Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- By 15 September 2026, list every string you are booked on through the fourth quarter and mark which ones have a second load port inside the same rotation; anything without one goes on a weekly watch list.
- Hold one sailing interval of safety stock, seven days at your normal run rate, at destination for every SKU with cover under three weeks, and release it the moment the vessel's actual departure is confirmed.
- For each affected sailing, get the actual departure date from the carrier in writing within 48 hours and re-book rather than wait; a rolled booking costs a week, and a speculative wait costs the same week with the option of losing it twice.
- Re-quote lead time to customers as a range tied to a named sailing rather than a fixed number, and re-check it at the eight-week mark when the arrival bunch reaches your discharge port.
Congestion is tying up roughly 5% of global deep-sea container capacity, equal to about 1.7 million TEU, with waits acute in Asia, Northern Europe and the Middle East. Shanghai faces up to 9 days of vessel waiting after typhoon shutdowns, Ningbo about 4 days, and Panama Canal queues reach 10 days. Shippers should build extra lead time into Q3 plans.
The Analysis
There is a number going around this week that sounds precise and is almost useless: about 5% of global deep-sea container capacity is sitting in port queues, roughly 1.7 million TEU. The details attached to it are these. Shanghai is showing waits of up to nine days after typhoon shutdowns, Ningbo about four days, and the Panama Canal queue is running to ten. The advice being handed to shippers is to build extra lead time into third-quarter plans.
Divide 1.7 million by 5% and you get about 34 million TEU, the implicit size of the deep-sea fleet this figure is drawn from. That denominator is the part worth keeping. Five percent of a fleet is not a rounding error; it is a fleet, parked, with crews aboard and fuel in the tanks, unavailable to anyone.
And the advice is not wrong. It is unsized. Nobody can act on more lead time the way nobody can act on driving more carefully, so in practice it gets nodded at and then ignored.
A ship at anchor is capacity that exists and cannot be booked. That is the whole mechanism, and it is worth stating plainly because it is usually described as a delay rather than as a withdrawal. When 1.7 million TEU sits in queues, the market loses 1.7 million TEU of supply for as long as it sits, and the loss is not shared evenly. It lands on the strings whose rotations touch the congested ports, while everyone else keeps bidding for the same slots at the same price. For about a decade the industry's shock absorber was the ship: slow steaming, idling, blank sailings. The absorber has moved ashore. Buffer is now consumed by ports rather than by ships, and a queue costs the terminal a small fraction of what it costs the cargo.
The five percent is an average over three phenomena with three different time constants, and averaging them is not conservative, it is destructive. A typhoon is days: shut, wait, reopen. Northern European congestion is weeks: berths, labor, inland evacuation. Panama is months to years, because the constraint there is water. Put all three into one number and you know how much capacity is parked and nothing at all about whether it is coming back. A pulse you schedule around. A structural constraint you route around. Those are different budgets, different owners inside your company, and different conversations with your carrier.
Nine days at Shanghai matters more than nine days' worth, because a weekly string has a seven-day cadence. Once the wait exceeds the interval, part of the book does not wait nine days. It waits nine days plus a full sailing, which is up to sixteen. Then the port reopens, every ship that was held arrives inside the same window, berth windows compress, departures bunch, and the bunching rides the rotation for three to four weeks after Shanghai is clear. Congestion is measured where the queue is and delivered where the ship arrives, and those are rarely the same port. The person loading in Asia and the person taking delivery in Europe are watching one event from two ends of a delay that neither of them controls.
Scale decides who absorbs this, and the split is not what people assume. A large buyer with fixed weekly allocation and its own equipment pool barely notices, because they bought allocation rather than space, and allocation survives a queue. A mid-sized importer buying on spot loses the sailing, and the loss is binary rather than proportional: one missed sailing is a week of stock, not five percent of a week. A small shipper moving less-than-containerload never sees the queue at all, only the consolidation cut-off moving, which is a different clock on a different calendar. Direction matters too. Exporters feel it at the load port as a rolled booking and a missed cut-off. Importers feel it at discharge as dwell time and as a storage clock that starts when the box hits the ground, whether or not anyone can collect it that day.
The sequence runs roughly like this. The shutdown is week zero. Rolling and re-booking is weeks one and two. The downstream arrival bunch is weeks three to five, which is when discharge ports fill up with cargo that was already late when it left. Dwell peaks at weeks six to eight, and normalisation lands somewhere around weeks nine to twelve. If your plan is a quarterly plan, the window you actually have to protect is weeks three through eight, and by then the typhoon is a memory and everyone has stopped looking.
Here is what the coverage has not said. Extra lead time is not a lever you own. Lead time is a promise you make to a customer; the lever you own is the buffer stock standing behind that promise. Quote nine extra days, watch the port recover in five, and you have handed away four days of competitiveness for nothing. Hold one sailing interval of safety stock at destination instead, and you can release it the week the schedule normalises. The second half of this is the part that decides whether any of it applies to you.
Five percent is a fleet share, and no shipper buys fleet share. You buy slots on a specific string, and a string's exposure is binary. If its rotation touches the affected port, you are in the queue at one hundred percent. If it does not, you are untouched at zero. The average of those two states is five percent, which is about as useful to you as the average shoe size is to a shoemaker.
Run it the other way and the same number changes owner. Had this typhoon arrived two weeks later, after the peak-season cut-off, the same 5% would have hit cargo that could not be re-scheduled, and the cost would have been carried by shippers as air freight or as lost sales rather than by carriers as a late sailing. Same queue, same tonnage, different payer. That is worth keeping in mind when someone tells you congestion is a carrier problem. The capacity loss is theirs. The timing risk is yours, and it is priced into the promise you already made.
A small account with the assumptions marked. Take a book of 40 FEU a week into Europe, which is 160 FEU a month. Assume, and this is my assumption rather than a figure from the report, that three of the four monthly sailings touch an affected load port, that each affected sailing slips seven days on average, and that a quarter of the boxes on those sailings roll to the next vessel for another seven. That is 120 FEU carrying one extra week, or 840 FEU-days, plus 30 FEU carrying two extra weeks, or 420 more. Call it 1,260 FEU-days a month, then multiply by your own carrying cost per FEU-day. I will not put a number on that for you, because the number that matters is the one in your books. The point is the shape of the exposure. It is a stock, not a rate, and it grows with the length of the delay and with the number of sailings you have concentrated on a single load port.
Ports are the part of this network that everybody draws as a dot and nobody draws as a constraint. The dots have been multiplying for thirty years. The water has not. When five percent of the fleet is parked, the network is telling you where its slack went, and it is telling you in the only language it has, which is time. The shippers who come out of quarters like this intact are not the ones who predicted the queue. They are the ones who knew, before the queue formed, which of their strings had a second path and which did not.
↑ Back to the key points