How rock dug from the earth becomes the single most important structural material of the modern world — and where the industry that makes 1.85 billion tonnes of it a year is heading next.
More of it is made each year than every other metal combined.
Dig up the ore, melt it down, then pour and freeze it into the shapes that become everything else. Tap any stage to look closer.
↓ tap any stage to explore ↓
Iron is hidden inside rock because it's chemically bonded to oxygen. Everything that follows is about breaking that bond.
Steel is deceptively simple: an alloy of iron and carbon, with carbon usually between 0.02% and 2.1% by weight. That small dose of carbon — plus pinches of manganese, chromium, nickel or molybdenum — is what turns brittle raw iron into a material strong enough for skyscrapers yet workable enough for a soup can. But iron doesn't sit in the ground as metal. It sits there locked to oxygen, and freeing it means moving mountains and building furnaces hot enough to separate iron from rock.
Mostly hematite (Fe2O3) and magnetite (Fe3O4). Raw ore runs 50–70% iron; the rest is waste rock called gangue. Chief sources: Australia, Brazil, China and India.
The furnace needs something that will pull oxygen away from the iron. Carbon does exactly that — so coal is baked at 1,000–1,100°C in air-free ovens for 12–18 hours until it becomes coke, a hard, almost pure carbon fuel that both heats the furnace and does the pulling.
Ore arrives mixed with unwanted minerals. Limestone combines with them to form a new molten material — slag — that is lighter than iron, so it floats on top and can simply be poured off. It's later reused in cement and roads.
Steel is really iron with less carbon than the stuff that leaves the blast furnace. Making it is a two-step job: first add carbon to free the iron, then take most of it back out.
Almost all the world's steel is born one of two ways. The primary route smelts fresh iron from ore in a blast furnace, then burns off its excess carbon in a converter. The secondary route skips the ore entirely and re-melts scrap steel with electricity. Globally the primary route still dominates — 69.4% oxygen-blown against 30.3% electric in 2025 — but the balance swings sharply by region: China runs 89% BF-BOF, while the United States is 71% EAF and Türkiye 72%.
That leaves 0.3% unaccounted for — about 5 Mt made in open hearth furnaces, a slow 19th-century method that the oxygen converter made obsolete in the 1960s and 70s. It survives almost entirely in Ukraine, where it still accounts for over 40% of output, and in parts of Russia. It's the last echo of how the world made steel before the modern process arrived.
Then they converge. Whichever route made it, liquid steel is fine-tuned in a ladle and then continuously cast. Instead of filling one mould at a time, a modern plant pours an endless ribbon of steel that solidifies as it travels and is cut to length while still glowing — the change that let steelmaking run around the clock. Reheated to ~1,200°C, it's hot-rolled into strip, bars, rails or wire, and sometimes cold-rolled, coated or heat-treated for the final finish.
Chromium protects steel by forming its own invisible armour — and rebuilding it whenever it's damaged.
Add at least 10.5% chromium to iron and carbon and you get stainless steel. The chromium reacts with oxygen in the air to form a film so thin it's invisible, which seals the metal underneath. The clever part is what happens when you scratch it: oxygen immediately rebuilds the film over the exposed metal. Ordinary steel, once its surface is broken, just keeps rusting inward — stainless repairs itself. Tuning the other ingredients splits it into distinct families, each a trade-off between corrosion resistance, strength and workability.
The most common family. Grade 304 is the classic "18/8" (18% chromium, 8% nickel); 316 adds molybdenum for "marine grade" resistance to saltwater and pitting.
10.5–27% chromium, low nickel — cheaper, magnetic, good corrosion resistance. Grade 430 shows up in kitchen utensils, appliance trim and automotive detailing.
More carbon means it can be heat-treated to great hardness and strength. Used where an edge matters — surgical instruments, cutlery and tooling.
A blend of austenitic and ferritic phases — stronger than 304 or 316 with excellent resistance to stress-corrosion cracking. Favoured in oil, gas and chemical plants.
The world has built far more steel plant than it can keep busy — and the gap is still widening.
