border, and all increasingly problems of proof. And the corridor has no institution built to handle any of them jointly. There is no binational entity that can convene the assembly plants of Juárez, the freight operators of El Paso, and the workforce that crosses every morning. There is not even a combined statistical picture of the metro. Juárez’s data come from INEGI, El Paso’s from the U.S. Census and Bureau of Labor Statistics, on different survey calendars and definitions; the two are reconcilable, since INEGI’s SCIAN codes and the U.S. NAICS are the same North American classification, yet no standing institution combines them. Coordination has never had a common, current factual basis to act on.
What exists is a set of partial institutions, none of which fits. Borderplex Alliance, the regional economic-development organization, is a U.S.-incorporated nonprofit headquartered in El Paso, funded by U.S. dues and governed by a U.S. board, with Juárez participation that is informal and advisory. It is the closest thing to a binational economic body on the corridor, and the point is that the closest thing is a US nonprofit. Plan Estratégico de Juárez, the natural counterpart, is civil society, built in the mid-2000s; it gives Juárez rare institutional continuity but holds no binational mandate and does not seek one. The North American Development Bank is the only genuinely 50/50 binational institution in the U.S.-Mexico relationship, with a substantive presence in Ciudad Juárez, but it was designed to finance environmental infrastructure and carries no economic-development mandate. The template exists. The application does not.
Figure 4. The governance gap. Each candidate institution measured against what governing the corridor as one unit would require. Source: author institutional analysis.
The obvious question is why. If a binational corridor institution is so clearly needed, why does it not already exist? The honest answer is that the obvious version of it is unconstitutional, and the affordable version has never been built.
The hard reasons are constitutional, and symmetric across the border. The three powers a genuine binational government would need, it cannot have. It cannot make foreign agreements: the U.S. Compact Clause bars a state, and therefore a city, from a binding compact with a foreign power without the consent of Congress, and Article 117 of the Mexican Constitution bars states from any treaty with a foreign power. It cannot run the ports: customs is a federal monopoly on both sides, run by CBP in the United States and by SAT and ANAM in Mexico, and no city can staff a lane. And it cannot count on continuity: three-year Mexican municipal terms against four-year terms in El Paso mean the same two principals rarely lead a joint effort for long. The softer reasons follow, above all that Mexican municipalities raise only about a fifth of their own revenue, so Juárez cannot reliably fund a standing secretariat out of its own budget.
But read the constraints carefully and they do not forbid coordination. They forbid one specific thing: a chartered cross-border government with binding authority over customs and foreign affairs. They say nothing against a body that only produces shared evidence and petitions jointly. That distinction is the whole opening.
Integration outruns governance
Juárez–El Paso is not an outlier. It is the sharp case of a pattern that runs the length of the border, and the pattern is clean: economic integration runs from high to very high, while binational governance capacity runs from moderate to low, and the two do not move together. If anything, trade value substitutes for institution-building, because the flows happen anyway.
San Diego–Tijuana is the most institutionally developed, with about 5.5 million people, the Smart Border Coalition since 1994, and the SANDAG studies. After three decades it still has no permanent secretariat and no shared budget, and its flagship commercial crossing took some 25 years to move from concept toward construction. High integration, moderate capacity. Laredo–Nuevo Laredo moves the most value of any U.S.-Mexico crossing, $354 billion in 2025, the largest U.S. land port, yet its institutional architecture is the thinnest of the three: a U.S.-side development corporation and the World Trade Bridge, with no binational body of shared authority. Governance never had to scale, because the value crosses anyway. Very high integration, low capacity.
Juárez–El Paso sits between them on integration and at the floor on capacity, with not one institution holding binational governance, a shared budget, and an operational mandate; the civic capacity that might have built one was absorbed, through the 2008–2012 violence crisis, by the more urgent work of keeping the city functioning.
Figure 5. Three binational metros by economic integration and binational governance capacity. The governance score is the author’s index. Sources: OEC trade 2025; SANDAG.
No metro on the border occupies the quadrant that matters, high integration with high governance capacity. That is the gap to fix. The Juárez–El Paso proposal that follows is therefore not an innovation. It is an overdue correction to a gap every binational metro on this border shares.
