Result: agreement on 4 of 5. On bauxite, cobalt, graphite and rare-earth
magnets the atlas matches USGS/IEA on mine and refining shares, and its refiner-illusion / processing-chokepoint
signals hold. On lithium the audit did its job — it surfaced a stale mine-share vintage (now corrected to
USGS 2024) and a product-definition nuance (the traded code is carbonate, not spodumene). External figures: USGS
Mineral Commodity Summaries 2025 and IEA Critical Minerals Outlook 2025.
Atlas figures are pulled live by
build_casestudies.py, so this page cannot drift from the data.Bauxite / alumina / aluminium
Strong agreementAtlas shows
Mine 🇦🇺Australia 24%, 🇬🇳Guinea 24%, 🇨🇳China 23%, 🇧🇷Brazil 8%
Refine 🇨🇳China 60%, 🇦🇺Australia 12%
Top exporters 🇬🇳Guinea 72%, 🇦🇺Australia 16%, 🇧🇷Brazil 2%
Network chokepoint 🇨🇳China (processing chokepoint)
Refine 🇨🇳China 60%, 🇦🇺Australia 12%
Top exporters 🇬🇳Guinea 72%, 🇦🇺Australia 16%, 🇧🇷Brazil 2%
Network chokepoint 🇨🇳China (processing chokepoint)
Authoritative (USGS 2024 / IEA)
Mine Guinea ~29%, Australia ~22%, China ~21%, Brazil ~7% (USGS 2024)
Refine Alumina: China ~59%, Australia ~13% (IEA); smelting China-dominant
Chokepoint Alumina refining and energy-intensive smelting — China
Refine Alumina: China ~59%, Australia ~13% (IEA); smelting China-dominant
Chokepoint Alumina refining and energy-intensive smelting — China
Commonly misread as: "Australia controls bauxite" — but China refines ~59% of alumina; mine tonnage ≠ value-added control
Verdict: Atlas refined share (China 58%) matches IEA (59%); mine shares align (Australia and Guinea are neck-and-neck — the atlas lists Australia first, USGS 2024 Guinea first). The network layer correctly identifies China as the processing chokepoint despite its small export share.
Verdict: Atlas refined share (China 58%) matches IEA (59%); mine shares align (Australia and Guinea are neck-and-neck — the atlas lists Australia first, USGS 2024 Guinea first). The network layer correctly identifies China as the processing chokepoint despite its small export share.
Cobalt
Strong agreementAtlas shows
Mine 🇨🇩Congo [DRC] 74%, 🇮🇩Indonesia 7%, 🇷🇺Russia 4%, 🇦🇺Australia 2%
Refine 🇨🇳China 82%, 🇫🇮Finland 6%
Top exporters 🇫🇮Finland 29%, 🇨🇳China 19%, 🇿🇲Zambia 17%
Network chokepoint 🇨🇳China (processing chokepoint)
Refine 🇨🇳China 82%, 🇫🇮Finland 6%
Top exporters 🇫🇮Finland 29%, 🇨🇳China 19%, 🇿🇲Zambia 17%
Network chokepoint 🇨🇳China (processing chokepoint)
Authoritative (USGS 2024 / IEA)
Mine DR Congo ~76%, Indonesia ~10% (USGS 2024)
Refine China ~70–80% of refined (IEA)
Chokepoint DRC mine concentration + Chinese refining/chemical processing
Refine China ~70–80% of refined (IEA)
Chokepoint DRC mine concentration + Chinese refining/chemical processing
Commonly misread as: "DRC produces most cobalt" — true at the mine, but DRC refines almost none; usable cobalt is Chinese-processed
Verdict: Atlas has mine DR Congo 76%, refine China 76%, yet the top customs exporter is Finland (29%) — and the origin trace re-attributes it to DR Congo. The refiner illusion is captured exactly.
Verdict: Atlas has mine DR Congo 76%, refine China 76%, yet the top customs exporter is Finland (29%) — and the origin trace re-attributes it to DR Congo. The refiner illusion is captured exactly.
Natural graphite
Strong agreementAtlas shows
Mine 🇨🇳China 77%, 🇲🇬Madagascar 6%, 🇲🇿Mozambique 6%, 🇧🇷Brazil 5%
Refine 🇨🇳China 94%, 🇯🇵Japan 3%
Top exporters 🇨🇳China 36%, 🇹🇿Tanzania 27%, 🇩🇪Germany 7%
Network chokepoint 🇨🇳China (processing chokepoint)
Refine 🇨🇳China 94%, 🇯🇵Japan 3%
Top exporters 🇨🇳China 36%, 🇹🇿Tanzania 27%, 🇩🇪Germany 7%
Network chokepoint 🇨🇳China (processing chokepoint)
Authoritative (USGS 2024 / IEA)
Mine China ~79% (USGS 2024; ~82% in 2025)
Refine China ~90%+ of battery-grade / spherical (IEA)
Chokepoint Mining of usable grades + battery-grade processing — China
Refine China ~90%+ of battery-grade / spherical (IEA)
Chokepoint Mining of usable grades + battery-grade processing — China
Commonly misread as: Assuming synthetic graphite fully substitutes — it competes only partly, and China leads that too
Verdict: Atlas: mine 78% / refine 95% / export 36%, all China-led, chokepoint China. Matches the authoritative picture closely.