Steel is a 1.85-billion-tonne-a-year business. But the more revealing number isn't what the industry makes — it's what it could make and doesn't. The world can produce roughly a third more steel than it currently sells, and that gap has been growing for five years.
The overcapacity problem, in one number. Enough steelmaking plant sits idle worldwide to supply more than three times India's entire annual output — furnaces built, financed and standing still. In figures: capacity of 2,445 Mt against production of 1,849 Mt in 2025, leaving about 596 Mt unused, or a utilisation rate near 76% — well below what the industry considers healthy.
Global capacity hit a record 2,445 Mt in 2025 — a fifth straight year of expansion — while production fell for a fourth consecutive year. The OECD puts excess capacity at 640 Mt in 2025, already more than 200 Mt above everything the entire OECD bloc produces, and projects 745 Mt by 2028 as planned additions of up to 139 Mt outpace demand growth of just 0.9% a year. Utilisation could slide to 74% or lower. The OECD attributes much of the expansion to subsidised capacity outside its member countries.
Steel's supply chain is one of the largest movements of material on Earth, and the countries at each stage are mostly different countries. Where the maps fail to line up is exactly where the ore ships, the exports flow, and the overcapacity bites.
Every one of those gaps is closed by a ship. Roughly 1.77 billion tonnes of iron ore and another 95 Mt of scrap cross borders each year, and China alone exported a record 133.6 Mt of finished steel in 2025. It is why the industry is so exposed to freight costs, tariffs and trade policy — and why one country's surplus capacity becomes everyone else's price problem.
Replacing carbon means changing the chemistry, not just changing the fuel. Carbon isn't only the furnace's fuel — it's the tool that frees the iron.
The blast furnace's chemistry has a built-in catch: carbon is the very thing that strips oxygen from iron ore, so CO2 is unavoidable in the traditional route. That makes decarbonising steel one of the defining industrial challenges of the century — and the reason so much money is flowing into new ways of making it.
Carbon does two jobs in a blast furnace: it heats the place, and it takes the oxygen off the ore. Swapping in hydrogen replaces both — and the second job doesn't behave the same way.
That's why a hydrogen furnace has to preheat its gas — HYBRIT plans to do it with electricity. Hydrogen isn't a cleaner fuel poured into the same process; it's a different reaction.
Hydrogen reduction is fussier about its raw material than a blast furnace is. It needs DR-grade pellets — purer ore, with less waste rock to get in the way. Hematite ores suit it best because the gas can work through them quickly; magnetite can be used but usually needs extra processing first.
The IEEFA ranks securing that ore as one of the two foremost obstacles to hydrogen steelmaking, level with building the hydrogen supply itself. Miners are responding by opening high-grade deposits such as Iron Bridge in Australia and Simandou in Guinea.
The transition is limited not just by how much green hydrogen exists, but by how much premium ore does.
Drive the process with clean electricity instead of coke.
Melt scrap in electric furnaces — no fresh ore needed at all.
Trap the CO2 from processes too hard to electrify.
Researchers modelled global steelmaking out to 2060 under three futures, each assuming a different level of climate ambition, and measured how much the CO2 released per tonne of steel would drop compared with 2020. The scenarios aren't predictions — they're a way of asking how much the technology can deliver if the world pushes hard.
The uncomfortable finding is the last one. Even the most aggressive pathway leaves roughly a fifth of today's emissions per tonne still in place by 2060 — and because the world is expected to be making more steel by then, total emissions fall by less than these per-tonne figures suggest. Technology alone doesn't finish the job.
The bottom line: carbon isn't just steel's fuel — it's the tool that frees iron from rock, so cleaning up steel means rebuilding the chemistry, not swapping the energy source. The technology to do that already exists and works. What decides whether it spreads isn't new demand — there isn't any — but replacement and regulation: which route gets chosen when today's ageing furnaces reach the end of their lives, and whether carbon costs enough by then to make the clean option the obvious one. The danger is that cheap, subsidised, overbuilt capacity keeps prices low, the retirements that should happen don't, and the decision keeps getting deferred. And even the most ambitious pathway leaves roughly a fifth of today's emissions per tonne in place by 2060 — so the rest will take clean electricity, better ore, more recycling and using less steel in the first place.