Non-Domiciled CDL Fight Splits 21 States as Driver Supply Keeps Contracting

The Analysis

Laredo, then the long dry pull up to San Antonio. El Paso, then the flat run north. Twenty-one states plus the District of Columbia on one side of a courtroom, a coalition of Republican-led states on the other, and the rule that started all of it has been in force since 16 March 2026. The rule limits non-domiciled commercial driver's licences to holders of H-2A, H-2B and E-2 visas. The D.C. Circuit denied an emergency stay two to one, so the rule stays in force while the litigation runs. Final briefs land 5 August 2026, oral arguments 15 September.

I have run the Laredo and El Paso corridors for years, and here is the thing people forget when they read about a rule: the rule is not on the road. Paper does not sit in a tractor at a border crossing in August heat. The lane where the rule stops being a paragraph and becomes a queue is the Laredo corridor, the El Paso corridor, and the multi-hour pull up to Fort Worth and Dallas and Houston behind it. English language proficiency enforcement alone has placed roughly 5,000 drivers out of service in two months. Mexico's freight transport chamber says the US revoked work visas for about 20,000 Mexican truck drivers between April 2025 and April 2026. Those are not abstractions. That is capacity walking off the board.

Every corridor has a story, and this one has a hole in it. On this lane, the bottleneck is not the bridge. It is the seat behind the steering wheel.

Start with what the numbers actually are, because a lot of the commentary I have read compresses the two tracks into one and gets the timing wrong. Track one is litigation. Twenty-one states plus the District of Columbia are challenging the non-domiciled CDL rule as of late July 2026; a coalition of Republican-led states backs the agency. The D.C. Circuit refused an emergency stay on a two to one vote, which means the rule keeps operating while the case proceeds. Final briefs are due 5 August 2026 and oral arguments are set for 15 September. Track two is enforcement, and enforcement does not wait for a calendar. The rule took effect 16 March 2026 and limits non-domiciled CDL eligibility to H-2A, H-2B and E-2 visa holders. English language proficiency enforcement by itself has taken roughly 5,000 drivers out of service in two months, and Mexico's freight transport chamber puts the number of Mexican truck drivers whose US work visas were revoked between April 2025 and April 2026 at about 20,000.

Why those two tracks matter separately is the part I keep having to explain to people who have never stood on a dock in Nuevo Laredo at six in the morning. A court date is weather on the horizon. Enforcement is the pothole you already hit. The 5 August filing and the 15 September argument are real dates and you should plan around them, but neither of them puts a driver back in a cab next week. If the D.C. Circuit eventually rules against the rule, the capacity comes back through a hiring funnel that has been narrowing for a year and a half, and a funnel does not refill the week after a ruling. If the court upholds the rule, what you have today is not a disruption event, it is the new floor. Either way, the thing you can control between now and mid-September is your exposure on the US inland leg behind the border crossing, not the outcome of an appeal.

Here is the chain the way I see it on the map. The rule narrows who can hold a non-domiciled CDL to three visa classes. That narrows the pool of drivers legally eligible to run US domestic legs. English proficiency enforcement pulls drivers off the road in the near term, and roughly 5,000 came out of service in two months. Visa revocations, about 20,000 Mexican drivers between April 2025 and April 2026 per Mexico's freight transport chamber, shrink the cross-border pool further and they apply to people who were already working the lane. Fewer drivers means fewer available trucks per load slot, longer dwell at the yard, slower turns, and eventually either a higher spot price for the dry pull or cargo that simply misses its delivery window because nobody could cover it. None of that shows up in a headline about a court. It shows up in your appointment sheet three weeks out.

Put this on the map node by node and you can see where it bites first, because a corridor is not one road, it is a string of nodes and each node fails its own way. The first node is the bridge and the queue in front of it, and that one is not where the driver rule bites, because the drivers on that stretch are largely local and cross-border and the enforcement that has taken 5,000 out of service is spread across the country, not concentrated on the bridge. The second node is the transfer yard on the US side, and that is where you first feel it: the pickup that used to leave the yard in the same shift starts waiting for a tractor, and waiting turns into dwell, and dwell turns into a missed slot down the line.

