There is a reassuring way to look at the Capesize market through 2030. Count the ships.
As of April 2026, Clarksons data reproduced in Himalaya Shipping’s Q1 presentation showed 268 Capesize+ vessels scheduled for delivery between 2026 and 2030: 54 in 2026, 73 in 2027, 85 in 2028, 42 in 2029 and 14 in 2030. At that point, the orderbook stood at 14% of the existing fleet.
That sounds comfortable. It may even prove comfortable. But ship counts answer only half the question. A vessel is useful not because it exists, but because of how much cargo-lifting work it can perform in a year. As more bulk cargo originates farther from China, Capesize capacity increasingly becomes a question of calendar rather than steel.
That distinction is where the interesting arithmetic begins.
Guinea has already demonstrated the mechanism
Guinea is no longer an interesting future case. Bauxite has already provided the experiment.
Guinea exported 182.8 million tonnes of bauxite in 2025, up 25% year on year, according to official data reviewed by Reuters. Chinese customs data showed that 74% of Guinea’s bauxite shipments went to China, while China’s total bauxite imports rose 26.4% to 200.5 million tonnes.
The effect on shipping has been disproportionate to the cargo volume. DNV estimates that bauxite accounted for around 16% of Capesize tonne-mile demand in 2025, equivalent to roughly 250 additional vessels compared with 2015 levels. DNV also calculates that Guinea–China bauxite shipments generate roughly three times the tonne-miles of Australian shipments.
The market has therefore already seen what happens when commodity growth moves to the wrong side of the map, at least from a fleet-utilisation perspective. Tonnes rise. Tonne-miles rise much faster. The vessel spends more days earning on one cargo and fewer days becoming available for the next.
The unit that matters is the vessel-year
This analysis treats vessel demand as vessel-years: the annual cargo-lifting capacity required to service a trade, not a forecast of how many ships owners will order.
A Capesize-equivalent carrying roughly 170,000 tonnes can complete about 3.5 to 4 Guinea–China round voyages per year. The range is supported by two independent industry methodologies. DNV’s Simandou analysis uses an average of four annual Guinea–China round trips for a standard 180,000-dwt Capesize. Breakwave Advisors estimates approximately 3.5 voyages, based on an 11,187-nautical-mile Morebaya–Qingdao route and a 108-day round voyage. The same analysis places the Australia–China trade at roughly nine voyages annually.
The exact number varies with port calls, weather, waiting time, vessel speed and ballast positioning. Pretending that a voyage calendar possesses decimal-point certainty would therefore be rather ambitious.
Table 1. How voyage geography changes Capesize capacity
| Trade | Assumed cargo | Round voyages/year | Vessels required per 10 Mt |
|---|---|---|---|
| Australia–China | 170,000 t | ~9.0 | ~6.5 |
| Guinea–China¹ | 170,000 t | ~3.5–4.0 | ~14.7–16.8 |
| Brazil–China² | 170,000 t | ~3.5 | ~16.8 |
¹ For Guinea–China, the vessel requirement moves inversely with voyage frequency: approximately 16.8 vessels per 10 Mt at 3.5 voyages annually and 14.7 vessels at four voyages annually.
² Brazil–China uses 3.5 annual round voyages as a modelling assumption rather than a directly sourced route-specific estimate.
This is the central mechanism. Ten million additional tonnes do not impose the same requirement on Capesize capacity when they originate in Australia, Brazil or West Africa. The commodity may be identical. The calendar is not.
Bauxite could consume another 29 to 34 vessel-years
Guinea’s extraordinary 2025 performance does not justify mechanically extending the same growth rate to 2030. Indeed, doing so would be difficult to defend.
Bauxite exports reached 60.9 million tonnes in Q1 2026, up 25.3% year on year, according to official data reviewed by Reuters. Conakry also plans export curbs intended to lift depressed prices and protect smaller producers. Separately, Guinea is pursuing a much broader move into domestic alumina processing, with the government targeting multiple new refineries by 2030.
That is why this analysis does not extrapolate recent growth. Instead, it tests an illustrative scenario in which annual China-bound Guinean bauxite exports in 2030 are 20 Mt above their 2025 level.
