AAR says US railroads averaged 296,741 weekly intermodal containers and trailers in August, up 4.4% year on year and a new monthly record, with August gains for a 7th straight year. Weekly carloads averaged 235,109, the highest since October 2019. For the week ending 5 September, total traffic reached 533,545 carloads and intermodal units, up 13.8%, with intermodal up 18% to 299,148 on a weak holiday base; year-to-date volume is up 3.6%. AAR credits higher trucking costs for pushing long-haul freight onto rail.
Supply Chain Action Points
What this means for your business — and what to do about it:
The Association of American Railroads reported that US railroads averaged 296,741 weekly intermodal containers and trailers in August, up 4.4 percent year on year and a new monthly record, with August gains for a seventh consecutive year. Weekly carloads averaged 235,109, the highest since October 2019. For the week ending 5 September, total traffic reached 533,545 carloads and intermodal units, up 13.8 percent, with intermodal up 18 percent to 299,148 on a weak holiday base; year-to-date volume is up 3.6 percent. AAR attributes the shift to higher trucking costs pushing long-haul freight onto rail.
The 18 percent weekly figure is not the planning number. It sits on a weak holiday comparison. Plan off the 4.4 percent August gain and the 3.6 percent year-to-date figure, and treat the 18 percent as evidence that capacity is being consumed, not as a trend line. The two durable facts are that rail is absorbing freight converted from trucks, and that carloads at a level last seen in October 2019 mean equipment and ramp capacity are tightening at the same time.
The consequence is that the inland rail leg, not the ocean leg, now carries the schedule and cost risk on US-bound cargo. The five sections below translate these numbers into dated actions for exporters, cross-border ecommerce sellers, factories, brands and procurement teams.
For Exporters
AAR's August print puts the rail leg of every US-bound booking under pressure. US railroads averaged 296,741 weekly intermodal containers and trailers, up 4.4 percent year on year and a monthly record, with gains in August for a seventh straight year. Weekly carloads averaged 235,109, the highest since October 2019. AAR attributes the volume shift to higher trucking costs pushing long-haul freight onto rail. For a Chinese exporter moving boxes to US inland points, that is the sentence that matters: cargo that used to run on trucks is now competing with you for the same well cars and the same ramp slots. Quote against last year's inland transit assumption and the gap lands inside your own cost, whether the Incoterm puts it there or not.
Work the arithmetic before the next price list. Assume you ship 40 FEU a month to US inland points, and 24 FEU of that moves through a West Coast port onto rail (assumption). Assume ramp and rail dwell stretches from 4 days to 7 days (assumption; the article does not publish dwell days). Three extra days at an assumed USD 150 per FEU per day in container usage and storage charges gives 24 x 3 x 150 = USD 10,800 a month. The same three days under a DDP contract with an assumed 0.3 percent per day late-delivery penalty on an assumed USD 80,000 cargo value per FEU is 24 x 3 x 0.3% x 80,000 = USD 17,280 a month of penalty exposure. The second number, not the first, is what a DDP quote actually risks.
Three dated moves. By 19 September, cut US-lane quote validity from 30 days to 14 days and break the inland rail leg out as a separate line item naming the ramp you priced against, such as Chicago or Dallas. By 22 September, put any customer shipping 10 FEU a month or more onto a FAK rate plus a space guarantee, with the ocean allocation locked for 30 days and the rail leg refreshed monthly. From 1 October, every DDP quote carries a congestion clause that moves charges above the agreed dwell days to the buyer. Give one named person ownership of the ramp dwell number and the switch decision.
The alternatives each carry a price. Routing via an East Coast port such as Savannah or Norfolk avoids the West Coast rail queue but adds seven to ten days of ocean transit, and East Coast inland rail is tight too. Switching to all-truck delivery only pays inside roughly 800 kilometres, and AAR has said trucking costs are themselves rising, so you would be paying on both legs. Three pitfalls recur. Pricing FOB cargo as DDP is the first: the inland leg never enters the quote and only surfaces on the invoice. The second is a mismatch between the origin declaration, the bill of lading consignee and the customs entry, which leaves the box sitting uncollected on the ramp. The third is free time counted from vessel arrival when you assumed discharge completion; two days of that mistake puts you into chargeable storage. By 30 September, have your documentation desk track four timestamps on every US lane shipment: vessel arrival, discharge completion, rail connection, and ramp release. Exporters who can show a buyer those four dates are the ones who get the price increase accepted.
