Six major Chinese container ports flagged 'heavily disrupted' after twin typhoons

The Analysis

Put the whole eastern seaboard on the table this time, not one bay: the arc that runs from Qingdao in the north, down past Shanghai and Ningbo, down the Fujian coast to Xiamen, and on south to the Pearl River Delta where Shenzhen and Guangzhou sit almost shoulder to shoulder. On 28 July 2026, a customer advisory marked all six of those hubs as heavily disrupted at the same moment. This is not one port having a bad week. It is the spine of Chinese container exports having a bad week together, and if you have watched this coast for a living, that sentence is the news.

The detail is still worth reading, because it tells you how deep the disruption ran. Shanghai's seven-day average delay reached 4.56 days. At Shanghai Yangshan, non-Gemini services were facing berthing waits of 144 hours; at Ningbo, 48 hours; at Qingdao, 72 hours. Industry estimates put the cumulative capacity affected at close to two million TEU. Mainline services were cancelling or reordering their Shanghai and Ningbo calls, and Asia-Latin America strings were transshipping through Busan and Singapore. And as if on cue, Typhoon No. 13 had formed, adding further schedule risk.

Ten years ago this coast did not move as one body in quite this way. It had more, smaller gateways, and when a weather system swept through, it took a slice rather than the spine. What has changed is not the weather. What has changed is that the coast now concentrates more of its export capacity in fewer, larger hubs, and a concentrated network is faster in fair weather and more exposed when the weather comes for all of it at once.

Start with the shape of the network, because the shape is doing most of the work here. Ten years ago, a weather system that crossed eastern China took a share of capacity, a few berths here, a few there, and the rest of the coast kept lifting. The system has since consolidated. The biggest gateways have grown bigger and fewer, and the strings that matter have grown wider, so that when weather sweeps the arc from Qingdao to Guangzhou it no longer clips the edges of the system, it reaches across the middle of it. That is not a failure of any single port. It is the arithmetic of concentration: the same fleet, the same schedules, the same weekly capacity now pass through a smaller number of larger doors, and a system with fewer doors feels a coast-wide event more sharply than one that had many.

It helps to stop thinking of these as six ports and start thinking of them as one instrument with six keys. Qingdao, Shanghai, Ningbo, Xiamen, Shenzhen and Guangzhou are not independent options you can swap at will. They are a division of labour, with the Yangtze delta, Shanghai and Ningbo together, carrying the heaviest mainline load, the north at Qingdao handling its own hinterland and feeder flows, and the south at Xiamen, Shenzhen and Guangzhou feeding the export engines of Fujian and the Pearl River Delta. When all six are disrupted at once, the network has not lost six alternatives. It has lost, for a period, the ability to choose at all, because the fallback you would normally use is standing in the same weather you are running from.

Take the delay figures and read them as network language rather than as a countdown. Shanghai's seven-day average of 4.56 days says the Yangtze delta hub is running roughly a working week behind where it should be, which means every downstream plan built on an on-time Shanghai call is now built on air. The 144-hour wait for non-Gemini services at Yangshan is a different kind of fact again. It is the clearest signal in the whole advisory about priority, because a wait that long tells you which ships are being given the berth first and which are being asked to circle. Ningbo's 48 hours and Qingdao's 72 hours are shorter, but they are not small. They are the difference between a vessel making its next pro-forma window and missing it, and a missed window is how a delay at one port becomes a delay three ports later.

The causal chain here is a sequence, not an accident. Bavi and Hongxia struck East and South China in succession, and the advisory of 28 July is simply the moment the network admitted how much of itself was still down. Now add the formation of Typhoon No. 13. This matters more than it looks, because it means the recovery window is not clean. Hubs recover fastest when they get a stretch of undisturbed days to drain their queues. A new storm forming on the horizon compresses that window, and it does so exactly when the vessels displaced by the first two storms are trying to berth. The system is not being given the quiet it needs to heal, and that is the difference between a disruption that lasts one week and one that lasts three.

