Rio Tinto's $25 billion EBITDA still rides one rock — and China is letting go

---

Read Rio Tinto's 2025 results release the way the company wants you to read it, and you come away convinced you have bought a transition story. Copper production up 11% year-on-year, a record copper EBITDA cited in the very first investor slides, the Oyu Tolgoi underground in Mongolia finally producing, the Arcadium lithium business folded in for $6.7 billion, aluminium, titanium, the Simandou iron-ore mega-project pouring its first ore in December. The narrative is decarbonisation, electrification, the metals that wire the future. The dividend — $6.5 billion paid at a 60% payout ratio, the tenth consecutive year at the top of the policy range — is offered as proof of a durable, diversified compounding machine.

Now read the same document as an auditor would, asking only one question: where does the money actually come from? The answer is the same answer it has been for two decades. Rio Tinto is, before it is anything else, an iron-ore company that sells a single commodity, priced in dollars, to a single country's steel industry, and almost everything else in the portfolio is a rounding error against that one exposure. The 2025 numbers are good. The problem is not the numbers. The problem is what the numbers depend on, and the direction that dependency is now moving.

The denominator nobody at Rio controls

Start with the arithmetic of a price-taker. Rio Tinto shipped 342 million tonnes of iron ore in 2025, with Pilbara shipments of 326.2 million tonnes on a 100% basis — actually down 1% on the prior year. Volume, in other words, did not grow. The headline 7% revenue increase and 9% EBITDA increase did not come from selling more rock; they came substantially from the price of the rock, from a record-volume run that began in April, from cost discipline, and from a genuinely strong copper year. But the single largest swing factor in any given year for this company is the realised iron-ore price, and the company sets none of it. It is quoted in Singapore, cleared against Chinese mill demand, and printed on a screen every morning whether Rio likes the number or not.

This is the structural tell that the brochure buries. A business with pricing power expands margin by raising prices into demand. A price-taker expands margin only by cutting unit costs or by riding a commodity tape it cannot influence. Rio delivered a 5% real-terms unit-cost reduction and roughly $650 million of annualised productivity gains in 2025 — real, creditable work — but those are defensive levers, the moves you make precisely because you cannot touch the top line. When the tape is friendly, as it was for much of 2025, the operating leverage looks like genius. When the tape turns, the same leverage runs in reverse, and there is nothing in the cost line large enough to offset a multi-dollar move in the per-tonne price across 340 million tonnes.

Cyclical earnings wearing a secular costume

The energy-transition framing is doing enormous narrative work here, and it deserves to be interrogated on its own terms. Copper had a record EBITDA year. Lithium is now in the portfolio. Aluminium is a transition metal. All true. But a transition exposure is not the same as a transition business model, and the distinction is the whole game. Rio's copper and lithium operations are themselves price-takers — copper clears on the LME, lithium prices have been brutalised through 2024-2025 — so adding them does not convert the company from cyclical to secular. It simply swaps one set of commodity prices the company cannot control for another. The "diversification" reduces single-commodity concentration at the margin; it does not create pricing power anywhere.

And the scale gap is the point. Copper's record EBITDA, impressive as the slide deck makes it, sits inside a $25.4 billion group EBITDA still anchored by iron ore. The Arcadium lithium deal closed in March 2025 into a collapsing lithium price; it is a strategic land-grab on future demand, not a 2026 earnings engine. So when the stock is discussed as a copper-and-lithium transition play that happens to also mine iron ore, the emphasis is exactly inverted. It is an iron-ore company that happens to also mine copper and, recently, lithium. The market is being invited to apply a secular-growth lens to an asset whose cash flows are dictated by the most cyclical end-market on earth: Chinese construction.

One customer, and that customer is selling

Here is the exposure the transition story is designed to make you stop thinking about. Iron ore exists to make steel, and the marginal tonne of seaborne iron ore is bought, overwhelmingly, by China. China's steel demand is not flat; it is contracting. Chinese steel production fell roughly 5% year-on-year in the first quarter of 2026, to approximately 250 million tonnes. Apparent consumption of the five major steel products dropped 3.1% week-on-week in early June 2026 — an acceleration from the prior week's 0.5% decline, the kind of sequential deterioration that says the slide is speeding up, not stabilising. Underneath it all sits the property crisis that will not end: Chinese real-estate investment fell 16.2% year-on-year in the first five months of 2026. Property is the swing consumer of construction steel, and property is in its fifth year of decline.

A company with genuine customer diversification could shrug this off. Rio cannot, because there is no replacement buyer for hundreds of millions of tonnes of seaborne ore at China's scale. India is growing but is a fraction of the volume. Southeast Asia is not big enough. The blunt reality of the iron-ore market is that it is a one-customer market, and that customer has entered a structural, policy-acknowledged downcycle. S&P has been writing about the Chinese steel downcycle as an ongoing, multi-year theme. This is not a weather event Rio waits out. It is the slow re-rating of the single demand curve under the single product that pays for the dividend.