- Cut US-lane quote validity from 30 days to 14 days by 19 September and list the inland rail leg as a separate line item naming the reference ramp
- By 22 September move any customer shipping 10 FEU a month or more onto a FAK rate plus space guarantee, locking ocean for 30 days and refreshing rail monthly
- From 1 October attach a congestion clause to every DDP quote, moving charges above agreed dwell days to the buyer
- Build a four-timestamp tracker by 30 September covering vessel arrival, discharge completion, rail connection and ramp release, owned by the documentation desk
- Review the DDP versus FOB price gap monthly to confirm the full inland cost is inside the quote
- Name one owner for the weekly ramp dwell figure and the routing switch decision
For Cross-Border E-commerce
Cross-border sellers restocking into the US are paying for a rail squeeze they did not order. August intermodal averaged 296,741 units a week, up 4.4 percent year on year, and the week ending 5 September showed intermodal up 18 percent to 299,148 units. That 18 percent was measured against a weak holiday base, so it is not a restocking signal; the 4.4 percent August figure and the 3.6 percent year-to-date figure are. What it does confirm is that the inland rail leg is being consumed by freight converted from trucks, which AAR names as the cause. Your first-mile cycle losing four days does not show up on the freight invoice first. It shows up in safety stock days and in whether your best seller is on the shelf.
Quantify it. Assume daily sales of 20,000 orders (assumption) with core SKUs at 30 percent of that, so 6,000 units a day, an assumed landed cost of USD 12 per unit and a USD 29 selling price. Four extra days of first-mile transit means carrying 4 x 6,000 = 24,000 extra units, tying up 24,000 x 12 = USD 288,000 of working capital. Now run the other side: if stockouts push 3 percent of core-SKU demand into pre-orders or cancellations (assumption), 6,000 x 30 x 3% = 5,400 orders a month, and at an assumed 45 percent gross margin that is 5,400 x 29 x 45% = USD 70,470 of lost contribution a month. Carrying USD 288,000 of extra inventory is the cheaper of the two.
Act on a calendar. By 18 September, raise US-lane safety stock from 30 to 34 days, applied only to SKUs selling more than 100 units a week. By 20 September, finish one round of SKU tiering: high-turnover, high-margin lines take the rail express service with pre-booked ramp windows, and slow movers take the cheapest sailing, even via an East Coast port ten days slower. By 25 September, attach a substitutable backup SKU to every best seller so a stockout does not leave the traffic slot empty. From 1 October, stop adding ad budget to any SKU that has been in transit more than 34 days until it lands.
There are three fallbacks and each costs something. Air freight compresses first mile to under seven days, but at an assumed five to nine times the ocean unit cost it only works for high-value, low-volume lines. Rerouting via an East Coast port avoids West Coast rail but makes the last mile more expensive. Building deeper stock moves the risk forward but parks your cash at sea. Three pitfalls matter most. Confirm in writing whether surcharges are calculated from sailing date or arrival date before you book. Read free time as calendar days, because weekends count. And keep a five-day buffer between the platform promise date and actual warehouse arrival, or late-delivery payouts and returns will rise together. Treat the rail leg as a product decision, not a freight decision. A SKU that cannot survive a 34-day first mile should either move to a premium lane or leave the US assortment this quarter.
- Raise US-lane safety stock from 30 to 34 days by 18 September, limited to SKUs selling over 100 units a week
- Complete SKU tiering by 20 September, routing high-margin best sellers on rail express with pre-booked ramp windows
- Attach an automatically substitutable backup SKU to every best seller by 25 September to protect the traffic slot
- From 1 October stop adding ad budget to any SKU in transit more than 34 days until it lands
- Confirm in writing the surcharge calculation basis date and the free-time convention, counted in calendar days
- Keep a five-day buffer between platform promise dates and actual warehouse arrival on every US-lane replenishment
For Manufacturing Plants
Factories drawing components from Asia should read the AAR numbers as a production scheduling input. Weekly carloads averaged 235,109 in August, the highest since October 2019, while intermodal averaged 296,741 units a week, up 4.4 percent. Both lines rising at once means well cars, flatcars and ramp capacity are all turning tighter. Rail equipment is not an unlimited resource, and AAR naming higher trucking costs as the driver says this is a structural diversion of volume, not a holiday spike that clears in a fortnight. For a plant, a three-day extra stop at the ramp is not a freight annoyance. It is a three-day slip in the production plan.
