Who produces the metals powering AI, defense and the energy transition — who refines and consumes them — where demand is heading, and where the supply chain breaks.
The minerals behind AI datacenters, defense systems and the energy transition are not scarce in the ground — they are scarce in the supply chain. The world mines critical minerals in many places but refines them in one. That asymmetry, not geology, is the defining risk of the decade.
Ranked selection of strategic mineral commodities by producer concentration. Share = approximate % of global mine / primary production, 2024 (USGS-derived). Import reliance = US net import reliance as % of apparent consumption. CRIT = on the USGS 2022 Final List of Critical Minerals.
| Mineral | Top producers (mine share, 2024) | US import reliance | China midstream | List | Key supply risk |
|---|---|---|---|---|---|
| Cobalt | DR Congo ~76%, Indonesia ~10%, Russia ~3% | 76% | >75% | CRIT | DRC concentration + Chinese refining; artisanal/ESG exposure |
| Rare earths | China ~69%, US ~12%, Myanmar ~8% | 80% | ~85–90% | CRIT | China dominates separation + magnet alloy; export licensing |
| Graphite (nat.) | China ~79%, Madagascar ~6%, Mozambique ~5% | 100% | ~98% | CRIT | China mining + near-total anode processing; Dec-2023 controls |
| Gallium | China ~99%, Russia, Japan/Korea | 100% | ~98% | CRIT | China banned all exports to US (Dec 2024); GaN/GaAs chips |
| Magnesium | China ~95%, others ~2% each | >75% | ~95% | CRIT | ~95% single-country; only US smelter idle since 2021 |
| Niobium | Brazil ~92%, Canada, DR Congo | 100% | — | CRIT | Single-country (Brazil) dependence; US 100% reliant since 1959 |
| Tungsten | China ~83%, Vietnam ~4%, Russia ~3% | >50% | dominant | CRIT | Chinese production + APT/carbide conversion dominance |
| Platinum | South Africa ~71%, Zimbabwe ~11%, Russia ~11% | 85% | — | CRIT | Extreme S. Africa concentration; grid/load-shedding risk |
| Vanadium | China ~70%, Russia ~21%, S. Africa ~8% | 40% | dominant | CRIT | China + Russia ~91%; no US primary production |
| Indium | China ~70%, S. Korea ~17%, Japan ~6% | 100% | ~61% | CRIT | 100% reliant; China dominates output + exports (ITO coatings) |
| Titanium sponge | China ~64%, Japan ~16%, Russia ~6% | >95% | dominant | CRIT | Aerospace-grade sponge China-concentrated; 1 small US plant |
| Fluorspar | China ~62%, Mexico ~13%, Mongolia ~13% | 100% | dominant | CRIT | 100% reliant; Chinese mining + HF/fluorochemical dominance |
| Antimony | China ~60%, Tajikistan ~17%, Russia ~13% | 85% | dominant | CRIT | China banned exports to US (Dec 2024); price ~tripled |
| Nickel | Indonesia ~60%, Philippines ~9%, Russia ~6% | ~100%* | controls Indo. | CRIT | Single-country supply; *ex-scrap reliance near 100% |
Source: USGS Mineral Commodity Summaries 2025 (mine/primary production, 2024 est.); refining/midstream shares IEA/industry. Shares rounded. Continued overleaf.