Rio is about to make its own oversupply worse

The most damning part of the 2025 story is that Rio is itself adding to the glut it is exposed to. Simandou — the vast Guinean deposit Rio has chased for thirty years — poured first ore in 2025, with first shipment from the WCS port in December and the cargo landing in China in January 2026. In total 2.3 Mt of crushed ore was produced in 2025 (100% SimFer basis), and Rio guides Simandou sales of 5–10 Mt for 2026 on a 100% basis, ramping toward a project that, fully built, will add roughly 20 million tonnes of high-quality, low-cost ore to a market the company already concedes is demand-constrained.

Read that sequence slowly. The consensus iron-ore price is forecast to fall from about $101 a tonne in 2025 to roughly $94 in 2026 — a 7% decline — and the analysts citing that forecast name two drivers: stagnant Chinese steel demand and the launch of Simandou. Rio's own trophy project is, by the market's own accounting, one of the reasons the price of Rio's own core product is expected to fall. The company has spent a decade and tens of billions of dollars to bring on supply that helps depress the price of the 340 million tonnes it already sells. Management will frame Simandou as premium, high-grade, decarbonisation-friendly ore, and there is truth in that. But in a market clearing on tonnage into a shrinking customer, new low-cost supply is not a moat. It is a margin headwind the company built itself.

The dividend is an output of the tape, not a promise

The 60% payout ratio is presented as a feature — discipline, consistency, ten straight years at the top of the range. Look at what the payout ratio actually is: a percentage of earnings, not a fixed commitment of dollars. That is the crucial sleight of hand for income investors. A 60% payout of a swinging earnings number is itself a swinging dividend. The $6.5 billion paid in 2025 was 60% of a year inflated by a friendly iron-ore price and a record copper run. Hold the payout ratio constant and shrink the earnings base — which is exactly what a falling iron-ore price and contracting Chinese steel demand do — and the dollar dividend falls in lockstep. The "ten consecutive years at the top of the range" is not a sign of safety. It is a sign that the dividend tracks the commodity cycle with a one-year lag, because by design it is a fixed fraction of a cyclical number.

Income investors who bought Rio for yield stability have, whether they realise it or not, bought a geared claim on the China iron-ore tape. When the tape was good, the yield looked like a bond with upside. When the tape turns, the same mechanism that paid them generously becomes the mechanism that cuts them. There is nothing dishonest in Rio's disclosure here — the payout policy is stated plainly. The dishonesty, if there is any, is in the market's willingness to treat a payout-ratio dividend as though it were a payout-dollars dividend.

Quality of earnings: good, but borrowed from the cycle

Give the 2025 print full credit on its own terms. Underlying earnings of $10.9 billion were stable year-on-year; net cash from operations of $16.8 billion was up 8%; the company paid $10.4 billion in taxes and government royalties, which is a real cash outflow that flatters the "underlying" framing by reminding you these are pre-tax operating economics of genuine scale. The cost-out was real. The copper record was real. This is not a company cooking its books; Rio's accounting is conservative by the standards of the sector.

But quality of earnings is not only about whether the numbers are clean — it is about whether they are repeatable. And the repeatability of Rio's 2025 earnings rests on assumptions that the company's own 2026 guidance and the external consensus already undercut: a sub-$100 iron-ore price, flat-to-down Pilbara volumes (2026 Pilbara guidance of 323–338 Mt on a 100% basis brackets 2025's outcome), and a Chinese steel market in acknowledged decline. Earnings that are clean but cyclically peaked are still earnings that mean-revert. The auditor's question is not "are these real?" It is "what happens to these when the one input you don't control reverts?" — and the honest answer is that a meaningful chunk of 2025's profit is borrowed from a price environment unlikely to persist.

The diversification that isn't quite there yet

There is a version of this company, perhaps five years out, that genuinely is a balanced transition house: copper from Oyu Tolgoi and Kennecott at scale, lithium from a built-out Arcadium platform into a recovered price, Simandou as a high-grade premium franchise, iron ore as one leg among several rather than the load-bearing wall. That company would deserve a secular multiple. The problem is that investors are being asked to pay for that company now, on the strength of a results release that still derives its cash overwhelmingly from the old exposure. Demonstration is not deployment. A lithium business bought at the bottom of the lithium cycle is a call option, not a cash flow. A copper record is a strong year in a price-taking segment, not a structural re-rating of the whole. The transition portfolio is a real and improving thing — and it is also, today, too small to move the group's fortunes against a multi-dollar swing in the iron-ore price.

