China's cold-chain network expands to 270 million cubic metres of storage

The Analysis

Two hundred and seventy million cubic metres. That is the number CCTV put on China's national cold-storage capacity at the end of June 2026, up 4.6% on a year earlier. I have read enough of these to know what a number like that is and what it is not. It is a photograph of buildings. It is not a promise about the container of durian you have crossing a land border on a Thursday night. I run import lanes, and the thing that has never once broken my chain is a national shortage of cold rooms. The thing that breaks it is ninety minutes with the power off at a transfer point nobody thought to put in the contract.

The route I have in mind is the one the report actually touches. Durian comes up from the south, crosses a land border into Yunnan, and moves inland on the new cold-chain international rail service, in the new long-range temperature-controlled containers, cold chain unbroken end to end. That sentence is a road, not a statistic. On the Yunnan side I mean the gates at Mohan and Hekou; over in Guangxi there is Youyiguan. Places like that are where a chain either holds or lets go, and every one of them is a handover, and every handover includes a few minutes in which the temperature is nobody's job.

What I want to do below is narrow. Take 270 million cubic metres, 4.6%, 105 national cold-chain logistics bases and three large cold stores being rushed up at Zhangzhou, and turn all of it into something you can act on today: how much room one container of yours actually eats, where the chain really breaks, who owns the temperature at each handover, and what arrival rhythm keeps you out of the queue. I am not going to talk about five-year plans. The NDRC said it will keep building cold-chain facilities under the 15th Five-Year modern logistics plan, and that is a sentence about 2030, not about the truck you have landing on Friday.

Start with the four things the report actually says, and be precise about what each one covers. CCTV reported on 26 July 2026 that total cold-storage capacity reached 270 million cubic metres at the end of June, up 4.6% year on year. A backbone of 105 national cold-chain logistics bases now spans the eastern, central and western regions and handles the bulk of fresh-produce transshipment and distribution. New cold-chain international rail services out of Yunnan, together with new long-range temperature-controlled containers, have made unbroken cold-chain movement of imported durian possible. At Zhangzhou in Fujian, three large cold stores are going up fast to meet the fishing season, when aquatic products come to market all at once. And the NDRC said it will build out cold-chain facilities further under the 15th Five-Year modern logistics plan. That is the whole of it. Everything below is arithmetic done on those numbers plus assumptions that I will label as mine.

The arithmetic comes first, because 4.6% is not a number anybody can feel. If 270 million cubic metres is 4.6% above a year earlier, then a year earlier the country stood at about 258.1 million, and twelve months added roughly 11.9 million cubic metres. Spread over the year, that is about 989,000 cubic metres a month, or about 32,500 cubic metres a day, on average. Spread over the 105 bases, the average base grew by about 113,000 cubic metres in the year. That sounds like a great deal until you divide the national total by the same 105 and find that the average base holds about 2.57 million cubic metres. So the average base added roughly 4.4% of its own size in twelve months. Which is another way of saying that the average base put on one cold room's worth of room, and nothing whatsoever changed about your Thursday night.

Here is the part shippers get wrong, and it is worth doing slowly. Capacity is a stock. Your problem is a flow. A cold room standing empty at three in the morning in a base you have never used will not cool a single one of your pallets. What you consume is not cubic metres, it is cubic-metre-days: room, multiplied by how long your goods sit in it. So let me put numbers on one consignment, with the assumptions written out first so that you can argue with them. Assume one 40-foot reefer of durian, 18 tonnes net, palletised, needing 40 cubic metres of cold room while it is held. Assume it is held four days at the gateway, two for inspection and release and two waiting on onward trunk transport. That is 40 cubic metres times four days, or 160 cubic-metre-days per container.

Now run a programme through it. Take 40 containers a week through one gateway, which is a size I recognise for a mid-sized importer in peak durian season. Forty containers a week at 160 cubic-metre-days each is 6,400 cubic-metre-days of consumption every week. By the usual queueing rule, the amount of room you occupy at any given moment is that weekly consumption divided by seven, so about 914 cubic metres standing at any time. Hold that against the average base of 2.57 million cubic metres and you are using about three and a half hundredths of one per cent of it.

Hold it against the national daily addition of about 32,500 cubic metres and the country was adding, on an average day, roughly thirty-six times the standing room your entire programme occupies. Those two numbers are the whole argument. Nobody is short of cold rooms in the abstract. What you can be short of is forty cubic metres at the right temperature, on the right dock, at two in the morning.

