Critical Minerals: The Sovereign-Gated Supply Chain

China processes roughly 91% of global magnet rare-earth production and manufactures 94% of the NdFeB permanent magnets inside every EV drivetrain, wind-turbine generator, and precision-guided munition — a structural concentration that export-control regimes, not geography alone, have converted into a durable pricing-power mechanism. The mine can be built almost anywhere; the solvent-extraction chemistry that separates individual rare-earth oxides from concentrate has been scaled industrially in one country, and that country has spent two years extending its licensing jurisdiction to magnets manufactured abroad. The question for this desk is not where the chain is most likely to break, but where the spread the bottleneck implies is being structurally captured.

The Bottleneck Is the Chemistry, Not the Mine

The Bottleneck Map frames this distinction cleanly: fragility describes a topology that could fail; a bottleneck describes a topology that has already chosen a winner. The supply-chain-fragility literature is the literature of single-point failure — it asks what breaks first and where redundancy is absent. This paper asks who is holding the part of the income statement that everyone else must flow through, and what mechanism makes them hold it durably.

For critical minerals, the answer is located in the midstream. Mining can be geographically distributed — the United States mines rare-earth concentrates at Mountain Pass, Australia at Mount Weld, Brazil from substantial deposits. But the separation and refining step, where solvent-extraction chemistry isolates individual rare-earth oxides from concentrate, has been scaled industrially in China and only narrowly elsewhere. China holds approximately 91% of global magnet rare-earth separation and refining, 94% of NdFeB permanent-magnet manufacturing, and an average refining share of roughly 70% across the 19 strategic minerals the IEA tracks (IEA Critical Minerals Outlook 2024). The pricing power sits with producer-country governments operating export-control regimes on top of that concentration. That combination — processing dominance plus sovereign enforcement — is the structure competing buyers cannot price out.

This is structurally distinct from IP-protected pricing power. Ozempic's patent expires in 2030; Mounjaro's in 2036; every drug class in the FDA Orange Book carries a visible terminus. Sovereign protection carries no expiry date. China's rare-earth processing dominance, the Chilean CORFO state contract governing SQM's Atacama volumes through 2030, the Indonesian in-country smelting requirement that turned the PT-FI smelter from a corporate choice into a regulatory condition — each took decades to construct and would require decades to deconstruct through market competition alone. The export-control escalation timeline, which runs August 2023 → December 2023 → April 2025 → October 2025, demonstrates that this regime is actively expanding rather than static.

The Arithmetic — Three Structural Conditions

Three primary-sourced conditions define the sector's architecture.

Processing concentration. The relevant share is not the mining share but the separation-and-refining share. Across the 19 strategic minerals the IEA tracks, China's average refining market share sits at approximately 70% (IEA commentary on critical-minerals export controls, 2025). Replicating the solvent-extraction chemistry at commercial scale elsewhere requires a process-engineering investment measured in years, not months, and the December 2023 ban on rare-earth extraction and separation technology export blocks the knowledge transfer that would accelerate that rebuild.

Sovereign gating in action. China's August 2023 gallium and germanium export controls are the cleanest empirical test of the mechanism. Chinese germanium exports fell from 7,965 kg in July 2023 to 590 kg in October 2023; gallium fell from 6,876 kg to 227 kg — an effective near-zero in the month the controls took effect, per USITC data (USITC executive briefing on germanium and gallium export controls). The April and October 2025 layers added restrictions on seven rare-earth elements and — the strategically significant step — on foreign-produced magnets containing heavy rare earths above 0.1%, extending Chinese jurisdiction beyond Chinese borders (CSIS — China's new rare-earth and magnet restrictions, October 2025). A magnet manufactured in Japan or the United States with even a trace of Chinese-sourced dysprosium now falls under the licensing regime. The control regime is the binding mechanism; the underlying minerals are abundant.

Ore-grade decline in copper. Global average copper ore grade fell from approximately 1.5% in 1900 to around 0.6% by 2020, with continued deterioration since; processing lower-grade ore drives roughly 13% higher energy intensity per tonne of copper produced, per ICSG data (ICSG selected copper statistics). Grade decline concentrates pricing power in the lowest-cost, highest-grade mines regardless of political risk profile, and makes smelter and concentrator capacity — not mine output — the binding supply constraint. New smelters take seven to ten years to permit and build. That construction lag is the mechanism by which refining concentration persists even when mine output is distributed.

First-Order Chokepoints — Named Holders by Mineral Class

Six entities define the first-order layer, clustered by mineral class.