At 170,000 tonnes per cargo, that additional 20 Mt per year requires about 118 laden voyages annually. At four round voyages per vessel per year, the trade would require approximately 29 vessel-years. At 3.5 voyages, the requirement rises to roughly 34.
Had the same annual 20 Mt increment originated in Australia, nine round voyages per vessel would require only about 13 vessel-years.
The difference, approximately 16 to 21 vessel-years, is the geography premium.
No extra tonnes are created by that calculation. Distance simply makes the same tonnes more expensive in ship time.
SimFer turns future geography into measurable capacity demand
Then comes Simandou.
The first shipment from Simandou left Guinea in December 2025. Rio Tinto’s SimFer joint venture is developing a mine with an annualised capacity of 60 Mtpa, while the co-developed rail and port infrastructure is designed to support exports of up to 120 Mtpa of iron ore from the SimFer and Winning Consortium Simandou mining concessions combined. Rio Tinto’s project disclosure makes that distinction explicit.
The timing matters as much as the scale. The December shipment marked the start of commissioning tests of the common rail-to-port infrastructure. Rio Tinto says that, once commissioning is complete, SimFer is expected to undergo an approximately 30-month ramp-up to full capacity.
This distinction matters. The model below concerns SimFer’s 60 Mtpa mine only, not the full 120 Mtpa Simandou system.
Where those tonnes ultimately go matters even more. China is central to the project’s commercial architecture, but assuming that every SimFer tonne will sail there would turn an analysis into a preference. We therefore model three destination cases.
Table 2. SimFer China-allocation scenarios and implied Capesize capacity
| China share | China-bound cargo | Gross vessel-years @ 3.5 v/y | Australian counterfactual | Geography premium |
|---|---|---|---|---|
| 50% | 30 Mt | ~50 | ~20 | ~30 |
| 75% | 45 Mt | ~76 | ~29 | ~47 |
| 90% | 54 Mt | ~91 | ~35 | ~56 |
The 75% case is not a Rio Tinto forecast. It is the central scenario used here for capacity testing.
At that level, SimFer alone requires approximately 76 Capesize-equivalent vessel-years. An equivalent 45 Mt moving on Australia–China geography would require roughly 29. The difference, approximately 47 vessel-years, is attributable to geography.
That number is more useful than a prediction about freight rates. It tells us what geography does before sentiment, congestion or scrapping enter the argument.
The source-locked core exceeds 100 vessel-years
Before adding Brazil, the two West African vectors can be combined on their own. Together, Guinea bauxite and SimFer provide the cleanest test of the geographical thesis developed above, allowing the core result to stand independently of the additional Brazil scenario introduced later.
A +20 Mt Guinea bauxite scenario requires approximately 29 to 34 vessel-years. A 75% China SimFer scenario requires another 76.
That produces a core requirement of roughly 105 to 110 vessel-years by 2030.
Yet an Australian counterfactual would need only about 42 vessel-years to lift the same incremental cargo. Therefore, approximately 63 to 68 vessel-years of the requirement arise from geography alone.
Table 3. The core capacity model
| Vector | Incremental China-bound cargo | Gross vessel-years | Australian counterfactual | Geography premium |
|---|---|---|---|---|
| Guinea bauxite | +20 Mt | ~29–34 | ~13 | ~16–21 |
| SimFer, 75% China scenario | +45 Mt | ~76 | ~29 | ~47 |
| Core total | +65 Mt | ~105–110 | ~42 | ~63–68 |
That is the finding worth keeping.
Roughly three in every five modelled vessel-years do not arise because more cargo exists. They arise because the cargo has farther to travel.
Brazil adds a third long-haul vector
Brazil adds a third vector, although this one requires a more explicit assumption.
Vale produced 336 Mt of iron ore in 2025, its highest annual output since 2018, while the company aims to produce approximately 360 Mt in 2030. Projects including Capanema, Vargem Grande 1 and Serra Sul +20 form part of the expansion and operational flexibility behind that trajectory.
That expansion is already moving from construction into operation. In July 2026, Serra Sul +20 began commissioning of its second long-distance conveyor belt system, marking the beginning of project start-up. Together with the Compact Crushing project, which Vale schedules for start-up in Q4 2026, Serra Sul +20 is expected to add 20 Mtpy of production capacity once fully ramped up.
