Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- Within 14 days, get a written per-kilo rate and a stated cut-off time for O'Hare, Kennedy and one west coast gateway, and stop contracting a single national US rate. An average is not a lane.
- Before the end of the month, run the New York drayage quote against your real ten-tonne profile. Switch only if the door-to-door saving survives the truck, the extra handoff and two extra days.
- Secure a standing overflow allocation on the west coast before September 30, even at zero committed volume, so the option exists on the morning O'Hare jams.
- For any shipment above 5,000 kg with a fixed shelf date, move the booking date forward by seven days and confirm the cut-off in writing rather than by phone.
- Ask your carrier for allotment by flight number, not by week. A weekly number hides which of the seven departures you actually own.
In the week ending Sep 1, average air cargo spot rates from Western Europe to the US rose 7% from four weeks earlier to $1.88/kg, per Xeneta. Inbound Eastern US led gains-Chicago up 17% to $2.20/kg, New York up 16% to $1.58-while West Coast gateways LAX and SFO fell 5% and 15%. Analysts warn AI-hardware demand and Middle East capacity loss could pull the peak season forward to September.
The Analysis
Two pallets land at six in the morning on the same Tuesday, one at O'Hare and one at LAX. Same weight, same commodity, same shipper. By Thursday one of them is on a dock in the western suburbs of Chicago with the stretch wrap already off, and the other is still sitting because nobody booked a door for it. The flight was never the interesting part.
The week ending September 1 put numbers on the gap. Average westbound transatlantic air spot from Western Europe to the United States came in at $1.88 a kilo, 7% up on four weeks earlier. Under that average the two coasts went opposite ways. Chicago up 17% to $2.20. New York up 16% to $1.58. Los Angeles down 5%. San Francisco down 15%.
I have loaded and re-loaded enough of these lanes to distrust an average sitting on top of that much disagreement. What those four numbers describe is not one market with a wobble. It is two physical systems that happen to share an ocean.
Read the four numbers for what they are. Chicago and New York are levels, dollars per kilo. Los Angeles and San Francisco are changes, percent, direction only. Nobody published the level at LAX or SFO, so any spread you calculate against the west coast is a number you invented. What holds up is narrower and more useful: two East Coast gateways, both up by roughly the same amount in percentage terms, sitting $0.62 a kilo apart. Same ocean, same week, same direction of travel, O'Hare is 39% above Kennedy.
A 39% level gap between two airports that both moved up 16 or 17% is not a demand story. Demand lifted both by the same amount. The gap was already there and this week only lit it up. It is a map. Kennedy sits inside a cluster: Newark, Boston, Philadelphia, all inside a night's drive, all able to take a maindeck and a truck the same morning. O'Hare sits in the middle of a catchment with nothing comparable within several hundred miles. A shipper in the Midwest who needs a pallet off a plane and onto a dock has one door, and everybody in the market knows there is one door. The premium at O'Hare is not being charged for the flight. It is being charged for the absence of a second option.
The part that gets lost in rate talk is what actually happens to the cargo. A high-yield commodity entering a lane does not raise the price for everyone and then everyone pays. It takes the space, and whatever cannot pay gets left on the ground. AI hardware and the data-centre build-out behind it can absorb $2.20 without blinking, because freight on a rack that costs six figures is a rounding error. General retail cannot. So the apparel consolidation that used to leave on Wednesday does not get a higher quote. It gets a later flight, or a split shipment, or a call on Thursday saying the space is gone. The box does not get expensive. The box gets stranded.
Who feels it first depends on what they bought. A shipper holding a direct allotment feels it as an allotment cut, which is quieter and worse than a price rise, because you find out when the booking is rejected rather than when you build the budget. A mid-size shipper buying spot feels the price, then a week later the cut-off moving earlier. A small shipper feels neither, because a small shipper is not quoted a number at all, just told there is nothing on Wednesday. Meanwhile anyone whose DC is in California is having the opposite month and should be using it. Direction matters as well: this is westbound, Europe into the US. Eastbound is a separate conversation and nothing in these four numbers speaks to it.
The order in which this reaches you is fairly fixed. Space tightens first, and it shows up as a cut-off time rather than a rate. Give it two to four weeks from a week like this one and your Wednesday departure becomes a Tuesday cut-off. The dock feels it five to ten days behind the flight, which is roughly how long a surge takes to clear and reach a receiving door. Your customer feels it two to four weeks after that, when a shelf date slips. Nobody feels anything on the day the rate prints, which is why rate-watching makes a poor early-warning system.
Analysts quoted on this story say AI-hardware demand and lost Middle East capacity could pull the peak forward into September. The direction is right and the mechanism is wrong, and the difference matters if you are the one booking space. An early peak does not add a month of demand. It stacks two demand curves onto the same four weeks. September was already the month when Q4 retail replenishment hits the same docks. Moving AI hardware into that window creates no capacity, it doubles the claim on the assets that cannot be rescheduled: dock doors, cargo-terminal lanes, and the two or three drivers per gateway who actually know the airport run. An aircraft can be re-timed inside a week. A door and a badged driver take six to ten weeks to add, and nobody adds them for a peak they expect to be over by October.
The second half of that point is the one I would act on this week. The west coast going the other way is not a pricing quirk, it is a statement that there are doors and trucks free at LAX and SFO right now. That is not a rate to chase, it is an option to buy, and it is worth buying only for cargo whose final destination can genuinely be served from the west. Put that cargo on a truck and drive it east for three days and you have spent the option on the road.
Run the trade on paper before anyone books it. Assume ten tonnes from Europe to a Midwest DC, and assume the choice is O'Hare at $2.20 against Kennedy at $1.58. The truck figure is my assumption, not a number from the source: a ten-tonne partial from New York to Chicago runs in the low thousands of dollars, plus one extra handoff and two or three days. On the air leg the difference is $6,200. After the truck and the handling you keep maybe a third of it, and you have spent a day you may not have. That is the whole trade, thin money against a real day. It pays for cargo with no deadline and a DC that can take it on a Friday. It does not pay for anything with a shelf date, which is most of what flies.
The expensive mistake in a week like this is not paying the premium. It is paying the premium and still missing the door. Keep everything flowing through O'Hare because that is where the DC is, and you do not pay 17% once. You pay it for the whole peak, on cargo that never needed to fly into that airport. The failure that actually hurts is a missed receiving appointment: the truck arrives, the door has no slot, the load sits, and the put-away slips five days into a shelf date that was tight already. That bill is larger than the freight delta and it never appears in a rate table.
Tomorrow morning, pull ninety days of arrivals into O'Hare and count how many were genuinely destined inside a day's drive of it. I would bet the number is lower than the booking pattern suggests, and every one of those was a pallet paying for a map it did not need.
↑ Back to the key points