The Report’s Importance: It’s Not Just an Event, It’s What It Reveals
On March 6, 2026, ArcelorMittal announced the release of its Annual Report 2025 on Form 20-F to the U.S. SEC and its 2025 annual report to the Luxembourg Stock Exchange, with a printed copy of the 20-F available upon request. It marks an administrative milestone, yes, but in corporate finance, the procedures are just smoke; what matters is the "electrocardiogram" they reveal.
Fiscal 2025 ended with sales of USD 61.352 billion (a slight decrease from 2024), EBITDA of USD 6.5 billion, and net income of USD 3.152 billion. The standard headline here would be "resilience." I translate that into risk language: the company managed to sustain cash and improve balance sheet quality in a business that punishes rigidity. And it did so by moving three levers that, when functioning together, reduce the likelihood of accidents: liquidity, capex discipline, and partial control of its critical raw material (iron ore).
There’s also a second layer. The report formalizes a narrative of energy transition (electric furnaces, renewables, electric steels) and returns to shareholders (dividends and buybacks). The proof is not in the narrative but in the structure: how much of the "plan" is financed by real cash versus promises, and how much leaves room for error without breaking down.
The Hard Numbers: Sufficient Profitability, Watchful Cash Flow, and a Cushioned Balance Sheet
The most useful data from 2025 isn’t revenue growth; in steel, revenues are largely dependent on the climate, not DNA. The useful figure is the ability to convert operations into cash without stretching the balance sheet.ArcelorMittal reported EBITDA of USD 6.5 billion and EBITDA per ton of USD 121, which the company describes as more than double the previous cycle’s lows. In a capital-intensive business, the margin per ton is like the "spread" in credit: when it compresses too much, the structure suffers; when it holds, the company can breathe and decide.
In terms of cash, the metric worth monitoring is the “investable cash flow” over the last twelve months: USD 1.9 billion as of December 31, 2025, practically in line with 2024 (USD 2.0 billion). Translated: there isn’t a cash leap showing extreme tailwinds; there’s consistency. And consistency is what lowers refinancing risk and the risk of desperate decisions.
In capital allocation, the company executed USD 1.1 billion in strategic capex and returned USD 0.7 billion to shareholders (USD 0.4 billion in dividends, USD 0.3 billion in buybacks). This is significant for an uncomfortable reason: a company can only sustain buybacks if its structure is not on the edge. Buybacks can be a façade when cash is artificial; here, cash doesn’t seem exuberant, but stable enough to sustain moderate returns without mortgaging flexibility.
The balance sheet picture supports this: net debt of USD 7.9 billion, with total liquidity of USD 11.0 billion (gross debt USD 13.4 billion and cash USD 5.5 billion). Additionally, in 2025 it achieved rating upgrades: Moody’s to Baa2 (stable) and S&P to BBB (stable). The upgrades are not moral medals; they are an external reading that credit risk has decreased. In a cyclical industry, maintaining investment-grade status serves as insurance: it lowers the cost of debt and, crucially, prevents a bad year from turning into a financing crisis.
There’s an additional point: the net income includes an exceptional gain of USD 871 million, largely associated with acquiring the 50% stake from Nippon Steel in ArcelorMittal Calvert, partially offset by a USD 0.4 billion charge related to a purchase price dispute in Brazil and other items. This necessitates separating “result” from “engine.” The engine seems healthy, but it’s prudent not to over-read the income as recurring.
Iron Ore Integration: When Risk Lies in Inputs, Margins Are in the Mine
Steel is a cost war masquerading as a heavy industry. In that war, iron ore is ammunition. ArcelorMittal reports iron ore reserves of ~3.7 billion tons and, more importantly, a leap in self-sufficiency to 72% in 2025, up from 58% in 2024, driven by expansion in Liberia and other assets. The declared goal includes ramping Liberia to 20 Mtpa.From a risk perspective, this matters for two reasons. First, it reduces exposure to price shocks and logistical frictions. It does not eliminate the commodity risk, but it changes its shape. Instead of being a buyer fully exposed to spot markets, it becomes a player with partial "natural" coverage. In portfolios, it’s akin to having an asset that pays you when your main cost rises.
Second, vertical integration introduces a new risk: execution and capex, as mining is also capital-intensive and susceptible to operational issues. The difference between natural coverage and a black hole is discipline in the pace of investment and the quality of assets. With the available data, what is observed is direction, not detailed performance by mine. Nonetheless, the leap from 58% to 72% in a year doesn’t sound like marketing; it sounds like effective expansion.
