China Leads With 60 Automated Terminals, Targets 90 Terminals by 2030

The Analysis

The number being repeated this week is sixty. China operates sixty automated container terminals, the most in the world, and the Ministry of Transport says the target is ninety by twenty thirty. Sixty to ninety, thirty more terminals in four years. Every headline has that pair in it and almost none of them ask the question I would ask first, which is what a terminal actually is in this count, because the answer decides whether sixty-to-ninety is a growth story or an accounting story.

Here is why that matters. An automated terminal in this context can mean a greenfield berth built from the ground up around automated equipment, or a conversion of an existing conventional berth where the cranes and the yard were automated on a footprint that already existed. Those two things have completely different costs, completely different timelines, and completely different operational profiles. A greenfield build takes years and is capital-intensive and starts with a clean layout. A conversion has to automate around the constraints of an existing yard, existing rails, existing vessel mix, and it often delivers a lower automation rate for the same investment. If the sixty includes a lot of conversions and the ninety is going to be reached mostly by more conversions, then the story is not about new capacity at all; it is about the existing fleet of terminals getting smarter, and that is a very different story from the one the headline tells.

The third thing the coverage skips is what automation actually buys once it is installed, and the release gives the number for that too. Qingdao Qianwan, the first fully automated terminal in the country, moves sixty-one containers per crane each hour, more than double a conventional berth. Sixty-one per crane per hour. That is the metric that matters, not the terminal count. A country with ninety automated terminals that each move forty containers per crane hour is worse off in real terms than a country with sixty that each move sixty-one, and the terminal count alone cannot tell you which one you have. So before we get excited about thirty more, let us get precise about what the first sixty are doing.

Start at the machine level, because the whole thing is a machine-level argument and it gets lost in the institutional language. A conventional container crane is driven by an operator sitting in a cab, looking down at the container, and placing it by eye and by feel. The operator's performance ceiling is set by human visual acuity and reaction time, and the container moves at the speed the operator's judgement allows. An automated crane is driven by a control system taking input from sensors, and the performance ceiling is set by the control loop's cycle time and the mechanical limits of the machine.

The gain from automating a crane is therefore not a fixed percentage; it is the ratio between the human ceiling and the machine ceiling on that specific task. Qingdao Qianwan's sixty-one per crane hour against "more than double a conventional berth" implies the conventional baseline on that task is somewhere in the twenties to low thirties per crane hour, and the automated ceiling is roughly double. That is a doubling of the core cycle, and it is the single most important number in this release.

Now here is the part that the press release never says out loud, because it does not fit the growth frame. A crane moving sixty-one containers an hour is only useful if the rest of the terminal can keep up. A container terminal is a chain of subsystems: quay crane, horizontal transport, yard crane, gate, and the vessel stowage plan that governs the whole sequence. If the quay crane is automated to sixty-one an hour but the horizontal transport, the automated guided vehicles or the autonomous trucks, can only deliver containers at a lower rate, then the quay crane spends part of its hour waiting and the effective rate falls back toward the transport rate.

This is the classic queue-and-buffer problem, and it is the reason two terminals with identical cranes can have very different real throughput. The release reports the crane rate, which is the headline-able number, and it does not report the buffer utilisation or the transport rate, which is the binding constraint. So a shipper reading "sixty-one per crane hour" should understand it as the burst rate of one piece of the chain, not the sustained rate of the terminal, and the sustained rate is what moves the boxes.

Let me put the modelling on the table, assumptions written out, because this is the calculation that separates a real automation gain from a brochure gain. Take a berth with five quay cranes, each capable of sixty-one container moves an hour in burst. Take the horizontal transport fleet and give it a delivery rate that keeps up at eighty per cent of the crane rate, because that is a realistic buffer assumption in a well-run automated yard but not a perfect one.

Take the vessel alongside for a twenty-four-hour call. If the cranes ran at their full sixty-one all day, they would move five times sixty-one times twenty-four, which is seven thousand three hundred and twenty moves. At eighty per cent transport-constrained effective rate, the real figure is five thousand eight hundred and fifty-six moves. So the buffer assumption alone removes about fifteen hundred moves a day from the theoretical ceiling, and on a two hundred and forty metre vessel whose call might be, say, three thousand moves, the difference between seven thousand three hundred and five thousand eight hundred a day is the difference between clearing the vessel in about ten hours and clearing it in about thirteen. Multiply that three-hour difference by every call in a year and it is real berth capacity, or real berth time released for the next ship. The buffer is where the money is, and the crane rate is where the press release is.

