Coal is the one industrial raw material nobody manufactures. Swamps, burial and time made it; all we do is find it, lift it and wash the rock off. What happens next splits three ways — burned for electricity, baked into coke to pull the oxygen out of iron ore, or torn apart into gas and rebuilt as plastic and fertiliser.
Rarely in the room with you — but behind a great deal that is.
Every tonne of coal takes the same first three steps: cut from a seam, hauled to the surface, separated from the rock it came up with. Only then does its destiny divide — and the three destinations ask completely different things of it. Tap any stage to look closer.
Coal is not one commodity but a family, and the branch point is decided underground, not in the factory. Whether a seam ends up in a boiler, a coke oven or a gasifier was settled tens of millions of years ago by how deep it was buried and how hot it got.
↓ tap a stage, or any of the three branches ↓
Coal is a sedimentary rock made of lithified plant remains. Understanding one word — rank — explains almost everything about what a given coal can and cannot do.
Coal isn't a substance with a fixed recipe; it's a snapshot of an unfinished process. Every coal is plant matter caught at a different point along the slow squeeze into carbon, and how far it got decides its price, its job and its buyer.
Start with a swamp. Dead plants pile up faster than they can rot, and what accumulates is peat — soggy, spongy, roughly 60% carbon. Bury it under younger sediment and pressure squeezes the water out while temperature climbs with depth. Over millions of years the peat gradually loses its water and volatile compounds, mostly as carbon dioxide and methane, and what stays behind grows steadily richer in carbon. Geologists call this coalification.
Where a coal sits along that sequence is its rank. Low-rank coals — lignite and sub-bituminous, the "brown coals" — weren't buried deep enough or long enough; they still hold a lot of water and give up little heat per tonne. Push further and you reach the "hard" black coals: bituminous, then semi-anthracite, then anthracite at over 95% carbon. The energy released when any of them burns is, quite literally, sunlight captured by plants before the dinosaurs.
Rank rises as burial drives out water and volatiles. Carbon content climbs from about 60% in peat to over 90% in bituminous coal and around 95% in anthracite.
Rank produces the split the market cares about: thermal coal, burned for heat, and coking coal (metallurgical or met coal), which the steel industry buys. The difference isn't just carbon content. Coking coals soften, swell and re-fuse when heated without air, which is what lets them form coke strong enough to hold up a blast furnace. Thermal coal can't do that trick — so the two aren't interchangeable, and met coal fetches the higher price. Met coal can always be burned as fuel; thermal coal can never be coked.
That division is a useful simplification rather than a hard rule. Modern blast furnaces also inject non-coking pulverised coal through the tuyeres, so even the steel route consumes coal that would fail a coking test. And a single mine — sometimes a single seam — can yield products for both markets depending on blending. End use is decided by a full specification: calorific value, ash, moisture, sulphur, strength.
Wet, crumbly and low in energy. Rarely worth transporting far, so it's usually burned in a power station built next to the mine that produces it.
Drier and hotter-burning than lignite but still a thermal fuel. A workhorse of power generation where it's mined cheaply at surface.
The versatile one, and the source of nearly all coking coal. Above 90% carbon. Graded by volatile matter — and the mid-volatile grades prized for coke are the scarcest.
~12% of US met coal output is mid-volHard, dense and nearly pure carbon. Too valuable and too low in volatiles to coke, it's injected into electric arc furnaces as a carbon source that cuts power use and cycle time.
~95% carbonThere are only two ways to reach a coal seam, and which one a mine uses is not really a business decision. It's dictated by how deep the coal sits.
Mining coal is mostly not about moving coal. It's about moving everything that sits on top of it, or holding it up while you cut underneath — and then separating the coal from the rock that inevitably comes with it.
When the seam lies close to the surface, the cheapest path to it is straight down. Blast and strip away the soil and rock above — the overburden — until the black layer is exposed, then dig it out and haul it to a stockpile. It recovers a very high share of the coal in the ground, but the earthmoving is on a scale that's hard to picture: excavators weighing up to 800 tonnes, haul trucks carrying more than 300 tonnes a load.
Past a certain depth, stripping the overburden stops making sense and you go in after the coal instead. In longwall mining a shearer runs back and forth along a face hundreds of metres wide, slicing coal onto a conveyor while hydraulic supports hold the roof over the machine. As the face advances, the supports step forward and the roof collapses under control in the mined-out area behind them. The coal rides a conveyor to the surface; nothing above ground is removed at all.
Coal never comes out of the ground clean. Mixed in with it is rock from the roof and floor, bands of shale inside the seam, and mineral matter that ends up as ash — dead weight no customer will pay to ship. Impurities like sulphur cause worse trouble: corrosion in the plant, and sulphur dioxide in the air, which drives acid rain.
