Congo mines three-quarters of the world's cobalt. It stopped shipping it

For most of a decade the Democratic Republic of the Congo mined about three-quarters of the world's cobalt and priced none of it. The metal was quoted in London and Shanghai, the margins were made in refineries in China, and when oversupply drove the price to a nine-year low of $24,343 a tonne at the start of 2025³, the country holding the deposit absorbed the loss. In February 2025 it stopped absorbing: the government suspended cobalt exports altogether¹.

The suspension held, in extended form, for eight months. Since 16 October 2025 it has been replaced by something more durable: an export quota system run by ARECOMS, the country's strategic-minerals authority, capping shipments at 18,125 tonnes for the rest of 2025 and 96,600 tonnes a year for 2026 and 2027². That annual cap is roughly half what the country exported in 2024. By the end of December the metal traded at $53,005 a tonne, up 130% on the year³.

How concentrated this market is

Few industrial metals depend this heavily on one country. Benchmark Mineral Intelligence put global mine supply at about 256,000 tonnes in 2024, with the DRC producing over 75% of it⁴. The next producer, Indonesia, mined an estimated 20,500 tonnes in 2024, a fifteen-fold rise in a decade but still under a tenth of Congolese output⁵. Everything else is single-digit thousands: by-product streams from nickel and copper mines in Australia, Canada, Cuba, and the Philippines.

Concentration repeats one level down. Cobalt in Congo is a by-product of copper, and the copper belt's largest operator is CMOC of China, whose two mines, Tenke Fungurume and Kisanfu, produced 114,000 tonnes of cobalt in 2024, 31% above their stated capacity and over half the country's output; CMOC alone was 41% of world supply⁴. Glencore, operating Katanga and Mutanda, is the other major. So the question of who controls cobalt has two answers that sit uneasily together: the Congolese state controls the border, and two foreign companies control most of what crosses it.

The artisanal share, the sector that gave Congolese cobalt its reputation for child labour and unrecorded digging, has shrunk to a historic low: under 2% of the country's production in 2024, about 1% of the world's⁴. Low prices did what audits alone had not; hand-dug ore stops being worth digging at $24,000 a tonne. Whether the quota-driven recovery pulls diggers back in is one of the things the new regime will be judged on.

What the ban was for

The stated purpose was price repair. Cobalt is mined as a by-product, which breaks the usual supply discipline: when copper is worth digging, cobalt comes out of the ground regardless of whether anyone wants it, and by early 2025 the market held years of accumulated surplus. A producer of three-quarters of world supply that cannot slow production can still slow exports. That is what the February suspension did¹: hydroxide piled up at the mines, refined markets tightened, and the price left its floor within weeks, trading above $34,000 by the end of March³.

The quota system that followed is the ban converted into an instrument. Each approved exporter receives an annual allocation based on its average production and exports over the previous three years²; CMOC's quota to the end of 2025 was 6,500 tonnes⁶. The state now decides the country's export volume the way OPEC decides a member's production ceiling, and for the same reason: to make scarcity where geology made abundance.

Who pays, and who adapts

The bill lands first on battery makers. Batteries are cobalt's dominant use, and the nickel-cobalt-manganese chemistries that use it power much of the western electric-vehicle fleet. A doubling of the cobalt price moves a cell's cost by single-digit percentages, which is absorbable but not ignorable, and manufacturers have spent years learning to ignore less of it: thrifted high-nickel chemistries use less cobalt per cell, and lithium-iron-phosphate, which uses none, has taken a growing share of new vehicles, especially in China. Recycling adds a slower counterweight, with metal recovered from spent cells and manufacturing scrap covering a growing slice of demand. Analysts made the consequence explicit as prices rose: the longer they stay elevated, the faster manufacturers move to low-cobalt and cobalt-free designs where they can³.

