Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- By 15 September, confirm a box-and-chassis pairing for every container arriving in September and October, and reject any booking that reserves rail without a chassis against it.
- Set chassis turn at five days or better, and cost out every day beyond six against the specific receiver holding the box rather than absorbing it in overhead.
- Pull September and October arrivals forward by five days; the brief says build extra lead time, five days is my number.
- Route anything within 300 miles of the destination by truck door to door, instead of competing for a chassis in order to save a rail leg.
- Track box-without-wheels hours weekly, and move to the backup yard once a single terminal goes above 24 hours.
KWE's week-35 brief notes U.S. intermodal volumes continue to grow as tight truckload capacity steers more shippers to rail, with domestic container traffic up 7.4% in the first half of 2026. While localized congestion and chassis shortages affect some terminals, the network remains fluid overall, though shippers are advised to build extra lead time ahead of the September peak.
The Analysis
A container without a chassis is a concrete block with a number painted on it. It has weight, it has a destination, it has a customer waiting, and it cannot move one inch. That is the whole story of a chassis pool in a tight market, and it is sitting in the middle of a week-35 brief that otherwise reads calm.
The brief says US intermodal keeps climbing, tight truckload capacity is steering more shippers to rail, and domestic container traffic is up 7.4% in the first half of 2026. It also says localized congestion and chassis shortages are hitting some terminals, the network remains fluid overall, and shippers should build extra lead time ahead of the September peak.
Those sentences do not sit comfortably together. A network that is fluid overall with chassis shortages in a few places is not a network in good shape. It is a network where the average is still fine and the tail has already started to fail.
A chassis pool is a circulating asset, and circulating assets fail at the tail, not at the average. The pool is sized on how long a box takes to come back, not on how many boxes there are. If the median turn is five days and a minority of boxes sit for twelve, the pool looks healthy on a dwell report and quietly runs dry, because every box that overstays is a chassis that is not available to the next truck. Average dwell can improve while the pool drains. That is not a paradox, it is arithmetic.
The mechanics never change. A box lands, the receiver's dock is full, the driver drops it and keeps the wheels, the chassis sits under a loaded container in a yard for a week. Multiply by a hundred and the pool is gone. Nobody did anything wrong, the dock genuinely was full. But a chassis is a shared object and a dock is a private one, and the private decision to hold the box removes a public asset from circulation. That is how pools die in a strong market, not in a weak one.
Who feels it first depends on who owns the wheels. An importer running a private chassis fleet barely notices, because those wheels come back to their own yard on their own schedule. An importer relying on the terminal pool feels it as a wait of one to two days that nobody will confirm in advance, which is worse than a known delay because you cannot plan around a number that has not been given to you. A terminal with on-dock rail and enough ground rides it out. An off-dock depot with a small footprint cannot, because it has nowhere to put a box that is waiting for wheels.
Direction of trade matters less here than the length of the inland leg. This is a domestic intermodal story, so an exporter shipping into the US is exposed through the same pool the moment the box is discharged, and an exporter shipping out of the US is not exposed at all. If your cargo moves inland, you are in this, whichever way the invoice runs.
The order in which the symptoms appear is the useful part. The pool drains first. Then boxes sit. Then the yard fills. Then the congestion becomes visible and everybody starts talking about it. Chassis shortage is not a symptom of congestion, it is the thing that happens two to three weeks before congestion. By the time a terminal is described as congested, the wheels problem is old news and the recovery is already late.
Which makes the reassurance in the brief the warning. Fluid overall, shortages in some terminals, is the exact signature of a pool being drained by outliers. A genuine capacity problem shows up everywhere at once, at every gate, on the same day. Localized is what a tail looks like from the inside.
There is a second thing the brief cannot say, because it is dated week 35. Its own advice is late. Build extra lead time before the September peak, it says, at the end of August. The first half already absorbed 7.4% growth, which means the peak is landing on a network that has been running fuller than last year since January, with less slack than the calendar implies. Lead time is not something you add in the last week of August. It is something you either bought in June or you are now paying a premium for.
Here is why a small dwell slip beats a volume increase. Assume a terminal doing 1,000 chassis turns a week with a five day average turn, which is my assumption, not a figure from the article. Divide 1,000 by 7 and multiply by 5 and the pool in use is about 714 chassis. Add 7.4% volume and it becomes about 767, a rise of 53. Now hold volume flat and let the average turn slip from five days to six. 1,000 divided by 7 times 6 is 857, a rise of 143. The dwell slip costs 2.7 times what the volume growth costs, and it happens without one extra container arriving at the terminal.
That is the number to put in front of a receiver who wants to keep the box one more day. It is not their one box. It is about a fifth of the pool, taken out by a hundred small decisions that each looked reasonable on the dock where they were made.
If the truck market had stayed loose, none of this would be binding, because a shipper with options does not queue for a chassis and a terminal with spare ground does not run out of stack. Tight truckload capacity is what turned a pool management problem into a service problem. Remove the tightness and the rail share falls, the pool turns faster, and the shortage disappears. That is not a forecast, it is what the mechanism does.
The trap is booking the rail move and treating the chassis as part of the deal. It is not part of the deal. The chassis is a separate reservation against a separate pool, and if you have not paired the two, you will have a box on the ground, a customer on the phone, and nothing in between. The second trap is measuring average dwell, which will tell you everything is fine right up to the week it is not.
Pair the box with the wheels before you book anything. That is the whole job this month.
↑ Back to the key points