Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- By 4 September, get ingate-to-ramp-availability days for each of your lanes over the last eight weeks, plus average chassis availability on a Monday against a Thursday at every ramp you use.
- From the week of 7 September, move at least 30% of weekly intermodal volume out of the Friday-to-Sunday ingate window and into Tuesday to Thursday, and hold that split through the end of October.
- From 1 September, extend the receiving appointment for every intermodal box by one day and pre-book the additional slots rather than requesting them on the morning of arrival.
- By 11 September, get written confirmation from the drayage partner of guaranteed chassis per day at each ramp, and a named fallback with a response time for the days the pool runs short.
- Track gate-out against appointment weekly, hold 90% on time by 30 September, and if it falls below 80% for two consecutive weeks, move the overflow back to road for the peak weeks and price that decision now.
Tight truckload capacity is pushing more U.S. shippers onto rail, with domestic container traffic rising 7.4% in the first half of 2026, according to KWE's week-35 market update. While localized chassis shortages and terminal congestion persist, the network remains fluid; shippers are advised to add buffer time ahead of the September peak season.
The Analysis
A rail ramp at six in the morning is the most honest place to watch American freight, because nothing standing on a ramp can be talked up. There is a train that arrived on schedule, there are boxes sitting on the ground, there are chassis, and there is a gate that opens at five and, at a great many facilities, effectively closes by three in the afternoon when the receiving docks stop taking trailers.
The figures from the week-35 market update: domestic container traffic in the United States rose 7.4% in the first half of 2026 as tight truckload capacity pushed shippers onto rail. Localised chassis shortages and terminal congestion persist, the network is still described as fluid, and shippers are advised to add buffer time ahead of the September peak.
I have no argument with the volume number. My argument is with the word shift, because intermodal does not replace a truck. It replaces a long-haul truck and adds two short ones, and the short ones are the scarce thing.
Seven point four per cent is easy to say and hard to picture, so picture it as boxes on the ground. A train arriving is a scheduled event with a published time. A ramp is a shared resource with an unpublishable time, because what happens on it depends on how many boxes showed up in the same four hours and how many of them can find wheels. When volume grows 7.4% across a network that did not add ramps, the trains keep their schedules and the ground absorbs the difference. That is why the update can say the network is fluid and still report congestion in the same paragraph. Both are true. The line is fine. The ground is where it hurts.
A chassis is an object with a location. That sentence sounds trivial until you are short of one, and then it is the only sentence that matters. A localised shortage does not mean the country is short of chassis. It means they are in the wrong place at the wrong hour, and the hour is predictable. Trains arriving Friday night, Saturday and Sunday unload into a yard whose customers all want wheels on Monday morning. So Monday and Tuesday become the pinch, not because demand is higher across the week, but because a weekend of arrivals compresses into two working days. Every drayage operator I have ever worked with plans around that, and every shipper who does not plan around it spends Tuesday on the phone.
Here is the part the coverage walks past. Intermodal does not remove a truck from the road. It removes the long-haul tractor and puts a dray tractor at each end of the move, one to get the box from your warehouse to the origin ramp and one to get it from the destination ramp to the distribution centre. So a 7.4% shift to rail is also a 7.4% increase in demand for short-haul drivers, chassis and gate appointments, in a year when the reason for the shift was that road capacity was tight. The mode consumes the resource it was supposed to escape. That is why the two symptoms in the update appear together: rising intermodal volume and localised chassis shortage are not two facts, they are one fact seen from two ends. The more successful the shift, the more expensive the last mile of it becomes.
Who feels this depends on the width of their receiving window. A large retailer with a third-party warehouse at the ramp, five-day receiving windows and a planning team that books appointments a fortnight out will barely notice a day of slip. It absorbs the delay in a buffer it already pays for. A mid-size importer with one distribution centre, a two-day appointment window and a receiving dock that also serves e-commerce fulfilment will feel it as a missed appointment rather than a late train, and a missed appointment is a different order of problem: the trailer goes back to the yard, the box keeps burning days, and the reslot is next week.
An exporter shipping into the United States feels none of this until the invoice arrives, because the pain is on the destination ramp and nobody on the origin side is watching it. A domestic shipper who moved to intermodal to get away from tight truckload capacity has bought a cheaper long haul and a new bottleneck, and will discover the exchange rate between those two things in about ten days.
The timing is the part I would circle in red. You do not feel this in the first week of September, when the trains are still running to schedule and the ramps are still clearing. You feel it in the second week, when peak-season volume lands on a network that is already carrying 7.4% more than it was built for, and the first thing to go is not the train but the appointment. What goes next is the chassis pool at whichever ramp your boxes happen to arrive at.
Put a small account on it, with my assumptions flagged. Say your distribution centre takes 25 containers a week and 40% of them now move intermodal, which is ten boxes. Suppose the ramp has six chassis free on a Monday morning, an assumption, and the weekend arrivals have put all ten of your boxes in the yard by then. Four of them wait a day. That is four box-days a week, roughly two hundred box-days a year, and every one of those days is a day your inventory is in the wrong place and your appointment is at risk. The number that actually costs you money is not the box-day. It is the reslot: at a grocery or general-merchandise DC with a booked calendar, losing your slot on Tuesday commonly means the next one is the following week. One day of chassis waiting turns into seven days of the goods not being on the shelf.
Which tells you where the lever is, and it is not at the destination. The day a box becomes available at the ramp is set by the day it was ingated at origin, three to five days earlier, another assumption of mine but a robust one across most domestic lanes. You do not control the ramp's Monday. You control the day you hand the box over. A shipper who spreads ingates across Tuesday to Thursday gets arrivals spread across the week, and the same network that bottlenecks on Monday morning is perfectly fluid on Wednesday. That is the whole trick, and it costs nothing except a change to a loading schedule that somebody set years ago and nobody has questioned since.
The advice to add buffer time is right, and it is also the most commonly misapplied advice in this business, because people add the buffer to the promise date instead of to the handover. Buffer in the promise date protects your reputation and does nothing for the box. Buffer at the ingate actually changes when the box arrives. And there is a timing problem with buying buffer at all: everybody reads the same peak-season warning in the same week, so the buffer purchased in the last week of August is available and the buffer purchased in the second week of September is a queue. The cheapest day to buy it was three weeks ago. The second cheapest is today, before the rest of the market finishes reading.
↑ Back to the key points