The third node is the dry pull up to San Antonio, Fort Worth, Dallas, Houston, and that is the node where the shortage becomes a price, because that is a long paid run that eats a driver's hours and every carrier in the corridor is bidding for the same finite bench. The fourth node is the turn back south, which is the node most people forget, and it is the one that decides tomorrow: if the return leg has no driver, the outbound leg does not happen either, and the yard fills up with empty trailers nobody can move. When I have seen this pattern before, on other corridors, the failure never announced itself at the bridge. It announced itself at the yard, three days later, in the form of a trailer that was supposed to be loaded and was not.

Where it lands depends entirely on which part of your book you are looking at. If you are moving full truckloads from Monterrey or Guanajuato or Querétaro into Texas and the Midwest, you are living the tightest part of this, because there is no way to move that freight that does not need a driver in a US cab at some point. If you are moving freight on the US inland leg of an intermodal box that came into a US port, you are exposed to the same driver pool but from a different angle: your drayage and your inland rail-to-door moves draw from the exact same labour market, so a tight truck market raises your door cost even though your box never touched Mexico.

If your product is time-flexible and you have real inventory cushion, this is noise. If your product is time-critical, this is the whole game, because the failure mode here is not a higher rate you can absorb, it is a missed appointment you cannot fix with money on the day.

The part the coverage is not telling you, and it is the one thing I would want every importer and exporter on this lane to hear: there are two supply clocks running at different speeds, and the market is pricing only the fast one. The docket clock runs on a brief filed 5 August and an argument on 15 September, and it can move the rule months out. The driver clock runs faster than a court and slower than a rate sheet.

About 5,000 drivers out of service in two months, and about 20,000 visa revocations in a twelve month window, do not reverse on a ruling. Somebody reads a decision in September and assumes the corridor breathes again in October. It does not. You cannot rehire a driver in a week, and you cannot un-revoke a visa with a docket entry. The capacity that comes back on the slow clock has to be re-sourced, re-documented, and re-trained, and the visa pipeline is not a pipe you can open and close on the day.

So the thing that will actually determine your cost from here to the end of the year is not what the D.C. Circuit decides. It is whether your trucking partners have a stable bench of drivers in the three eligible visa classes, and whether they have been holding it. Ask that question now, while everybody else is arguing about the court.

Take the counterfactual and run it out, because it tells you why this is a capacity story and not a politics story. Suppose the rule had never been issued. The non-domiciled driver pool would still exist at roughly the size it had in early 2025, and the English proficiency enforcement campaign would still have taken its 5,000 in two months, because that enforcement does not depend on the rule for its existence. So the rule is not the whole shortage. It is the part that makes the shortage structural instead of a bump.

That distinction matters for your planning. A bump is something you wait out. A structural change is something you re-plan around. If the rule is struck down, you go back to roughly the pre-2026 driver market, but the practical floor under your cross-border truck cost stays elevated until the visa pipeline refills and drivers actually return, which is not a September question. If the rule stands, the tightening is permanent, and what you should do about it changes from wait to restructure.

Now the arithmetic, and I will say out loud which numbers are assumptions because most of what I need here is not in the article. Assume a cross-border dry van from a Mexican border city into the US interior. Assume that because the driver pool is thinner, covering that move now requires a premium of 15 percent over a stable-market rate, and assume that stable-market rate is about 1,000 dollars, so the premium is about 150 dollars per move. Assume you run 40 such moves a month. That is about 6,000 dollars a month, roughly 72,000 dollars a year.

But the rate is the cheap part of this, so do not stop there. Assume that in one month out of six, the load cannot be covered on schedule and slips a day. Assume 200 dollars in per diem, detention and rescheduling for a one day slip. With 240 moves a year, that is 40 slips a year and 8,000 dollars in slip costs, on top of the 72,000 in premium. Roughly 80,000 dollars a year of truck-side exposure on this one lane, before anything goes wrong in a way that costs you a customer.

Now look at the number the whole industry quotes and check it against your own book, because that is the one that decides whether you have a problem. About 5,000 drivers out of service in two months is roughly 2,500 drivers a month. Nationally that is a thin slice, on the order of one percent of the US driver pool, assuming a pool in the neighbourhood of 3.5 million. And 20,000 visa revocations over twelve months is roughly 1,700 drivers a month, again a modest national share. I know how that reads. It reads like noise.

