Cobalt — A Complete Market Guide (2026)

Data as of 6 September 2026. Prices are quoted as multi-year and full-year averages, not a single day’s snapshot, so this report stays useful over time. Reserves, production splits, balances and historical series are estimates from agency and industry data, rounded for clarity. This report is for information only and was prepared with AI assistance — see the disclaimer at the end.

Cobalt is the metal nobody sets out to mine. Roughly 99% of it comes out of the ground as a by-product of copper and nickel, which means the cobalt market’s own price has almost no power to call forth new supply — and that single fact explains every boom and bust in its history. It is also the commodity where politics has most visibly seized the wheel: in February 2025 the Democratic Republic of the Congo, source of three-quarters of world output, simply banned cobalt exports, then replaced the ban with a hard export quota running to 2027 — and the price roughly tripled. This report is the free, big-picture primer on how the cobalt market actually works: where it comes from, why supply is deaf to price, who refines it, what it is used for, and which economic regimes favour it. For the company-level data behind the charts — every producer screened by production, resources and cost — go to Metal Pilot.

TL;DR & Key Takeaways

• What it is: a hard, blue-grey metal valued for two unrelated jobs — stabilising the crystal structure of lithium-ion cathodes, and holding strength at extreme temperature in superalloys. It is consumed in use, so demand is industrial flow, not a hoard.
• Market structure: the most concentrated of any major metal at both ends of the chain — the DRC mined ~74% of world supply in 2025 and China refines ~79% of it — and one company, CMOC, produced 117.5 kt in 2025 from two mines.
• Supply is a by-product, and that is the whole story: roughly 76% comes from copper mines and 23% from nickel, so output is set by copper and nickel economics, not by the cobalt price. Supply is close to perfectly inelastic, which is why cobalt overshoots in both directions.
• Demand story: batteries take ~76% — EV cathodes ~43%, portable electronics ~30% — with superalloys the largest industrial use at ~8%. The threat sits inside the boom: cobalt-free LFP chemistry passed half the global battery market in 2024.
• Price regime: violent and mean-reverting around a 2000–2025 average near $19/lb: peaks of ~$39/lb (2008) and ~$37/lb (2018), a ~$6.9/lb trough in 2002, ~$16.8/lb in 2024 — then ~$21/lb in 2025 after the DRC intervened.
• Biggest swing factor: DRC export policy. The 2026–27 quota of 96,600 t/yr is roughly 42% of recent mine output — the first serious attempt in cobalt’s history to manage the market on purpose.

Figure 1. Cobalt at a glance

Figure data: USGS Mineral Commodity Summaries 2026 for mine production, shares and price; Cobalt Institute Cobalt Market Report 2025 for demand and refining; see Sections 1.3–2.7.

Why it matters now: for a decade cobalt traded as a derivative of the electric-vehicle story, and that trade failed. What replaced it is a policy commodity — supply deaf to price but exquisitely sensitive to two governments’ decisions, demand broadening into defence and aerospace.

How to read this guide: new to cobalt? Read straight through from Section 1. Here for the investment case? Jump to Section 4 — how to own it, how to screen the miners, and which regimes favour the metal. Every technical term is defined in the Vocabulary (Section 7).

1. Cobalt & the market basics

1.1 What cobalt is — physical basics & quality

Cobalt (chemical symbol Co) is a hard, lustrous, blue-grey metal that sits between iron and nickel in the periodic table and behaves a little like both. Two properties pay for it. First, it is stable at extreme temperature, which makes it the backbone of superalloys — the nickel- and cobalt-based metals inside jet-engine combustors and turbine blades — and of cemented carbides (“hard metals”), where cobalt powder binds tungsten-carbide cutting tools. Second, and far more importantly for the modern market, cobalt stabilises the layered crystal structure of lithium-ion cathodes: in NCM and NCA chemistries it stops the cathode degrading as lithium ions shuttle in and out. It is also the “Co” in samarium-cobalt (SmCo) magnets, which hold their magnetism at temperatures that destroy cheaper neodymium magnets — small in tonnes, loud in missiles and avionics.

Unlike copper or nickel , cobalt has no ore body of its own worth mining at scale. Its quality story therefore starts with the host rock, and there are three families — each defined here once and used throughout this report:

• Sediment-hosted stratiform copper-cobalt ore — the Central African Copperbelt, through the DRC’s Lualaba and Haut-Katanga provinces into Zambia. Cobalt occurs as heterogenite (a cobalt oxide-hydroxide) and sulphides at roughly 0.3–0.8% Co alongside 2–4% copper: by far the richest cobalt endowment on Earth, and the reason one country dominates.
• Nickel laterite ore — shallow tropical weathering blankets in Indonesia, the Philippines, New Caledonia, Cuba and Madagascar. Cobalt concentrates in the upper, iron-rich limonite layer at roughly 0.1–0.15% Co and is recovered only through HPAL (high-pressure acid leaching); the saprolite layer beneath it, smelted into nickel pig iron, yields effectively none. Which layer a country processes decides whether it is a cobalt producer.
• Magmatic nickel-copper sulphide ore — Norilsk, Sudbury, Western Australia. Modest grades, but the ore is already being smelted and refined, so the cobalt comes almost free.

Everything else is a rounding error: Bou Azzer in Morocco is the world’s only mine of scale where cobalt is the principal product, and the one American attempt at a primary cobalt mine — Jervois’s Idaho Cobalt Operations — was idled in 2023 weeks before opening (Section 4.3).

On the product side the vocabulary is short but essential. Cobalt hydroxide is the crude DRC intermediate, roughly 25–40% Co, shipped in bulk bags to Chinese refineries — what the DRC exports and what the export quota counts. MHP (mixed hydroxide precipitate) is the Indonesian analogue, a nickel-cobalt intermediate at a nickel-to-cobalt ratio near 8:1 to 10:1. Cobalt sulphate is the battery salt cathode makers buy, typically 20.5% Co. Refined cobalt metal — cathode, briquettes and rounds, minimum 99.3% Co — is the smallest slice by volume but the price reference for everything else, because superalloy and magnet buyers take nothing less. The three routes below all converge on a refinery, and in four cases out of five a Chinese one.

Figure 2. The cobalt value chain — three routes, one refinery

Source: industry value-chain primers and Cobalt Institute process descriptions; conceptual diagram.

1.2 Units & measurement conventions

This report uses the base-metals convention throughout, stated here so every later number is unambiguous. Cobalt quantities are metric tonnes (t) and thousand tonnes (kt) of contained cobalt — the pure-metal content inside ore, hydroxide, MHP, sulphate or refined metal — mirroring the convention of the U.S. Geological Survey (USGS) and the Cobalt Institute . Cobalt is a small market by tonnage, so kt is the working unit and Mt almost never appears except for reserves.

The price convention is the one place cobalt differs from its base-metal siblings. Cobalt is quoted in US dollars per pound ($/lb) of refined metal, not per tonne, because the market grew out of the US superalloy trade; the conversion is 1 tonne = 2,204.62 lb, so $20/lb ≈ $44,100/t. Both units appear in the sources, and this report leads with $/lb and gives the tonne equivalent where it helps. Ore and product grades are % Co.