The two-million-TEU figure deserves a moment, because it is the number everyone will quote and the number most easily misread. That is capacity affected, not capacity lost. Ships do not evaporate; they wait, they reorder their calls, they reroute. The distinction matters because the two have completely different consequences. Lost capacity is a supply problem: fewer slots, higher pressure on everything that floats. Affected capacity is a time problem: the slots still exist, but they are in the wrong place at the wrong time, and the cost shows up as delay and as dwell, not as scarcity. For a shipper, that is actually the better news in the whole advisory. The boxes will move; the question is only when, and through which door. But it is also where the hidden cost lives, because time, unlike capacity, is very hard to buy back once it is lost.

Now watch what the network does when six of its hubs are down, because this is the part that shows you its intelligence. It does not wait. Mainline services cancel or reorder their Shanghai and Ningbo calls, which is the network deciding to skip a full hub rather than queue inside it. Asia-Latin America strings push their cargo through Busan and Singapore instead, which is the network routing around the damage. Both moves are the same move: when the primary hubs are unavailable, you do not stand still, you find another node and you pay it in extra time. That is what a healthy network looks like under stress. Not intact, but adaptive.

There is a chessboard quality to what happened this week, and it is worth naming even though nobody moved a piece on purpose. When the biggest hubs go dark, the network does not simply lose capacity; it redistributes it, and the distribution follows the paths that are still open. Mainline services that skip Shanghai and Ningbo are not only avoiding a queue, they are re-weighting the whole coast, pushing pressure onto the ports that remain reachable and pulling it off the ones that are not. The Asia-Latin America strings that swing through Busan and Singapore are doing the same thing at a smaller scale. Nobody designed this. It is the network finding the least-bad arrangement under constraint, the way water finds the lowest ground. But the arrangement it finds is a fact about the shape of the map, and a shape that gets used tends to get used again.

Here is what I did not see in the headlines, and it is the point I would underline for anyone planning the next three months. Everyone focused on the six ports that failed. The story that matters is the two that absorbed the spill. Busan and Singapore are the relief valves of this network, and when six Chinese hubs falter at once, those valves do not merely take up the slack, they get tested, they get paid, and they get noticed. Capacity that reroutes through a hub tends to remember the route. If Asia-Latin America strings learn this month that Busan and Singapore can carry them, some of that traffic may not fully come home when the weather clears.

That is how a hub quietly gains position: not by winning a contract, but by being the door that stayed open on the week the others closed. The delay the media counted sat at the six Chinese ports. Part of the real delay is riding, uncounted, at Busan and Singapore, because a transshipment adds dwell at the intermediate hub, and that dwell never shows up on the origin port's clock.

There is a second layer to the same point. When you reroute a string through a transshipment hub, you are not removing the delay, you are relocating it and usually enlarging it. The container now has to be discharged, staged and reloaded at an intermediate port, and every one of those steps has its own queue, its own cut-off and its own risk of the connecting vessel being full. So the schedule you publish to your customer may show the same sailing date, but the box may arrive several days later, because the extra time is hiding in a port that appears nowhere in your plan. If you only track your origin port's delay, you are tracking the visible half of the problem. The invisible half is the dwell at the valve.

The thing about a relief valve is that it only works if it is not itself under pressure, and this is where I would keep a close eye over the next fortnight. Busan and Singapore can absorb a surge of transshipment, but they are not infinite, and they too have yards, cranes and cut-offs. A network that leans hard on two valves for several weeks runs the risk of congestion simply relocating, from the ports that failed to the ports that saved them. If that happens, the delay does not disappear; it spreads, and it spreads to the very hubs everyone was relying on to stay clean. That is the quiet risk in the current picture, and it is the one most people are not watching, because the headline is still about the six Chinese ports.

None of this means the alternative hubs are better than Shanghai or Ningbo. On most days they are not, and the concentration of this coast in the Yangtze delta exists for good reasons of hinterland and depth. The point is narrower and more durable: when the primary hubs are unavailable, the network discovers that it can function without them, and a discovery like that does not fully disappear when the weather does. It settles into the market's memory as a proven detour, waiting for the next reason to be used.