| Mineral | Top producers (mine share, 2024) | US import reliance | China midstream | List | Key supply risk |
|---|---|---|---|---|---|
| Lithium | Australia ~37%, Chile ~20%, China ~17% | >50% | ~60–65% | CRIT | Chinese chemical-conversion concentration; price-driven cuts |
| Tantalum | DR Congo ~42%, Nigeria ~19%, Rwanda ~17% | 100% | metal/powder | CRIT | Conflict-affected Central-Africa sourcing; US 100% reliant |
| Manganese | S. Africa ~37%, Gabon ~23%, Australia ~14% | 100% | alloy/battery | CRIT | US zero domestic ore; Chinese alloy + battery-grade processing |
| Germanium | China leading (~68% output), Russia, others | >50% | ~68% | CRIT | China banned exports to US (Dec 2024); fiber/IR optics |
| Dysprosium / Terbium | China + Myanmar feedstock (heavy REE) | (REE 80%) | near-total | CRIT | Myanmar ion-clay → China separation; named in 2025 controls |
| Tin | China ~23%, Indonesia ~17%, Myanmar ~11% | 73% | + Indonesia | CRIT | Concentrated SE-Asian smelting; Myanmar (Wa) + Indo. policy |
| Zinc | China ~33%, Peru ~11%, Australia ~9% | 73% | dominant | CRIT | China mining + smelting dominance; 2024 refined deficit |
| Aluminum / Bauxite | Bauxite: Guinea ~29%, Australia ~22% / Al metal: China ~60% | 47% | ~60% | CRIT | China ~60% of (coal-powered) smelting; Guinea bauxite risk |
| Palladium | Russia ~39%, S. Africa ~38%, Canada ~8% | 36% | — | CRIT | Russia sanctions + SA grid; 2 countries ~77% of output |
| Titanium (mineral) | China ~37%, Mozambique ~21%, S. Africa ~15% | 86% | led | CRIT | China-led mining/consumption; TiO₂ pigment feedstock |
| Arsenic | Peru ~47%, China ~41%, Morocco ~10% | 100% | US imports | CRIT | Zero US output since 1985; ~96% of US metal from China |
| Tellurium | China ~75%, Russia ~7%, Japan ~7% | <25% | Cu-byprod. | CRIT | Byproduct of Cu refining (inelastic); CdTe solar |
| Copper | Chile ~23%, DR Congo ~14%, Peru ~11% | 45% | ~44% | — | Chinese smelting concentration; falling ore grades |
| Silver | Mexico ~25%, China ~13%, Peru ~12% | 64% | — | — | Inelastic byproduct supply; 2024 consumption > supply |
| Uranium | Kazakhstan ~39%, Canada ~24%, Namibia ~12% | ~92% | (Russia enr.) | fuel | ~92% imported; Russian enrichment dominance |
The single most important structural fact in critical minerals: ore is mined widely, but turned into usable material in very few places. The IEA finds China is the dominant refiner for 19 of 20 energy-related minerals, with an average processing share around 70%. The geographic mismatch between mine and midstream is where supply risk actually lives.
| Mineral | China processing / refining share | Mining reality (for contrast) | Source |
|---|---|---|---|
| Rare earths | ~85–90% of separation | China ~69% mined — the chokepoint is separation, not ore | IEA / USGS |
| Graphite (anode) | ~98% battery-grade | China ~79% mined; ~74% of full anode chain | Benchmark |
| Gallium | ~98% low-purity | ~50% at refined high-purity stage | USGS / CSIS |
| Cobalt | >75% refined | DR Congo ~76% mined but refines almost none | IEA |
| Germanium | ~68% of output | Few Western refiners remain | USGS / CSIS |
| Lithium (refined) | ~60–65% | Australia/Chile dominate mining; China dominates conversion | IEA |
| Aluminum (primary) | ~60% smelting | Bauxite spread across Guinea, Australia, China | USGS |
| Copper (refined) | ~44% | vs ~8% of mine output — the clearest mine-vs-midstream split | USGS / IEA |
| Nickel (Indonesia) | >75% of Indo. capacity is Chinese-controlled | Indonesia processes ~45% of refined nickel | NBR / ORF |
AI's mineral footprint is not mainly in the chips — it is in the power. Building, connecting and energising datacenters is a copper, gallium and rare-earth story before it is a silicon one.
Datacenters are becoming less copper-intensive per facility, but getting power to them — substations, transmission, distribution — is where the incremental copper sits. Estimates vary by scope, so the scope matters more than the number:
| Scope | Incremental copper | Source |
|---|---|---|
| Datacenter-direct, peak | ~400 kt/yr avg, peak ~572 kt (2028); >4.3 Mt cumulative by 2035 | BloombergNEF, Aug 2025 |
| Datacenter + grid connections | ~1 Mt added by 2030 | Trafigura |
| Grid-to-datacenter (broad) | up to ~5 Mt in new T&D through 2030 | Wood Mackenzie |
| Datacenter, 2040 | 2.5 Mt/yr (range 1.7–2.7) | S&P Global |
Intensity: roughly 30–40 tonnes of copper per MW of datacenter capacity (S&P). Against this, the IEA projects a ~30% copper supply shortfall by 2035 under current policy — widening on faster-decarbonisation paths.