Three distinctions matter constantly in cobalt and are flagged throughout:

• Mined vs. available supply. Mined supply is cobalt contained in ore extracted in a given year; available supply is what actually reaches buyers. In every other metal the gap is small. In cobalt since 2025 it is the whole story — see Section 2.4.
• Payables, not prices, for intermediates. Cobalt hydroxide and MHP are quoted not in dollars but as a payable percentage of the refined-metal price: 80% payable at a $20/lb metal price means $16/lb for the contained cobalt. When feedstock is scarce, payables rise toward 100% and refiners’ margins vanish; when it is abundant they fall below 60%.
• Flow vs. stock. Production, demand and export quotas are flows per year. Reserves (~12 Mt, the economically mineable subset of resources), refiner and consumer inventories, and China’s state stockpile are levels at a point in time.

Table 1. Cobalt units and conversions

Source: USGS Cobalt Statistics and Information , 2026; Cobalt Institute product and process descriptions, 2025. Grades are typical ranges, not specifications.

Numbers intuition: the world mines about 310 kt of cobalt a year — roughly a twelfth of the nickel market by tonnage — and at the 2025 average price, refined output of ~240 kt is worth about $11 billion a year, less than a single large gold miner’s revenue. A big cobalt mine makes 20–30 kt/yr; an HPAL plant contributes 2–5 kt/yr as a by-product; an electric car needs 5–10 kg. That miniature scale is why a market this strategically important can be moved so far by one company’s ramp-up or one ministry’s decree.

1.3 Pricing & benchmarks

Cobalt has an exchange price, but it is not really an exchange market. The London Metal Exchange (LME) has listed a physically-settled cobalt contract since February 2010 and a cash-settled LME Cobalt (Fastmarkets) contract since 2019, and the CME lists cash-settled Cobalt Metal and Cobalt Hydroxide CIF China futures, the hydroxide contract growing fastest as producers and refiners hedge the feedstock leg. But liquidity is thin and the physical market never migrated to the screen. The real reference is assessed: a standard-grade cobalt metal price (min 99.3% Co, EXW Rotterdam and in-warehouse US), a cobalt hydroxide CIF China payable, and a cobalt sulphate price in China. The USGS series used throughout this report is the long-running U.S. spot cathode assessment.

Because cobalt sells in three physical forms to three customer sets, the spreads between them carry most of the information. Superalloy and magnet buyers need metal and will not touch sulphate; cathode makers need sulphate; refiners live on the gap between hydroxide payables and the finished product. When DRC feedstock dried up in 2025, payables squeezed toward metal parity and Chinese conversion margins collapsed — the same event that lifted the metal price made the refining business worse. A reader who quotes only “the cobalt price” is quoting the smallest of the three markets.

Table 2. Key cobalt benchmarks & price markers

Source: LME cobalt contracts , 2026; CME Cobalt Hydroxide CIF China (Fastmarkets) futures , 2026; USGS price notes (S&P Global Platts Metals Week cathode assessment), February 2026.

The long-run price story is one of violent mean reversion around a surprisingly stable centre. Across 2000–2025 the annual average U.S. cathode price has been ~$19/lb, and cobalt has spent almost none of that period near it. The early 2000s bottomed at ~$6.9/lb in 2002 after the tech bust; Chinese industrialisation and a stockpiling scare drove ~$24.5/lb in 2004; the pre-crisis commodity mania produced the all-time annual high of ~$39/lb in 2008 before an 80%-plus collapse. The 2010s were a long grind — ~$12/lb in 2016 — broken by the electric-vehicle mania of 2017–18, when the price nearly tripled to ~$37.4/lb in 2018 (monthly assessments exceeded $43/lb in March of that year) on forecasts of cathode demand that did not arrive on schedule. The bust that followed took the annual average to ~$15.7–17/lb, a brief post-pandemic recovery lifted it to ~$30.8/lb in 2022, and then CMOC’s Kisanfu ramp-up (Section 2.1) crushed it to ~$16.8/lb in 2024 and to multi-decade lows in real terms early in 2025. The DRC’s export ban reversed that inside a year: the 2025 average recovered to ~$21/lb, and for reference, metal was assessed near $25.5/lb (~$56,200/t) in late April 2026 — a dated snapshot, not the number to plan on.

Table 3. Average annual cobalt price, 2000–2025 (US spot cathode, USD/lb)

Source: USGS Mineral Commodity Summaries , U.S. spot cathode annual averages, compiled from the 2005, 2010, 2015, 2017, 2021, 2024 and 2026 editions; the most recently published vintage is used for each year. Figures are calendar-year averages, not spot; 2025 is a USGS estimate. Monthly assessments were far more extreme than these averages — above $43/lb in March 2018 and below $10/lb in early 2025.

Figure 3. Cobalt price, annual averages 2000–2025 (USD/lb)

Figure data: Table 3. Eleven years selected from the full 2000–2025 series to carry the two peaks, the two troughs and the recent cycle.

2. Supply, demand & the market balance

2.1 Where cobalt is mined — deposits & geology

Cobalt’s geography is an accident of geology so lopsided it barely qualifies as a market. The Central African Copperbelt — an arc of sediment-hosted copper deposits running through the DRC’s Lualaba and Haut-Katanga provinces into Zambia — carries cobalt at grades an order of magnitude above anything else on the planet, because copper- and cobalt-bearing brines were trapped in the same Neoproterozoic basin. Everywhere else cobalt is a trace passenger: a fraction of a percent in tropical nickel laterites, less still in the magmatic sulphides of Norilsk, Sudbury and Western Australia.

The result: the DRC mined an estimated 230 kt of cobalt in 2025 — about 74% of the world total of 310 kt — with Indonesia a distant but fast-growing second at 44 kt (14%) and nobody else above 3%. Indonesia’s rise is a pure spillover from its nickel build-out: every HPAL autoclave built to make battery-grade nickel produces MHP, and MHP carries cobalt. In a decade Indonesia went from a rounding error to the world’s second cobalt producer without a single project being built for cobalt.

Table 4. Leading cobalt-mining countries, 2025 (estimated)

Source: USGS Mineral Commodity Summaries 2026: Cobalt , February 2026. Figures are USGS estimates for 2025, rounded; country rows and the world total are the USGS’s own; shares are computed against the rounded 310,000 t world total, so the listed rows sum to about 314,600 t.

Figure 4. Leading cobalt-mining countries, 2025 (kt Co)

Figure data: Table 4. Values are in thousand tonnes; “Rest of world” is the residual of the rounded 310,000 t world total after the seven named countries, and covers Papua New Guinea, Cuba, China, Turkey, the United States and the USGS “other countries” line.

At the asset level, concentration goes from remarkable to almost absurd. CMOC Group’s two DRC mines — Tenke Fungurume (TFM) and Kisanfu (KFM) — produced 114.2 kt of cobalt in 2024 and 117.5 kt in 2025, which is more than a third of world mine supply from one company and roughly half the DRC’s output. Kisanfu is the pivotal asset: bought from Freeport-McMoRan in 2020 and brought into production in 2023, it has an unusually low copper-to-cobalt ratio of about 2.2:1 on a reserves-and-resources basis, against a DRC average near 10:1, meaning every tonne of copper it mines drags out several times the normal quantity of cobalt. CMOC ran both mines roughly 31% above stated capacity in 2024 — cobalt as the collateral damage of a copper strategy.

