DP World Commits $90m to Double Luanda Terminal Design Capacity to 1.2m TEU

The Analysis

Spread the map out and run a finger down the western bulge of Africa until it stops at a bay that faces Brazil more than it faces Europe. Ten years ago Luanda was an oil port with a container yard attached to it, a place where boxes arrived on ships too small to matter, transshipped from somewhere else, cleared slowly, and left. If you were moving cargo into Angola you planned around the port rather than with it. That is the sentence the announcement from DP World this week is trying to retire.

The terms are these: US$90 million over the next two years at the multipurpose terminal at the Port of Luanda, alongside a ten-year concession extension that carries the agreement to 2051. Design capacity goes from 500,000 TEU to 1.2 million, a gain of 700,000. The quay lengthens by 222 metres and the operating area grows by 10 hectares. Three ship-to-shore cranes and twelve semi-automated rubber-tyred gantry cranes come with it, enough for the terminal to berth two Post-Panamax vessels at the same time, with vessel turnaround cut by up to half.

Ninety million dollars is not a large number in this industry. Savannah has just cleared a permit for US$2.3 billion, and a single large container ship costs more than this entire programme. That disproportion is where the story is, because it tells you what kind of project this is: not a new hub being willed into existence, but a comparatively cheap upgrade to a node that already exists, in a place where concrete, land and labour cost a fraction of what they cost elsewhere. Which leaves the only question that matters to anyone shipping boxes. Not whether the cranes will arrive, but whether the cargo will.

Start with the unit arithmetic, because it frames everything that follows. Ninety million dollars for 700,000 TEU of additional annual design capacity is about US$129 per TEU of capacity. Savannah's US$2.3 billion for 3.5 million TEU is about US$657 per TEU. One is roughly a fifth of the other, and no, the comparison is not clean: different labour costs, different land, a different concession structure, different crane specifications, and the American figure carries years of permitting and environmental mitigation with no counterpart here. Even so, a gap that wide is not explained by accounting. What it says is that capacity on this coast is cheap to build, and cheap capacity changes how a network behaves, because it lowers the threshold at which a carrier is willing to try a direct call instead of feeding cargo through somebody else's hub.

Look at what the money actually buys and the project reads differently from its own announcement. Two hundred and twenty-two metres of quay is the most consequential line in it, and the cranes, the yard and the turnaround claim are all downstream of being able to put two big ships alongside at once. A Post-Panamax vessel runs 300 metres or more in length, so two of them working simultaneously, with a working gap between them, needs something close to 650 metres of usable quay; the 222-metre extension is what closes that distance, which tells you the existing quay is already in that neighbourhood and what was missing was not a whole new terminal but the last few hundred metres. Ten years ago this was a port that could not take a Post-Panamax vessel without rearranging itself around the visit, and the binding constraint was never really the stacking area. It was the length of concrete at the water's edge. That is the bone that moved.

Now the number that deserved more scrutiny than it got: ten hectares carrying 700,000 TEU. Work it out with the assumptions stated. Ten hectares is 100,000 square metres. A forty-foot container occupies roughly 30 square metres of ground; stack four high and each ground slot is worth four boxes, so about 7.5 square metres of land per box. That gives something in the order of 13,300 ground slots. If the average box sits in the yard seven days, those slots turn over roughly 52 times a year, which is about 690,000 TEU. That is almost exactly the 700,000 in the announcement.

This is the part nobody printed, and it changes what the headline means. The 1.2 million figure is not a statement about cranes or quay length. It is a statement about dwell time. It holds at seven days of average dwell and it does not hold at ten. At ten days the same 100,000 square metres yields roughly 486,000 TEU, and the capacity gain shrinks by about a third before a single new crane has been bolted down. Average dwell across much of West and Southern Africa runs above a week, and not because terminals are badly run: customs release, inspection regimes, and a shortage of trucks able to move a box inland all add days the terminal operator cannot remove. The honest reading of the headline is therefore this. DP World has published a design number that is conditional on a dwell time it can only partly control.