Size the exposure. Assume the plant consumes 12 FEU of feedstock a week (assumption), one day of output is worth USD 150,000 and contribution margin is 25 percent (assumption). If ramp congestion adds three days and forces two reduced-shift days, the loss is 2 x 150,000 x 25% = USD 75,000 of contribution. On the hedging side, three days of cover at 12 FEU a week is roughly 7 FEU; at an assumed USD 60,000 of value per FEU that ties up 7 x 60,000 = USD 420,000. That is a balance-sheet item, not a profit-and-loss loss, and it unwinds the moment dwell normalises. Lost output never comes back, so as long as the pattern holds the buffer is cheaper than the stoppage.
Three dated actions. By 21 September, sort every Asia-origin feedstock by days of cover and put everything under ten days onto a single list that the planning desk refreshes each Friday. By 25 September, convert the inbound corridor from a single West Coast rail route into a dual corridor: rail as the base lane, an East Coast port with truck delivery as the backup. From 1 October, raise safety stock on single-sourced components from ten days to 13. By 15 October, move critical spares and every line-stopping item, motors, hydraulic pumps, sensors, to air or express regardless of freight cost.
Three alternatives, each with a cost. An East Coast gateway adds seven to ten days of ocean transit but sidesteps the West Coast ramps. Entering through a Canadian or Mexican port and moving overland adds a second customs clearance. Reshoring or nearshoring a supplier raises unit price but brings lead time back under control. The pitfalls sit in three places. Ramp free time and port free time are counted separately, so never treat them as one clock. Equipment and spare-part origin declarations and tariff codes must match the substitute model you are switching to, or the shipment stalls in classification. And air-freighting lithium-battery spares requires dangerous goods paperwork; discovering that at the airport adds two to three days. Put inbound lead time on the same weekly scorecard as line output, so the schedule decision is made before the material is late rather than after.
- List every Asia-origin feedstock with under ten days of cover by 21 September, refreshed by the planning desk each Friday
- Build a dual inbound corridor by 25 September with West Coast rail as the base lane and an East Coast port plus truck delivery as backup
- Raise safety stock on single-sourced components from ten to 13 days from 1 October
- Move critical spares and all line-stopping items to air or express by 15 October, regardless of freight differential
- Verify origin declarations and tariff codes line by line before switching to a substitute part or route
- Add inbound lead time to the weekly production scorecard alongside line output with a named re-planning owner
For Brand Owners
For a brand selling into the US, the AAR data translates into two commercial questions: can you keep the delivery promise, and is margin leaking out of it. August intermodal averaged 296,741 units a week, up 4.4 percent and a monthly record, weekly carloads hit 235,109, the highest since October 2019, and the week ending 5 September carried 533,545 carloads and intermodal units, up 13.8 percent. But the 18 percent intermodal jump inside that week sits on a weak holiday base. Do not write it into a multi-year fulfilment target. The risk you actually carry is inland dwell variance, which never appears on the freight invoice and always appears in late orders and returns.
Put a number on the customer side. Assume 20,000 US orders a month (assumption) against a five-day promise. A four-day inland slip that pushes 6 percent of orders past the promise (assumption) is 1,200 orders. Assume a quarter of those customers claim an assumed USD 10 credit, which is USD 3,000. Assume returns on the same cohort rise two percentage points, 400 orders, at an assumed USD 60 average order value and 40 percent margin: 400 x 60 x 40% = USD 9,600 of lost contribution. The combined USD 12,600 a month usually exceeds the freight premium for a priority rail or East Coast routing. None of this counts the customers who quietly stop reordering.
Three dates. By 18 September, tier US delivery promises by region, giving inland states a window two days longer than the coasts, and retire the single national number. By 22 September, require your 3PL or forwarder to report ramp-level rail dwell and on-site hours weekly, as a standing report rather than an ad hoc question. From 1 October, apply a stock allocation rule giving direct-to-consumer and member orders first claim on any container that has sat more than five days. Put dwell days, promise achievement and return rate on the same weekly scorecard so they are read together rather than in isolation.