Table 5. Major cobalt mines & operations (approximate recent annual output)

Source: CMOC capacity and output per Cobalt Institute / Benchmark Mineral Intelligence, Cobalt Market Report 2024 (TFM 37 kt and KFM 50 kt stated capacity; 114 kt actual in 2024, 31% above capacity) ; CMOC’s 2025 figures (cobalt 117,500 t, copper 741,100 t) are from its FY2025 annual results announcement, with the company’s cobalt product page for asset detail; Glencore and ERG figures per the same report; Sherritt per company reporting. Indonesian figures are derived, not filed: they are inferred from each plant’s published MHP nickel capacity at a typical 8–10:1 nickel-to-cobalt ratio and are indicative only. Bou Azzer and Norilsk are author estimates from country totals — treat both as approximate. Because CMOC’s two mines run above nameplate, their capacity figures understate actual output.

2.2 Demand & consumption

Cobalt demand is a battery story with an aerospace floor. Total demand reached 222 kt in 2024 and 276 kt in 2025, growing by double digits in both years — but the composition matters more than the total. Lithium-ion batteries take about 76% of world cobalt, split between EV cathodes (~43%) and portable electronics (~30%) — a segment most investors forget until they notice that every laptop, phone and power tool holds an LCO cathode roughly 60% cobalt by cathode mass. Growth in 2024 came almost entirely from batteries: 94% of the year’s demand growth, with EVs alone contributing 61%.

The non-battery quarter is smaller but strategically loud, and it is the part that does not care about EV subsidies. Superalloys — jet-engine combustors and turbine blades — are the largest industrial use at roughly 8% of demand (18.5 kt in 2024), and grew 5% on commercial-aviation deliveries and defence spending. Below that sit cemented carbides (cobalt binder in cutting tools), catalysts (petroleum refining and PET plastics), ceramics and pigments (cobalt blue is the oldest use of all), hard facing, and SmCo magnets. NATO added cobalt to a supply-chain security roadmap in 2024 as a high-risk material for military applications — an unusual event for a metal usually discussed in the language of consumer electronics.

Table 6. Global cobalt demand by end use, 2024 (share of demand)

Source: Cobalt Institute / Benchmark Mineral Intelligence, Cobalt Market Report 2024 , May 2025 — total demand 222 kt in 2024, batteries 76%, superalloys 18.5 kt. Shares are rounded and volumes derived from the 222 kt total; “other industrial” aggregates the report’s ceramics/colours (5.3 kt), hard facing (4.5 kt), tyres, soaps and paint driers (4.3 kt), magnets (3.5 kt) and other (3.3 kt) lines.

Figure 5. Global cobalt demand by end use, 2024

Figure data: Table 6.

The substitution problem is inside the growth story, not outside it. Cathode chemistry is where cobalt demand is won and lost, and it has moved against cobalt for a decade. LFP (lithium iron phosphate), which contains no cobalt at all, took 51% of the global battery market in 2024 and drove 79% of that year’s battery-demand growth, while within the EV segment cobalt-bearing chemistries fell from 60% of demand in 2023 to 53% in 2024. On top of that runs thrifting: NCM cathodes have moved from a 1:1:1 nickel-cobalt-manganese ratio to 8:1:1 and beyond, halving cobalt intensity per kilowatt-hour. Cobalt demand therefore grows only because the battery market grows faster than cobalt’s share of it shrinks — fragile, and the reason the 2018 EV thesis did not pay.

Geographically, cobalt is consumed where cathodes and refined products are made, overwhelmingly in China — the world’s leading refined-cobalt producer and its leading consumer, with nearly 87% of Chinese cobalt consumption going into lithium-ion batteries (the share the USGS last published, in its 2024 edition). No agency publishes country-level end consumption on a contained-cobalt basis, so refining location (Section 2.5) is the best available proxy. The one clean national datapoint is a revealing contrast: estimated U.S. consumption was about 8,000 t in 2025, of which 51% went into superalloys, 25% into chemicals, 15% into other metallic uses and 9% into cemented carbides. China’s cobalt goes into batteries; America’s goes into jet engines.

2.3 Supply: production and reserves

World cobalt mine production has roughly sextupled since 2004, from about 47 kt to 310 kt, and almost all of that growth came from two step-changes: the industrialisation of the DRC Copperbelt through the 2010s, and Indonesia’s HPAL build-out since 2021. There was no cobalt price signal behind either. The DRC’s expansion was a copper story; Indonesia’s was a nickel story.

Table 7. World cobalt mine production, selected years (t Co)

Source: USGS Mineral Commodity Summaries , world totals from the 2005, 2010, 2015, 2017, 2021, 2024 and 2026 editions; each year uses the most recently published vintage. Years chosen at roughly five-year intervals to 2020 and annually thereafter, where the market changed fastest. USGS mine production is contained cobalt in ore and runs roughly 11–19% above the Cobalt Institute’s mined-supply series, which nets out downstream yield losses — see the methodology note in Section 8.1.

Figure 6. World cobalt mine production, selected years (kt Co)

Figure data: Table 7.

Reserves tell the same story of concentration with an important footnote. The DRC holds 6 Mt of the world’s 12 Mt of cobalt reserves — half the planet’s economically mineable cobalt in one country — with Australia second at 1.7 Mt (of which only about 680 kt is JORC-compliant, the rest being sub-economic at recent prices) and Russia and Indonesia next. At 310 kt of annual mine output, world reserves imply a reserve life of roughly 39 years, and identified terrestrial resources are far larger at about 25 Mt. Cobalt is not geologically scarce. What is scarce is cobalt outside the DRC that can be produced at a price the market will pay — and, for Western buyers, cobalt that has not passed through a Chinese refinery.

Table 8. Cobalt reserves by country, 2025 (t contained Co)

Source: USGS Mineral Commodity Summaries 2026: Cobalt , February 2026. Reserves are estimates of economically mineable material and are revised annually; the listed country rows sum to about 11.5 Mt against the USGS’s rounded 12 Mt world total. Australia’s JORC-compliant reserves are about 680,000 t within the 1.7 Mt figure; the Russian figure is from the Russian Ministry of Natural Resources. The same edition puts identified world terrestrial resources near 25 Mt and reports very large but speculative cobalt content in seafloor polymetallic nodules — treat seabed figures as resource potential, not reserves.

2.4 The supply–demand balance

Cobalt’s balance is normally a straightforward industrial one: by-product supply arrives regardless of price, demand grinds upward, and the difference sits in Chinese refiners’ warehouses. Between 2023 and 2024 that arithmetic produced the deepest surplus in the metal’s history. Mined supply jumped from 208 kt to 254 kt (+22%) on CMOC’s Kisanfu ramp-up alone, demand grew fast but not that fast, and Benchmark Mineral Intelligence put the market surplus at +25 kt in 2023 and +36 kt in 2024 on a refined-availability basis. Prices went to historic lows and smaller DRC producers switched their cobalt circuits off entirely.

