July Sets All-Time U.S. Import Record at 2.47M TEU Before Forecast Turns Down

The Analysis

Spread the map out and start on the American side this time, with San Pedro Bay. The latest MARAD and Marine Exchange counts put the vessel queue at Los Angeles and Long Beach at zero. Not low, not improved, zero. No ship waiting for a berth at the two biggest container gateways in the United States. Now hold that picture against the other number from the same week: the National Retail Federation and Hackett Associates put July 2026 at 2.47 million TEU across the thirteen major US container ports they track, an all-time monthly record, ahead of the 2.4 million TEU set in May 2022. A record month with an empty anchorage. Ten years ago those two facts could not have lived in the same week.

They can now, and that is the interesting part. An empty anchorage during a record month is not luck. Berths, yards, cranes, gate systems and the rail spurs behind them are decade assets, and the ones that were built, deepened and re-equipped over the last ten years are now quietly doing their job on the busiest month the country has ever had. Nobody sends out a press release about a berth that was ready. You only find out on a month like this one, when the volume peaks and the roadstead stays empty. The constraint has moved. The queue data tells you it is not the berth anymore, and that means the pressure is showing up somewhere inland instead: in yards, on chassis, in warehouse labour and in appointment windows.

Then read the second half of the same tracker, because it inverts. August is projected at 2.22 million TEU, down 4.5 percent year on year. September and October are both projected at 1.99 million. November is projected at 1.92 million. From July's 2.47 million to November's 1.92 million is a drop of 550,000 TEU a month, roughly 22 percent, in four months. The tracker itself uses the word front-loaded for what July absorbed. If you run a planning process that takes the record month as its baseline, you are about to spend four months paying for capacity you do not use, and then spend the following year wondering why your working capital never came back.

Start with what the peak actually was. 2.47 million TEU against 2.4 million in May 2022 is an increase of 70,000 TEU, about 2.9 percent, over four years. That is the record, and it is a thin one. A record that thin is not evidence of a demand boom; it is evidence that the system is running near the top of what it can physically push through thirteen gateways in thirty-one days. When a record is set by three percent, the useful question is not how high demand went but where the ceiling is, and the empty anchorage says the ceiling moved up rather than the cargo piling against it. This matters for how you read the rest of the year, because a thin record set against a raised ceiling behaves differently from a spike set against a hard wall.

Now the descent, which is the part with money in it. The forecast runs 2.22 in August, 1.99 in September, 1.99 in October and 1.92 in November. Assume the tracker is right, which is an assumption and should be written down as one. The four-month average from August to November is (2.22 + 1.99 + 1.99 + 1.92) / 4, which is 8.12 / 4, or 2.03 million TEU. Against July's 2.47, that is a fall of 17.8 percent. Assume your own import volume through the tracked ports moves with the index, and assume you were running 900 TEU a month in July. Then August is 900 x (2.22 / 2.47), which is about 809 TEU. September and October are 900 x (1.99 / 2.47), about 725 TEU each. November is 900 x (1.92 / 2.47), about 700 TEU. Your July number is not your run rate. It is your high water mark, and everything you signed against it is now mispriced.

Here is the bill. Assume you contracted warehouse space, handling capacity and drayage capacity off the July peak, which is what most people do because July is the month they were standing in when the conversation happened. Assume a flat twelve-month commitment at 900 TEU a month at 120 dollars a box all in for storage and in-and-out handling. Assume you pay for the committed volume whether or not you use it. Under a flat commitment the August to November cost is four months at 900 TEU at 120 dollars, which is 432,000 dollars. Actually used in those four months: 809 + 725 + 725 + 700, which is 2,959 TEU. So you paid for 3,600 TEU of capacity and moved 2,959. The 641 TEU of unused commitment is 76,920 dollars of pure waste across four months, money spent on air in a warehouse.

Now price the alternative. Assume you go back and renegotiate into a stepped commitment: 809, 725, 725 and 700 across those four months, with a ten percent swing allowed either way. Assume the warehouse and drayage providers charge eight percent more per box for the privilege of a stepped deal, so 129.60 dollars instead of 120. The stepped cost is 2,959 TEU x 129.60, which is about 383,486 dollars, against 432,000 flat. That is a saving near 48,500 dollars, roughly 11 percent of the four-month bill. Solve for the breakeven and the threshold appears: the stepped deal stops being cheaper once the premium exceeds about 20 percent.