The alternatives all have a price tag. Pre-positioning inventory in US warehouses moves risk forward but adds storage and working capital. Extending the promise window for inland states is free to negotiate but costs conversion. Air-freighting high-value orders protects the promise at an assumed five to nine times the ocean unit cost. Two pitfalls dominate. Changing a promise without changing the storefront is the first: marketplace listings, ad creative and customer service scripts must ship in the same release, or agents spend every shift explaining why the page says three days when the box takes seven. The second is treating surcharges as one-off costs when they accrue per shipment against a calculation basis you never confirmed. Read the rail dwell number as a customer experience metric, not a logistics metric, and put it in front of whoever owns the promise.
- Tier US delivery promises by region by 18 September, giving inland states a window two days longer than coastal states
- Require weekly ramp-level rail dwell and on-site hours reporting from your 3PL or forwarder from 22 September
- From 1 October give direct-to-consumer and member orders first claim on containers delayed more than five days
- Release marketplace listings, ad creative and customer service scripts in the same batch as any promise change
- Review dwell days, promise achievement and return rate together in one weekly scorecard with a named owner
- Quantify the promise-protection value of a priority rail or East Coast routing before rejecting the freight premium
For Procurement Teams
Procurement should negotiate off the trend, not the weekly headline. August intermodal averaged 296,741 units a week, up 4.4 percent year on year, weekly carloads averaged 235,109, the highest since October 2019, and year-to-date volume is up 3.6 percent. Those three figures describe a moderate upward pull. The 18 percent intermodal gain in the week ending 5 September came off a weak holiday base and should not be used as evidence in a rate discussion. What does support the carrier's position is AAR's own explanation: higher trucking costs are pushing long-haul freight onto rail, so demand is still moving toward the rail network. Taking 3.6 percent year-to-date and 4.4 percent for August into the room is a materially better position than conceding 18 percent.
Model the mix properly. Assume an annual framework covering 600 FEU of US inland rail (assumption) at an assumed contract rate of USD 2,000 per FEU. If spot rates firm 12 percent to USD 2,240 per FEU and 30 percent of your volume sits in spot, your exposure is 180 FEU x 240 = USD 43,200 a year. Cutting the spot share to 20 percent reduces it to 120 x 240 = USD 28,800, saving USD 14,400. Locking everything into contract looks safer until you test the other side: if actual volume comes in at 70 percent of the framework, the resulting shortfall or minimum-quantity penalty, assumed at 10 percent of the undrawn volume, can exceed USD 120,000, which is more than the spot premium you avoided. The right ratio depends on how confident you are in the volume forecast, not on how nervous you are about rates.
Three dated moves. By 24 September, put the US inland rail leg on the table as its own negotiation item and require three tiers, contract price, spot price and the trigger conditions between them, instead of one blended number. By 30 September, write an index-linked clause into the contract, tied to publicly published weekly rail volume or rate indices, with a review trigger when the movement exceeds 5 percent (assumed threshold, negotiable). By 10 October, make the relief and adjustment clauses concrete: named treatment for congestion, ramp delay and equipment shortage, with time-definite relief and partial refunds. By the same date, qualify a second rail corridor and cap any single corridor at 60 percent of volume.
Three alternatives, each with a different cost. Converting the inland leg to all-truck removes rail dwell uncertainty, but AAR has said trucking costs are rising, so you simply swap one risk for another. Moving the delivery point back to the port under FOB or FCA hands the inland leg to the buyer, at the likely cost of a price concession. Adding an East Coast gateway as a second corridor costs seven to ten days of ocean transit but breaks the single-route concentration. The most expensive mistake is signing a minimum-quantity commitment as if it were a free clause; when volume falls short, the penalty can consume everything you negotiated. The second is leaving the index source and review date undefined, which turns a rate increase into a unilateral notice. Have your legal and operations teams read the terminology schedule together before the contract is finalised. Price the contract against the value of the cargo on the box and the delivery date it protects, not against the freight line alone.
- Put the US inland rail leg on the table as a separate item by 24 September, requiring contract, spot and trigger-condition tiers
- Write an index-linked review clause by 30 September tied to published rail volume or rate indices, triggered above a 5 percent movement
- Define relief and partial refunds for congestion, ramp delay and equipment shortage by 10 October
- Qualify a second rail corridor by 10 October and cap any single corridor at 60 percent of volume
- Stress-test the minimum-quantity commitment against a conservative volume forecast before signing
- Have legal and operations read the terminology schedule together, including index source and review dates