Then 2025 broke the model. On 22 February 2025 the DRC banned cobalt exports outright; on 16 October the ban was replaced by a quota of 18,125 t for the remainder of 2025 and 96,600 t a year for 2026 and 2027. Mine production carried on — the mines are copper mines, and copper was profitable — but the metal could not leave the country. Total supply for 2025 was 295 kt, of which the Cobalt Institute reckons only 183 kt actually reached the market: roughly 112 kt of cobalt was produced and stranded, against demand of 276 kt. This is the single most important fact about the cobalt market today, and it has no precedent in any other major metal.

Table 9. World cobalt supply, demand & balance, 2023–2025 (kt Co)

Source: Cobalt Institute / Benchmark Mineral Intelligence, Cobalt Market Report 2024 for 2023–24 (mined supply, secondary supply, refined production, demand and the published surplus) and Cobalt Market Report 2025 , May 2026, for 2025 (total supply 295 kt, mined 270 kt, refined 240 kt, demand 276 kt, availability 183 kt). Two bases appear here and must not be mixed: “balance on available supply” is total supply less demand, derived arithmetic; the “published market surplus” is Benchmark’s own balance struck on refined availability after downstream yield losses, which is why it is smaller. 2023 demand is derived from the 2024 total less that year’s reported +26.7 kt growth; 2025 secondary supply is the residual of the Cobalt Institute’s own total (295 kt) and mined (270 kt) figures, and the report’s separate description of secondary as “around 10% of total supply” does not reconcile with those two totals. Benchmark revised the 2024 demand base upward to roughly 244 kt in the 2025 report (derived: 276 ÷ 1.13), so the +13% growth to 276 kt is measured on the revised base, not on the 222 kt shown here.

Figure 7. Cobalt market balance, 2023–2025 (kt Co, total supply less demand)

Figure data: Table 9. All four bars are struck on the same basis — total supply less total demand — so the two 2025 bars differ only in whether stranded DRC material is counted as supply.

Where the shock absorber sits. Cobalt has no LME stock series worth watching — exchange inventories are trivial next to the market. The real buffers are three, and none is published in real time: Chinese refiners’ and cathode makers’ working inventories, the NDRC state reserve, a discretionary buyer whose purchases twice marked the price floor and which took in roughly 16.6 kt in 2024, about 7% of world demand or three and a half weeks of it; and, since 2025, the growing pile of unexported hydroxide in DRC warehouses — roughly 112 kt, or 21 weeks of world demand, the largest buffer the market has ever carried. That last one is the most consequential, because ARECOMS controls the tap on it: the market has priced a deficit while more than a hundred thousand tonnes of cobalt sits behind a border — an inherently unstable configuration.

The country picture is one asymmetry: the DRC produces roughly 230 kt and consumes essentially none, Indonesia produces 44 kt and consumes little, and China refines about four-fifths of world supply and consumes most of it, importing every tonne of feedstock. There is no diversified trade in cobalt — a pipe from central Africa and a newer one from Sulawesi and Halmahera, both emptying into the same Chinese refineries.

2.5 Supply structure: by-product, primary & recycled

This is the section that explains cobalt. Roughly 99% of mined cobalt is a by-product: about 76% comes out of copper mines — overwhelmingly in the DRC — and about 23% out of nickel operations, split between Indonesian HPAL and the older sulphide and laterite producers. Only Morocco’s Bou Azzer runs at scale as a primary cobalt mine; the USGS noted, before the latter was idled, that it and Jervois’s Idaho project were “the only mines in the world where cobalt was the principal product” (Mineral Commodity Summaries 2024).

The investment consequence is severe and often missed. Cobalt supply is close to perfectly inelastic to the cobalt price. A copper miner in Lualaba decides how much ore to move on the copper price; the cobalt is whatever comes with it, and the only marginal decision is whether to recover it into hydroxide or leave it in the tailings — a small incremental processing cost, not a cobalt-price call. So a price collapse closes almost no cobalt mines, because there are almost none to close; and a price spike summons no new supply, because nobody builds a copper mine for the cobalt. All the adjusting therefore happens on the demand side, through substitution and thrifting — exactly what happened after 2018. It is also why an export quota is such a powerful lever here: it does the job the price cannot.

The two genuinely price-elastic slices are small. Artisanal and small-scale mining (ASM) — hand-dug Congolese cobalt, the subject of most of the sector’s human-rights scrutiny — supplied roughly 10% of DRC output when prices peaked in 2018 and had collapsed to under 2% of DRC production, about 1% of global supply, by 2024 as industrial volumes swamped it. It rises with price but can no longer reclaim its old share. Recycling is the second: secondary cobalt reached 22 kt in 2024, 8% of total supply, growing in absolute terms while losing share to the primary flood. Most of it is production scrap from Chinese cell plants and end-of-life portable electronics rather than EV batteries, whose first large retirement wave is still ahead.

Table 10. Cobalt supply by origin, 2024

Source: Cobalt Institute / Benchmark Mineral Intelligence, Cobalt Market Report 2024 , May 2025 — mined supply 254 kt split roughly 76% copper-hosted and 23% nickel-hosted, secondary supply 22 kt of 276 kt total. Shares here are restated on a total-supply basis so they sum to 100%; volumes are derived and rounded. The primary-cobalt line covers Bou Azzer and small PGM-associated output.

Figure 8. Cobalt supply by origin, 2024

Figure data: Table 10.

Refining is the market’s second concentration, and the tighter one. Refined production reached 222 kt in 2024 and 240 kt in 2025, and China made about 79% of it in both years. Finland — home of Kokkola, the West’s largest refinery — is a distant second near 7%, followed by Canada, Japan and Madagascar, with Indonesia overtaking Canada in 2025 in the same downstream migration that reshaped nickel. Refining is where cobalt’s strategic dependency actually lives: a Western automaker can buy Congolese or Indonesian cobalt units, but almost every path to battery-grade sulphate runs through a Chinese plant.

Table 11. Refined cobalt production by country, 2024 (share of world)

Source: Cobalt Institute / Benchmark Mineral Intelligence, Cobalt Market Report 2024 , May 2025 (refined production 222 kt in 2024). China’s share held at 78.6% in both 2023 and 2024 and was ~79% in 2025, when Indonesia passed Canada into third place per the Cobalt Market Report 2025 .

Figure 9. Refined cobalt production by country, 2024

Figure data: Table 11. Segments are sorted largest-first. “Other” aggregates Indonesia (1.7%), Norway (1.4%), Australia (1.4%) and the rest of the world (3.2%).

2.6 Trade flows & the chokepoints

Cobalt has no Strait of Hormuz — no sea lane whose closure severs supply. Its chokepoints are of three other kinds, and all three bind harder than a strait would.

First, the roads out of a landlocked country. The DRC’s Copperbelt sits roughly 2,000 km from any port. Cobalt hydroxide leaves it in bulk bags on diesel trucks, historically south through the Kasumbalesa border into Zambia and on to Durban, or east to Dar es Salaam — 28 to 40 days door-to-port, with queues at Kasumbalesa running to kilometres. The Lobito Corridor, the rehabilitated Benguela railway running west from Kolwezi through Angola to the Atlantic, is the structural answer: the shortest route from the Copperbelt to any African port, with a first shipment reported at about seven days against roughly 28 via Dar es Salaam. It is also, not coincidentally, an American-backed project competing with Chinese investment in the eastward TAZARA line.

