U.S. Rail Intermodal Up 12.6% in Week 41 as Diesel and Truck Capacity Push Freight onto Rail

The Analysis

Lay a map of the United States flat on a table and put a finger on the intermodal ramp at Commerce, east of downtown Los Angeles. That slab of asphalt is where a container stops being the port's problem and becomes the railroad's. From Commerce a box goes north over Cajon Pass, across the Mojave to Barstow, east along the old Santa Fe route through Arizona and up the Rio Grande to Belen, across the Texas panhandle into Kansas City, then on to the Chicago yards, where it turns back into a truck for the last sixty miles to a distribution center in Joliet or Elk Grove Village.

Put a second finger on Lathrop outside Stockton, a third on Marion in Ohio, a fourth on the terminals feeding the I-85 corridor between Charlotte and Atlanta. Every finger lands on the same asset: a rectangle of ground with crane lanes, a gate, a chassis pool and about two days of parking. About two days of parking is the phrase to hold on to.

On 21 October the Association of American Railroads published traffic for the week ending 17 October. U.S. railroads moved 540,135 carloads and intermodal units, up 8.5 percent from the same week a year ago. Carloads, meaning the boxcars, hoppers and tank cars that carry grain, coal, chemicals, aggregates and forest products, came to 232,120, up 3.5 percent. Intermodal, meaning containers and trailers, came to 308,015, up 12.6 percent. Across the nine reporting railroads of North America the week totaled 731,715 units, up 7.2 percent, with intermodal up 10.8 percent. Through the first forty-one weeks, U.S. intermodal has moved 11,687,964 units, up 4.6 percent, and carloads 9,374,717, up 2.7 percent.

The 12.6 percent is the number that traveled, and it earned the trip. A week is a photograph while the year is the film, and the two say different things: one week at 308,015 against forty-one weeks averaging 285,072 is a pulse, not a trend. Pulses matter more than trends if you are shipping in November, because a pulse gets absorbed by the parts of the system with no slack. What follows is a walk along that Los Angeles to Chicago line joint by joint, a price on going around each joint, and an answer to the only question a shipper really has: does this route still work, and what do I have to buy in the next three weeks to keep using it.

Start with the arithmetic the headlines skipped. Last year's comparable week can be backed out of the growth rates: intermodal at 308,015 against a 12.6 percent gain means the same week in 2025 ran about 273,548 units, so the year-over-year difference is roughly 34,500 containers and trailers in seven days. Carloads at 232,120 against a 3.5 percent gain mean about 224,271 last year, a difference near 7,850 cars. Total U.S. volume was roughly 497,819 units a year ago, making this year's 540,135 about 42,300 higher, and intermodal supplied roughly 34,500 of that — about 81 percent of every incremental unit the rail system handled this week. The carload side, up 3.5 percent and only 2.7 percent across forty-one weeks, is flat enough that week-to-week noise explains most of it.

Hold the week against the year. Forty-one weeks of intermodal at 11,687,964 units is an average of 285,072 a week, and this week ran 308,015, about 8.0 percent above the year's own average. Carloads average 228,653 and ran 232,120, about 1.5 percent above. The honest sentence is not that rail is booming; it is that one product on the rail network is running about eight percent hotter than the network was staffed and equipped to run, and doing it just as the industry walks into its busiest eleven weeks.

Extrapolate, with the assumption stated. Assume the remaining eleven weeks run at exactly the week-41 level of 308,015 and the year closes near 15,076,000 units; assume they settle back to the 41-week average of 285,072 and it closes near 14,824,000. The gap is about 252,000 units, under two percent of the full year. On the 2025 side, the 41-week figure of about 11,174,000 implies an average week near 272,538, and if last year's final eleven weeks ran at that average, 2025 finished near 14,172,000 — which puts full-year growth at close to 6.4 percent rather than the 4.6 percent booked so far. Every sentence there rests on an assumption I wrote down, and that is the point: the number you plan against this week is not the number you get in December.

Why it is happening now has two engines, and only one sits on the railroad. Diesel near a five-year high lifts the cost of a long truck move faster than it lifts the cost of a rail move, because a locomotive burns far fewer gallons per ton-mile than a tractor and because the fuel surcharge formulas in most motor carrier agreements pass the increase through inside a week. The second engine is capacity leaving the truck market, and the freight that leaves first is the freight nobody wants: the 1,800-mile runs that keep a driver out four or five nights. Those are the lanes where rail is competitive, and a shipper who was indifferent in June is not indifferent in October.

Peak season is the third engine, and the one that turns a cost calculation into a service problem. Produce out of the Central Valley, seasonal replenishment for the big-box DCs in the Inland Empire and around Chicago, holiday inventory that has to be on shelves by early December, and the year-end push to get goods inland before the Lunar New Year factory shutdown all land on the same ramps in the same six weeks. The print is dated 21 October for a week that ended 17 October, which means the surge was already sitting in the terminals before most 2027 contract conversations opened.

