Cai Mep-Thi Vai Opens 1,200-Metre Joint Berth, Adding Up to 900,000 TEU a Year

The Analysis

Spread the map out on the floor the way we used to, the whole South China Sea and the Gulf of Thailand with it, and run a finger down the Vietnamese coast until it stops at the mouth of the Thi Vai river. Ten years ago this was another story entirely. Cai Mep was a name that came up when a box missed its connection in Singapore, and the honest routing plan for anything made outside Ho Chi Minh City was to send it south, into somebody else's hub, and let that hub decide which mother vessel it eventually rode.

On 9 October there was a change that fits in one sentence and will take a decade to read. CMIT and Tan Cang-Cai Mep Thi Vai opened what is being described as Vietnam's first joint berth: two adjacent deep-water terminals stitched into about 1,200 metres of continuous quay. Three large mother vessels can now work alongside where 2 could before. Each terminal expects its own capacity to rise by 30 to 40 per cent, or 400,000 to 500,000 TEU a year each and close to 900,000 TEU between them. Hapag-Lloyd's AIN SNAN EXPRESS took the first lift. Berth planning, equipment and labour are pooled from now on, and the maritime administration is considering writing joint operation into the draft code.

I have read a dozen versions of this story and every single one of them counted boxes. 900,000 TEU is a fine thing to say out loud and almost useless to plan against. What changed on the ninth was not throughput. It was the berth window, and the berth window is the unit in which shipping lines actually think. Where does that leave anybody moving containers through southern Vietnam over the next few years, and is this the step where the country stops being a spoke? That is what I want to open up.

The geography comes first, because it explains why a piece of concrete deserves this much attention. Southern Vietnam sits at the wide end of a manufacturing base that grew faster than its ports for fifteen years, and almost everything it makes wants to go east across the Pacific or west through Suez. Deep water close enough to the industrial belt is rare on that coast and Cai Mep has it, which is why it became one of the few places in the country able to take a large mother vessel directly instead of feeding one. Go back 10 years and the picture is unrecognisable: the standard plan for a garment container out of Binh Duong or an electronics load out of the north was to truck or barge it to an inland terminal, feeder it several hundred miles down to Singapore or across to Tanjung Pelepas, and buy space on somebody else's mainline. Ten years ago this was another business altogether. Every container leaving Vietnam paid a transit toll to a foreign port simply to reach its own customer, and everybody in the trade treated that toll as a fact of geography rather than a fact of scheduling.

Follow the causal line from the ninth rather than from the press release, because they point in different directions. 2 terminals used to sit side by side with a seam between them, each planning its own berth windows, its own crane allocation, its own yard slots and its own version of tomorrow morning. 2 halves with a seam behave nothing like one continuous quay of 1,200 metres, mainly because a large mother vessel does not arrive in instalments that respect a property line. Being able to work 3 at once instead of 2 is not merely extra width.

It is a different kind of freedom, because a schedule planner no longer has to fit an arrival into two separate diaries and can move a window by a few hours in either direction without asking anybody to break a plan that was already made. Whether a Vietnamese port can hold a transpacific or Asia-Europe rotation at all is decided inside those few hours. Capacity tells you how much can be poured through a throat. Windows tell you how often the heart gets to beat. Networks grow on beats.

Since everybody insists on counting boxes, let me do my own version, with assumptions you can throw out but cannot ignore. Assume a big mother vessel needs about 350 metres of berth plus working gaps, which is how 1,200 becomes 3. Assume an average working stay of 22 hours including manoeuvring and the ordinary run of delays. Assume an occupancy of 88 per cent, which is the kind of number terminals quote when they are feeling confident and few achieve for a full year. One berth-year is 8,760 berth-hours, and at 22 hours a stay that means roughly 398 potential calls at full occupancy, about 350 at 88 per cent. To turn 350 extra calls into 900,000 TEU, every one of those calls has to exchange around 2,571 TEU at Cai Mep, which is 29 per cent more than a defensible estimate of 2,000 today.

Loosen the occupancy instead, to the 70 to 80 per cent that real terminals run year-round, and the same extra berth produces between 279 and 319 calls; at 2,000 TEU apiece that is an increment of 558,000 to 638,000 TEU a year, somewhere between 62 and 71 per cent of what has been announced. My number sits below theirs over one line in the assumption box, and that line is occupancy. It is also the line nobody ever publishes.