Verdict: Atlas: mine 78% / refine 95% / export 36%, all China-led, chokepoint China. Matches the authoritative picture closely.
Lithium
Nuance + correctedAtlas shows
Mine 🇦🇺Australia 48%, 🇨🇱Chile 24%, 🇨🇳China 18%, 🇦🇷Argentina 5%
Refine 🇨🇳China 74%, 🇨🇱Chile 16%
Top exporters 🇨🇱Chile 75%, 🇦🇷Argentina 18%, 🇰🇷South Korea 2%
Network chokepoint 🇨🇳China (processing chokepoint)
Refine 🇨🇳China 74%, 🇨🇱Chile 16%
Top exporters 🇨🇱Chile 75%, 🇦🇷Argentina 18%, 🇰🇷South Korea 2%
Network chokepoint 🇨🇳China (processing chokepoint)
Authoritative (USGS 2024 / IEA)
Mine Australia ~37%, Chile ~20%, China ~17%, Zimbabwe ~9%, Argentina ~8% (USGS 2024, Li content)
Refine China ~60–80% of battery-grade conversion (IEA)
Chokepoint Chemical conversion capacity — China-heavy
Refine China ~60–80% of battery-grade conversion (IEA)
Chokepoint Chemical conversion capacity — China-heavy
Commonly misread as: Australia is top miner, but spodumene is not battery-ready until converted (mostly in China); a mine-vs-export comparison ignores the product/conversion step
Verdict: Two findings, both now handled. (1) The traded HS code is lithium carbonate — Chile’s product (75% of trade) — so the mine-vs-export gap is a PRODUCT difference (spodumene vs carbonate), not a refiner illusion; the atlas now states this. (2) This audit caught a stale mine-share vintage (the atlas had Australia 52%); it is corrected here to USGS 2024 (Australia 37%, with Zimbabwe ~9% added).
Verdict: Two findings, both now handled. (1) The traded HS code is lithium carbonate — Chile’s product (75% of trade) — so the mine-vs-export gap is a PRODUCT difference (spodumene vs carbonate), not a refiner illusion; the atlas now states this. (2) This audit caught a stale mine-share vintage (the atlas had Australia 52%); it is corrected here to USGS 2024 (Australia 37%, with Zimbabwe ~9% added).
Rare-earth permanent magnets (NdFeB)
Strong agreementAtlas shows
Mine 🇨🇳China 69%, 🇺🇸United States 12%, 🇲🇲Myanmar [Burma] 11%, 🇦🇺Australia 5%
Refine 🇨🇳China 84%, 🇺🇸United States 5%
Top exporters 🇨🇳China 64%, 🇯🇵Japan 9%, 🇻🇳Vietnam 6%
Network chokepoint 🇨🇳China (processing chokepoint)
Refine 🇨🇳China 84%, 🇺🇸United States 5%
Top exporters 🇨🇳China 64%, 🇯🇵Japan 9%, 🇻🇳Vietnam 6%
Network chokepoint 🇨🇳China (processing chokepoint)
Authoritative (USGS 2024 / IEA)
Mine China ~60–70% REO, US ~12%, Myanmar significant (USGS 2024)
Refine China ~85–92% separation, ~90–95% of NdFeB magnets (IEA/industry)
Chokepoint Midstream separation + magnet fabrication — China; heavy REE (Dy, Tb) tighter still
Refine China ~85–92% separation, ~90–95% of NdFeB magnets (IEA/industry)
Chokepoint Midstream separation + magnet fabrication — China; heavy REE (Dy, Tb) tighter still
Commonly misread as: Mine shares look diversified; separated and magnet-grade material is far more concentrated
Verdict: Atlas: mine China 69% / refine 90% / export 64%, all China-led; Myanmar (11%) is included and flagged as the worst-governance producer (g 0.84). Captured well.
Verdict: Atlas: mine China 69% / refine 90% / export 64%, all China-led; Myanmar (11%) is included and flagged as the worst-governance producer (g 0.84). Captured well.
Computed from data.json + flows_2024.json + the layer outputs → casestudies.json. External shares are rounded public figures; mine production ≠ exports ≠ refined output — that distinction is the atlas's whole point.