The two-track market hiding inside one price

There is a refinement Rio's defenders reach for when the China-demand argument lands, and it deserves a fair hearing because it is partly right — and then a closer look at why it does not save the thesis. The argument runs: iron ore is not one product but two markets stacked on top of each other. There is the high-grade, low-impurity premium ore that decarbonising steelmakers increasingly need to run cleaner, more efficient blast furnaces, and there is the lower-grade bulk tonnage that clears against raw construction demand. Rio, the bulls say, is tilting toward the premium end — Simandou is high-grade, the Pilbara blends are sought-after — and the premium market has its own demand curve, decoupled from Chinese property starts. As the world's mills are pushed by carbon regulation to use better feedstock, premium ore commands a widening price premium, and Rio sits on the right side of that split.

It is a genuine structural tailwind, and it is the most sophisticated piece of the bull case. But it does not rescue the 2026 setup, for two reasons the company's own numbers make plain. First, the premium spread is a margin story layered on top of a volume-and-base-price story, and the base price is what moves the $25 billion EBITDA. A wider grade premium on 342 million tonnes cannot offset a multi-dollar fall in the underlying benchmark; it can only soften it. Second, the decarbonisation-driven premium demand is a multi-year, gradual re-weighting — exactly the kind of slow secular shift that does not show up in any single year's cash flows. The premium thesis is real, and it is also precisely the sort of "someday" the market is being asked to pay for today, while the "now" — falling Chinese steel output, falling property investment, a sub-$100 consensus price — is the part that actually sets next year's dividend.

The asymmetry the multiple ignores

Step back from any single segment and look at the shape of the bet. At today's price, Rio is valued as though 2025's earnings are a reasonable run-rate and the transition optionality is gravy on top. But the distribution of outcomes is lopsided. On the upside, copper keeps compounding, lithium recovers, the grade premium widens, and Rio slowly earns its transition multiple — a good outcome, but a slow one, measured in years, with most of the value far out in the future where discounting bites hardest. On the downside, the China steel downcycle deepens, Simandou and peer supply collide with shrinking demand, the iron-ore price overshoots the gentle $94 consensus to the downside as cyclical bottoms tend to do, and the dividend — a fixed 60% fraction of a falling number — contracts visibly and quickly, repricing the income investors who form a large part of the holder base. The near-term, high-conviction path is the painful one; the rewarding path is distant and uncertain. That is precisely the asymmetry a priced-for-continuation multiple is built to ignore, and it is why a clean, profitable, well-run company can still be a poor place to be standing when the dominant input mean-reverts.

What the bulls genuinely get right

The bull case here is not a fantasy, and any honest short thesis has to concede where it is strong. Rio Tinto is one of the lowest-cost iron-ore producers on earth; even at $94 a tonne, the Pilbara generates enormous free cash flow, because Rio's unit costs sit far below the marginal seaborne producer. A falling price hurts the marginal tonne and the high-cost competitor long before it threatens Rio's cash generation. That cost position is a genuine, durable advantage.

The copper growth is real and well-timed: with Oyu Tolgoi's underground ramping and copper production up 11%, Rio is adding exposure to a metal with a credible long-run supply deficit and structural electrification demand — and unlike iron ore, the copper story has a demand tailwind, not a headwind. The Arcadium acquisition, bought into a depressed lithium price, may prove to be a shrewd counter-cyclical entry rather than the value trap it looks like today; buying a resource base at the bottom is how the best mining returns are made. Simandou, for all that it adds to near-term oversupply, is a multi-decade, high-grade asset that strengthens Rio's structural position in premium ore as steelmakers decarbonise. The balance sheet is strong, the dividend is covered, the management is disciplined on cost and capital. And a 60% payout, even if it swings, is a real cash return from a real business — this is not a cash-burning speculation. Bulls who buy Rio as a cheap, well-run, cash-generative miner with optionality on copper and lithium are not wrong about what they own. The disagreement is narrower and sharper: it is about what you are paying, and which way the dominant exposure is about to move.

The kicker

The genius of the Rio Tinto pitch is that every individual claim in it is true. Copper EBITDA really did hit a record. Lithium really is in the portfolio. Simandou really did pour first ore. The dividend really has been paid at the top of the range for ten straight years. None of it is a lie. It is, rather, a carefully arranged set of true facts that draws the eye away from the one fact that determines the outcome: that the majority of the cash still comes from selling iron ore to Chinese steel mills, and Chinese steel demand is in structural decline while Rio itself adds new supply into the shortfall. You are not buying a transition company that happens to mine iron ore. You are buying an iron-ore company that has bought, at fair prices, a portfolio of options on someday becoming a transition company — and being asked to pay today for the someday.

The market keeps pricing Rio Tinto as a diversified energy-transition miner, but the cash register still rings to a single sound — Chinese steel demand against the seaborne iron-ore price — and that one sound is now fading exactly as Rio's own Simandou turns the volume of new supply up.