Where does coal come from?

Coal is not one product. Thermal coal, also called steam coal, is burned to raise steam for electricity and industrial heat. Metallurgical coal, also called coking coal, is baked in the absence of air to make coke, the carbon and structural support that lets a blast furnace turn iron ore into pig iron. The two trade separately, at different prices, and a mine usually sells into one market or the other. Coal is also graded by rank, the degree to which burial and heat have driven off water and volatile matter: lignite (brown coal) at one end, then sub-bituminous, bituminous and anthracite at the other. The Energy Institute figures used on this page count commercial solid fuels by weight, hard coal plus lignite and sub-bituminous coal together, so a tonne of wet lignite and a tonne of anthracite count the same even though they carry very different amounts of energy.

In 2025 China mined 4,663 million tonnes, 52% of the world's 9,053 million tonnes (Energy Institute). That single share is the most important fact on the page: no other commodity of this size is so concentrated in one country, and Chinese mine output is set by domestic power demand and by safety and permitting rules rather than by any world market. India follows with 12% and Indonesia with 8.7%, then United States (5.3%), Australia (4.8%) and Russia (4.7%).

Geology set the map. Most bituminous coal formed in the swamp forests of the Carboniferous and Permian periods, which is why the great basins lie in ancient continental interiors: the Ordos and Shanxi fields of northern China, the Gondwana basins of Jharkhand, Odisha and Chhattisgarh in India, the Kuznetsk basin in Russia, the Highveld of South Africa, and Appalachia and the Illinois basin in the United States. Younger, softer coals dominate elsewhere. Indonesia's Kalimantan and Sumatra deposits are mostly sub-bituminous with high moisture, which is why the country ranks high by tonnes and lower by energy. Australia is unusual in holding both a large thermal resource in the Hunter Valley and the premium hard coking coal of Queensland's Bowen basin.

Output has been moving between countries rather than disappearing. 34 countries reported production in 2025, and everything outside the top ten came to 6.9%. World output changed +14% over the ten years to 2025 and +18% over five, while the composition shifted: United States contracted at -5.1% a year over that decade and Germany at -7.3%, against +4.9% for India and +15.9% for Mongolia, whose coking coal goes overland to Chinese steel mills.

Coal is the least traded of the big bulk commodities relative to how much of it is produced. Most of it is burned in the country that mined it, close to the pithead, because it is heavy, low in value per tonne and expensive to move. The export table therefore describes a different industry from the production table: a seaborne market supplying countries that burn far more coal than they dig. In 2024 the largest exporter of coal (HS 2701) was Australia, with 34% of the world's $179 billion in export value (CEPII BACI), followed by Indonesia (19%), Russia (13%), United States (11%) and Mongolia (4.6%), which is landlocked and moves its coking coal to China by truck and rail rather than by ship.

The buyers are the industrial economies of East and South Asia. China took 22% of the world's $179 billion of imports in 2024, ahead of India (18%), Japan (15%) and South Korea (8.6%) (CEPII BACI). China and India both mine more coal than anyone except each other and still import heavily, because inland mines cannot always reach coastal power stations more cheaply than a ship can. Netherlands appears high on the import list mainly as a transit and blending point for the Amsterdam-Rotterdam-Antwerp hub rather than as a final consumer. Processed coal shows a different pattern again: the leading exporter of coke and semi-coke (HS 2704) in 2024 was Poland with 20% of that $10.4 billion trade, just ahead of China, and the largest importer was India (17%).

There is no world price for coal. There are regional benchmarks for cargoes of a defined specification, and everything else is priced as an adjustment to one of them. The two contracts that matter for seaborne thermal coal both trade on ICE Futures Europe. The Rotterdam Coal contract (API2, ticker ATW) covers 1,000 tonnes per lot, is quoted in US dollars and cents per tonne, and is financially settled against the API 2 index published in the Argus/McCloskey Coal Price Index Report for cargoes delivered into Amsterdam-Rotterdam-Antwerp. The globalCOAL Newcastle contract (ticker NCF) is also 1,000 tonnes per lot in US dollars per tonne and settles against the Newcastle index for coal loaded in New South Wales, which is the reference for the Pacific basin.