Rare earths. MP Materials operates Mountain Pass in California, the only US-domiciled rare-earth concentrate producer at scale. The company produced 45,455 metric tonnes of rare-earth oxide in concentrate in 2024 — a record — and 1,294 mt of NdPr oxide, with magnet production commencing at its Independence facility from late 2024 and 2025 (MP Materials Q4/FY 2024 results). Department of Defense and General Motors offtake agreements function, in practice, as sovereign buyer-of-last-resort for US-sovereign rare-earth feedstock. The desk currently tracks MP as the only listed US pure-play on the NdFeB magnet bottleneck that binds robotics, EV motors, and defence simultaneously; the Independence facility ramp is the supply-side enabler if these applications scale, but the stock has run materially on bottleneck recognition, meaning the structural read and the crowded trade are beginning to converge.

Lynas Rare Earths is the largest rare-earth producer outside China, with NdPr family production capacity around 10,500 tpa and processing at Kalgoorlie and Malaysia — the only commercial-scale rare-earth separation operator outside China (Lynas FY2024 Annual Report). Its structural value is being the named non-Chinese option for OEMs constrained by the October 2025 MOFCOM extraterritorial rules.

Lithium. SQM holds approximately 65% of Chilean lithium output, producing 201,000 mt of lithium carbonate equivalent from the Salar de Atacama in 2024 — the largest single lithium brine operation in the world (SQM investor relations and Benchmark Minerals coverage). The CORFO state contract governs volumes and pricing through 2030. Sovereign-gated pricing power sits with the Chilean state structure as much as with SQM itself — a distinction that matters when assessing the durability of the moat.

Cobalt. CMOC produced approximately 114,000 mt of cobalt in 2024 — roughly 31% of global market share — across TFM and KFM in the Democratic Republic of Congo, which supplied over 70% of global mined cobalt output in the same year (Cobalt Institute Market Report 2024). CMOC's first-mover position in the DRC's new export-quota system consolidates the access advantage that Chinese state-linked operators hold in concession negotiations.

Copper. Freeport-McMoRan sold approximately 4.1–4.2 billion pounds of copper in 2024 across Grasberg, Cerro Verde, and Morenci (Freeport-McMoRan 10-K 2024). Grasberg's underground operation is one of the world's highest-grade remaining copper-gold deposits. The new PT-FI smelter in Eastern Java, mandated by Indonesian in-country processing rules, tightens concentrate supply to third-party smelters globally — sovereign requirement reshaping the global concentrate market as a structural by-product.

Uranium. Kazatomprom produced approximately 21% of primary global uranium output in 2024; sovereign-state ownership gives Kazakhstan direct influence over term-contract pricing, which has moved from sub-$30/lb in 2020 to $90+/lb in the current cycle (Kazatomprom Integrated Annual Report 2024). Combined Kazatomprom and Cameco share of primary supply exceeds 40%.

Second-Order Chokepoints — The Separations-and-Conversion Layer

Below the named first-order holders sits the separations and conversion chemistry where the bottleneck binds tightest.

Neo Performance Materials operates the Silmet plant in Estonia — the only commercial rare-earth separation operation in the European Union — and opened Europe's first mass-production NdFeB magnet facility at Narva in September 2025, with initial capacity of 2,000 mt/yr targeting 5,000 mt/yr (Neo Performance Materials — Estonia facility overview). Neo also commissioned a heavy-rare-earth separation line to handle dysprosium and terbium (Neo HRE separation commissioning). Controlling the full EU value chain from separation to finished automotive-grade magnet, Neo is positioned to satisfy up to 15% of EU magnet demand under the October 2025 MOFCOM rules — and is, at present, the only entity outside China and Japan/Korea that makes that claim credible rather than aspirational.

The Chinese magnet complex — JL Mag Rare-Earth alongside named non-Chinese producers Proterial and TDK (both Japan) — manufactures 94% of global NdFeB permanent magnets. The October 2025 MOFCOM controls extend Chinese export jurisdiction to foreign-produced magnets containing heavy rare earths above 0.1%, and the December 2023 technology export ban blocks new entrants from replicating separation chemistry without years of process-engineering build (CSIS). The Proterial and TDK positions outside China are real but remain within the licensing perimeter the October 2025 controls define.

Albemarle runs the lithium conversion and separations chemistry that feeds US-sovereign battery supply chains. The Kings Mountain mine restart in North Carolina is backed by approximately $240 million in federal grants, targeting full-scale operations late 2026 (Albemarle Kings Mountain factsheet). The IRA's "Buy American" provisions gate EV battery tax credits through domestic lithium conversion; Albemarle is one of the very few credible US-sovereign counterparties at scale.