Second, the Chinese refinery bottleneck (Section 2.5). A Western battery maker seeking non-Chinese sulphate has a very short list and pays for the privilege — what the EU’s Critical Raw Materials Act and the US critical-minerals programme are trying to change, so far at the margin.

Third, and newest, sulphur through Hormuz. Indonesian HPAL plants consume enormous quantities of sulphuric acid, and the Cobalt Institute flags that roughly 75% of the sulphur feeding them originates in the Middle East and transits the Strait of Hormuz. Cobalt’s only real maritime exposure is therefore not its own shipments but an input two steps upstream — a reminder that in a by-product metal, risk arrives through someone else’s cost curve.

Table 12. Cobalt trade roles, volumes & chokepoints

Source: USGS Mineral Commodity Summaries 2026 for mine output, US apparent consumption and the U.S. net import reliance figure (79% of apparent consumption in 2025); Cobalt Institute, Cobalt Market Report 2025 for refined production, refining shares and the sulphur and Hormuz exposure; corridor transit times per EGC shipment reporting and DRC logistics accounts, 2026. Refined volumes are derived from each country’s share of the 240 kt (2025) and 222 kt (2024) refined totals. Cobalt has no broad exporter or importer set to rank — three countries and one refining bloc carry essentially the whole trade — so this table stands in for the separate top-exporter and top-importer tables the guide series uses elsewhere. Transit times are indicative.

2.7 Market organisations & supply coordination

Cobalt has never had a cartel, a study group with real authority, or an OPEC. What it has, since 2025, is something stranger: a single producing country with three-quarters of world supply that has decided to manage the market directly, and no institution capable of stopping it.

The instrument is ARECOMS — the Authority for the Regulation and Control of Strategic Mineral Substances’ Markets — created by the DRC to administer exactly this. Its 2026–27 regime caps national cobalt exports at 96,600 t a year, of which 87,000 t is allocated to commercial producers pro rata against historical export volumes and 9,600 t is routed to a national strategic reserve. The allocations, as reported, are startlingly concentrated: CMOC 31,200 t and Glencore 22,800 t for 2026, with CMOC, Glencore and ERG together holding more than 60% of permitted exports across five operations. Whether this holds is the market’s central question: quota regimes reward cheating, and the incentive to smuggle grows with every dollar the price rises.

Around ARECOMS sit three weaker but real coordinating forces. China’s NDRC (formerly the State Reserve Bureau) is the buyer of last resort, its stockpile decisions unannounced and unaudited (Section 2.4). The DRC’s Entreprise Générale du Cobalt (EGC) holds exclusive rights to buy and export hand-dug artisanal cobalt. And the Cobalt Institute, with Benchmark Mineral Intelligence, publishes the market’s authoritative statistics — data, not quotas. On the demand side, Western governments shape the market through rules rather than tonnes: cobalt sits on the US critical-minerals list, the EU Batteries Regulation is phasing in due-diligence and recycled-content mandates, and third-party assurance — principally the Responsible Minerals Initiative (RMI), alongside the Copper Mark — covered 78% of mined and 88% of refined cobalt output by 2025.

Table 13. Who shapes the cobalt market

Source: DRC quota framework per Benchmark Mineral Intelligence and USGS , 2025–2026; the per-company allocations from press reporting of the ARECOMS decisions, the circulars themselves being unpublished; NDRC purchasing, EGC rights and assurance coverage per the Cobalt Institute market reports, 2025–2026.

3. The companies & the value chain

3.1 The largest cobalt companies

There is no cobalt sector in the way there is a gold sector. Because the metal is a by-product, the corporate league table is a list of copper and nickel companies that happen to produce a lot of it — and one of them is so large it is effectively the market. CMOC Group, a Chinese-listed miner that bought Tenke Fungurume from Freeport-McMoRan in 2016 and Kisanfu in 2020, produced 117.5 kt of cobalt in 2025, more than a third of world mine supply, while cobalt remains a secondary line item next to its 741 kt of copper. Glencore is second at 36.1 kt in 2025 (down 5% as it prioritised copper under the DRC export restrictions), and is the only major with both African and non-African cobalt plus its own Norwegian refinery. Third is Eurasian Resources Group, privately held, whose Metalkol RTR reprocesses a century of Copperbelt tailings.

Below the miners sits a Chinese refining and recycling tier that is arguably more strategically important — Huayou, GEM, Jinchuan — and, increasingly, Indonesian producer-refiners like Harita Nickel. Two names round out the picture for very different reasons: Sherritt International, whose Cuban Moa joint venture and Canadian refinery make it a rare Western integrated cobalt producer, and Managem, which operates Bou Azzer, the world’s only primary cobalt mine of scale.

Table 14. Leading cobalt producers and refiners (approximate)

Source: CMOC 2025 output (117,500 t cobalt, 741,100 t copper) per its FY2025 annual results announcement, with the company’s cobalt product page for asset detail; Glencore per its Full Year 2025 Production Report (36.1 kt own-sourced, −5% y-o-y); ERG, Huayou, GEM and Harita per Cobalt Institute / Benchmark Mineral Intelligence , May 2025. Listing notes: ERG is majority-owned by the Government of Kazakhstan alongside private shareholders; Nornickel trades in Moscow and is constrained by Western sanctions; Jinchuan is Gansu-province state-owned with a Hong Kong-listed subsidiary. Figures marked “derived” or “est.” are author estimates from country totals and plant capacities, not filed numbers; refiners’ rows show no mined tonnes because they buy feedstock rather than mine it. No market-capitalisation figures are shown by design.

3.2 Company archetypes along the value chain

Cobalt exposure is unusually hard to buy cleanly, and the archetypes explain why. Primary cobalt producers — a category with essentially two members worldwide — offer the purest leverage and the worst survival odds, because they must cover full mine costs from a metal that everyone else produces as a credit. Explorers and developers are option value on a Western supply chain that policy keeps promising and economics keeps refusing. Cobalt-weighted nickel and copper producers are the practical middle: companies where cobalt is a meaningful revenue line rather than a rounding error, so the equity actually moves with the price. Diversified copper-cobalt majors — CMOC, Glencore — give real cobalt volume inside a business whose earnings are driven by something else entirely; you get the metal without the torque. Refiners and recyclers earn a conversion spread and are hurt, not helped, by feedstock scarcity — the 2025 squeeze compressed their margins even as the metal price tripled. And royalty and streaming exposure barely exists: the one serious attempt, Cobalt 27 Capital, ended in a take-private (Section 4.1).

Table 15. Cobalt company archetypes

Source: company filings and the Metal Pilot project-type taxonomy, 2026.

Figure 10. Cobalt company archetypes by price sensitivity

Source: company filings; conceptual. Bar length is an ordinal rank on a six-step qualitative scale — from survival leverage at the top to blended stability at the bottom — not a measured sensitivity; refiners and royalties rate the same. Ratings as in Table 15.

3.3 Infrastructure & balance-sheet assets

Reading a cobalt company’s accounts means reading someone else’s mine. The resource and reserve base is reported under JORC, NI 43-101 or S-K 1300 as contained cobalt at a stated grade and cut-off — but for a by-product it is the copper-to-cobalt or nickel-to-cobalt ratio that decides how much is actually recovered, which makes it the single most useful number in a Congolese or Indonesian filing. Kisanfu’s 2.2:1 Cu:Co ratio against a 10:1 DRC average is precisely why one mine reshaped the world market.