At a 21.7 percent premium the per-box price reaches 146 dollars and the stepped total hits 432,000, exactly flat. So when your 3PL quotes you a stepped commitment with a surcharge, the question is not whether the surcharge sounds fair. It is whether it clears 20 percent. Below that, you take the stepped deal and you take it now, while the forecast is on your side of the negotiation rather than after the volumes have already fallen and the provider has no reason to move.

The larger question underneath all of this is what the descent actually is, and you will not know the answer until September's number is printed. Assume the August to November average of 2.03 million represents the underlying level and that July's excess was borrowed from the months after it. On that assumption July ran 440,000 TEU above the underlying level, and if that volume was pulled forward evenly out of the following four months, each of them is running about 110,000 TEU below where it would otherwise have been, meaning the underlying level is nearer 2.14 million than 2.03. That gives you a clean test with a clean threshold. If September prints below 1.88 million, the shortfall is bigger than the pull-forward can explain and you are looking at genuine demand weakness: re-base the year, cut the buy, and do not rebuild inventory ahead of a recovery that the tracker is not forecasting.

If September prints above 2.10 million, the pull-forward reading was too pessimistic and this is a timing effect: resume normal replenishment and treat July as a borrowed month rather than a new floor. Anything between those two numbers means keep the plan and wait for October. Write the threshold down now, in August, while you are calm, because the temptation in September will be to explain whatever number appears.

There is a second consequence of the front-loaded pattern that gets less attention than the volume drop, and it is the one that costs importers real money. If July absorbed a chunk of what would normally have shipped in September and October, then the goods are already in the country. They are in a warehouse, or in a yard, or on a chassis, and they will be sold through over the months when the tracker says the ports go quiet. Your inbound pipeline and your outbound sales curve have come out of phase.

Assume a three to five week mismatch between goods arriving in July and goods selling through in September and October, and assume your carrying cost is 18 percent a year on landed value. On 900 TEU at 40,000 dollars a box that is 36 million dollars of landed value, and a four-week phase mismatch carries about 1.24 million dollars of annualised carrying cost. That number is not a rounding error, and it is entirely a consequence of timing rather than of demand.

One place where a front-loaded year quietly costs you money is the annual volume commitment. Assume you negotiate a twelve-month minimum quantity commitment with a carrier or a 3PL, and assume the negotiation happens in the autumn, when July's record is the freshest number in the room and the forecast descent is a table nobody reads aloud. On that basis you commit to a volume derived from a peak month, and you spend the following year either short of the commitment and paying deadfreight, or hitting the commitment by shipping goods you did not need in months you did not want them. The fix is unglamorous and it works: commit to a quarterly profile rather than an annual flat number, with the tracker's own monthly figures as the schedule.

Assume the quarterly profile costs you a slightly worse rate band than a flat annual commitment would. Ask for both quotes, in writing, and take the quarterly one unless the flat one is cheaper by more than the carrying cost of the inventory you would otherwise be forced to move. That is a comparison you can do on one page, and it is the difference between a commitment that describes your business and one that describes your busiest month.

The other thing to insist on is the port-level split rather than the national total. The tracker is a thirteen-port number, and a national record tells you nothing about which of those gateways took the volume or which of them will give it back. Assume the descent is not evenly distributed: some of those ports will hold their July level into the autumn, and others will fall faster than the 17.8 percent average implies. Your exposure is to the two or three gateways you actually use, not to the aggregate. Ask your forwarder for the monthly volumes at your own gateway for the last twelve months alongside the national figure, calculate the ratio between them, and apply that ratio to the tracker's monthly forecast rather than applying the national percentage. If your gateway has been running above the national trend, the national descent will understate your own, and you will under-plan the cover you need. If it has been running below, you have room you did not know you had.

So what do you actually do, and who owns it. This week, before Friday 7 August, the demand planning owner re-bases the second half forecast off 1.99 million rather than off July's 2.47, and circulates the revised monthly volumes to procurement, warehousing and finance on the same day. Target: every downstream plan in the business uses the tracker's own numbers by close of business Friday, and no plan, spreadsheet or headcount request still carries July as the baseline.

By Friday 14 August, the logistics lead pulls every warehouse, handling and drayage commitment signed on or after 1 June and lists the committed monthly volume against the re-based forecast. Anything with a flat commitment above 800 TEU on a lane now forecast below 750 goes back for renegotiation, with the 20 percent premium threshold as the walk-away rule. Owner: logistics, with finance sign-off on the revised commitment. Target: at least 80 percent of the identified gap closed or re-priced before the September volume lands.

By Friday 21 August, the procurement owner freezes new purchase orders that would land in October and November at July volumes, and re-phases them against the tracker's 1.99 and 1.92. Trigger: any SKU whose on-hand plus in-transit cover exceeds twelve weeks gets no new order until cover falls below ten weeks. This is the single most important item on the list, because inventory ordered now against a peak month is inventory you will still be paying for in February.