That leads to the piece I have not seen anyone write about this specific milestone, and it is the one that matters most for a shipping line choosing a terminal. The binding constraint in an automated terminal shifts over its life, and it usually shifts from the quay crane to the horizontal transport to the yard and back to the gate as the terminal loads up. In the first years of a new automated terminal, the quay cranes are the constraint because the yard and the gate are underutilised. As volume builds, the yard and the gate start to bind, and the symptom is a falling average crane rate even though no crane has changed.

So a shipper or a carrier evaluating a terminal should not ask what the peak crane rate is; they should ask what the sustained crane rate is at the terminal's current volume, because that is the number that governs their vessel's turnaround today, and the gap between the two is the terminal's buffer against a bad day. A terminal reporting a sixty-one peak and a forty-two sustained is a different terminal from one reporting a sixty-one peak and a fifty-nine sustained, even though the press releases are identical.

Now bring in the second number that the release offers and that the count-based coverage ignores, which is the export-of-standards angle. Westwell autonomous trucks are running at Felixstowe, and an AidData report counts about four point eight billion dollars of Chinese financing for port equipment across sixty-nine ports in forty-seven countries since the year two thousand. Read those two facts together and you get the real competitive story, which is not that China has sixty terminals. It is that China is exporting the equipment and the operating model that make an automated terminal work, and the leverage of the export is not in the terminal count at home, it is in the protocol standard abroad.

When Westwell trucks run at Felixstowe, the terminal-control integration that makes them work is a Chinese-designed integration, and that integration is a standard that has to be maintained, updated and supported over time. The terminal count is the display; the export of the control stack is the business. For a port operator in a third country choosing equipment, the question that decides the vendor is not who has the most terminals at home, it is whose control stack has the deepest long-term support bench and the widest installed base, because a terminal is a thirty-year asset and the control system has to be supported for all thirty. The four point eight billion dollars across sixty-nine ports is the installed base being built, and the installed base is the moat.

Now connect the two threads, because they meet in a very practical place for a shipper. If the automated terminal's real constraint is the buffer and the integration, then the benefit to a shipper is not a new peak crane rate; it is a lower variance in the vessel turnaround time. And lower variance is worth more to a shipper than a higher peak, because the shipper's planning system cares about the spread of possible arrival dates, not the best-case arrival date. A terminal that clears a vessel in twelve to fifteen hours reliable is more valuable to a supply chain than one that clears in eight to twenty, because the first lets you commit to a delivery date and the second does not.

This is where automation pays the shipper: not in the best case, which almost never happens on the day you need it, but in the tail, which happens exactly when your cargo is the cargo that got caught. So the metric to demand from an automated terminal, if you are a carrier or a large shipper, is the standard deviation of vessel turnaround, not the peak move rate, and the release does not offer it, which means you have to ask for it.

The counterfactual is worth naming, because it bounds the whole automation claim. If the buffer and the integration keep up with the crane rate, the terminal delivers close to the doubling and the whole corridor benefits. If they do not, the terminal delivers a fraction of the doubling, and the peak crane rate becomes a marketing number rather than an operating reality. The variable that flips the conclusion is not the automation rate; it is the sustained-to-peak ratio of the actual throughput, and that is measurable at every terminal and reported almost nowhere. Countries racing to add automated terminals should be tracking that ratio per terminal, not the total count, because a power plant of terminals running at sixty per cent of their rated throughput is not the same infrastructure as one running at ninety, even if the count is identical.

There is one more structural consequence of the sixty-to-ninety target worth spelling out, and it is about where the growth can physically come from. A greenfield automated terminal needs deep water, a large land footprint and years of construction. The good greenfield sites on the main corridors are largely taken, and the ones that remain are deeper, longer and more expensive. Conversion, by contrast, can happen at an existing terminal on an existing footprint, and it needs less new land and less new dredging.