The physical assets follow the three routes: open pits with leach and solvent-extraction circuits and, at Metalkol, a tailings retreatment plant cheap to feed but constrained by the grade of century-old waste; Indonesian HPAL autoclaves, measured by nickel capacity, uptime, acid consumption and delivered capex per tonne; and downstream hydrometallurgical refineries and black-mass recycling plants. Two accounting cautions run throughout: by-product credits can make a reported cobalt cost look like almost anything, and gross versus attributable tonnes diverge sharply in a sector built on joint ventures with free-carried state partners.

Table 16. Cobalt-company asset types and metrics

Source: company reserve statements (JORC / NI 43-101 / S-K 1300) and annual reports, 2024–2025.

4. Investing in cobalt

4.1 Ways to own it — exposure vehicles

Physical cobalt is impractical for individuals — refined metal trades in tonne lots and has no retail market — but not for a company, and the one serious attempt is instructive. Cobalt 27 Capital raised roughly C$500 million across a 2017 IPO and a 2018 bought deal, and bought roughly 2,900 tonnes of physical cobalt near the top of the EV mania; after the price halved it was taken private by Pala Investments in October 2019, its non-cobalt assets spun into what became Nickel 28. A physical cobalt trust is a real structure; the one that existed did not survive a cycle.

That leaves paper and businesses. Futures exist across the LME and CME, but liquidity is thin and they are professional hedging tools. There is no meaningful futures-based cobalt ETP, and that absence is itself the lesson: such a fund must roll expiring contracts, and in contango (futures above spot) each roll sells cheap and buys dear — a recurring drag that compounds badly. The canonical demonstration is oil’s 2020 super-contango, when crude-tracking funds lost enormous ground against the spot price they nominally tracked (worked through in the oil guide ); in a market as illiquid as cobalt’s it would bite harder, not less.

For long horizons the practical routes are equities, and the honest framing is that a pure cobalt equity barely exists: the realistic choices are cobalt-weighted nickel and copper producers, diversified majors where cobalt is a credit, battery-chain refiners and recyclers, and explorer-developers that are bets on Western industrial policy rather than on the metal.

Table 17. Ways to own cobalt

Source: contract structures per LME cobalt and CME specifications, 2026; Cobalt 27 structure and 2019 arrangement per company announcements. Structural attributes only — no fees or prices, which change.

4.2 How to value & screen cobalt producers

Screening cobalt starts from an awkward truth: you are usually screening a copper or nickel company. The framework therefore inverts the normal order — establish the host-metal business first, then ask how much cobalt torque sits inside it.

Four lenses do the sorting. First, cobalt revenue share: what percentage of revenue comes from cobalt at a normalised price? Below roughly 10% the equity will not move with the metal, whatever the marketing says. Second, the host-metal ratio — Cu:Co or Ni:Co — which sets how much cobalt a tonne of ore yields and whether the company keeps recovering it at low prices. Third, cost, read twice: a by-product producer’s cobalt “cost” depends entirely on how joint costs are allocated, so read the host metal’s all-in cost with and without the cobalt credit, and treat any headline cobalt AISC below a few dollars a pound as an allocation choice rather than a fact. Fourth, quota exposure, which in 2026 is quantifiable: what is this company’s ARECOMS allocation against what it can produce? A DRC producer whose mines can make more cobalt than its quota permits has a stranded-inventory problem, not an upside case.

For developers, the killer metric is delivered capex per tonne of annual capacity measured against Chinese-Indonesian HPAL benchmarks, plus an honest answer to the question of why the project survives when the marginal tonne of world supply carries a near-zero allocated cost. For recyclers, it is feedstock security and cobalt recovery rate. These are exactly the screens — resource base, production, cost position, jurisdiction — you can run across every listed producer on Metal Pilot, where the DRC copper-cobalt names sit alongside their peers.

Table 18. Cobalt screening metrics

Source: company MD&A, reserve statements and offtake disclosures, 2024–2026; DRC quota allocations per Benchmark Mineral Intelligence , 2025–2026. Thresholds are practical rules of thumb, not published standards.

Figure 11. Illustrative cobalt cost curve (allocated cash cost vs. cumulative output)

Source: illustrative and stylised, not company data. Tiers, left to right: DRC copper-cobalt majors (~42% of supply), DRC tailings retreatment and smaller Cu-Co operations (~20%), Indonesian HPAL nickel by-product (~14%), nickel-sulphide and laterite by-product (~12%), artisanal and small-scale mining (~6%), primary cobalt mines (~6%). Costs are allocated — for a by-product they depend on the joint-cost convention used, so read the whole curve as indicative. The reference line is the 2024 average price of $16.77/lb from Table 3, the level at which the 2023–25 shakeout happened.

The curve carries this section’s central point. On an allocated basis, nearly all cobalt supply sits below any plausible price, which is why the 2023–24 collapse closed almost nothing: DRC majors kept producing because copper paid, and Indonesian HPAL kept producing because nickel paid. The only tiers that responded were the two on the right — artisanal miners, who stopped digging, and primary producers, who went bankrupt.

4.3 Macro regimes, rates & correlations

Cobalt is the least macro-driven of the major industrial metals, and understanding why is most of the investment case. Its supply cannot respond to its own price, its demand runs on consumer-electronics and aerospace cycles rather than on construction or GDP, and its largest single variable is now a ministry in Kinshasa. Being dollar-priced, it drifts higher when the dollar weakens and suffers when tightening chokes electronics and vehicle sales, but the macro signal is thin next to the idiosyncratic one. (For how the major commodities compare across the same regimes, see the macro regime guide .)

Historically cobalt does best in technology-demand booms, supply-policy shocks and weak-dollar expansions; it does worst during its own supply waves and during chemistry-substitution phases, when the same battery growth that is supposed to help it quietly designs it out. Table 19 dates each regime to a real episode with the figures behind it. The pattern the last decade taught is the uncomfortable one: a metal can be destroyed by the growth of its own end market.

On past performance, cobalt has rewarded timing and punished patience. Across 2000–2025 the annual average rose only about 39% nominal — well below cumulative US inflation — while delivering two booms of more than 200%, a drawdown of nearly 70% from 2008 to 2016 and two further declines of 45–60%. It is a trading market that occasionally masquerades as a growth story. Past performance is not indicative of future results.

The equity amplifier. The 2022–2025 bust is the concluded case study, and a brutal one. Cobalt’s annual average price fell 46%, from $30.78/lb in 2022 to $16.77/lb in 2024, and roughly 72% from its April 2022 monthly peak. The equity offering the purest exposure did far worse: Jervois Global, owner of the only primary cobalt mine in the United States, fell about 99% from its 2022 peak. It idled the Idaho Cobalt Operations in March 2023, laying off 250 workers weeks before the mine was due to open, and in January 2025 signed a pre-packaged Chapter 11 rescue with Millstreet Capital that injected $145 million, converted more than $100 million of loans to equity and wiped out all existing shareholders. The lesson generalises: in a by-product metal, the businesses with the most price torque are exactly the ones with no host-metal revenue to survive the trough on. (Data: USGS annual average prices; ASX price history, company announcements and Chapter 11 filings, 2022–2025 — a concluded episode.)