The DRC is therefore running a timed experiment. Restricting supply raises the value of the deposit today and raises the incentive to engineer around the deposit tomorrow. Indonesia sits on the other side of the bet: its cobalt, a by-product of the nickel industry built by Chinese capital, is projected to keep multiplying through 2026, and every tonne of quota the DRC withholds is a tonne of demand its competitor can bid for⁵.

This market has crashed before

The 2025 intervention makes more sense against the last cycle. Cobalt peaked at $94,500 a tonne in March 2018 on early electric-vehicle enthusiasm, then lost 70% of its value in a year as Chinese supply from the DRC flooded out and battery makers publicly promised to design the metal down⁷. In August 2019 Glencore mothballed Mutanda, then the world's largest cobalt mine with about 27,000 tonnes a year, a fifth of world production; the price rose more than 30% on the announcement, and the mine stayed shut until 2021⁷.

That episode taught the copper belt two things. A single withheld mine can move the world price, and the party doing the withholding collects none of the benefit unless it also holds the export gate. In 2019 a Swiss trading house made the cut and absorbed the loss on its own book. In 2025 the state made the cut, spread it across every exporter by decree, and kept the price gain inside the border as higher royalty and tax receipts on each remaining tonne. The quota system applies that lesson, this time in law.

The refinery is the second border

Mining concentration is only half the dependency. Cobalt hydroxide is not battery material until it is refined, and 79% of the world's refined cobalt came out of China in 2024; Chinese refined output rose 8% that year to 131,000 tonnes even as the mine-side surplus was collapsing the price⁸. The IEA's 2025 critical-minerals outlook measured the broader pattern: across the key energy minerals, the top three refining nations held 86% of the market in 2024, up from 82% in 2020, and China supplied about 90% of the growth in refined cobalt, graphite, and rare earths⁸.

That distribution decides who feels the quota first. The DRC's export cap constrains Chinese refineries before it constrains anyone in Detroit or Stuttgart, because the feedstock goes to China first. But it also means the DRC's leverage is partial: a state that controls the ore but none of the refining can set the volume, not the destination, of its own exports. The 2025 restrictions are best read as an attempt to change that bargaining position, alongside publicly floated plans for domestic processing joint ventures that so far exist mostly on paper.

A railway redraws the map

The other structural change is a route. Congolese cobalt has historically left the continent eastward and southward by truck, through Zambia and Tanzania to Dar es Salaam or through South African ports, journeys of weeks on congested roads. The Lobito Corridor, roughly 1,700 kilometres of rehabilitated railway from the copper belt to Angola's Atlantic coast, cuts the Kolwezi-to-port journey to about seven days⁹. The first Congolese copper train ran in December 2023; by 2025 the line carried close to 200,000 tonnes of international cargo on twelve trains a week⁹.

The money behind it says what it is for. The US development finance corporation lent $553 million to the operating consortium in 2023, its largest rail investment in Africa, explicitly to give Atlantic markets a mineral route that does not pass through Chinese-financed infrastructure. In September 2026 the DRC granted a 30-year concession for its section to Mota-Engil, with up to $1.8 billion of investment and a target of 13.7 million tonnes of annual capacity⁹. None of that changes who refines the cobalt today. What it changes is what a future western refining industry would need in order to exist: a way to get the ore without asking anyone's permission.

What to watch

Three numbers will say whether this worked. The first is compliance: a quota system with porous borders is a price subsidy to smugglers, and enforcement in the copper belt is not a settled question. The second is the 2026 average price; analysts' projections cluster near $55,000³, and a figure far below that means the quota is leaking or demand is thrifting away faster than expected. The third is Indonesia's output number for 2026: the gap between roughly 60,000 projected tonnes and the DRC's 96,600-tonne ceiling⁵ is the distance between a monopoly rationing its market and a duopoly forming.

The precedent travels either way. Indonesia banned raw nickel exports and got a processing industry; the DRC has banned raw cobalt volumes and wants a price. Other single-commodity states are watching both results closely.