So if the ninety target is going to be met on the four-year timeline the release implies, a large share of it will have to come from conversion rather than greenfield, and conversion is where the harder engineering lives because you are automating around existing constraints. That means the interesting part of the next four years is not the new terminals; it is the conversions, because they are the ones that will show whether the automation technology can be retrofitted into the messy, irregular terminals that make up most of the world's port stock.

That retrofit capability, if it works, is the thing that has export value, because most of the world's terminals that need automating are old and irregular, not new and greenfield. The greenfield showcase proves the technology can work in ideal conditions. The conversion proves it can work in real ones. Watch the conversions.

Put a final number to the export angle, because the four point eight billion dollars deserves a per-unit frame. Four point eight billion across sixty-nine ports is about seventy million dollars per port on average, and that is a plausible total-project figure for a mid-sized automated terminal fit-out, which tells you the Chinese financing is not dabbling; it is placing complete automation packages at scale in foreign ports. For a port planning its own automation, that means the vendor field is increasingly shaped by whoever has the deepest installed base and the most reference projects, and reference projects are what reduce the risk of a first-of-its-kind project. So the sixty terminals at home are also a reference library for the export sales effort, and the ninety target expands that library. The count is not the goal. The reference library is the goal, and the count is how you build it.

There is a data-governance angle here that almost nobody connects to a terminal count, and it decides how fast the ninety can actually be reached. An automated terminal's control system learns from the operational history it accumulates: the vessel stowage patterns, the yard utilisation shapes, the weather and labour patterns that repeat across the years. A terminal that has been running for one year has one year of history, and its control system's accuracy is bounded by that history. A terminal that has been running for five years has a control system that has seen five seasonal cycles and can anticipate them. So the automated terminals that deliver the best results are not necessarily the newest; they are the ones with the deepest data history.

This has a direct implication for the sixty-to-ninety plan: the newly added terminals will, by definition, start with thin history, and their performance in year one will undershoot their performance in year three, not because the equipment is worse but because the data is thinner. A plan that counts terminals like identical widgets and expects the ninetieth to perform like the first is a plan that will be disappointed in the near term. The honest expectation is a ramp curve per terminal, and the ramp is set by how fast operational history accumulates.

That is also why I keep saying the conversion story matters more than the new-build story, and the data angle sharpens the point. A conversion takes an existing terminal that already has years of operational history, and wraps automation around it. In principle, that existing history is a head start if it can be fed into the new control system, which means a conversion can plausibly ramp faster than a greenfield of the same size, because the terminal's behaviour is already known. Whether the conversion actually captures that head start depends entirely on whether the data is migrated and integrated, and data migration is the step that automation projects most often underestimate.

So the question to ask a port doing a conversion is not which equipment it bought, it is what happened to the ten years of gate and yard records, because that is the asset that decides the ramp. A conversion that migrates its history well can beat a greenfield on time to full performance. A conversion that throws the history away starts from zero and pays the greenfield ramp without the greenfield layout. Same equipment, opposite outcomes, and the difference is somewhere in an integration schedule that nobody publishes.

One more practical consequence, and then I will bring it home. If the automation benefit shows up as lower variance in vessel turnaround rather than a higher peak, then the shipper's gain is realised in the planning system, not on the quay. That means a shipper only captures the automation benefit if their own planning horizon is long enough to use a tighter delivery window. A shipper who books on a two-day horizon cannot use a one-day tighter window; the tighter window only turns into value when it lets you commit to a customer, hold less buffer stock, or consolidate a shipment.

So automation at the terminal is worth nothing to a shipper whose planning is short, and worth a great deal to one whose planning is long, and the difference is entirely inside the shipper's own process. When a carrier advertises an automated terminal as a service enhancement, the shipper should ask what it lets them do differently in their own operations, because if the answer is nothing, the benefit is theoretical. The terminal can only hand you a tighter window. Whether that window becomes money is your job.

Bring it to the operating reality. If you are a carrier or a large shipper routing through an automated terminal, ask for the sustained vessel turnaround and its variance, not the peak crane rate. If you are a port operator choosing equipment, weight the installed base and the long-term support bench over the peak specification sheet. If you are watching this sector, watch the conversions, not the new builds, because conversion is where the real capability gets proven. And if you are reading the headline, remember that sixty terminals moving forty containers per crane hour is a slower country than sixty moving sixty-one, and the count alone will never tell you which one you are looking at.

↑ Back to the key points

— By Ravi Chandran