By Friday 28 August, put the September test into the monthly review calendar as a standing item with a named owner and a pre-agreed decision. Below 1.88 million triggers the demand-weakness plan: cut the buy by a further 10 percent, hold warehouse commitments at the renegotiated step, and stop all seasonal builds. Above 2.10 million triggers the timing plan: resume replenishment at the underlying rate and release the seasonal build. Between the two, hold. The value here is not the accuracy of the thresholds, it is that the decision is made before the number arrives rather than argued about afterwards.

Then there is the empty anchorage, and it deserves its own action rather than a footnote. Queue at zero at Los Angeles and Long Beach while the country sets an all-time import record means the berth is not your constraint this year, and every dollar you were going to spend hedging berth congestion at San Pedro Bay should be redeployed inland. By Monday 31 August, the operations owner should have three things confirmed for the fourth quarter: appointment availability at your primary and overflow warehouses, weekend and second-shift receiving capacity, and chassis and drayage availability for the weeks when the July arrivals and the September arrivals overlap in the same yards. Target: documented surge capacity of at least 15 percent above your peak week, pre-agreed in writing, before the seasonal build starts landing.

It is worth being explicit about why I treat the descending part of this forecast as the planning baseline rather than the record. A single record month is one observation, and one observation is the weakest possible basis for a commitment. Four consecutive projected months are a shape, and a shape is worth more than a point. Assume the tracker is directionally right and month-level wrong, which is the realistic case: September may come in at 2.05 rather than 1.99, and November at 1.98 rather than 1.92. Run the arithmetic again on that softer set and the picture does not change.

August to November averages (2.22 + 2.05 + 2.05 + 1.98) / 4, which is 8.30 / 4, or 2.075 million, still about 16 percent below July. The flat commitment still overpays, the stepped commitment still wins below a 20 percent premium, and the September test still works at the same thresholds. That is the test of a plan: if it survives being wrong by three percent a month, you can execute it without waiting for certainty, and certainty is not coming.

On alternatives, and on where this particular forecast can hurt you. Shifting gateway is the standing temptation when volumes look lumpy, and this year the argument for it is weaker than usual for one specific reason: with no queue at the San Pedro Bay ports, the thing you would be buying by shifting east or south is berth availability you already have. If you move volume to a different coast now, do it for inland freight economics or for risk spread, not to escape a congestion that is not there. Shifting to a faster mode is the expensive hedge and, on a declining forecast, the one that makes least sense: paying air freight to accelerate goods into a month the tracker says will be 20 percent lighter is buying speed in the wrong direction.

Shifting inventory policy is the lever that actually fits this forecast, and it cuts both ways: if you are long, the falling volumes are an opportunity to work the excess off without markdowns, and if you are short, the quiet months are the cheapest capacity you will get all year to rebuild cover before the next peak.

The traps, which are mostly about reading the wrong reference point. The first is contracting off a record month, which is what July was and which is why the flat commitment above cost 76,920 dollars across four months. The second is reading the year-on-year figure alone: August is projected down 4.5 percent against last year, which sounds mild, but it is down 10 percent against July, and the month-over-month drop is the one that hits your warehouse and your cash. The third is treating a forecast as a fact. The Global Port Tracker is a projection, and the September and October figures can move; plan against them, keep the thresholds, and re-run the arithmetic when the actual prints.

The fourth is assuming an empty anchorage means an empty supply chain. Ships not waiting at the berth are discharging, and everything they discharge is now competing for yard space, chassis, appointments and dock doors inland. The fifth is the one that bites importers at year end: a front-loaded year distorts the comparison base. If July borrowed from September and October, then next July will be measured against a distorted month, and anyone setting targets off it will set them wrong. Note the distortion now, in writing, while you still remember which month borrowed from which.

I have been watching these thirteen ports long enough to remember when a record month meant ships stacked up to the horizon off San Pedro and the photograph ran on the front page for a week. July set a record and the anchorage was empty. That is not a small thing, and it is not an accident; it is what ten years of concrete, dredging, cranes and gate systems looks like when it finally shows up in the numbers. What has actually changed is the net. The map I would hold in my head this autumn is not the one with a queue on it. It is the one where the water is clear and the pressure has moved inland, onto yards and appointments and the working capital tied up in goods that arrived in July and will not sell until October. Plan against the descent, not against the record. The record is already behind you, and the goods it brought are already in the building.

↑ Back to the key points

— By Tomás Ferreira