One caution on reading these prints: the weekly AAR number is raw counts, not seasonally adjusted, and a four-week moving average is the honest unit of observation. A single week still tells you something — not the level of demand, but where the slack went.

Walk the route joint by joint and price each one. The origin dray comes first in the sequence and is the joint most shippers never look at. A box discharged at San Pedro Bay is trucked twenty to forty miles to a ramp in Commerce or the Inland Empire before a crane touches it. Assume that dray costs about 450 dollars for roughly sixty miles, around 7.50 a mile, and note that the driver pulling it comes out of the same national pool that is short everywhere. If the pool tightens, you see it in the dray, not on the rail.

The gate and the lift are next. A ramp is a parking lot with a crane and a fence. Assume a mid-size facility handles 900 units a day across two shifts and a five-day week, or 4,500 a week; a week running eight percent above the annual average puts about 360 extra units through it, roughly half a day of additional lifts, out of the same gate lanes, the same hostlers and the same painted stripes. Crews can be added in two weeks. Ten acres of parking cannot, and parking is what actually limits a ramp when volume spikes.

The line-haul is the part shippers worry about and the part that usually holds. A train from Barstow to Kansas City does not need a new crew or another locomotive because one customer added forty boxes a week; it needs slots on a service that already runs, and filling slots is cheap right up to the tonnage or length limit. The mainline has slack in October that it will not have in the first week of December, but it is not where the queue forms first.

The destination dray, the chassis and the receiving appointment are where the line seizes. Assume the ramp sits seventy miles from the DC and that leg runs about 550 dollars. The tractor needs a chassis under the box and a door at the far end, and the receiving desk at a DC running at peak has exactly as many doors as it had in August. If the box waits, it waits on the railroad's clock: assume 48 hours of free time and 75 dollars a day after that. Two days of storage across twenty boxes is 3,000 dollars, annoying but survivable; a missed replenishment window that becomes a stockout is a different order of magnitude, and it never shows up in the freight line.

So the joint that bends is unglamorous: not the mainline, but the two dray pools at the ends and the parking in the middle. Everything you do about this week should aim at those three things and nothing else.

Who feels it and how much depends on three variables: how long your lane is, whether one Class I serves your ramp or two, and whether you control the dray or buy it through someone else. Lane length decides most of it. Below 500 miles intermodal rarely wins even at this diesel price, because two drays at roughly 450 and 550 dollars eat what the rail leg saves and the truck is four days faster door to door. Between 750 and 1,500 miles the arithmetic turns in rail's favor, and above 1,500 miles the line-haul saves enough that even a bad week of dwell leaves you ahead. Los Angeles to Chicago at about 2,100 miles is deep in that territory; Savannah to Charlotte at 250 miles is not, and that shipper should not read a 12.6 percent intermodal week as a signal to move anything.

One Class I or two is the second splitter, and it decides whether you have any leverage in November. If one railroad reaches your ramp, you have one bid and no fallback when its terminal fills, and it will fill. If two reach it, you can price them against each other and push overflow across town for about 40 extra dray miles, around 300 dollars a container. Three hundred dollars a box is cheap insurance against a two-day slip. Control of the dray is the third variable, and it is the difference between a plan and a phone number: buying drayage through the railroad's program buys convenience and gives up dispatch, while a dedicated dray carrier costs a little more and answers the phone at six in the morning when the appointment slips. In a week like this one, the phone call is the product.

Carload shippers live in a different economy. Their number is 232,120, up 3.5 percent, and 9,374,717 for forty-one weeks, up 2.7 percent. Grain, coal, aggregates, chemicals and forest products move on a network this print says almost nothing about, and reading a 12.6 percent intermodal week as a tight carload market means misreading two separate businesses that happen to share track.

Timing splits three ways. On cost the lag is short: fuel-driven modal shift shows up in spot truck quotes within one to three weeks, in contract tenders at the next bid cycle, and in your 2027 intermodal RFP four to eight weeks after the peak breaks. On transit the lag is immediate: dwell at a congested ramp adds one to three days inside the same week the volume lands, and it does not come out until volume falls. On inventory the lag is a full planning cycle, because four extra days of pipeline on a 2,100-mile lane means four extra days of safety stock you either fund or run without, decided before the conversion rather than after the first late delivery.

Here is the part the headline leaves out, and it is the part that costs money. The 12.6 percent week is being read as evidence that freight is shifting onto rail. It is not a shift, it is a pulse, and the size of the pulse is what hurts: a network equipped for 285,072 units a week absorbs 308,015 by borrowing from the slack at the three nodes that have none, and no railroad added 34,500 units of terminal capacity in seven days. The published number describes the mainline, the one part of the system with room to spare. The number that decides whether your November freight moves is the number nobody published: how many appointment slots are left at your ramp, how many tractors are left in the dray pool serving it, and how many hours of free time you actually negotiated. Buy the rate if you like; buy the slot, or you will not use the rate.

