ArcelorMittal (MT)

The Global Integrated Benchmark

ArcelorMittal shipped 54.0 Mt of steel in FY2025 from 34 integrated and mini-mill facilities across 14 countries, which is why it sits at the centre of almost every steel comparison: integrated cost structure, multi-region earnings through a cycle, and the multiple the market assigns. The company describes itself as the world's leading integrated steel and mining company, and for once the self-description is analytically useful, because both parts matter.

The footprint splits three ways. Of the 54.0 Mt shipped, Europe took 28.4 Mt, Brazil 13.9 Mt and North America 10.3 Mt, with a small remainder in downstream and tubular units; the AMNS India joint venture's further 7.9 Mt is reported separately, outside the group total. On the production route, FY2025 crude steel of 55.6 Mt ran 74% through blast furnace and basic oxygen furnace (41.2 Mt) and 26% through electric arc (14.4 Mt). So this is a BOF-dominant business with a growing EAF leg, against a global route split of roughly 69% BOF.

At the 9 June 2026 close the equity was worth about $50.4 billion, against net debt of $7.9 billion at year-end (gross debt $13.4 billion less $5.5 billion of cash). That is roughly 1.2x FY2025 EBITDA, comfortably inside the normal zone of our leverage screen.

EBITDA per Tonne: The Middle of the Ladder

FY2025 EBITDA was $6,541 million on 54.0 Mt shipped, which is $121 per tonne. On its own that number means little. Set against the two US extremes, the three companies bracket the entire steelmaking cycle in one fiscal year.

Two definition traps hide in that table. ArcelorMittal reports metric tonnes; Nucor and Cleveland-Cliffs report short tons, and a metric tonne is roughly 10% more steel. Put Nucor's $210 onto a metric basis and it becomes about $232 per tonne, so correcting the units widens the gap rather than narrowing it. And Nucor discloses both total mill shipments (25.3M tons, giving ~$165/ton) and external shipments (19.8M tons, giving ~$210/ton); the external basis is the one comparable to ArcelorMittal's reported group shipments. Our EBITDA per tonne guide works through both traps in detail.

The gap to Nucor is the integrated route and where the steel gets sold. On the corrected metric basis, $232 against $121, an EAF steelmaker selling into the tariff-walled US market earned nearly twice ArcelorMittal's per-tonne EBITDA in FY2025. The market pays for it: Nucor's ~$57.9 billion market cap exceeds ArcelorMittal's on less than half the tonnage. ArcelorMittal earns almost none of that US spread. It sold its US mills to Cleveland-Cliffs in December 2020 and kept only the Calvert finishing complex in Alabama, taking full ownership of that in June 2025; just over half its steel now goes to Europe, where imports set the marginal price. The gap to Cleveland-Cliffs is what the same year did to a US integrated producer without ArcelorMittal's diversification: the spread that left Nucor at $210/ton left Cliffs at roughly break-even. How the spread drives those outcomes is the subject of our steel spread guide.

Where the Cash Goes: The Capex Split

ArcelorMittal's FY2025 capital expenditure is the cleanest capex disclosure in the sector; the split below is why steelmaker multiples stay low.

Read the first row against the EBITDA line. Sustaining capex of roughly $54 per tonne shipped consumed about 44% of the $121/t the business earned in FY2025. Before a dollar reaches growth projects, dividends or buybacks, nearly half of EBITDA goes back into simply keeping the existing plants running at existing volumes. That is the structural reason steelmakers carry low multiples on normalised EBITDA: the headline earnings number overstates the cash an owner can actually take out.

The $0.3 billion decarbonisation line is small today but worth watching. The 74/26 BOF/EAF split is the company's carbon problem in one ratio, and the pace at which that capex bucket grows tells you how quickly the route mix will shift.

Owning the Ore: 72% Self-Sufficiency

ArcelorMittal mined 48.8 Mt of iron ore in FY2025, covering 72% of what its own furnaces consumed, up from 58% a year earlier as the Liberia expansion ramped. An integrated steelmaker consumes roughly 1,370 kg of iron ore per tonne of crude steel, the worldsteel average for that route, plus 780 kg of coking coal. So captive supply is a partial hedge on the input side of the integrated spread, the steel price minus the ore-and-coal basket the mills consume: when iron ore rallies, the mines earn back part of what the mills lose.

The hedge is real but incomplete. The remaining 28% of ore and effectively all of the metallurgical coal are bought at market, so the integrated spread still moves the earnings, just with a damped input leg. It also cuts the other way; cheap ore helps a non-integrated competitor more than it helps ArcelorMittal, because the captive mines earn less at the same time the steel input cost falls.

Valuation: The Through-Cycle Question

Enterprise value of roughly $58.3 billion ($50.4 billion of equity plus $7.9 billion of net debt) against FY2025 EBITDA of $6.5 billion puts ArcelorMittal at about 9x trailing EBITDA, well above our ~4-6x through-cycle band for steelmakers. That gap carries the whole analytical question: how much of it an earnings recovery explains, which turns on whether FY2025's $121/t sits below mid-cycle. The answer requires a normalised EBITDA built from a planning spread rather than last year's print, which is exactly the exercise our through-cycle EV/EBITDA guide walks through, with Cleveland-Cliffs' $37M year as the proof that trailing multiples fail at the extremes.

One adjustment belongs in that exercise. Most of the captive ore never leaves the group, so its value shows up as a lower input cost inside the steel segments rather than as mining profit; the reported Mining segment, which covers only the Canadian and Liberian operations, earned $422 million in FY2025. An iron ore earnings stream does not deserve a steelmaker's multiple, so pulling it out and valuing it separately moves the answer.

The balance sheet does not complicate the question. Net debt of $7.9 billion is about 1.2x FY2025 EBITDA, inside the 1.0-2.5x normal zone of our screen, and FY2025 was not a peak earnings year to be flattering the ratio.

Key Risks

European exposure. Just over half of shipments land in Europe, the region most exposed to Chinese export pressure: Chinese hot-rolled coil, the sheet steel benchmark, was offered at $496-504/mt FOB in June 2026 while the US domestic price sat near $1,193/short ton behind the tariff wall. ArcelorMittal's largest market is the one with the thinnest protection, and European trade policy is the swing variable for its realised prices.

Trade policy on both sides. The tariff wall that shelters Nucor works against ArcelorMittal. Its North American mills sit mostly in Canada and Mexico, so steel they send into the US pays the Section 232 duty, 50% on primary steel since June 2025 and not waived by the North American trade agreement. The company named the tariffs as one reason North American earnings fell in FY2025. The same wall pushes tonnes that would have gone to the US toward Europe, where ArcelorMittal already sells more than half its steel.

Cycle leverage. A $10/t move in the realised spread is roughly $540 million of EBITDA on 54 Mt of shipments, against a sustaining capex bill that does not move at all. The same arithmetic that made FY2025 EBITDA $6.5 billion can halve it without anything going operationally wrong.

The decarbonisation bill. Converting 41 Mt of BOF capacity toward lower-carbon routes will eventually cost far more than the current $0.3 billion a year, and most of that spending defends existing volumes rather than adding new ones. Capital that behaves like sustaining capex but gets reported as transformation deserves the sceptical treatment in any free cash flow model.