So when somebody tells you the cold chain is short of capacity, ask them where. Not nationally, because nationally the stock grew by 11.9 million cubic metres in twelve months. Ask at your gateway, at your base, at the specific plug your container is on. The 105 bases handle the bulk of fresh-produce transshipment, which is exactly why the pressure collects there, because that is where everything flows. And if your produce lands at a place that is not one of the 105, or at one of the 105 that added nothing last year, then your share of the 4.6% is zero. I would rather you found that out in July than in September, when the fishing season lands on top of it.

Which brings me to where chains actually break, and there are four places, and I have paid for every one of them at some point. Start at the origin, with pre-cooling. Fruit picked at thirty degrees and loaded warm will fight the reefer for two days and arrive with the pulp two degrees above where you wanted it, and no downstream capacity ever takes out field heat that was never removed at the packing house. Next comes the border transfer, and this is the one that costs the most. Your container arrives on one chassis and leaves on another, or is transloaded into a rail box, and for however long that takes the reefer is either unplugged or running on a genset that may or may not match the plug standard at that yard.

Then there is the container swap on the trunk leg, the same problem again at a dry port. And at the far end there is the last leg, where a pallet that has been held at thirteen degrees for nine days goes into a non-refrigerated van for a two-hour run to a distribution centre. A chain is exactly as strong as its worst handover, and every handover is also a handover of responsibility.

One thing about gates that I learned the expensive way, and it holds at whichever crossing you use. Do not underestimate that border crossing. A land gate is not a road with a building beside it; it is a queue with a clock on it. Inspection happens in working hours, in the order the trucks arrive, and that order is not something you can buy your way up. If your container lands at four on a Friday afternoon, it is not being inspected on Friday. So the question that actually matters is not how long an inspection takes. It is what the gate's working day is, what the cut-off is for same-day handling, and whether there is a night shift at all. Get those three answers in writing and you will have removed more risk than any quantity of additional cold storage removes. I have watched importers spend a year negotiating warehouse rates without once asking what time the gate stops working.

Let me put a price on one of those handovers, with the assumptions on the table. Assume your 40-foot reefer carries 18 tonnes of durian. Assume a landed cost of 30 yuan per kilo, which is my assumption and not a number from the report, so one container is worth about 540,000 yuan. Assume a transfer where the power is off for ninety minutes in twenty-eight degree ambient air, and assume that excursion costs you 0.8% of the consignment's value in downgraded fruit. That 0.8% is my number too, and you should replace it with your own claims history, because your claims history is the only figure here that actually belongs to you. Eighteen thousand kilos at 30 yuan is 540,000 yuan; 0.8% of that is 4,320 yuan, on one container, for ninety minutes that were nobody's fault in particular. At 40 containers a week that is 172,800 yuan a week, and across 52 weeks about 8,985,600 yuan a year.

Now put that next to what the fix costs. Assume a single-use temperature logger at 150 yuan, one per container, set to record every five minutes. Across 2,080 containers a year that is 312,000 yuan. I am not going to claim a logger prevents an excursion. A logger does not cool anything. What it does is turn an argument into a document. Right now, when fruit arrives soft, the dispute is about whose ninety minutes it was, and the dispute takes longer than the fruit lasts, and everybody loses. With a curve that travelled with the goods, the answer is in the file and the conversation lasts ten minutes. If logger data lets you recover the cost of even one excursion in ten, that is about 898,560 yuan back for 312,000 yuan spent, and the rest of the spend is insurance you were already paying for in phone calls.

The second lever is rhythm, and it costs nothing. Take the same 40 containers a week and assume they currently land on Tuesdays and Thursdays, which is what happens when everybody books against the same two rail departures. That is 14 containers a day on peak days, each needing a transfer. Assume the yard has two transfer bays and each transfer takes 45 minutes: 14 containers at 45 minutes is 630 minutes of work, and over two bays that is 315 minutes, or five and a quarter hours of queue on a peak day. Spread the same 40 containers across five days and you get 8 a day: 8 times 45 is 360 minutes, over two bays that is 180 minutes, or three hours. The peak-day queue drops by two and a quarter hours, with no money spent and no facility built. All it costs is a conversation with whoever books the rail slots, and the willingness to accept that some of your fruit arrives on a Wednesday.

Then fix the ownership, because temperature responsibility is the thing everybody agrees about in principle and nobody writes down. The rule I use is blunt: whoever controls the plug owns the temperature. On the origin leg that is the shipper, and it ends when the container crosses the border. At the transfer it is the terminal or the carrier, and that is the piece you need in the contract with numbers in it rather than adjectives: a single unplugged period of no more than 30 minutes, a cumulative total of no more than 60 for the whole handover, and a stated consequence when it is exceeded.