The series charted on this page is Australian thermal, Newcastle. It stood at $135.2/t in August 2026, up 20% from a year earlier (World Bank Pink Sheet). Its nominal record was $430.8/t in September 2022, during the European energy crisis; adjusted for United States inflation the real high was $456/t in September 2022, and the lowest monthly average in a series that begins in 1970 was $7.8/t in January 1970. Over ten years the series changed +101% and over five -20%. The companion Atlantic reference, South African thermal, Richards Bay, was $96.8/t in August 2026, +4% from a year earlier, and its own record was $302/t in April 2022.

A quote only means something once you know what it describes. Newcastle and Richards Bay prices are FOB, free on board at the loading port, so the buyer pays freight; API 2 is CIF, cost, insurance and freight included, into northwest Europe. Both refer to a standard energy content, conventionally 6,000 kilocalories per kilogram net as received, and cargoes of higher or lower energy are priced pro rata, with penalties for ash, sulfur and moisture written into the contract. Two ambiguities follow. First, both price series on this page are thermal coal; metallurgical coal is quoted separately, usually off premium hard coking coal FOB Australia assessments, and can trade at a large multiple of the thermal price. Second, production here is measured in tonnes across all ranks, so a country that mines wet lignite looks larger by weight than it does by the energy it actually produces.

China mines and burns roughly half the world's coal, so administrative decisions in Beijing move the seaborne price more reliably than any demand signal. When output is restricted for safety inspections, consolidation or environmental targets, Chinese utilities buy imported tonnes and the Pacific benchmark rises; when the state pushes mines to produce, imports are displaced and prices fall. The 2016 rule capping mine working days at 276 a year is the clearest case, cutting domestic output and lifting prices across Asia within months.

Coal is the swing fuel in most Asian power systems, so it absorbs the residual after hydro, nuclear, wind and solar have run. A hot summer in China or India, a cold snap in northeast Asia, or a weak monsoon that empties hydropower reservoirs all raise burn and draw down stockpiles at power stations. Because utility stock levels are reported and watched, a run of unusual weather shows up in the price long before it shows up in annual consumption data.

In Europe and parts of Asia, gas and coal plants compete to run, and the switching point depends on the relative cost of fuel plus the price of carbon allowances. When gas becomes expensive, as it did across Europe in 2022, utilities dispatch coal units harder and bid up API 2 cargoes; when gas is cheap, coal plants sit idle. This links the coal price to LNG and pipeline gas markets even in years when coal supply itself is unremarkable.

Coal is worth little per tonne, so ocean freight is a large share of the delivered cost and small changes in capesize and panamax rates move the economics. When the gap between the Atlantic and Pacific benchmarks exceeds the cost of the voyage, Colombian, South African and United States cargoes head east and Australian cargoes head west, closing the spread. Congestion, canal restrictions and longer routings all raise the effective freight cost and widen regional differences.

India mines more coal every year and still imports, because its domestic coal is high in ash, its inland railways are congested and its coastal power stations and cement plants can often buy an imported cargo more cheaply than a domestic tonne railed a thousand kilometers. Indian buying is price sensitive rather than fixed: when the benchmark falls, imports rise sharply, which puts a floor under the market and makes India the marginal buyer for Indonesian and South African coal.

Coal trade responds to politics faster than most bulk markets because cargoes are fungible and voyages are short. China's unofficial halt to Australian coal purchases from late 2020 and the European Union's embargo on Russian coal from August 2022 did not change how much coal existed; they changed who bought it from whom, lengthened voyages and widened the spread between benchmarks. Re-routing raises freight costs and creates persistent regional discounts and premiums.

Metallurgical coal is priced by blast furnace economics, not by power demand. When steel margins are good, mills run hard and bid for premium hard coking coal with low ash, low sulfur and strong coke strength; when construction slows, coking coal falls even if thermal coal is firm. Supply is far more concentrated than for thermal coal, so a cyclone in Queensland or a rail outage on the Bowen basin lines moves the coking price on its own.

Policy affects supply through finance more than through demand. Banks, insurers and export credit agencies have withdrawn from new thermal coal projects, and permitting has become slower in most jurisdictions, so replacement capacity is not being built at the rate depletion would imply. That makes the supply curve steeper: when demand rises unexpectedly, there is less idle capacity to bring on, which is one reason coal prices have spiked more violently in recent cycles than in the 1990s.

Coal begins as peat in a waterlogged forest floor where dead plant matter accumulates faster than it rots. Burial under sediment applies heat and pressure, driving off water, oxygen and volatile compounds and concentrating carbon. That progression defines rank, and rank determines use: lignite is burned close to the mine because it is too wet to ship economically; sub-bituminous coal, such as Indonesia's, is a cheap power fuel; bituminous coal covers both high-energy thermal grades and the coking coals; anthracite is a small, specialized market.