On correlations, cobalt is the outlier of the battery complex. Its strongest links are to lithium, through shared cathode-chemistry demand, to nickel, through shared HPAL supply, and to copper, the host whose economics set DRC output. Links to broad risk assets and to the dollar are real but weak; the link to gold is effectively nil. Every relationship weakens further whenever DRC policy is in play, which since 2025 has been most of the time.

Table 19. Cobalt across economic regimes

Source: price history from USGS annual averages (Table 3); balance and chemistry data from the Cobalt Institute market reports, 2025–2026; author analysis. Regime characterisations are historical observations, not predictions.

Table 20. Cobalt correlations (approximate, monthly, 2010–2025)

Source: author analysis of published monthly cobalt, lithium, nickel, copper, equity and dollar price series over 2010–2025 — the window across which continuous battery-metal assessments exist. These are approximate magnitudes and signs, not computed betas: cobalt’s price is partly assessed rather than exchange-traded, and its supply shocks are idiosyncratic, so correlations are unstable, sample-dependent, and break down entirely during policy events such as 2025.

Figure 12. Cobalt correlations, approximate, monthly 2010–2025

Figure data: Table 20. Values are approximate author estimates, not computed coefficients. Shading runs on a fixed ±0.75 “strong correlation” scale, so a tile means the same thing here as in any other guide, rather than being re-ranked within this figure.

4.4 Price drivers & cycles

Strip out the noise and cobalt’s price is set by a short list of forces, best read through concluded episodes. On the supply side: DRC policy, copper and nickel output decisions taken for reasons that have nothing to do with cobalt, and — at the margin — artisanal supply and recycling. On the demand side: cathode chemistry, the consumer-electronics cycle, defence and aerospace, and Chinese state stockpiling. The recurring pattern is the classic inelastic-supply signature: long, flat stretches punctuated by violent overshoots in both directions, because when the market needs to clear, price is the only variable that can move.

The settled case studies are four, and Section 1.3 carries their price arithmetic. The 2007–2008 mania and crash was a demand shock — Chinese industrialisation, Congolese supply disruption and a speculative squeeze, unwound by the financial crisis. The 2017–2018 electric-vehicle mania was a forecasting failure rather than a directional one: EVs arrived, but with less cobalt each, and Glencore mothballed Mutanda in November 2019 trying to defend the price on the way down. The 2023–2025 Kisanfu flood was a pure supply event — one company running 31% above stated capacity shut smaller DRC cobalt circuits down entirely and bankrupted the only US primary producer, with the cobalt price influencing its decision at no point. And the 2025 DRC intervention was the first time in the metal’s history that a producing state successfully managed it.

Table 21. Cobalt price drivers

Source: agency and industry data (USGS , Cobalt Institute ) and the long-run price history in Table 3. All case studies referenced above are concluded historical episodes.

4.5 Risks, controversies & ESG

Cobalt’s bull and bear cases share a root, and it is the same root as nickel’s: concentration. Policy risk cuts both ways. The DRC’s quota is the reason the price is high; widening it, or failing to enforce it against smuggling, would be the reason it falls. Every cobalt thesis written today is, knowingly or not, a forecast of ARECOMS’s behaviour — and quota regimes have a poor record once the incentive to cheat grows large enough. China’s refining position compounds it: with roughly four-fifths of conversion capacity, China holds an export-control lever it has already used on other critical minerals.

Substitution is the structural bear case, and Section 2.2 quantifies it. Sodium-ion, if it scales, would remove cobalt entirely from parts of the storage and entry-level EV market. Working the other way, recycling will displace primary demand from the 2030s as the first large EV fleets retire — good for supply security, bad for miners. And because supply is inelastic, cobalt has no defence against any of this: it cannot shut capacity to defend a price.

The ESG ledger is heavy, contested, and central to the investment case rather than adjacent to it. Artisanal mining — hand-dug cobalt, worked in unsafe conditions and documented to involve child labour — has been the sector’s defining reputational problem for a decade, and the proportions matter: at under 2% of DRC production it is a small share of tonnes and a very large share of risk, because one traceability failure contaminates a buyer’s whole supply chain. The industry’s answer has been assurance at scale (Section 2.7); the DRC’s has been EGC’s monopoly channel. Reasonable people disagree about whether formalisation improves artisanal miners’ lives or merely launders the supply chain.

Two further points deserve fair statement. Indonesian HPAL cobalt is not the low-carbon alternative it is sometimes assumed to be: the Cobalt Institute’s life-cycle assessment puts its global warming potential more than 70% above the DRC-to-China route, because the ore is so much leaner, with acid-neutralisation emissions alone accounting for nearly a third of the total; tailings and residue disposal remain contested. And the conflict in eastern DRC, which resurged in 2024–25 around Goma and the Kivus, is roughly 1,100 km in a straight line from the cobalt-producing Copperbelt (more than 2,000 km by road) and has had little direct effect on output — coltan and the 3TG minerals, not cobalt, are the conflict-mineral story, and conflating them misprices the risk in both directions.

Figure 13. Cobalt risk map — likelihood vs. impact

Source: author’s qualitative assessment. Each point is scored 1–5 on likelihood and impact (printed as likelihood · impact) and nudged within its cell so the marks read separately; each risk is developed with figures in Section 4.5.

5. Future outlook & forecasts

Cobalt’s next five years turn on a question no other metal has to answer: whether a supply cartel of one can hold. Forecasts here are scenarios, not measured facts — they depend on ARECOMS’s decisions, on Indonesian construction schedules and on cathode chemistry, and the last decade shows how badly all three can be misjudged.

5.1 Demand

Demand reached 276 kt in 2025, but its composition is shifting in a way that arguably improves cobalt’s quality as an investment. Batteries still take about three-quarters, and there the tug-of-war continues: more vehicles and devices against falling cobalt intensity per unit. The newer, more durable pillar is defence and aerospace — superalloys, SmCo magnets, drone and soldier-system batteries. That demand is not priced off subsidy schedules and does not substitute to LFP, and the Cobalt Institute recorded it rising through 2024 and 2025.

5.2 Supply and the balance

Supply is where the scenarios diverge sharply. Indonesia is the fastest-growing source of cobalt supply, with more than 95% of its growth to 2030 expected to arrive as MHP from HPAL plants — Benchmark’s analysis sees Indonesian MHP cobalt with the potential to rise roughly fivefold by the end of the decade, which alone would exceed the DRC’s entire 2022 output. Set against that, the DRC quota described in Section 2.7 holds national exports at roughly 42% of what the country mined in 2025 through 2027. Two very large forces therefore point in opposite directions, and the near-term balance depends on which moves first: Fastmarkets and others entered 2026 expecting a deficit as the pre-quota feedstock pipeline drained, while a widening of the quota or faster Indonesian construction would refill it quickly.

One wildcard sits outside that arithmetic. Government stockpiling could add a non-cyclical demand layer: the United States published a solicitation in August 2025 to buy about 7,500 t of cobalt over five years for the National Defense Stockpile, then cancelled it that October — an instructive near-miss, since a purchase of that size would be roughly 3% of annual world demand.