So the coal goes to a preparation plant, where separation exploits one simple difference: coal is lighter than rock. Suspend the crushed material in a dense fluid and the coal floats while the waste sinks. What emerges is cut to a specification the buyer's plant expects — because a power station or coke oven tuned for one coal quality performs badly on another.
This is a physical change, not a chemical one. Washing separates coal from rock by density; it does not alter the coal itself. Ash and sulphur bound up inside the organic structure of the coal survive washing untouched — which is why seam chemistry, decided in the ground, still sets the ceiling on product quality.
About 5.5 billion tonnes a year go into power stations. This is coal at its simplest: a carbon store, oxidised to release heat.
Coal's share of the world's electricity is falling while the absolute tonnage burned keeps setting records. Both are true at once, because global electricity demand is growing faster than coal is being displaced.
The mechanism barely needs explaining. Pulverise the coal, blow it into a furnace with air, and the carbon reacts with oxygen to release heat. The heat boils water, the steam spins a turbine, the turbine turns a generator. What's interesting is the position it occupies: coal-fired generation hit its highest recorded level in 2024 at 10,766 terawatt hours, and remains the single largest source of electricity worldwide.
Combustion is complete oxidation. Almost every carbon atom in the coal ends up as a molecule of carbon dioxide — which is precisely why coal is the most carbon-intensive way to make a unit of heat, and why the tonnage burned maps so directly onto emissions.
The real shift is in what the fleet is for. Coal's share of the global electricity mix stood at 35% in 2024, and the IEA forecasts 27% by 2030 as solar and wind take most new supply. But installed capacity isn't falling nearly as fast as the share. The plants simply run less — ramping up when the wind drops and reservoirs run low, idling when they don't. Coal is drifting from baseload workhorse to system backup: a smaller job, but a stubborn one.
That drift explains why year-to-year numbers look contradictory. In the first half of 2025 US coal demand rose an estimated 12% and EU demand around 5% — not because either is building coal plants, but because low hydro and wind output and higher gas prices pushed the existing fleet harder. Weather and gas prices now move coal consumption more than any structural trend does.
Around 1,114 million tonnes a year of met coal go to the steel industry. Here, coal's job is chemical, and the heat is almost a side effect.
In a blast furnace, coal isn't burned for its energy — it's consumed for its carbon, which grabs the oxygen out of iron ore. Nothing else currently does that job at scale and at cost, which is why this use of coal is so much harder to replace than burning it.
Iron ore is not iron. It is iron locked to oxygen, and turning it into metal means finding something that wants the oxygen more than the iron does. Carbon does — and coal is where the carbon comes from.
But raw coal won't do. Dropped into a blast furnace it would crumble, choke the gas flow and collapse under the burden above it. So it is baked into coke first: heated above 1,000 °C in a sealed oven with the air excluded, so it cannot burn. The volatiles boil off, leaving a hard, porous lump of nearly pure carbon. Porous lets gas percolate upward; hard carries the column of ore stacked above. This is the whole reason only coking coals work — they must soften and re-fuse into that structure rather than fall apart.
Coke burning at the base makes carbon dioxide; that gas meets more hot coke and becomes carbon monoxide, the actual reducing agent, which strips oxygen off the ore as it rises. The carbon leaves as CO2 either way — these emissions are inherent to the chemistry, not a sign of inefficiency.
The quantities are large: roughly 750 kilograms of steelmaking coal per tonne of steel. A single offshore wind turbine needs about 1,050 tonnes of steel, and so carries around 790 tonnes of coal behind it. The energy transition is, in this narrow sense, built on coal.
The electric arc furnace route mostly sidesteps this by melting scrap with electricity — about 72% of US output. But even EAFs aren't coal-free: carbon must be added to the melt to control chemistry and foam the slag, and anthracite is one of the materials used. Less coal, not no coal.
What really makes this use resist substitution is scarcity. Coke-making prizes mid-volatile coals, the rarest grade — only about 12% of US metallurgical output, so plants blend low- and high-vol coals to approximate it. Supply is geographically concentrated too: roughly 40% of the world's exported steelmaking coal comes from one Australian state, Queensland.
Over 400 million tonnes a year — overwhelmingly in China — are gasified rather than burned, and rebuilt into the molecules a petrochemical plant would normally make from oil or natural gas.
Gasification is not combustion. Feed coal steam and just enough oxygen to react but not fully burn, and it breaks into carbon monoxide and hydrogen — a mixture called syngas that is a chemical building block, not a fuel. From syngas you can assemble methanol, ammonia, plastics and diesel. Coal becomes a way to make things, not a way to make heat.
The purpose here is substitution, and it's strategic before it's economic. A country with a lot of coal and not much oil can use gasification to build a chemical industry that doesn't depend on imported crude. That is essentially why China built the sector at scale. In 2024 about 7% of Chinese coal — the world's largest coal consumption — went to chemicals rather than energy.