There is a third arrangement that produces the same 900,000 and would mean something completely different. Suppose those extra hours go to feeder calls rather than mother vessels: 10-hour stays instead of 22, about 800 TEU exchanged instead of 2,000. At realistic occupancy the third berth then absorbs somewhere north of 600 feeder calls a year, close to two a day on top of what is already there, and 500,000 TEU appear in the annual tally without a single additional direct string touching the country. The count moves. The map does not. A wider quay sending boxes south for somebody else to load is still a spoke in every sense that matters to the person paying the bill, which is why I would not build one planning assumption on 900,000 until somebody publishes the split between mother vessel calls and feeder calls.

Here is the arithmetic that actually belongs in your own spreadsheet, and again, assumptions stated so you can argue with them. Assume you move 800 boxes a month out of southern Vietnam into the US West Coast. Assume 40 per cent of them ride a direct mother-vessel call today and 60 per cent travel via a transshipment hub, and that any journey with a transshipment in it costs 4 more days door to door than a direct sailing: 2 days waiting for the connection, 2 days of handling and variance.

Assume your cargo is worth USD 38,000 per loaded box and your cost of carrying inventory is 12 per cent a year, which is 38,000 multiplied by 0.12 divided by 365, call it 12 dollars 50 a day. Four days saved is 50 dollars a box, and that is not a freight rate, it is your own money spent less time floating. If additional weekly strings out of Cai Mep lift your direct share from 40 to 65 per cent, then 200 boxes a month convert, which at 50 dollars each is 10,000 dollars a month, roughly a 120,000 dollars a year, before anybody has won or lost an argument over rates. Note what this does to the frame. The value of this project to you is measured in how many more of your own boxes get a direct sailing, not in how many total boxes this terminal can absorb.

Who gets something out of the ninth, layer by layer. The carriers and the alliances are closest to it, because a window that can be moved without a negotiation is worth more to a schedule planner than to anybody in the supply chain. If a mainline operator can lengthen or shorten a stay without crossing 2 sets of planners, rotations become cheaper to design and more robust when one of their ports misbehaves, and you should expect somebody to test fairly quickly whether Cai Mep can now hold a slot that used to belong to Singapore.

Vietnamese shippers come next, and they do not all experience this the same way. Garment producers around Binh Duong and Dong Nai gain something they have never really had, which is a genuine choice between paying for speed and paying for volume, because their cargo historically could not justify a direct call. Electronics assemblers further north have less to gain per box, since their goods already command space on the strings that exist, and they may quietly lose something, because every factory converted from feeder to direct is another bidder for the same booking. Lower-value manufactured goods sit at the other edge: those containers only ever travel on a direct call if there is space left over, so their economics improve the moment surplus berth capacity appears and sour again the moment everybody else notices it.

Below that layer sit the people nobody invites to the announcement. Feeder operators, Mekong barge lines, the truckers moving boxes between industrial parks and the water. A terminal that grows its mainline capacity by something like 500,000 TEU a year needs somebody to bring those boxes to the yard, and if the barges cannot carry what factories actually load, because of draught restrictions upriver, or the inland container depots have no rail slots to sell, then the additional berths fill with nothing more profitable than air. This is the old vascular problem in its plainest form: widening the artery does nothing for circulation if the capillaries were never built.

Then there are the other hubs looking at the same coastline. Singapore and Tanjung Pelepas are not going to lose any sleep over one Vietnamese joint berth, but every string that stops at Cai Mep instead of Singapore is a transshipment box with somebody else's margin stapled to it, and both have every incentive to price keenly to keep it. Laem Chabang is reading the same board. None of it is zero-sum within a single year, and it turns into a real contest within 3.

Now the part I have not found written anywhere, and I went looking through everything I could open. Every report has counted TEU. Not one of them has noticed that since the ninth there is one fewer counterparty on this stretch of coast. Until this week, an operator negotiating a berth window at Cai Mep had the oldest lever in port economics sitting in their pocket: the terminal next door. Joint berth planning takes that lever and puts it in a drawer, because the party across the table can now refuse without losing anything.

There is no next door any more. Exactly the same applies to you. If you have been booking through both CMIT and Tan Cang-Cai Mep Thi Vai, you have been running a quiet hedge without necessarily calling it one, and that hedge was quietly cancelled on Thursday. What makes this stickier than usual is the second half of the announcement: the maritime administration may write joint operation into the draft code. A commercial arrangement between two companies can be unwound the day it stops suiting one of them.