It is not noise on the border, because the headline number is somebody's average and the border corridor is where the variance lands and it lands hard. Assume your own book draws from a corridor pool of 2,000 cross-border drivers. Even a low single digit percentage tightening there is enough to change who covers your Tuesday morning appointment, and the coverage you lose will not be the cheapest load, it will be the one with the tightest window. That is what the aggregate number hides. Average looks calm. Your appointment sheet looks like a fight.

So the way I would run it from here to the middle of September, and I would run it in two phases because there are exactly two dates on the calendar and I would not treat them as one event. Phase one runs now through 5 August 2026, the final briefs date. Phase two runs from 6 August to 15 September 2026, the oral argument. Before 5 August, the job is measurement, not negotiation. Identify who actually hauls you on the cross-border legs and how many of your committed moves depend on drivers in the three eligible visa classes.

If that number is a guess, it is the most expensive guess you will make this year. Get your carrier list, get the actual percentage of your volume on the Laredo and El Paso corridors, and get an answer on how each of them is holding drivers. From 6 August to 15 September, with an actual count in hand, convert that into contract language. Fixed volume commitments with capped premiums on the tightest 20 percent of your lanes, and an agreed fallback appointment window so a slip does not turn into a lost customer.

That count gives you your trigger, and I would write the threshold on the wall rather than carry it in your head. If your border-lane truck cost is running 10 percent or more above the same week last year, you are already in this, stop modelling and start locking. If coverage of your booked moves at seven days out falls below 90 percent, that is the trigger to activate the fallback window before your customer hears about it from their own dock. On the routing side, the alternative to the dry pull up from Laredo or El Paso is to shift what you can onto the rail and let the box move inland on a schedule instead of on a driver's availability, which fits cargo that tolerates a two to five day longer door-to-door.

The other alternative is to shuffle the mix: move the time-critical 20 percent of your volume onto a premium service with a named commitment, and let the patient 80 percent ride the slower path and absorb the variability. Do not move everything. If you move everything at once you pay the premium on cargo that never needed it, and you have traded money for peace of mind on loads where peace of mind was free.

There is also the week itself, and it is where most plans quietly fall over. A docket moves in months. A corridor moves in weeks, and inside that week it moves on fixed days that nobody rearranges for you. Pick one standing slot and keep it, the same hour every week. I use Monday mornings, before anybody starts calling about late trucks, and I run it as a headcount rather than a feeling: how many of this week's booked border moves are covered by a named driver whose papers sit in one of the three eligible classes, and how many are covered by nothing firmer than somebody's word.

Whatever sits on somebody's word gets re-quoted to the customer on Tuesday, not discovered on Friday afternoon. Then hang the whole thing on the two dates that are actually written down: before 5 August you are counting, and from 6 August to 15 September you are turning that count into commitments. Run that rhythm for six straight weeks and you will know your true corridor exposure before the judges have even taken their seats, while plenty of your competitors are still reading about who filed what.

The traps in the water, and I will give you the four that have cost people I know real money. Truck capacity is not uniform, so a national average rate tells you nothing about your lane in a given week; if your contract is indexed to an industry average, read the index definition before you sign it, because the average will look fine while your corridor is on fire. Truck and rail are substitutable in your planning and not on your dock; rail does not rescue a load that needed to arrive Tuesday, and pretending otherwise just moves the failure to a different calendar line. Contingency capacity has a clock on it: if a partner can pull a truck from another region to cover you, that truck takes days to arrive and the price you agree today is not the price on the day, so either pre-agree the number in writing or do not count it in your plan.

And the visa and driver side is not a switch: a favourable ruling does not refill a corridor in weeks, so do not build a plan that only works if September goes your way. Build one that works either way, and then you are not betting on a court, you are just running the corridor.

Closing it out, and this is the question I always come back to: can you still run this road. Yes, you can still run it. It is not closed, it is narrower, and narrower is a different kind of problem than closed. A closed road you go around once and you are done. A narrow road you drive every week and it costs you something every week. So treat the last week of July 2026, before the 5 August briefs, as the week you find out who is actually driving your freight and lock what you can lock. Treat 15 September as a checkpoint, not a starting gun. And whatever you do, do not confuse the court with the corridor. The court is where this gets argued. The corridor is where this gets paid for, and the corridor runs on Mondays whether there is a hearing or not.

↑ Back to the key points

— By Ömer Kaya