Report 01 · August 2026

Breaking the chokepoint: from finding to decision

Critical Materials Atlas · independent public-data research · public data, trade year 2024 (2002–2024 panel for the emergence test)

Abstract. The dominant fact of critical-material supply — that a handful of countries, above all China, control the refining stage — is by now well known. It is also only half an answer. This report turns that fact into decisions across 32 materials: it classifies what kind of moat holds each chokepoint (because the moat type decides what can work), identifies who could realistically break it, tests whether the world’s remaining capacity could absorb a cut, and measures how exposed each importing country is. Two results stand out: the chokepoint is usually a capability, not an ore endowment, so export controls on ore cannot touch it; and for 8 of 31 materials, removing the leading exporter would make global concentration worse, not better — diversification, done naively, backfires.

Key findings

Capability, not endowment. For 10 of 32 materials the chokepoint is a processing capability — the leader imports the ore and exports the refined metal, or extracts the metal as a by-product of a domestic host. No embargo on ore reaches it. Gallium is the clean case: China refines ~99% and mines almost none.
Diversifying can backfire. Modelling a shock as the loss of the leader’s exports and reallocating optimally, 4 of 31 materials cannot be covered even if every other exporter doubles; and for 8 of 31, removing the leader hands the chokepoint to a more-concentrated runner-up (phosphorus → Kazakhstan, fluorspar → South Africa).
Exposure is uneven and personal. The East-Asian manufacturing belt is the most locked-in — South Korea is ≥50%-China-sourced on 12 of 31 materials — while the US and Germany are far more diversified.
Alternatives exist, but the buildout lags. For most chokepoints there are real, publicly-announced diversification projects (Perpetua, Almonty, Lynas, Euro Manganese…), but they are early-stage and thin relative to the concentration they would need to offset.

1 · What kind of moat

A chokepoint is not a chokepoint is not a chokepoint. If a leader imports the ore and exports the refined metal (Japan’s titanium sponge, Indonesia’s nickel, China’s tungsten), the moat is the furnace — tariffs or export controls on ore are useless; only new refining capability elsewhere breaks it. If the leader extracts the metal as a by-product of a domestic host it already controls (China’s gallium from its own alumina, germanium from zinc), there is no ore stream to redirect at all. Only where a leader genuinely mines and refines its own resource does ore access become a real lever. The same ~99% share can mean three completely different things to do.

China refines what it doesn't mine — capability, not endowment — Figure 1. The chokepoint is a capability, not an ore endowment — which is why export controls on ore cannot touch it.

2 · Can the rest of the world cover a cut?

The usual question — “is there another exporter?” — is the wrong one, because a fallback exporter is already serving its own customers. We model a shock as the loss of the leader’s exports (the only supply that was ever redirectable) and reallocate its freed demand across every surviving exporter by optimal transport, on the real bilateral trade matrix. The results are sobering and, in places, counterintuitive.

Diversifying away from the leader can make concentration worse — Figure 2. For 8 of 31 materials, removing the leading exporter raises concentration — the runner-up is the real problem. Phosphorus: HHI 0.34 → 0.44.

Reallocation map: if China's magnet exports stop, supply reroutes through Japan, Korea and SE Asia — Figure 3. Where supply would have to come from if a chokepoint were cut — the min-distance reallocation for rare-earth magnets, with China removed.

3 · Who is exposed

The global finding becomes actionable only when it is made personal: for each importing country, how much of its supply runs through a single dominant source, material by material. The exposure is strikingly uneven — concentrated in the economies that manufacture the most.

Leverage map: South Korea is China-sourced on 12 of 31 critical materials — Figure 4. The leverage map. South Korea is ≥50%-China-sourced on 12 of 31 materials; the US and Germany are diversified.

4 · A note on what we do not claim

“Who could break it” is estimated from product-space capability-adjacency where a clean signal exists, and from current alternative-supplier shares otherwise — always labelled. We separately tested whether product-space proximity actually forecasts which countries develop downstream capability, and found that the apparent signal is largely a diversity artifact (already-diversified economies diversify further); that honest control is reported in the product-space work rather than overstated here. The reallocation is a statement about today’s capacity, not a forecast.

Method

Bilateral trade is reconciled from UN Comtrade via CEPII BACI (validated against the official series: top-1 exporter 25/30, 3.5% share error). Moat type is read from the trade feedstock signature (imports ore / exports refined) plus BGS/USGS physical output; refining concentration from refined-export HHI, cross-checked against the IEA Critical Minerals Dataset, EU CRM 2023, and the USGS World Minerals Outlook. The reallocation is an entropic optimal-transport (Sinkhorn) solve on the bilateral matrix; coverage assumes surviving exporters can scale to a stated ceiling. Full methods and all data are open. Figures are approximate and rounded; read each material as an overlay of distinct measures, not one observed pipeline.

How to cite. Critical Materials Atlas (2026). Breaking the chokepoint: from finding to decision. Report 01. Zenodo. https://doi.org/10.5281/zenodo.21948855

@techreport{cma_report01_2026,
  title       = {Breaking the chokepoint: from finding to decision},
  author      = {{Critical Materials Atlas}},
  institution = {Critical Materials Atlas},
  number      = {Report 01},
  year        = {2026},
  publisher   = {Zenodo},
  doi         = {10.5281/zenodo.21948855},
  url         = {https://criticalmaterialsatlas.org/report-decision-layer.html}
}

Archived on Zenodo — concept DOI 10.5281/zenodo.21948855 (always resolves to the latest version).