Three things would flip this, and you should be able to name them before signing. Diesel falls thirty to forty cents and stays there a month, the truck quote drops, the indifferent shipper becomes indifferent again, and the weekly number decays toward the 4.6 percent trend inside six to eight weeks. Import volume at the coast gateways softens through November and December, ramp slack returns, dwell normalizes, and what you locked at a premium looks expensive by February. A Class I adds terminal capacity in your market, or another ramp opens appointments, and your leverage comes back without you doing anything. The reverse holds too: if the last eleven weeks run at 308,015 or better and diesel stays put, that 6.4 percent stops being a projection and becomes the base you negotiate 2027 against.

Run the money on a real shape. Assume a shipper moves 40 containers a month from a Southern California origin to a Midwest DC, 2,100 miles door to door, cargo worth 55,000 dollars a container, and an inventory carrying rate of 20 percent a year. Truck, all in: assume 2.05 a mile including the fuel surcharge, which is 4,305 dollars a container and four days transit with a team, or 172,200 dollars a month for forty boxes.

Intermodal, door to door: assume a quoted 3,180 dollars a container, built from a rail line-haul of 2,180, an origin dray of 450 and a destination dray of 550, with eight days transit. Forty boxes comes to 127,200 dollars a month. The saving is 1,125 dollars a container, 45,000 dollars a month, about 26 percent off the freight line, or 540,000 dollars across a year.

The penalty is four extra days. Carrying cost is 55,000 times 20 percent divided by 365, about 30 dollars a container per day, so four days costs roughly 120 dollars a container and 4,822 dollars a month; net saving lands near 40,200 dollars a month. Break-even is 1,125 divided by 30, about 37 days: intermodal would have to be more than a month slower before inventory cost ate the freight saving. That margin is wide enough to make the modal decision at this lane length easy, which is worth knowing, because it tells you to spend your attention on service rather than rate.

Now make the truck cheaper, because that is the real risk. Assume trucking softens to 1.60 a mile: 2,100 times 1.60 is 3,360 dollars, the saving per container drops to 180, and break-even falls to about six days. Add two days of dwell on top of the four-day penalty and the modal decision is dead even. Freight is moving onto rail not because rail got wonderful but because truck got expensive, and it walks back the other way the moment truck gets cheap.

Now make drayage expensive, the other real risk. The two drays are 1,000 dollars of the 3,180, about 31 percent of door-to-door cost for the last 130 miles of a 2,100-mile move; six percent of the miles carrying a third of the cost is the character of intermodal in one line. Assume peak-season drayage runs 25 percent over contract, 250 dollars a container and 10,000 dollars a month, and the saving drops to 35,000; assume two days of ramp storage at 75 dollars a day on half the boxes, another 3,000, and the saving is still around 32,000 a month. The arithmetic says stay on rail, and the arithmetic was never the risk. The risk is that the box cannot get an appointment and the DC misses its window, a cost that lands in lost sales and expedite freight rather than in transportation.

Grace Yang does the cost-structure work on our side and would put that last sentence into the landed-cost model rather than the freight budget, which is the right place for it: the reason to lock a ramp slot in October is not the 250 dollars of drayage it might save but the replenishment window it protects, and a protected window is worth more than the freight line says.

Alternatives, priced honestly. Changing ramps inside the same gateway is the cheapest move, about 40 extra dray miles or 300 dollars a container, plus a day or two re-qualifying your appointment pattern at a new gate. Changing gateways is a different animal: moving a box from San Pedro Bay to Oakland or the Pacific Northwest changes which Class I serves you, changes the rail miles, and typically adds one to three days; do that for a season, not a week. Going back to truck is the most expensive and most reversible option, 1,125 dollars a container and one phone call away, which makes it the right release valve when a lane breaks. Transloading at the destination is a fourth path, and it only pays when the box was going to be broken down anyway.

The pitfalls arrive in a predictable order. Locking rail line-haul without locking drayage is the classic one, and the contract you signed in October does nothing when the dray carrier cannot find a chassis. Assuming appointment capacity came with the rate is the second, and appointment capacity is the scarce thing. Signing a 2027 contract off a week-41 print is the third: 12.6 percent is a peak-week number against a year running 4.6 percent, and the base you price is the year. Believing the mainline is the constraint is the fourth, and it almost never is.

Does this route still work? For a 2,100-mile domestic leg, yes, and the arithmetic is not close: 37 days of break-even against a four-day penalty, 40,200 dollars a month net on forty boxes. What changed is not the answer but the thing you have to buy. In an ordinary October you buy a rate. In this October you buy a ramp appointment, a dray carrier who answers the phone at six in the morning, and forty-eight hours of free time you have actually read. The railroad will sell you 2,100 miles of line-haul all year, and will keep selling it while your box sits in a stack at Commerce waiting for a tractor. It will not sell you a parking space there in the first week of December unless you asked in October. Ask in October.

↑ Back to the key points

— By Ömer Kaya