In a cyclical industry, controlling the critical input enhances competitive positioning during the low points of the cycle. When steel prices fall, a producer with cheaper inputs can last longer without destroying their balance sheet. It’s corporate natural selection: it’s not the biggest that wins; it’s the one that doesn’t run out of oxygen.
But let’s be clear: integration is not synonymous with immunity. If the cycle turns down and the market punishes everyone, having your own mine doesn’t mean you “win”; it merely means you lose less. And losing less in markets with high operational mortality is a huge advantage.
Energy Transition: Real Investments, But Risks Lie in Sequence and Modularity
ArcelorMittal positions itself as an enabler of energy transition and outlines concrete objectives: 2.8 GW of renewable generation capacity by 2028, expansion of electric arc furnaces (EAF) by an additional 3.4 million tons by the end of 2026, and growth in electric steels for automotive targeting 0.4 million tons of NOES by 2028. Additionally, it reports USD 335 million in R&D spending in 2025 across 14 sites in 9 countries.Thus far, the temptation is to applaud the "green future." I prefer a portfolio approach: these investments are options. Some will succeed, others won’t. The technical question is not formulated with question marks; it’s answered through financial design: if the plan partially fails, the company remains alive.
The EAFs and the renewable portfolio aim for two things: a smaller carbon footprint and potentially more controllable energy costs. In a volatile energy pricing environment, that may stabilize margins. However, there exists the typical heavy industry risk: large projects that are delayed and consume cash. The reassuring data, for now, is that the company maintained stable investable cash flow and retained liquidity of USD 11.0 billion. That doesn’t guarantee perfect execution; it does reduce the likelihood that a tactical deviation becomes a solvency event.
The most interesting component is the electric steels (NOES), as it is a bet on product mix, not merely on process. In steel, climbing the value chain is the cleanest way to defend margin without praying for the cycle. The risk is market-driven: demand, competition, and technical approvals. Without contract or penetration numbers, it’s best viewed as an exploration with uncertain probability, backed by a not-inconsequential R&D muscle.
Concurrently, the company communicated progress in its three-year safety program, with tangible progress in KPIs and prevention of fatalities. No specific figures are available in the extract provided, meaning there’s nothing to quantitatively audit. In terms of operational risk, it does signal that management understands that in heavy assets, a serious incident is not just a human and legal cost; it’s also production interruption, risk premium, and reputational damage.
Returns to Shareholders: Discipline When the Business Does Not Fall in Love with Its Own Cycle
The report includes a more explicit returns policy. The board proposes a FY 2026 dividend of USD 0.60 per share, up from USD 0.55 in 2025, and reaffirms the commitment to return at least 50% of free cash flow post-dividend via buybacks.In a cyclical business, this kind of promise can be dangerous if it becomes too rigid. When the cycle turns, rigidity kills. What can be assessed with the available data is whether, at the cutoff date, ArcelorMittal has the capacity to sustain a returns framework without increasing fragility: controlled net debt, investment-grade rating, ample liquidity, and positive investable cash.
The key is the order of operations: first, maintain the core running; then invest selectively; and only then return capital. In 2025, they returned USD 0.7 billion, a material but not aggressive figure compared to the size of the group and its EBITDA. It’s a return that seems calibrated, not euphoric.
There’s also an implicit detail of financial governance: when a company commits to repurchasing shares based on free cash flow, it imposes a rule that competes against the expansion of capex for capex’s sake. It’s a way to limit the classic bias of industrial management to build more assets simply because they can.
None of this eliminates the structural risk in the industry: steel will remain cyclical, exposed to global prices, energy, and macroeconomic factors. What does change is the probability of a shock finding them with a fragile balance sheet. In terms of portfolio management, ArcelorMittal is rebalancing towards less tail risk through input integration, liquidity, and allocation discipline.
What This 20-F Makes Clear About Structural Survival
The announcement of the 20-F is a formality; the content is the signal. ArcelorMittal closed 2025 with sufficient operational profitability, stable investable cash, moderate returns to shareholders, rating improvements, and a significant step in iron ore self-sufficiency to 72%. Meanwhile, it maintains an investment plan regarding energy transition and R&D that, for now, doesn’t appear to be financed by stretching the balance.The cool reading is this: the company is buying resilience with textbook tools —liquidity, investment-grade status, and partial input integration— and that architecture reduces the typical fragility of the sector in light of cyclical deterioration. Structural survival seems reasonably protected as long as capex discipline and liquidity cushions remain real restrictions in the system.