The first reaction absorbs heat rather than releasing it — the opposite of burning — which is why a gasifier has to be fed energy to run. The second, the water-gas shift, is where most of the carbon problem lives: coal syngas has too little hydrogen for polymer chemistry, so carbon monoxide is deliberately traded away for hydrogen, and the discarded carbon leaves as CO2.
The output is genuinely competitive: China's plants make chemicals and polymers of comparable quality to the petrochemical route, using methanol-to-olefins technology to reach the plastics that oil crackers normally supply. Sustained high oil prices since 2018 made the economics work.
The cost is carbon, and it's worse here than burning. Because only a fraction of the coal's carbon ends up in the product — the rest is shed in the shift reaction — making ammonia from coal instead of natural gas roughly triples the carbon dioxide released. Chinese coal gasification alone emits over 1.1 billion tonnes of CO2 a year. Sasol's Secunda complex in South Africa, which consumes more than 30 million tonnes of coal annually, is reported to be the largest single-point source of greenhouse gases in the world.
And the sector is growing. Planned coal-to-chemical plants nearly doubled in a year to 47 — 21 in China, 14 in India, the rest in Indonesia, Kazakhstan, Botswana and Pakistan — together consuming at least 145 million tonnes of coal a year. Advocates point to a real technical advantage: gasification yields a concentrated CO2 stream before synthesis, making it one of the cheaper places to bolt on carbon capture. The NGO coalition that compiled the count argues none of the plants would be viable without subsidy.
The headline figures for coal look like a contradiction until you separate absolute volume from relative position.
The striking thing about that chart is how little of this coal travels. Most is burned in the country that mined it — China and India both treat domestic coal as an energy-security asset rather than a traded commodity. International trade passed 1.5 billion tonnes for the first time in 2024, but that is still only about 18% of global demand. Coal is a local business with an international fringe.
Split by end use, the pattern is stable. Roughly 5.5 billion tonnes a year go to electricity; heavy industry, chiefly steel and cement, takes around 2.5 billion tonnes; gasification for chemicals consumes over 400 million tonnes. Metallurgical coal demand sat at about 1,114 Mt in 2025, broadly flat against 2024, with China alone accounting for two-thirds. The IEA expects it to drift to roughly 1,061 Mt by 2030 as electric arc furnaces gain ground — a decline, but a slow one.
Coal is usually discussed as a single question. It is really three, and they have very different answers.
The easy part of coal to displace is the biggest part. The hard parts — coke for iron ore, carbon as chemical feedstock — are smaller, but they are the uses where coal is doing something other materials currently cannot do at all.
Burning coal for electricity is, technically, the solved problem. Solar and wind are taking most new generating capacity, and coal's share of the mix is forecast to fall eight percentage points inside six years. What keeps the tonnage high isn't that the alternatives don't work — it's that electricity demand is growing so fast that displacing coal's share doesn't yet displace coal's volume. That gap closes eventually. It just hasn't yet.
The steel problem is different in kind. Coal in a blast furnace isn't supplying energy something else could supply — it's supplying carbon atoms that strip oxygen from iron ore. Replace the energy and the reduction still has to happen. The credible routes are scrap-based electric arc furnaces, limited by how much scrap exists, and hydrogen direct reduction, which is real but early. Meanwhile the US Department of Energy argues the opposite case from a security angle: that metallurgical coal should be classified a critical material, because its supply is concentrated, hard to expand, and essential to the steel that underpins every piece of energy infrastructure built.
The chemicals problem is the most uncomfortable, because it's the one actively getting bigger. Emissions per tonne of product are worse than burning coal outright, yet planned capacity nearly doubled in a year. It persists not because it's cheap or clean but because it substitutes a domestic resource for an imported one — and energy security has proven a more durable motivation than cost.
Renewables plus storage take over baseload, leaving coal plants as backup capacity that runs rarely rather than constantly.
35% → 27% of the electricity mix by 2030Electric arc furnaces on scrap, and hydrogen direct reduction for fresh ore, remove coke from ironmaking — where the inputs exist.
Met coal ~1,114 Mt now → ~1,061 Mt by 2030Gasification yields a concentrated CO2 stream before synthesis, making capture cheaper here than at almost any other industrial source.
Against >1.1 Gt CO₂/yr from Chinese gasificationWhat makes coal hard to reason about is that none of these tensions resolve on the same timescale. Power-sector coal is declining in share while growing in volume; metallurgical coal is nearly flat; coal-to-chemicals is expanding outright. "Coal is in decline" is true of one and false of another, depending entirely on which number you pick up. The honest version: the world is mining more coal than it ever has, in an energy system where coal matters relatively less every year — and the last uses to go will be the ones where coal was never really a fuel at all.