All figures below are modeled scenarios from named sources, not forecasts of price or returns. IEA multiples are robust on direction and magnitude; absolute tonnages by scenario live in the IEA Critical Minerals Data Explorer.
| Mineral | Demand trajectory | Primary drivers | Source / scenario |
|---|---|---|---|
| Lithium | ~5× by 2040 (STEPS); up to ~7× by 2035 (NZE). 1→>3 Mt LCE by 2030 | EVs/batteries ~80% of growth; grid storage | IEA; Benchmark |
| Copper | 28→42 Mt by 2040 (+50%); ~30% supply gap by 2035 | Grid + AI datacenters + defense (each ~3× by 2040); EVs | S&P; IEA |
| Rare-earth magnets | NdFeB demand ~4× to >880 kt by 2040; ~206 kt/yr shortfall by 2035 | EV motors, wind, robotics (~29% CAGR), defense, datacenter motors | Adamas; IEA |
| Graphite | Battery-sector ~+250% (≈3.5×) 2023→2030; total 2× by 2040 | EV/battery anodes (~86% of demand), grid storage | Benchmark; IEA |
| Gallium | Datacenter demand >10% of today's supply by 2030; power-GaN market ~42% CAGR | GaN power chips (datacenter PSUs, EV inverters), defense radar/EW | IEA; Yole |
| Cobalt | +50–60% by 2040 (STEPS); up to ~3× by 2035 (NZE) | Batteries; aerospace superalloys/defense. LFP shift = headwind | IEA |
| Nickel | ~2× by 2040 (STEPS) — a more balanced market | EV batteries; stainless steel baseline | IEA |
A composite read of structural risk. High = single-country dominance and high US import reliance and active export-control exposure. Elevated = two of three. Moderate = concentrated but with diversification or balanced supply.
| Mineral | Producer concentration | US import reliance | Policy / control risk | Composite |
|---|---|---|---|---|
| Gallium | China ~99% | 100% | US export ban (Dec 2024) | High |
| Germanium | China leading | >50% | US export ban (Dec 2024) | High |
| Rare earths / heavy REE | China ~69% mine, ~90% sep. | 80% | 12/17 licensed (2025 peak) | High |
| Graphite (natural) | China ~79%, ~98% anode | 100% | Export licensing (Dec 2023) | High |
| Antimony | China ~60% | 85% | US export ban (Dec 2024) | High |
| Tungsten | China ~83% | >50% | Added to China controls | High |
| Magnesium | China ~95% | >75% | No active ban | Elevated |
| Cobalt | DR Congo ~76% | 76% | China refining leverage | Elevated |
| Niobium | Brazil ~92% | 100% | Allied supplier | Elevated |
| Platinum / Palladium | S. Africa / Russia | 85% / 36% | Russia sanctions exposure | Elevated |
| Lithium | Australia ~37% | >50% | China conversion concentration | Elevated |
| Nickel | Indonesia ~60% | ~100% ex-scrap | Chinese-controlled capacity | Elevated |
| Copper | Chile ~23% (dispersed) | 45% | China ~44% smelting | Moderate |
| Uranium | Kazakhstan ~39% | ~92% | Russian enrichment | Elevated |
Since 2023, policy — not geology — has been the dominant price signal in critical minerals. The timeline below is corroborated across multiple sources (Reuters, CSIS, CSET/Georgetown, IEA, European Commission, US Treasury/White House).
Mine/primary production tonnages and US net import-reliance percentages are quoted from the USGS Mineral Commodity Summaries 2025 (published Jan 2025; data year 2024, estimated), with country shares derived arithmetically from USGS world-production tables and rounded. Refining/midstream shares are not from the USGS (which tabulates mine output) — they are IEA and industry estimates and are labelled as such throughout. Critical-list membership refers to the USGS 2022 Final List of 50 Critical Minerals (Federal Register 87 FR 10381); a newer 2025 list exists but is not used here. Demand trajectories are modeled scenarios from the named sources.
USGS Mineral Commodity Summaries 2025 · IEA Global Critical Minerals Outlook 2025 · IEA Energy and AI (2025) · S&P Global Copper in the Age of AI (Jan 2026) · BloombergNEF (Aug 2025) · Adamas Intelligence · Benchmark Mineral Intelligence · Yole Group · CSIS · CSET / Georgetown · European Commission / Consilium · Wood Mackenzie · CRU · Trafigura · World Nuclear Association & EIA (uranium).
Important — not investment advice. This report is produced by MetalsIntel, an editorial & data publication of Atlas Markets, a critical-minerals data and analytics company; this report and the data it references may relate to Atlas products, an affiliation we disclose openly. The report is provided for general information and education only. Nothing in it is investment, financial, legal or tax advice, a research recommendation, a price quote, a solicitation, or an offer to buy or sell any security, commodity, instrument or derivative. Demand trajectories are modeled scenarios, not forecasts of price, supply or returns. Figures may be delayed, estimated, illustrative, or subsequently revised; third-party figures are reproduced as cited and not independently audited. Commodity and derivative markets carry substantial risk of loss. Do your own research and consult a licensed professional before making any decision. © 2026 Atlas Markets. All rights reserved.