The arithmetic from 336 Mt to approximately 360 Mt implies roughly 24 Mt of additional annual production by 2030. But production growth is not the same thing as China-bound cargo growth.
China provides a useful benchmark. Reuters reported in the context of an interview with Vale CEO Gustavo Pimenta that China accounts for around 60% of Vale’s sales. We therefore use 60% as a destination scenario, not as a Vale forecast for where its incremental 2030 production will be sold.
Applying that share to the additional 24 Mt gives 14.4 Mt of incremental China-bound cargo. At 170,000 tonnes per cargo and 3.5 Brazil–China round voyages per vessel annually, that would require approximately 24 Capesize-equivalent vessel-years. The same 14.4 Mt moving on Australia–China geography would require roughly nine.
The resulting geography premium is therefore approximately 15 vessel-years.
Brazil is deliberately treated differently from Guinea and SimFer. The production expansion is company-guided; the 60% China allocation is a scenario. If a smaller share of Vale’s incremental tonnes moves to China, Brazil’s contribution to the model falls accordingly. The core geography argument does not depend on it.
What the model measures, and what it does not
This is not a fleet forecast. It is a capacity-accounting exercise.
Model A asks how many Capesize-equivalent vessel-years are required to lift incremental cargo on the routes where that cargo is expected to originate. Model B asks how much of Model A exists solely because those routes are longer than an Australia–China counterfactual.
The two building blocks are already established above: the Guinea bauxite scenario and the SimFer capacity case provide the source-locked inputs for the core model.
Table 4. Model A versus Model B
| Case | Model A: gross incremental requirement (vessel-years) | Model B: pure geography premium (vessel-years) |
|---|---|---|
| West African core | ~105–110 | ~63–68 |
| Core + Brazil scenario | ~129–134 | ~78–83 |
The distinction prevents a tempting analytical mistake: labelling the gross requirement itself as geography. The 105–110 vessel-year figure is not a distance-only number. Some of that vessel employment would be required even if the same tonnes moved on a shorter route, simply because more cargo exists. Model B isolates only the vessel-years attributable to the longer routes.
On both the core and expanded cases, roughly three-fifths of the gross incremental requirement comes from distance itself, not from cargo volume. The stability of that share, whether or not Brazil’s conditional scenario is included, is the thesis in its cleanest form.
Then there is the orderbook
The supply side looks reassuring at first glance.
Clarksons Shipping Intelligence Network data, published by Himalaya Shipping, show 268 scheduled deliveries through 2030.
It is a meaningful supply response. Nobody needs to pretend otherwise.
There is, however, another column in the same dataset that deserves at least equal attention. The delivery wave peaks at 85 vessels in 2028, just before the ageing profile turns sharply: 110 vessels reach 20 years of age in 2029 and another 212 in 2030.
Table 5. Scheduled deliveries meet the ageing fleet
| Year | Scheduled deliveries | Vessels turning 20 |
|---|---|---|
| 2026 | 54 | 58 |
| 2027 | 73 | 56 |
| 2028 | 85 | 45 |
| 2029 | 42 | 110 |
| 2030 | 14 | 212 |
| 2026–2030 | 268 | — |
Note: Ageing figures are not summed because reaching 20 years marks a fleet-age threshold, not a scheduled removal.
That does not mean 322 ships will disappear in 2029 and 2030. Age is not a demolition order. A profitable old Capesize has a remarkable capacity to survive predictions of its demise.
But those vessels create optionality. Owners must decide whether ageing ships merit further surveys, dry-docking and capital expenditure. Himalaya Shipping’s Q1 2026 filing states that about 24% of the total Capesize fleet will be due for dry dock or Special Surveys in 2026.
That matters because dry-docking removes capacity temporarily even when a vessel is not scrapped.
The orderbook is hard data. Future removals are not.
Treating both as equally certain would give a spreadsheet considerably more foresight than its owner.
So, won’t the market still loosen?
Quite possibly.
On headline numbers, the 268-vessel delivery schedule looks larger than either the 105–110 vessel-year core requirement or the 129–134 expanded requirement. But the two cannot simply be netted against each other.
The 268 is a gross delivery count staggered across five years, before any allowance for vessels leaving the fleet, precisely the removal side this analysis has declined to forecast. The 129–134 figure measures the incremental annual vessel employment required once the expanded cargo scenario reaches its modelled 2030 level.
