China's container plants run flat out as longer turn times drain empty stocks

The Analysis

CCTV Finance put five numbers on the screen on 28 July 2026 and I have not stopped turning them over since. Chinese container plants have moved from one shift to two. A Ningbo plant says daily output has doubled against pre-May levels, to between 500 and 550 boxes, with production booked through November. Ningbo customs counted 64,000 newly built containers exported through the port in June, up 28.9% on the year. Reefer container exports passed RMB 10.5 billion in the first half. And more than 95% of the world's reefer boxes are built in China, with one Qingdao producer reporting first-half output up about 15.7%. That is the whole evidence base. It is a television report, not a balance sheet.

I read box news differently from rate news. Rates are a price problem, and you can buy your way out of a price problem in an afternoon. Boxes are a lead-time problem, and you cannot buy your way out of a lead-time problem in an afternoon, because the factory's order book is now the constraint. That Ningbo plant is booked through November. Everything else in this story hangs off that one sentence.

So let us run the arithmetic before anybody panics, or worse, before anybody relaxes. Numbers first, then opinions.

Start with the plant, because it is the only number here with a factory floor behind it. Five hundred to five hundred and fifty boxes a day, doubled from before May. Assume it runs 26 days a month under two shifts, and one plant is putting 13,000 to 14,300 boxes a month into the world, call it 156,000 to 172,000 a year. Hold that as the unit of account while we look at everything else.

Now the customs number. Sixty-four thousand newly built containers went out through Ningbo in June. Against one plant at 13,000 to 14,300 a month, that is somewhere around four or five plants' worth of monthly output clearing a single port in a single month. The arithmetic is loose and I will not pretend otherwise, but it gives the June figure a size that the raw number does not carry on its own. Annualised, 64,000 a month is 768,000 boxes a year of new equipment moving through one Chinese port.

And here is the part that gets skipped. Those 64,000 are exports. A newly built box loaded out of Ningbo in June is not a box available to a shipper in Ningbo in June. It is capacity joining somebody else's fleet, somewhere else, weeks later. If you are an exporter in Ningbo reading a headline about 64,000 boxes, you are reading about supply leaving the country, not supply arriving at your gate.

Then there is the gap between two growth rates, and it is worth a minute. Output at that plant doubled, which is 100%. June exports through Ningbo rose 28.9%. One is a supply response and one is a flow measurement, and they are not measuring the same thing, but the shape still tells you something. If supply is doubling while measured exports grow by less than a third, then either demand is bigger than the export count shows, or the industry is working through a backlog that the doubling has only now made visible. The order book running to November is evidence for the second reading. A plant does not book four months out because it is idle.

The reefer numbers are where I put my foot down. Reefer container exports exceeded RMB 10.5 billion in the first half. That is a value. It is not a count, and I have watched what happens when a value gets repeated as though it were capacity. You cannot divide 10.5 billion by anything useful until you know the unit price, and reefer unit prices are not the sort of thing a television report hands you. This number has a problem if you try to turn it into a fleet size. It is perfectly useful if you only ask it the question it can answer, which is where reefer equipment comes from and how concentrated that supply is: more than 95% of the world's reefer boxes are made in China, and one Qingdao producer's first-half output was up about 15.7%.

One more thing worth pulling out of that pair. Ningbo's new-box exports, mostly dry equipment, grew 28.9%. The Qingdao reefer producer grew 15.7%. Same country, same half year, different slopes. If that holds, dry box supply is being expanded faster than reefer supply, which is what you would expect when the constraint is general turnaround rather than cold chain specifically. Anyone whose programme is reefer-heavy should be watching that slope, not the headline.

Here is the calculation that actually changes what you do on Monday. Assume your boxes used to complete a full round in 60 days, from stuffing at the factory to the next stuffing. Assume that round is now 70 days, because turnaround cycles have stretched. A box on a 60-day cycle runs 365 divided by 60, which is 6.08 trips a year. On 70 days it runs 5.21 trips a year. The same annual volume therefore needs 6.08 divided by 5.21, or 1.167, times as many boxes. About 17% more boxes for the same cargo. Push the cycle to 80 days and it is 4.56 trips a year, or 1.33 times: a third more boxes for the same cargo.