How it is mined depends on how deep the seam sits. Where coal lies near the surface, as in Indonesia, the Powder River Basin and much of Australia, it is stripped: overburden is removed by dragline, shovel or truck, the seam is drilled, blasted and loaded, and the ground is reshaped behind the pit. Deeper seams are worked underground. Longwall mining shears a face several hundred meters wide while hydraulic supports hold the roof and let it collapse behind, and it is the highest productivity underground method; room and pillar mining, older and more flexible, leaves pillars of coal standing to hold the roof and recovers less of the seam.

Raw coal from the pit is rarely what the customer buys. At a preparation plant it is crushed, sized and washed, usually in dense medium baths or cyclones that float coal off the heavier rock and pyrite. Washing raises the energy content and cuts ash and sulfur, but it also loses mass, so a mine's saleable output can be well below what it digs. Thermal coal is then blended to hit a contract specification for calorific value, ash, sulfur, moisture and volatile matter. Metallurgical coal goes further: in a coke oven battery it is heated to around 1,000 degrees Celsius without air for about a day, driving off volatiles to leave porous, strong coke, and capturing coke oven gas, coal tar, ammonia and light oils as by-products.

Seasonality in coal is a demand and logistics phenomenon rather than a harvest. Stockpiles build in shoulder seasons and draw down in summer and winter peaks; the Australian cyclone season from November to April interrupts Queensland loading; the monsoon floods Indian and Indonesian open pits; and river levels on the Rhine and the Barito determine how much coal can move by barge in any given month.

The dominant use of coal is electricity generation, and the second is iron and steel, where coke is both the reductant that strips oxygen from iron ore and the permeable skeleton that lets gas flow through a blast furnace charge (International Energy Agency, Coal 2022). Cement is the third large user: coal fires the kiln that turns limestone into clinker, and the ash often ends up in the cement itself. Beyond those three, coal supplies industrial process heat, district heating in parts of Europe and Asia, and a chemical industry that is largely Chinese, where coal is gasified to make methanol, ammonia for fertilizer, and olefins that elsewhere come from oil or gas.

The by-products of coking are a small but old chemical industry in their own right: coal tar for pitch, creosote and carbon black feedstock, ammonia for fertilizer, and benzene and toluene for solvents and plastics. Fly ash and bottom ash from power stations are sold into cement and concrete, and pulverized coal injection displaces some coke in the blast furnace itself. None of these is large next to the two big uses, and none of them changes the essential fact that coal is bought for its energy and its carbon, in that order.

The seaborne coal chain is short and rigid: mine, rail, port, ship, discharge berth, stockyard, boiler. Because coal is bulky and cheap, the rail and port links are usually purpose built and owned by consortia of the miners themselves, and they run close to capacity by design. Australia loads through Newcastle in New South Wales for thermal coal and through Hay Point, Dalrymple Bay, Gladstone and Abbot Point in Queensland for coking coal. South Africa concentrates almost everything through the Richards Bay Coal Terminal, fed by a single heavy haul line from the Mpumalanga Highveld. Indonesia is different again: much of its coal is barged down the Barito and Mahakam rivers in Kalimantan and transshipped to ocean vessels at anchorage, which makes river levels and barge availability part of the supply picture.

The other routes are equally identifiable. Colombia ships from Puerto Bolívar and Ciénaga on the Caribbean coast; the United States loads Appalachian coal at Baltimore and Norfolk and Illinois basin coal through New Orleans; Russia moves Kuznetsk coal thousands of kilometers east on the Trans-Siberian and Baikal-Amur lines to Far East ports and west to the Baltic and Black Sea; and Mongolia trucks and rails coking coal across the Gashuunsukhait and Gantsmod crossings into Inner Mongolia. On the buying side, Amsterdam-Rotterdam-Antwerp is the blending and storage hub for northwest Europe and the physical basis for the API 2 contract.

The single points of failure are almost all inland. A derailment or maintenance shutdown on the Richards Bay line, cyclone damage to a Queensland rail corridor, flooding in a Kalimantan pit, or congestion at a Chinese border crossing removes cargoes that no other supplier can replace within the loading window, because export capacity is fixed and stockpiles at ports are small relative to annual flows. Concentration is also structural on the demand side: with China and India together taking 22% and 18% of world import value in 2024 (CEPII BACI), a policy change in either country resets the seaborne market on its own.