The same lens applies to the turnaround claim. Three new ship-to-shore cranes across two Post-Panamax berths is about 1.5 additional cranes per berth, on top of whatever is already there. Large vessels want three to five cranes working them at once to reach the moves per hour that make a short port call possible, so three more units are an improvement and not a transformation. Where turnaround time actually halves is rarely at the ship's side; it is when the crane never has to stop because the yard keeps taking boxes away, which is precisely what twelve semi-automated gantries are for. Read the equipment list as one system and the message is clear. This is a quay-and-yard project that will be judged on dwell, and the cranes are the photograph rather than the engine.

Two words in the announcement are carrying more weight than they can bear, and they are the words up to. A turnaround cut of up to 50% describes the best case, which is the vessel that gets a berth on arrival, two or three cranes on her, and a yard with somewhere to put the boxes. The average call gets less than that. If a port call runs four days today and the best case takes it to two, the realistic outcome across a year of calls lands somewhere in between, and it is the average rather than the best case that a transit plan has to be built on. So ask for the number behind the number: current average turnaround at this terminal, in hours, discharge to sailing, and hold it as the baseline the operator is measured against once the works finish. Without that, up to half is a slogan nobody can audit.

Layer the impact, because it does not land the same way on everyone with cargo in this corner of the Atlantic. An Angolan importer whose constraint today is berth availability and yard space benefits from the quay and the yard almost as soon as the works finish, because their problem was never customs philosophy. A shipper moving boxes to the interior beyond Angola benefits far less, because the road out of the port is the binding link and no crane changes it. A carrier deciding whether to insert a direct call benefits most of all, and their decision is what eventually reaches you, one or two years after the steel arrives.

And a shipper currently routing through a Mediterranean or South African transshipment hub sits in an odd middle: they gain only if a direct call appears on a string they actually use, which is a decision made in a head office they have no relationship with. The gap between those four positions is where the real value of this announcement sits, and it is much narrower than the headline suggests.

The concession extension is the line that will matter longest and got the shortest treatment. Twenty-five more years, to 2051, means the operator is now planning a network around this node on a quarter-century horizon, and that changes what a terminal is for. A terminal on a short lease optimises for throughput it can monetise quickly. A terminal held to 2051 can afford to think about transshipment, about feeder networks, and about the slow business of persuading carriers to restructure a string. On this coast that means Luanda positioning itself against Lobito, Pointe-Noire, Matadi, Walvis Bay and Namibe for the boxes of the central and south-western Atlantic seaboard. It is a chess move on a board that only became interesting in the last decade.

There is a harder version of that point, and it runs directly against the interest of anyone whose boxes will sit in this yard. A concession held to 2051 has to be financed, and one of the ways a terminal finances itself is through storage charges, which are at their most productive exactly when boxes stay on the ground. A revenue model that leans on storage is a revenue model with no incentive to shorten dwell, and dwell is the single variable the entire 1.2 million figure depends on. That tension is not unusual and it is not a scandal; it is simply how the arithmetic of a long concession works. For a shipper the practical consequence is narrow and specific: watch the tariff schedule rather than the press release. If storage charges stay steep and free time stays short after the expansion, the yard is being run as a revenue asset, and you should plan your dwell assumptions on ten days rather than seven.

Which brings me to the question I ask every time a coast announces capacity. Is this a pulse or a turn? The ten-year pattern across West Africa is hard to read any other way than as a turn. Lekki, Kribi, Tema, Lobito and now this: a coastline that used to be fed by small ships out of Europe and South Africa is being rebuilt to take big ships directly. The reason is not mysterious. Ships got bigger, the Suez and Cape routings were repriced repeatedly, and a 14,000-TEU vessel does not want to dribble five hundred boxes at a time into a port that cannot berth it. Design capacity is the visible symptom. The underlying change is that the skeleton of the map rearranged itself and the ports are now catching up to it.

The caveat is that design capacity is not throughput. A terminal built for 1.2 million that handles 450,000 is not a hub; it is a well-financed empty yard, and the risk in a project of this shape is that capacity arrives a decade before demand and the operator spends those years pricing to fill it. For shippers in the short run that is not a bad outcome, because cheap capacity tends to mean competitive tariffs. It does mean you should not plan routing around 1.2 million TEU of capability. Plan around dwell time, because that is the number the whole structure rests on.