Copper smelter and concentrator infrastructure sits between upstream mines and downstream end-users. The Freeport PT-FI smelter and Codelco's Chilean assets are the named instances; Metso and FLSmidth are the equipment vendors who gate how fast new processing capacity can be built globally (Freeport-McMoRan operations — Indonesia PT-FI). Each new greenfield smelter takes seven to ten years to permit and commission, and as ore grades decline, refining capacity tightens regardless of what the mine output curve does.

Integration with the Broader Bottleneck Loop

The chip stack cannot scale without the grid; the grid cannot scale without the materials; the materials cannot scale without sovereign export licensing on the upstream chemistry. A constraint at any one layer transmits to the others.

The energy-intensity consequence of copper ore-grade decline connects the materials sector directly to the energy bottleneck. Processing lower-grade ore drives approximately 13% higher energy intensity per tonne of copper produced — which bids up grid capacity at precisely the moment AI capex is bidding up that same grid. PJM capacity has cleared at the FERC ceiling of $329.17/MW-day for two consecutive years (PJM Reliability Pricing Model) — the textbook empirical signature of a binding physical constraint that capital has not yet fully repriced. The rare-earth-to-magnet chain connects the materials sector to the compute layer downstream: NdFeB permanent magnets are required for EV drivetrains, wind-turbine generators, and defence applications — all of which sit downstream of the processing chokepoint China controls and the October 2025 MOFCOM extraterritorial rules extend.

The Opportunity Signal

The Opportunity Framework identifies the allocation signal from this bottleneck structure. Stanley Druckenmiller's Q4 2025 Duquesne Family Office 13F is the strongest single public expression of where the diagnostic arithmetic points an allocator with a four-decade record: a portfolio rotated systematically out of the post-1980 default and into the configuration the regime change rewards (Duquesne Family Office LLC — 13F Filings). The Alcoa position at approximately $73 million rests explicitly on the aluminium-and-AI-capex energy-intensity thesis: the energy-and-materials layer of the AI capex cycle earns the revenue the platform layer is failing to translate into earnings. If inference cost commoditises — the DeepSeek-class efficiency curve is the empirical evidence that it is doing exactly that — value accrues to the physical infrastructure layer rather than to the platform layer.

Marc Faber's commodity super-cycle framework provides the temporal context. Commodity cycles run 45–60 years trough to trough, with a peak in 1980, weak prices through the 1980s and 1990s, and a pickup starting approximately 1999–2000 that peaked in 2008 (Marc Faber, Tomorrow's Gold, CLSA Books, 2002; MarcoPolis interview, August 2015). On that framework, the current upswing started approximately 1999–2000; six to eight years into a cycle with twenty to thirty-plus years remaining, EM commodity producers are not a mean-reversion trade but a secular rotation. The critical-minerals sovereign-gated structure is the materials-layer analogue to the PJM pricing signal: the constraint is binding, and capital has not yet fully repriced the bottleneck holders.

What Would Change My Mind

Three named thresholds with named mechanisms and named primary sources.

The first is non-Chinese NdPr separation capacity exceeding 30% of global supply, versus the approximately 9% that sits outside China today per the IEA Critical Minerals Outlook (IEA Critical Minerals Outlook 2024). At that threshold, China's export-control leverage becomes partial rather than near-total — competing buyers have credible alternative feedstock sources, breaking the processing chokepoint that makes export restrictions effective. The three projects whose collective ramp would close on that threshold are MP Materials' Independence facility, Lynas's Kalgoorlie processing buildout, and Neo Performance's Estonia/Narva integration. Until their aggregate capacity approaches 30%, the chokepoint holds.

The second is the substitution pathway maturing: LFP cathode displacing NMC above 50% of new global EV production (cobalt and high-grade nickel processing gate weakens); sodium-ion battery chemistry reaching commercial-scale EV and grid storage (lithium-conversion gate weakens); rare-earth-free permanent-magnet motor architectures reaching mainstream EV drivetrain adoption (rare-earth-for-EV claim weakens, though defence and wind-turbine demand persist regardless). Each substitution has named primary sources at the IEA Global EV Outlook and battery-technology trade publications, and each has a different distance to the threshold.

The third is DRC sovereign action on CMOC's TFM/KFM operating licences. The DRC government holds the precedent — Tenke Fungurume was effectively nationalised in 2022 when Kinshasa disputed Freeport's original sale to CMOC. A repeat would remove 31% of global cobalt supply simultaneously. The DRC's new export-quota system, launched late 2024, suggests Kinshasa is currently extracting rents rather than nationalising — which would update the read in the opposite direction, consolidating rather than threatening CMOC's position. Either outcome is falsifiable in advance against observable DRC policy actions. That is the discipline: describe the regime, name the triggers, refuse to call the dates.