Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- By 10 August, convert the September forecast into pallet positions and net it against the positions you hold in writing; cover any gap above 15% before the end of the month rather than in the first week of September.
- Rank the September air list by value per kilo and move everything under 40 dollars per chargeable kilo to ocean this week; that is a working threshold of mine, not a figure from the report, and it should be adjusted to your own margin.
- Tender 72 hours before cutoff instead of 24, so a late truck costs a rebuild instead of a lost position, and tell the forwarder that is the standard you are working to.
- Split September's air volume across two gateways so that one cutoff moving earlier cannot take the whole month with it.
- Write the weekly position count into the block agreement as a specific number, with a stated remedy when it is not delivered, before next quarter's rates are signed.
IATA data show global air cargo demand rose 8.5% year on year in June while capacity added just 4.4%, tightening the market ahead of the traditional July-October peak. North America led with 13.1% growth and Asia-North America was the fastest-growing lane. DHL has confirmed 13 Mammoth-converted Boeing 777-200 freighters, four entering service this year.
The Analysis
The peak used to be a date you could put in a planner. Volume built through August, tightened in September, and the last two weeks of October were the wall everybody ran into. That calendar is now a quarter early: June demand up 8.5% against capacity up 4.4%, North America leading at 13.1%, and Asia to North America the fastest lane of the bunch.
The order book is answering, just not on this timeline. DHL has confirmed 13 Mammoth-converted Boeing 777-200 freighters, four of which enter service this year.
What follows is not a demand story. It is a story about a supply curve that arrives in steps and a demand curve that arrives as a ramp, and about the shipper standing where the two cross in the last week of September.
Put the two percentages on the same base and something ugly shows up. If the market was running at 90% load factor before June, my assumption and not a number in the report, then demand at 108.5 against capacity at 104.4 leaves about 93.5%. Three and a half points of slack gone in a month, across a global network.
What slack is for matters here. At 90%, there is room for a freighter that goes technical, a two-day closure, a charter somebody else bought, a night when the ramp is two loaders short. At 93.5%, none of those is absorbed. Each one is passed down to whoever is last in the queue. A market at 93.5% does not clear on price. It clears on allocation, and allocation is decided by who booked first, not by who bid highest.
The physical version of that sits on the ground, not in the air. A pallet position is an aluminium sheet, a net, a crew, and a truck that arrives inside a window. Once load factor passes the low nineties, the constraint stops being whether space exists and becomes whether the ramp can build the pallet before cutoff. A freighter that lands at two in the morning needs its pallets built by ten the night before, weighed, screened, documented and staged. One late truck and the position flies empty while the cargo waits a day. That empty position is the most expensive seat in the chain, and it explains why forwarders get unpleasant about cutoff in a tight market. They are not defending a rule. They are defending a hole in the aircraft.
The other physical input is the unit load device pool. Pallets and containers are steel, the stock at any station is finite, and they come back on the aircraft they left on. When utilisation rises the ULD turns faster, which means a delay at the far end turns into a shortage at the near end within days rather than weeks.
Four converted aircraft this year is a step, not a ramp. A conversion arrives on one date and adds positions to one lane, and between the order and the delivery there is a queue of its own: a conversion slot, a crew that needs a type rating, a maintenance check, a loading system. That queue does not compress because demand is strong. The aircraft meant to fly in September tends to fly in November, which lands it after the peak it was ordered for and in front of the next one. Thirteen airframes is a serious commitment and a 2027 answer.
There is another reason not to read an order book as capacity on your lane. An operator decides where a converted freighter flies in the week before delivery, and that decision follows revenue per position. The lane that pays best that week gets the aircraft. Your lane is one bid among several, and if your cargo is light, your bid is weak.
Value density is what decides who flies in a tight market, and it appears in none of the headlines. A position earns what the cargo on it is worth per kilo. AI hardware and high-value electronics are dense and expensive, so they outbid light assembled goods for the same position, every time, without anyone making a decision about it. An exporter of bulky parts is not competing with another exporter of bulky parts. It is competing with a product worth ten times more per kilo, and it loses that auction in September no matter how early it calls.
The other asymmetry is directional. North America grew 13.1% while the world grew 8.5%, and Asia to North America is the fastest lane. The return leg is not the fastest lane and does not carry the same premium. An importer bringing cargo into Asia should not be paying a peak surcharge on a lane where no peak is happening, yet most will, because the quote arrives as a round-trip blend.
The calendar is the one thing nobody moved. Peak was pulled forward into July through October, but the contracts covering it were signed under the old assumption that October was the pinch. Positions for the last week of September were committed in the first week of July. The shortage a shipper discovers in the third week of August is not an August failure. It is a July failure arriving late.
What the coverage misses is that the peak did not get bigger, it got steeper. The same tonnage that used to spread across the season now has to clear inside a narrower window, against build-up capacity hired on a weekly roster. A bigger peak fails on price, with everybody paying more and everybody seeing it. A steeper peak fails on cutoffs, with cargo simply not flying while the price looks almost normal. That is the harder failure to notice, because no single number jumps.
If the peak had stayed in October, the July booking window would have been adequate and the load factor would have topped out at a level the ramp could build. Pulling it forward compresses the tonnage into a window where the labour has not been hired yet. That is why the pain arrives as missed cutoffs rather than as one price shock.
The trap is reading 13 new freighters as 13 aircraft of space on your lane. They are 13 decisions made close to delivery, on whichever lane pays best that week. Sort the September list by value per kilo and draw a line where the air budget runs out. Everything below the line goes by sea, and it has to be booked this week. In September the line gets drawn for you, by somebody else's booking.
↑ Back to the key points