On timing, be precise about what arrives when. The money is committed across the next two years, which puts the steel in place in 2027 and 2028, and a terminal does not change its behaviour the day a crane is delivered. Quay works inside an operating terminal are done in stages alongside live cargo operations, which makes them a source of short-term disruption before they become a source of capacity. For an importer or exporter the practical shape of this is that nothing about Luanda improves for the next twelve to eighteen months, and it may get slightly worse while the works are under way. What you can usefully do inside that window is not booking. It is measuring. Establish the dwell number now, from your own boxes, discharge to gate-out, so that when the capacity does arrive you are comparing it against a baseline you collected rather than one somebody quotes at you.

Direct calls also have a chicken-and-egg problem that deserves naming, because it is the mechanism that decides whether any of this ever reaches you. A carrier inserts a call when the volume is there; the volume is there when the service is reliable; and the service becomes reliable about a year after the call has been running. Something has to break that circle, and on this coast what usually breaks it is one of two things: the terminal operator pricing the first couple of years of calls cheap enough to make the experiment worth a carrier's while, or an anchor shipper committing volume in writing. Neither appears in the press release. If you move enough through Angola to be that anchor shipper, your volume is worth something as negotiating currency right now, and it is worth considerably more before the cranes arrive than it will be in 2028.

What would turn this into something larger is not more cranes. It is customs reform that pulls average dwell under seven days, and a hinterland road and rail link that clears the yard fast enough to keep it empty enough. The Benguela corridor through Lobito is the live alternative, and it is the condition that would change this reading: if that railway starts carrying real container and mineral volume out of the interior, Luanda and Lobito become a genuine two-node system competing for the same hinterland, and the geometry of this investment changes. Watch dwell time, and watch Lobito, in that order. If dwell falls and Lobito stays slow, this is the hub. If dwell sits at ten days and Lobito moves, this is a very good terminal serving one country.

Here is the account, assumptions first. Suppose you import thirty forty-foot containers a month into Luanda, average goods value USD 38,000 a box, carrying cost 8% a year, and today your boxes reach Angola through a transshipment hub, Tanger Med or Algeciras or round the Cape via a South African port, with two extra handling events and something like twelve to eighteen extra days of transit. A direct call that delivers even the low end of that saving takes twelve days out of the pipeline. Thirty boxes a month is 360 a year, roughly USD 13.7 million of goods flow, and twelve days of that flow is about USD 450,000 of inventory no longer sitting on the water, worth roughly USD 36,000 a year at 8%. Add two fewer handling events at, say, USD 150 a box across 360 boxes and you have another USD 54,000 a year. Call it USD 90,000 a year on a modest volume.

That USD 90,000 exists only if the ship berths when it arrives. If a direct call saves twelve days at sea and the box then sits nine days in a yard waiting for release because dwell has not improved, you have swapped a predictable ocean delay for an unpredictable landside one, and the second is harder to plan against than the first. So the decision is not whether to use Luanda once it is ready. It is what dwell time makes it worth using, and on the volume above the answer runs roughly like this: under seven days the direct call wins outright, between seven and ten it wins only for the time-sensitive part of your book, and above ten days the transshipment routing remains the more honest plan.

The traps follow from that, and every one of them is a version of confusing a design number with an operating one. The expensive version is committing volumes on the strength of 1.2 million TEU that will not exist for years. Close behind is the assumption that two Post-Panamax berths mean your carrier will call there, because a berth is an invitation rather than a service and strings get restructured on volumes that exist today rather than volumes that have been designed. The concession to 2051 is not a stability guarantee on your own rates either; it guarantees the terminal operator's horizon, not your freight bill. And the one people forget is the inland leg, because for most practical purposes Luanda's hinterland is its own country and the road out of the port has historically been the slowest hour of the whole journey.

I have watched enough of these announcements to be careful with the word hub. A hub is not a place that builds capacity. It is a place that other people's routing decisions converge on, and those decisions get made slowly, by carriers watching volumes and by shippers watching whether the last box came out on time. Luanda has just bought itself the right to be considered: 222 metres of it, and ten hectares, and twenty-five more years. Whether the map redraws around it will be settled by something far less photogenic than a ship-to-shore crane, which is how long a box waits on the ground. Ports are built out of concrete. Networks are built out of time.

↑ Back to the key points

— By Tomás Ferreira