A line in a code is a template, and templates travel. Ba Ria-Vung Tau this month, the Cat Lai and Phu Huu cluster or Hai Phong and Lach Huyen next, then Da Nang. Capacity along this coast is going up by around 900,000 TEU at the same moment the number of people you can play against each other goes down. Both facts were printed in the same release. Only one of them made it into any headline I saw.

The strongest argument against my own reading is to imagine the other build. Suppose each terminal had extended its own quay independently. The arithmetic would probably have landed in a similar place: 2 timelines of 400,000 to 500,000 TEU each, close to 900,000 between them, with 2 sets of planners and 2 boards and 2 negotiators facing each other. That path buys approximately the same throat and keeps everybody's optionality intact, which is precisely why it appeals to me in the abstract. What it does not buy is those continuous 1,200 metres, and that difference is larger than it sounds on paper.

A seam in the middle of a quay obliges nobody to compromise, whereas a single planning desk can slide a vessel from one end to the other while it is still at sea, re-cutting a rotation around weather or congestion as though the quay were one machine. Efficiency of that sort is always bought with somebody's autonomy, and the people who pay are seldom the people at the signing table.

Then the disagreement that decides whether this is a chess move or simply a bigger table. The generous reading says this is the moment Vietnam stops paying a feeder toll and starts behaving like a genuine load centre, with enough water and enough windows to hold direct strings to the US West Coast, the US East Coast and North Europe, and with enough manufacturing density behind it to justify every one of them. The reading I keep returning to when the numbers refuse to add up is the opposite one: capacity first and cargo afterwards has been the regional habit for twenty years, and what tends to follow is a very handsome quay, a discount war to fill the yard, and a handful of services that return to their old routing once the incentives run out.

3 things will settle it and none of them is a berth. Channel depth, and whether the dredging programme keeps pace with the ships operators are actually ordering. The capillaries: whether Mekong barges can carry what the factories load, whether there are rail slots out of the inland depots, whether a truck gets through Ho Chi Minh City in under three hours on a bad Thursday. And the cargo itself, meaning whether the export mix climbs far enough up the value ladder to tolerate a direct service at all, because a box of electronics can carry a premium string and a box of low-value goods simply cannot.

Timing, in the only units that are useful to you. Nothing observable changes for the next quarter: the joint berth exists, AIN SNAN EXPRESS has taken the first lift, and your containers still move on last year's rotations. The next real evidence comes with the following schedule filing cycle, which for the large alliances runs about 6 months ahead, so anything decided for the 2027 summer season gets filed early in the first quarter and appears as published rotations somewhere around April; that is the moment you find out whether any additional weekly direct string out of southern Vietnam actually exists or whether the extra windows were filled by feeder calls nobody will mention.

Through 2027 keep an eye on the draft code, because if joint operation becomes law the pattern replicates along this coast and the shape of the whole board shifts rather than one square on it. Across 3 to 10 years the verdict is readable in three numbers you can track without asking anybody's permission: a written 12-month dredging programme, barge and depot capacity rising at least as fast as berth capacity, and above all the count of weekly mainline strings calling southern Vietnam for the US West Coast, the US East Coast and North Europe, taken from the operators' own published schedules and nothing else. Keep a quarterly count. If that number has not risen within 4 consecutive quarters, what opened on the ninth was additional berth capacity rather than additional service, and those two are not remotely the same line in your budget.

Maps are more honest than press releases. A map shows you a seam between two terminals, and then one day it does not, and everything downstream of that little change is a decade of work. What I want to know about this particular seam is simple to ask and takes years to answer: it changed the tide, or it merely widened the mouth of the harbour. The tide is what governs this trade. It comes in as direct calls, as feeder networks thinning behind them, as schedules written around your factory rather than somebody else's hub, and it goes out again just as quietly the moment the dredging stops or the inland loses the plot.

900,000 TEU is a number about concrete. 3 mother vessels where 2 used to work is a number about time, and time is the one currency here that nobody printed. Watch your own lane, count the strings, and keep one route out of this cluster genuinely open so you always have somewhere to put a box when the tide turns.

↑ Back to the key points

— By Tomás Ferreira