On the rail leg it is the rail operator, and you want the same two numbers there. On the last leg it is whoever you hired to deliver, and in my experience that is the leg nobody thinks to specify, which is how a nine-day cold chain ends up in a hot van. Four legs, four owners, four thresholds, one logger that never leaves the pallet. Put it in the rate sheet, not in the email.

The two things in the report that are genuinely new for a shipper are the international cold-chain rail service out of Yunnan and the long-range temperature-controlled containers. Together they do one job: they cut the number of handovers where your box depends on somebody else's power. A conventional reefer on a long inland run needs plugging in at intervals, every plug point is a place the chain can fail, and the more plug points there are the more places there are to fail.

A container that can hold its set point over a long distance without that takes a whole string of risks down to two ends. That is the real gain in this story, and it is worth more to you than the capacity figure. Note what it does not do. It does not touch the ninety minutes at the transfer, and it does not touch the last mile. The chain got shorter, so the two handovers that remain now carry a larger share of the risk than they did before.

Which is why the first thing I would go and get is the departure timetable in writing: how many cold-chain trains a week, and if you miss one, how many days to the next. On a lane measured in days, missing a weekly departure is not a delay. It is seven days, and seven days on durian is not a logistics problem any more.

One more timing point, and it is the one that bites first if you are on aquatic products rather than fruit. Three large cold stores are being rushed up at Zhangzhou to meet the fishing season, when the catch comes in all at once. Whatever the opening date turns out to be in a given year — it is published by the authorities each season and you should look up the current year's notice rather than trust a date from memory — the effect on a shipper is the same. For a few weeks, cold rooms and refrigerated trucks in the coastal provinces are contested, and rates and queues move together. If your own seasonal peak falls inside that window, commit room before the season opens rather than during it. As a working target I would want committed space equal to 120% of my peak-week arrivals, in writing, at least three weeks before the season starts, plus a named alternative base if the first one fills.

The alternatives, and what each one costs. A different gateway sounds like the easy answer and usually is not: another border crossing means new cut-off times, new transfer crews and a road you have not run, and the first three shipments through a new gate always cost more than the last three through the old one. Shifting part of the volume from rail to road buys flexibility and costs money per tonne-kilometre, and it puts back the plug-point problem the long-range containers just took away. Holding more inventory is the option everybody reaches for and the one I trust least on fresh produce, because safety stock of something with a shelf life is only spoilage you have not recorded yet.

And the pitfalls, which are boring and expensive in equal measure: cold-store contracts billed by the day rather than by the pallet, free time that starts at train arrival rather than at release, temperature data that belongs to the terminal and cannot be produced when you need it, and a genset clause nobody checked against the plug standard at the far end.

Who does what, then, because a plan without a name on it is a plan that does not happen. The arrival rhythm belongs to whoever books the rail slots, and that person should be working from a timetable rather than from habit. The logger belongs to whoever packs the container, and it goes in before the doors close, not at the gate. The thresholds in the transfer contract belong to whoever signs the carrier's rate sheet, and if that is a procurement colleague who has never stood in a transfer yard, then take them to one, because nobody writes a 30-minute limit into a contract for a handover they have never watched.

The seasonal space belongs to whoever owns the sales forecast, and the two of them need to be in the same room at least three weeks before the season rather than the week it opens. And the fishing-season notice is a two-minute lookup that nobody in most organisations owns at all, which is exactly why it gets missed every year.

Strip it down and the week says one thing. The country added about 11.9 million cubic metres of cold storage in a year and the average base grew by about 113,000, and none of that will cool a box that sat unplugged for ninety minutes at a transfer nobody specified. You cannot buy your way out of a handover with somebody else's building. What you can do is cheap and specific: one logger per container, four thresholds written into four contracts, a departure timetable in writing, and an arrival rhythm that spreads your peak across five days instead of two. This lane I have run, and what I would tell anyone standing where you are standing is that the cold chain is not a warehouse problem at all.

It is a road problem. The cold rooms are the car parks. The handovers are the junctions, and junctions are where things go wrong. You do not check the national count of parking spaces before a trip. You check the tyre pressure, you note which roads are open, and you keep one spare in the boot — and on this lane the spare is a second gateway you have already qualified, before you need it.

↑ Back to the key points

— By Ömer Kaya