The orderbook counts scheduled hull deliveries. The model measures annual transport employment.
This analysis therefore does not demonstrate that a Capesize shortage is inevitable. Nor does it attempt to.
It demonstrates something subtler and, for freight markets, more useful.
What the model shows is that a substantial amount of additional vessel employment could be absorbed by trade geography itself. New vessels do not enter a market in which trade geography has remained static. They enter one in which the long-haul Atlantic Basin routes modelled here allow each vessel to complete considerably fewer annual voyages than on the Australia–China counterfactual.
Consequently, the relevant question is not whether new vessels are scheduled to enter the fleet. They are.
The question is how much of that additional fleet capacity ultimately becomes genuinely available transport capacity after longer-haul cargo growth has consumed its share of the calendar.
That is why the headline orderbook and the eventual degree of market looseness are not the same number.
Additionality remains the hardest question
The largest uncertainty in the model is not voyage arithmetic. It is additionality.
Simandou could displace Australian ore. It could displace Brazilian ore. Higher-quality imported material could replace lower-grade Chinese domestic production. Guinea’s bauxite exports could also slow as domestic refining expands. Every tonne displaced elsewhere reduces the net effect of the new long-haul tonnes modelled here.
There is, however, a useful external benchmark.
Himalaya Shipping’s earlier modelling estimated that 170 Mt of additional Atlantic Basin iron ore could require 272 Capesize-equivalents, while explicitly cautioning that the actual requirement would depend on Chinese domestic production and on the extent to which new Atlantic volumes replace Australian exports.
The absolute numbers are not directly comparable with this model because the cargo assumptions are different. The implied vessel intensity, however, is remarkably close. Himalaya assumed 180,000-dwt Capesizes at 95% utilisation and 3.65 round voyages per year. Its resulting transport requirement sits in much the same range as the voyage-calendar intensities used here.
That is not proof that either outcome will occur. It is evidence that the order of magnitude of the transport-intensity assumption is not eccentric.
More importantly, Himalaya’s own caveat points to exactly the same unresolved variable.
The question is not simply how much new Atlantic cargo appears. It is how much of it is genuinely additional to the seaborne system, and which existing production it displaces when it arrives.
There are limits to how much comfort one should demand from a shipping model.
Three variables will decide how much slack survives
Guinea: How much cargo actually leaves?
The country’s 2025 bauxite expansion was exceptional, and Q1 2026 remained strong. Yet falling prices, export controls and the government’s drive toward domestic alumina processing make straight-line extrapolation indefensible.
If China-bound exports exceed the +20 Mt scenario, the trade requires more Capesize employment. If domestic refining captures more ore, the opposite follows.
SimFer: Where does the cargo go?
Few variables move the result as quickly as SimFer’s destination mix.
At 50% China, the model requires about 50 vessel-years and generates a geography premium of about 30 vessel-years. At 90%, those figures rise to roughly 91 and 56.
The uncertainty is therefore not simply how much SimFer produces. It is how many of those tonnes travel the long haul to China.
Ageing ships: How much of the fleet stays?
The 322 vessels reaching 20 years in 2029–2030 are not scheduled removals. They are decisions waiting to be made.
Strong earnings can keep older ships trading. Weak earnings make surveys and dry-docking rather less romantic.
No single 2030 fleet number can tell us how much genuine slack will remain.
The map will have its say
The Capesize orderbook is substantial. It should add meaningful supply, and under several plausible combinations of cargo growth, substitution and vessel retention, it can create genuine slack.
But its headline size says less about eventual slack than the raw delivery count suggests.
The source-locked core requires 105–110 additional vessel-years to handle incremental Guinea bauxite and a 75% China SimFer case. Roughly 63–68 vessel-years of that requirement exist because of geography alone.
Include a clearly labelled Brazil scenario and the figures rise to 129–134 gross vessel-years, of which 78–83 represent the pure distance premium.
None of this proves that rates must rise. It does something more modest.
It measures how much more of each ship’s year the trade map may claim before fleet growth becomes surplus capacity.
For a market accustomed to counting hulls, that distinction matters. The next Capesize cycle may not be decided simply by how many ships yards deliver, but by how much of their working year the Atlantic has already claimed.
The ships are there. The distance has changed.