Put it on your own book. Assume you ship 1,200 boxes a month, which is 14,400 box-trips a year and a daily flow of about 39.5 boxes. At 60 days you need 14,400 divided by 6.08, call it 2,368 boxes working for you. At 70 days you need 14,400 divided by 5.21, call it 2,763. The gap is 395 boxes. At 80 days it is 3,157 boxes and the gap is 789.

Now invert it, because this is the version that gets budget approved. Every day you take out of the average cycle frees one day of flow, and at 39.5 boxes a day that is about 40 boxes back in service. Cut ten days out of the cycle and you have freed 395 boxes. That is the same 395 boxes you were about to order. One route costs you a purchase order and a wait until at least January. The other costs you discipline at the destination end. Only one of them is available this quarter.

Price it, with the assumption stated out loud, because I am not going to put a factory number in your mouth. Assume a new dry box lands at your yard at USD 2,300. Three hundred and ninety-five boxes is roughly USD 908,500 of capital, plus whatever your finance people charge for carrying it. Against that, ten days of extra cycle on a 2,368-box fleet is 23,680 box-days of equipment you are paying for one way or another, through detention, through demurrage, through storage, or simply through stock sitting still. Assume your free time is seven days and the excess runs at USD 25 per box per day. On the boxes that tip over, ten extra days is not a rounding error on most books. Substitute your own rates. The shape of the comparison does not move: buying boxes is a capital decision with a four-month queue behind it, cutting cycle is an operating decision with a ninety-day payback.

The calendar is what should worry you. Today is 28 July. The plant is booked through November. Assume an order placed now takes a December slot, and assume 30 days to move and position new equipment where you actually need it. Usable boxes in January. A decision made in the first week of August is therefore a decision about January capacity, not about August capacity, and anyone who tells you otherwise has not read the order book. Which means the five months from August through December are a bridge you cross with the boxes you already have, plus whatever you can free, rent or borrow.

So where do the days go? They go to destination dwell. They go to the two or three days an empty sits before anyone authorises the return. They go to yard congestion. And they go to the quiet habit of using a box as a warehouse, because warehouse space has a price and a box looks free. That last one is the most common and the least visible. At a flow of 39.5 boxes a day, one box parked as storage for a month is not one box lost, it is thirty box-days, and thirty box-days at your fleet size is real capacity. I would rather audit that than argue with a factory about a delivery date it has already told me it cannot beat.

Watch what the carriers say next, because this is where a box shortage actually shows up. Nobody announces that they are short of equipment. It arrives as an equipment imbalance surcharge, as a change to free time, as a restriction on shipper-owned boxes, or as a booking that gets confirmed without a release number attached. Read those four as one signal. If two of them land on your lanes within the same month, your cycle assumption is no longer 70 days, it is worse, and the line has been crossed. Cross that line and you lock the 2027 terms rather than waiting for a quieter market that the order book says is not coming.

Empty repositioning is the cost nobody budgets. Assume you have to move 200 empties from an inland point back to a port at USD 180 a move. That is USD 36,000, and it does not appear as a freight rate anywhere, so it never gets negotiated. If your cycle has stretched, you are moving more empties further, and that line is growing while you are looking at the ocean freight number.

The risk for an exporter comes down to one question: will there be a box at my stuffing point in the fourth quarter, and on what terms. Carriers with their own equipment allocate to their own priorities first, and a shipper-owned box is not always accepted, or not always accepted on the same free-time terms. The opportunity sits on the other side of the same fact. If you control boxes, or if you run a short cycle, you are holding something the market is short of, and that is worth raising in a 2027 negotiation even if nobody puts a line item on it.

There are four ways out of a box gap and they are not equally available. Buy new, which is the longest queue and the largest cheque. Lease, which is faster and usually available in the wrong place at the right price, so check pick-up locations before you check the rate. Buy used, which is quickest of all if you can find stock and if the boxes pass whatever condition standard your carrier applies, and that is a condition question rather than a price question. Or shrink the cycle, which is free in cash and expensive in management attention. Most people reach for the first because it looks like a purchase, and treat the fourth as somebody else's job.

