Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- This week, reprice every booking due to sail before 20 August against the four-day Yangshan-Waigaoqiao gap, and plan on that gap closing within seven to ten days.
- By 15 August, get the per-voyage intake ceiling in writing from each carrier on the Shekou and Yantian strings, and build rolling assumptions on that figure rather than on the schedule.
- Spread gate-in across five days instead of two in any week where your forwarder's daily allocation sits within 20 percent of your peak day; the cut-off margin is what you protect, not the trucking cost.
- For import cargo arriving 25 August to 10 September, reserve warehouse and drayage at 1.2 times normal weekly volume, assuming the restart compresses three weeks into five days.
- Track empty repatriation weekly; if empties fall below your normal three-week cover, treat it as an export capacity event rather than a yard problem.
Shanghai handled a record 203,881 TEU in a single day on 1 August, yet congestion persists: waits at Yangshan Phase 3 reached about 12 days and Waigaoqiao eight. Yantian has capped daily export bookings at 9,000 boxes under an ETB-7 gate rule, while Shekou added a 120% per-voyage intake ceiling from 3 August as Typhoon Baihaitun approaches.
The Analysis
Shanghai moved 203,881 TEU in a single day on 1 August, a record, and in the same release reported berth waits of about twelve days at Yangshan Phase 3 and eight at Waigaoqiao. Most readers will take that as a triumph followed by a caveat. They are the same sentence read twice.
A port does not print a daily record because it has room to spare. It prints one because the berths are at the physical limit while a stock of waiting ships feeds the quay without pause. The record measures how fast a queue is being drawn down, not the size of the network underneath it. When the queue empties, the record does not repeat.
So let me spread the map. Three nodes moved inside ten days, and the throttles around them are not three answers to three problems.
Shanghai's deep-water capacity sits offshore at Yangshan; a box landing there still has to cross water before it touches a mainland yard, a truck, a rail ramp. Waigaoqiao sits on the river, closer to the city and to the export base that feeds it. Two nodes, one authority, four days between their queues in the first week of August. Four days is the most tradeable number here, and it will not survive the month.
The rest of the week is the same mechanism. Yantian has run an ETB-7 gate rule since 27 July with a booking ceiling of 9,000 boxes a day. Shekou added a 120 percent per-voyage intake ceiling from 3 August. Baihaitun is on approach. Four items, one instrument, four points on the same pipe.
Twelve days of waiting means ships arrive faster than the quay can serve them; the terminal answers the only way it can, feeding the cranes from the anchorage at full rate, and that is exactly what produces a 203,881 TEU day. The record is a symptom of the queue, not a rebuttal of it. Then the queue has to go somewhere. Cap the gate at 9,000 and the wait moves from water to road, becoming a booking problem instead of an arrival problem. Cap intake at 120 percent and the boxes that do not fit roll to the next vessel as new demand for the same gate slots.
The typhoon is the fourth term and the most misread. A scheduled stop ahead of a peak is not a pause, it is a compression: everything that would have moved in the stopped days arrives in the days after, into a system already throttled. The ceilings published before the storm are a decision about how much of that wave each vessel may carry, and somebody made it deliberately. Had they not been set, the same backlog would have landed as one spike at the yard and importers and exporters would have been hit in the same week. The cap turns one sharp shock into a longer, shallower one, paid in time instead of in a collapse. It is why the restart will look orderly and feel slow.
Who hurts first depends on place in the allocation order, not on volume. A small exporter booking late at Yantian, with no contract allocation, meets the 9,000 box ceiling as a wall: their booking is the one that does not clear. A large buyer meets the same ceiling as an inconvenience, because the forwarder holds slots and part of the flow moves to Waigaoqiao and buys back four days. In between sits the shipper moving two to five hundred boxes a month: able to switch on paper, but with paperwork, trucking and customs built around one gate, so a week of administration to save four days is a wash that feels like a loss.
Importers are in the same queue and rarely notice. The wait has no direction. Twelve days holds the discharge of import boxes exactly as it holds the loading of export boxes, and the import boxes hold the empties, and the empties are the export supply. That chain runs three to four weeks behind the headline and almost never shares a paragraph with the record.
The calendar matters more than the magnitude. The berth wait is already inside this week's ETA, a fact about ships at anchor, not a forecast. The gate cap surfaces two to four days later, when a driver is turned away. The booking cap surfaces at cut-off, five to seven days before sailing, so the miss is visible a week before the vessel leaves and ten days before anyone downstream believes it. For a European importer the date to circle is not the typhoon but the arrival week thirty to thirty-five days later, when three weeks of boxes try to land in five and the constraint becomes the warehouse.
One coupling I have not seen put in a single place changes what you do this week. A per-voyage intake ceiling is a carrier instrument, not a terminal one: it makes the carrier the shock absorber, and boxes above 120 percent come back as fresh bookings. So Shekou's ceiling does not reduce demand for gate slots, it relocates part of it to Yantian, which is already capped at 9,000. Two announcements that read as separate capacity measures are one queue being pushed uphill. The same reporting has missed that the four-day gap is self-liquidating: nothing keeps an arbitrage open in a port system for more than a week, because the alternate node has no spare berth, only a shorter queue today.
A small number, with the assumptions flagged as mine and not from the article. Take an exporter moving 500 FEU a month through Shanghai with six days of added delay. That is 3,000 FEU-days sitting in the system instead of a customer's warehouse; at an assumed 1.50 dollars per FEU per day, about 4,500 dollars a month that never appears on a freight invoice. If the forwarder holds a sixty-box daily gate allocation, 120 FEU now spreads across five days instead of two, and the cut-off margin goes from two days to zero. The rate did not move. The working capital did.
The trap is expensive and quiet: reading a throughput record as evidence of capability. Records printed during a queue are drawdown numbers. The other trap is treating the typhoon as the event; the storm is weather, the throttles are policy, and policy decides how the restart is sequenced. A port that caps intake before a closure has already shown you the next three weeks.
Ten years from now this reads as a footnote, one of those summers when the network was asked to swallow more than its berths and gates could take at once, and answered by moving the wait around until it found somewhere to sit. The map is not changing. Where the map puts its patience is. For an exporter the useful question is narrower: which of my flows can buy four days this week, and which will still be standing in the same queue in September.
↑ Back to the key points