Chokepoints in motion: the mechanism predicts the trend

Key findings

• Of the six key energy minerals the IEA projects to 2035, only two refining chokepoints are easing — rare earths and graphite — and both are built-capability moats being deliberately rebuilt outside the dominant supplier (MP Materials, Lynas; African graphite and new anode plants).
• The three entrenching — nickel, copper and cobalt — are held by policy (Indonesia’s ore-export rules), thermodynamics (China’s continuous-hot smelter build-out) and by-product coupling. None can be quickly out-built, and two are getting worse.
• The direction of change is set by the mechanism: capability erodes (slowly, with capital and years); policy reverses fast in either direction because a rule made it; by-product and thermodynamic chokepoints are pinned by physics and stay put.
• This makes the taxonomy falsifiable and predictive: from the mechanism alone it forecasts which of the IEA’s six chains ease and which entrench — and the IEA’s asset-by-asset projections agree.

1 · Concentration is not a fixed fact

A snapshot market share — “China separates 84% of magnet rare earths” — is alarming but static. It does not say whether that share is rising, falling, or stuck, and therefore does not say whether to wait, stockpile, or build. Yet the direction is knowable in advance, because it follows from the same property that set the concentration in the first place: the mechanism of the binding stage. A chokepoint held by an accumulated capability can be rebuilt, just slowly. One held by a rule can vanish or double overnight. One pinned by thermodynamics or by a by-product coupling barely moves, because no policy repeals physics.

2 · The test: six minerals, mechanism vs 2035 trajectory

The IEA’s Global Critical Minerals Outlook 2026 projects supply for six key energy minerals to 2035 from an asset-by-asset review of announced projects — an independent, forward-looking dataset the atlas does not itself produce. We compute each mineral’s single-country top-refiner share in 2025 and 2035 from those projected tonnages, and compare the direction to the mechanism the atlas assigned that chokepoint before seeing the projection.

The alignment is the result: every capability chokepoint eases, and the two that move up are thermodynamic and policy — the mechanisms the taxonomy says can concentrate further. Rare earths, the one chain the IEA independently singles out as declining, is the clearest capability case: its fall is driven by exactly the deliberate capacity-building (US separation at MP Materials, Malaysian separation at Lynas) that the taxonomy says a capability moat invites.

3 · The rule

4 · Why it matters

Diversification is expensive and slow, so it must be aimed. Sorting chokepoints by trajectory-and-mechanism tells a buyer or a policymaker where time is on their side (capability chokepoints that are already eroding — support them, don’t overspend), where a rule is the lever (policy chokepoints — trade and diplomacy), and where nothing self-corrects (by-product, thermodynamic, geological — build capacity or stockpile). The 2035 numbers will move; the mechanism logic that predicts which way they move will not. This is the taxonomy earning its keep as a forecasting tool, tested against a second, independent source.

Method

Top-refiner shares are single-country shares of refined output, computed from the IEA GCMO 2026 annex supply tables (pp. 347–364): 2025 = historical, 2035 = base-case projection. Rare earths = magnet-RE separation; lithium = chemical conversion; nickel = Indonesia’s refined share. Mechanism tags are assigned in the atlas’s Chokepoint Map (Report 04) from process physics, independently of the IEA data. “Easing / entrenching” thresholds at ±3 percentage points. The IEA’s forward projections rest on proprietary asset-level data (S&P Global, Wood Mackenzie) that the atlas cannot itself reproduce; this report uses them as an external test of the taxonomy’s predictions, not as atlas-generated figures. IEA data © IEA, CC BY 4.0. The interactive version, with the slope chart, is at trajectory.html.