The traps are predictable. Sixty-four thousand is one port in one month; do not let it become a national run rate inside a board pack. Five hundred to five hundred and fifty a day is one plant talking to a camera; do not build a sourcing plan on one plant. Booked through November tells you the order book is full, it does not tell you the factory is at physical capacity, and those are different things when you are deciding whether to pay a premium. The reefer figure is a value in RMB and must never be converted into units in front of a customer. And the one that will actually cost you money: 2027 contracts negotiated on 2025 free-time assumptions. If your cycle is ten days longer than it was, seven days of free time is now a bill, and the bill is per box per day.

If you want one sentence to carry into a Monday meeting, it is this: the box market has stopped being a price market and become a queue market. In a price market you shop; in a queue market you reserve. That changes who in your organisation needs to be in the room. Shopping is a procurement job. Reserving is a planning job, because the decision you make in August is a claim on January, and planning is the function that owns January.

When you make that call before September, ask for four things in writing and nothing else. The earliest slot. The location where the boxes will be available, down to the depot. How long the quoted unit price stays valid, because a quote that expires before the slot arrives is not a price. And what happens if the slot slips, in writing, with a date. A supplier who will not put the fourth one in writing is telling you the queue is longer than the slot he just sold you.

Reefer deserves its own paragraph because it does not behave like dry equipment. More than 95% of the world's reefer boxes are made in China, so a reefer shortage is not a regional inconvenience, it is a single-origin problem, and single-origin problems do not have a second phone number to call. A reefer also has to be inspected and plugged in before it earns anything, and a box that has passed through a ramped-up two-shift line still has to pass that inspection. If your cold chain programme is seasonal, place the equipment request before the season is on you, not when the season is on you, because by then the queue is the season.

There is an inventory angle to all of this that finance will understand faster than operations will. A box that turns in 60 days is inventory that moves six times a year. A box that turns in 80 days is the same asset moving four and a half times. You have not lost the box, you have lost a third of its working life per year, and the way you get it back is not a purchase order, it is a clock at the destination.

Write the allocation rule before the gap bites, not during it. Rank customers by contract, not by volume. State in writing which programmes get boxes when supply covers only 80% of the book, and who signs off on the exception. A rule written in October from a calm desk is a policy. A rule written in November with three people shouting is a concession, and concessions get renewed.

One caution on the supply response itself, because a doubling in four weeks is not a neutral event. A line that goes from one shift to two is adding people, adding hours and running equipment harder, and the first months of a ramp are when condition problems show up at the far end, not at the factory gate. I am not saying the boxes are bad; nobody has told me that and I am not going to invent it. I am saying that if you are taking new equipment this year, inspect on arrival rather than on paper, and count the days that inspection costs you in your cycle, because inspection days are cycle days too.

If you sit on the importing side of this, the exposure looks different but the arithmetic is the same. Your risk is not that you cannot buy a box, it is that empties do not come back to you when your plan says they will, so your receiving calendar and your de-stuffing capacity fall out of sync. Put a number on it: at 39.5 boxes a day, a de-stuffing backlog of 20 boxes is half a day of flow, and half a day of flow repeated every week for a quarter is a fleet you did not budget for. Whoever owns de-stuffing capacity owns your cycle, whether or not that is written anywhere.

What I would do this week, in the order things fall apart. Pull ninety days of box movements and get a real turnaround number by lane rather than a feeling. Run the two-cycle calculation on your top five lanes and write the gap down as a number of boxes, not as an adjective. Call your equipment suppliers before September and ask two questions only: when is the earliest slot, and where will the boxes be. Then go to the destination side with the ten-day target and a named owner, because that is the only lever that does not wait on a factory.

Boxes are boring right up to the moment they are not, and then they are the only thing anybody talks about. My own read is that the queue is real, the cycle is the part you control, and the factory will be fine either way. It is a comfortable business to be in when your order book runs to November and your customers' order books run to Friday.

↑ Back to the key points

— By Vivian Zhao