Figure 14. DRC cobalt output versus the export quota (kt Co)

Source: USGS Mineral Commodity Summaries , 2024 and 2026 editions — DRC mine production of 170,000 t (2023), 226,000 t (2024) and 230,000 t (2025e), and the quota of 96,600 t per year for 2026 and 2027 inclusive of 9,600 t for national strategic reserves. Values are in thousand tonnes. Mined output and permitted exports are shown on one scale; the gap is what accumulates inside the country. The producer-quota bar is the commercial share of the 96.6 kt cap above it, not a separate flow.

Table 22. Cobalt supply, demand and policy outlook

Source: Cobalt Institute Cobalt Market Reports 2024 and 2025; quota per USGS , February 2026. The 2023 demand figure is derived (see the note to Table 9); directional columns are scenario judgements, not forecasts, and the 2025 demand base differs in vintage from 2023–24.

5.3 Catalysts to watch

The near-term watch-list is dominated by one capital: Kinshasa. ARECOMS’s quota decisions, its treatment of unused allocations and — above all — whether the 2027 regime is extended, tightened or abandoned are the fastest-moving price drivers in the market. After that, watch CMOC’s quarterly production reports, because one company’s ramp decisions have moved the world price twice in three years; Indonesian HPAL commissioning and sulphur costs, the supply growth that no policy controls; the LFP-versus-NCM share in monthly battery data, which decides demand intensity; NDRC stockpile activity, historically the marker of a price floor; and any revival of the US National Defense Stockpile purchase. What would confirm the bull case: a quota held tight into a recovering electronics and defence cycle. What would break it: enforcement failure, a widened quota, or Indonesian MHP arriving faster than the DRC withholds.

Table 23. Cobalt catalyst calendar

Source: Cobalt Institute , USGS and company reporting calendars, 2026.

6. Summary

Cobalt is the metal that comes along for the ride. It is prized for two unrelated jobs — stabilising lithium-ion cathodes and holding strength in jet-engine superalloys — but almost nobody mines it on purpose: roughly 76% arrives as a by-product of copper and 23% of nickel, and only Morocco’s Bou Azzer runs at scale as a primary cobalt mine. That single fact drives everything else, because it means cobalt supply is nearly deaf to the cobalt price, so all the adjustment happens through demand — substitution and thrifting — or through policy. Its price is best read as a multi-year average, and those averages describe a violent, mean-reverting market centred near $19/lb since 2000, with peaks of $39.01/lb in 2008 and $37.43 in 2018 around a $16.77 trough in 2024 and ~$21 in 2025. It is mined 74% in one country and refined 79% in another — a double concentration no other major metal approaches — and consumed three-quarters as batteries, with superalloys the durable industrial floor beneath them. Its balance ran a record +36 kt surplus in 2024 and then inverted in the strangest way in commodity history: in 2025 the world produced 295 kt but only 183 kt reached buyers, because roughly 112 kt was stranded behind a closed Congolese border. Coordination, absent for a century, now exists as ARECOMS, whose 96,600 t annual export quota for 2026–27 is about 42% of what the DRC mines. The companies are copper and nickel businesses with cobalt attached — CMOC above all — plus a Chinese refining tier and a very thin set of pure plays, compared on production, resources, cobalt revenue share and cost position, never market cap. The regime profile is the least macro-driven of the industrial metals and the most policy-driven; the durable lesson of 2023–25 is that in a by-product metal the price cannot discipline supply, so a government eventually tries to. The variable to watch is DRC export policy, with cathode chemistry close behind.

To go from this big-picture view to the actual companies — screening every copper-cobalt and nickel-cobalt producer by production, resources and cost position — explore Metal Pilot.

7. Vocabulary

Every specialised term used in this report, defined once at first use in the body, collected here for reference.

Table 24. Cobalt vocabulary

Source: definitions follow JORC and SEC S-K 1300 reserve standards and Cobalt Institute terminology, 2026.

8. Sources, methodology & disclaimer

8.1 Sources, methodology & data vintage

Agencies & official data: USGS Mineral Commodity Summaries 2026: Cobalt and the 2005, 2010, 2015, 2017, 2021, 2024 and 2026 editions for mine production, reserves, prices and US statistics; USGS Cobalt Statistics and Information for the long-run series.

Industry & exchanges: Cobalt Institute , whose market reports are prepared with Benchmark Mineral Intelligence — the Cobalt Market Report 2024 (May 2025) and Cobalt Market Report 2025 (May 2026) — for demand by end use, mined and refined supply, balances, recycling, artisanal-mining shares and life-cycle assessment; LME cobalt contracts and CME battery-metals futures for contract structure; Benchmark Mineral Intelligence for the DRC quota framework and allocations.

Company filings & statements: CMOC production reporting for 2024 and 2025; Glencore Full Year 2025 Production Report ; ERG, Harita Nickel, Sherritt, Managem and Nornickel figures as reported in the Cobalt Institute market reports and company disclosures, 2024–2025.

Methodology: prices are calendar-year averages of the USGS U.S. spot cathode assessment (2000–2025), never spot snapshots; the single 2026 price reference in Section 1.3 is explicitly dated. Two supply series appear here and are deliberately not merged: USGS mine production counts contained cobalt in ore and runs roughly 11–19% above the Cobalt Institute / Benchmark mined-supply series, which nets out downstream yield losses. Tables 4, 7 and 8 and Figures 4, 6 and 14 use the USGS basis; Tables 9, 10, 11 and 22 and Figures 7, 8 and 9 use the Cobalt Institute basis, and each source line says which. Demand carries a vintage break too: Benchmark revised its 2024 base upward between the two reports, so the 2025 growth rate is measured on the revised base. The 2025 “available supply” figure of 183 kt is the Cobalt Institute’s; the deficit derived from it is arithmetic, labelled as such. Two artifacts this guide series normally carries are adapted, because the data does not exist for cobalt: no agency publishes country-level end consumption on a contained-cobalt basis, so refining location (Section 2.5) stands in for a top-consumers table; and cobalt’s trade has three centres rather than a rankable top five, so Table 12 carries volumes and net positions in place of separate exporter and importer tables. Company outputs sit on differing bases (attributable vs. 100%, mined vs. refined) as noted under each table, and Indonesian per-asset figures are derived from plant capacity and are indicative only. Correlations are approximate author estimates over 2010–2025, not computed coefficients. Reserves, resources and forecasts are estimates, not measured facts; where sources conflict, the more primary and more recent figure is used and the discrepancy is stated.

Data as of: 6 September 2026. Intended update cadence: annually, after the USGS Mineral Commodity Summaries (January/February) and the Cobalt Institute market report (May).

8.2 Disclaimer & disclosure

This report is for informational purposes only and is not investment advice, a recommendation, or an offer to buy or sell any security or commodity. Cobalt prices are exceptionally volatile and policy-driven, and the figures here are estimates as of the stated date that will change; reserves, resources, correlations and regime characterisations are estimates and historical observations that may not persist. Do your own research and consult a licensed financial adviser before acting. This report was prepared with the assistance of AI; its figures were sourced from the references above and reviewed, but readers should verify any number before relying on it. The author holds no position disclosed as a conflict in respect of the companies named.