Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- By 10 September, pull eight weeks of gate-in to available dwell by ramp; wherever it grew more than half a day, move 15% of volume to a named secondary ramp before 30 September.
- Convert all drayage at your two busiest ramps from weekly spot booking to contracted capacity with a stated daily appointment allocation, signed before 1 October.
- Shift 30% of pickups into the first two hours of the gate day and 20% into the last two hours, then measure the dwell difference at the four-week mark.
- Set one written rule with a same-day clock: any box not grounded within 24 hours of arrival triggers a call to the terminal and a case number, logged by 17:00 that day.
- If intermodal keeps growing at the week-34 rate, plan one extra day of average dwell per quarter and size safety stock for that on a 1 November review, not a January one.
U.S. railroads moved 543,212 carloads and intermodal units in the week ending 29 August, up 4.1% year on year, with intermodal volume of 303,191 containers and trailers gaining 5.7% and carloads up 2.2% to 240,021. Metallic ores and metals led carload growth at +11%, while chemicals (-3.8%) and motor vehicles (-6%) fell; cumulative 2026 traffic is up 3.3%.
The Analysis
Rail is the part of the supply chain people stop watching, because it looks like a timetable rather than a decision. Week 34 of 2026 came in at 543,212 carloads and intermodal units, up 4.1% year on year. Intermodal alone was 303,191 containers and trailers, up 5.7%. Carloads were 240,021, up 2.2%. Cumulative traffic for the year is up 3.3%.
Those four numbers read like one story about a network holding up. They are two stories pointing in different directions, and the one that decides whether your container makes its vessel is the quieter one. Metallic ores and metals led the carload side at plus 11%, while chemicals fell 3.8% and motor vehicles fell 6%.
Boxes need a lift, a chassis, a truck and a gate appointment. Cars need a siding and a switch crew. This network just moved weight from the second bucket into the first, and the first bucket is the one with the narrow throat.
Three hundred three thousand one hundred and ninety-one containers and trailers in seven days is not a number anyone can picture. Turn it into a delta and it becomes physical. Against the same week last year that is roughly 16,400 additional units, and every one of them needed a lift onto a railcar, a lift off at the other end, a chassis underneath, and a truck at both ends of the rail move. The line-haul is the part the railroad is good at and the part with room to spare. The lift and the truck are the parts with a ceiling.
A ramp is a fixed asset. It has a certain number of tracks, a certain number of lifts per hour, a certain number of gate lanes, and a fixed number of hours when the gate is open. Push 5.7% more boxes through that and you do not get 5.7% more waiting. You get more than that, because a queue does not behave linearly near its ceiling. Boxes that used to sit two days start sitting three, and the ones that used to sit four start missing their vessel. Reported volume describes the middle of the distribution. The pain lives in the tail.
The headline also flattens something worth separating. Chemicals down 3.8% and motor vehicles down 6% is not only a demand signal. It releases yard tracks, switch windows and inspection capacity at terminals that handle boxes as well. Metallic ores and metals at plus 11% went the other way, so the car side did not shrink evenly; it re-weighted toward heavy bulk. What happened physically is that the network got busier at the box end and quieter in particular carload corners, at the same facilities. A 4.1% blend of 5.7% and 2.2% is not a measure of pressure. It is a measure of transfer.
Who feels it first. An importer running a thirty-day reorder cycle through a single inland ramp with a Tuesday and Thursday cut will feel it as appointment scarcity, then as dwell, then as a missed cut. A bulk chemical shipper on a dedicated siding feels the opposite: fewer cars, emptier track, possibly better service than last quarter. Size matters more than category. A large shipper with contracted drayage and a dedicated chassis pool is insulated from the ramp, because its trucks are already allocated to it. A mid-size shipper buying drayage week by week is the one who finds out on Wednesday that there is no appointment before Monday. Direction matters too. Import-heavy flow concentrates on the coastal-to-inland leg where the ramp is the gate, while export flow stuffed at an inland warehouse sits on the same ramp but usually with more lead time to absorb a day.
Timing deserves a closer look than the percentage. Week 34 ended on 29 August, and the week after it is Labor Day, the unofficial end of the American summer. That placement carries more information than the 5.7%. An intermodal gain printed in the last full week before a holiday is a build-ahead number, not a peak number. Those boxes were pulled forward to clear before the long weekend, which means they are sitting on the ground right now, and September will read as a release of stored volume rather than fresh demand. The dwell spike belongs to the ten days after the holiday, not to the week being reported.
Here is the part I have not seen anyone put down. Read the two growth rates against each other and a structural fact falls out. Intermodal grew at more than twice the rate of carloads while two of the largest carload categories contracted. The growth is landing in the one segment that needs a truck at both ends. A railroad can add a train. It cannot add a trucker, a chassis pool or an appointment slot in the same week. The bottleneck inside this number is not on the steel, it is on the rubber. There is a second-order move that gets missed as well: intermodal substitutes for long-haul trucking, so every box converted to rail releases over-the-road capacity on lanes long enough to justify a ramp, and competes for the same drivers on lanes too short for one. The same 5.7% can loosen truck capacity in one market and tighten drayage in another. Reading the average as a single market signal is reading the wrong object.
Do the small arithmetic. Assume you move thirty FEU a week through one ramp, and assume your dwell from gate-in to available stretches from 2.4 days to 3.1 days because the appointment book is full. That is twenty-one extra container-days a week. At an assumed fifty-five dollars a day for chassis and yard combined, which is my assumption and not a number from the release, that comes to about eleven hundred and fifty dollars a week, roughly five thousand a month. The figure that actually hurts is smaller in count and larger in cost. If one box a week misses the cut, you rebook a week later, and on a thirty-day reorder cycle one lost week is a stockout nobody planned for.
Switching the whole flow to over-the-road is the reflex and it is the wrong reflex on long lanes, because that is exactly where capacity has been leaving. What works is decoupling. Take the ramp out of the critical path by pre-pulling the boxes you know you will need, taking the off-peak gate window, and owning or contracting the chassis so the ramp pool stops being your constraint. Name a second ramp before you need it, because the week you need it is the week it is full.
The trap I would flag is treating one week as a trend and re-forecasting inventory off it. Week 34 carries a holiday build, and the next two weeks will look worse for reasons that have nothing to do with demand. The other trap is expecting the terminal to tell you when it is full. It will not. It will simply stop confirming appointments, and by the time you notice, your drayage is booked for the wrong day.
↑ Back to the key points