Who carries the pain of a rebalance like this, and how unevenly? Mainline and alliance services, the ones with the volume and the leverage, reorder their calls first and get served first. Non-Gemini services, as the Yangshan figure shows plainly, wait behind them. Regional and feeder services, and the strings that transship rather than call directly, absorb the reroute and the extra dwell. Then the shippers behind them sort out by size again: the largest can pre-book space on the alternative routing, the smallest discover that the alternative is already full. A hub failure does not fall on the cargo that is easiest to move. It falls hardest on the cargo with the least room to manoeuvre. That is true at the port level and it is true at the customer level.

Step back to the decade view and something quietly structural is happening. A port's rank in a network is not set by how much it can handle on a good day; it is set by how reliably it can be reached on a bad one. For ten years the competition among these hubs has been a competition of scale: deeper water, bigger cranes, more throughput. Weeks like this one shift the axis of that competition, at least a little, toward resilience, which is the ability to stay open, or to reopen fast, when the weather comes for the whole coast. The hubs that handle this month well will not advertise it as a reason to route through them. They will simply be the ones whose calls were not cancelled, and over a few years, that is how a network's centre of gravity can drift without anyone announcing that it moved.

Seen over a decade, this is how a network's hierarchy really gets rewritten: not in a boardroom, but in the accumulated memory of which door stayed open. A shipper who rerouted through Busan this month and got a clean delivery will not forget it, and the next time the coast wobbles, that memory becomes a habit. Reliability, repeated, becomes preference, and preference, repeated, becomes structure. That is why a month of disruptions is never only a month of pain; it is also a slow auction over which hubs the cargo will trust, and the bid is paid in the simple currency of whether the box arrived when it was promised.

Let me put an illustration on the table with the assumptions stated, because only a handful of numbers here come from the advisory: Shanghai's 4.56-day average, the 144, 48 and 72 hours at Yangshan, Ningbo and Qingdao, the roughly two million TEU of affected capacity, and the shift of the Asia-Latin America strings through Busan and Singapore. Assume the two million TEU is spread across six ports with the Yangtze delta carrying the largest share, and assume your own book is six hundred containers moving on mainline services out of Shanghai and Ningbo. If the Yangtze hub is running 4.56 days behind, that is the baseline drift you should plan on before any reroute, call it a working week of schedule risk built into every promise you make.

Now suppose a quarter of your book, one hundred and fifty boxes, is rerouted through Busan or Singapore to protect a hard customer date. Each of those picks up extra dwell at the intermediate hub, and the only number I am willing to put on that is qualitative: it is days, not hours, and it is variable, because the connecting vessel's reliability is now part of your schedule. So the honest plan is not one arrival date. It is a range, and the width of that range is set by how much of your cargo you choose to send through a valve.

The traps in a week like this are predictable and expensive. The first is believing a reroute is free. It is not; it buys schedule certainty with extra dwell, extra handling and a new dependency on a hub you do not control. The second is rerouting everything at once: if enough of the market moves the same string the same way, Busan and Singapore fill up and the valve you were relying on starts to look like the port you left. The third is treating the two million TEU as lost capacity and over-buying space in a panic; it is affected capacity, and much of it will simply come home late, so overshooting on the spot market is paying twice for the same delay. The fourth is planning against a single number, the average, when the whole point of this week is that the spread is what hurts. Plan to the date your customer actually needs, not to the average you can compute.

And the counterfactual, because it is the boundary of everything above. Suppose Busan and Singapore had no spare capacity this month, suppose their yards were as full as Shanghai's. Then the network would have had no relief valve, the rerouted strings would have queued at the alternative hubs as well, and the two million TEU would have stopped being a time problem and started being a capacity problem. That is the condition that flips the whole picture. Which is why the single most useful thing to watch over the next few weeks is not the Chinese coast. It is the two hubs the coast is leaning on. If those hold, this remains a story about delay. If they fill, it becomes a story about what the network does when it runs out of doors altogether.

The coast will dry out; it always does. The typhoons are older than the terminals, and the terminals were built in the knowledge that the weather comes. What is newer is the way the whole spine can flinch at once: six hubs, one month, one network discovering that its strength and its exposure are the same property, the concentration that made it fast. The doors reopen. The question the next decade will answer is whether the network reopens exactly as it was, or whether this month was the moment a little of its traffic learned a different way home. Networks do not announce their turns. They quietly grow another artery, and the map is never quite the same again.

↑ Back to the key points

— By Tomás Ferreira