What two mayors can do, and why it is lawful
The mayors of Juárez and El Paso cannot renegotiate the USMCA, fund a new bridge, or merge two customs authorities. They can do something narrower and, because it is narrow, lawful and immediate. Each instrument below is built to fit through the constitutional gap identified above: it produces shared evidence and coordinates advocacy, and it governs nothing.
The first and foundational instrument is a Joint Evidence Unit, and its first deliverable should be concrete: a Juárez–El Paso Verification Dossier for the ten product families most exposed to the USMCA review, built from customs records, plant-level employment data, and direct engagement with firms, and designed so corridor producers could carry it to USTR, the Secretaría de Economía, or their own headquarters. Producing shared evidence is the one binational function that requires no constitutional authority at all, which is why it comes first: it fills the missing-data gap directly, and it needs no one’s permission.
The second is a Port Reliability Compact: a shared dashboard of commercial-crossing performance at the two cargo bridges, the Bridge of the Americas and Ysleta-Zaragoza, with quarterly publication and joint petitions to CBP and to SAT and ANAM where wait times breach agreed benchmarks. It commissions the wait-time number that Juárez–El Paso, like Laredo, still lacks. It touches no port of entry, because it operates none; it measures, and it petitions. The need is immediate: the U.S. General Services Administration has proposed permanently removing commercial cargo from the Bridge of the Americas during its reconstruction, a change that would reroute thousands of trucks a day, and business leaders on both sides have been contesting it without a shared channel to do so.
Both fit inside a ninety-day compact anchored to the July review. The permanent version is a binational economic council, formally chartered by both city governments but anchored in the institutions that survive the electoral cycle: Plan Estratégico de Juárez, the two universities, and El Paso’s professionally managed economic-development office. It would run on staff time and existing budgets, not a new joint payroll the Mexican side could not reliably carry. The North American Development Bank, the one genuinely 50/50 binational body on this border, proves the model can exist. It was simply pointed at the environment instead of the economy.
The verification border
Return to the truck on the bridge. It will cross freely; the tariff is not what holds it. What is scarce, and what determines whether the corridor’s deepening integration is worth anything, is the proof that travels with it: proof of origin, proof of content, and before long proof that the power behind it was clean. The USMCA moved the binding constraint on North American integration from liberalization to verification, and verification is produced where firms, lanes, and substations physically sit, which is to say in the metropolis, not the capital.
Juárez–El Paso already has the labor, the capital, and the infrastructure. The missing factor of production is institutional: a body that can convene firms, reconcile their data, and stand behind a claim. The July 1 review is the clock. The corridor is deepening integration faster than its institutions can verify it, and that is a choice about institutions, not a fact of geography. The mayors of Juárez and El Paso, between now and summer, have the smallest and sharpest version of the chance to choose differently.
REFERENCES
This article draws on the Federal Reserve Bank of Dallas’s April 2026 Southwest Economy analysis, “Ciudad Juárez retools amid job losses”; the University of Texas at El Paso Hunt Institute’s 2025 research on Juárez sectoral employment; INEGI maquiladora wage data as reported by El Diario de Juárez (January 2026); OEC trade data for the El Paso customs district and for Chihuahua state (2025); reported 2025 trade volumes for the Laredo customs district; US Bureau of Labor Statistics QCEW wages for El Paso County (2024); USMCA Annex 4-B and Article 34.7; SENER PRODESEN 2024 and PLADESE 2025 electricity-capacity data (via https://lineupmx.com/, NZIPL Mexico energy platform) and the CNE generation-permit registry; published hyperscaler clean-power commitments (the Climate Neutral Data Centre Pact and the Google 24/7 carbon-free-energy program); the SANDAG cross-border wait-time studies; the US Compact Clause and the Mexican Constitution (Articles 89, 115, and 117); OECD and World Bank analyses of Mexican municipal finance. Energy-share figures compare a state series (Chihuahua, December 2023 baseline) with a national series (December 2024). The three-metro governance tiers are the author’s index; the underlying scale and trade figures are measured. Full source notes available on request.
