The 17 rare-earth elements (the 15 lanthanides plus scandium and yttrium) are, despite the name, not geologically rare — but economically concentrated, hard to separate, and structurally controlled by China. They are the substrate for NdFeB permanent magnets and a long tail of semiconductor, photonic, and optical materials. The investable position built on this concept is Rare Earth Supply Chain; the geopolitical roof is Sovereignty Semi Bifurcation.
The elements that matter
- Light REE (LREE): lanthanum, cerium, praseodymium, neodymium, samarium. Cheap-ish, China-dominated but more substitutable.
- Heavy REE (HREE): dysprosium, terbium, and the rest — scarcer, far more China-concentrated, and the highest-leverage: Dy/Tb give NdFeB its high-temperature coercivity (stability above ~150 °C), so EV-traction and generator magnets cannot drop them without performance loss.
- The magnet four-plus-two: Nd + Pr (the bulk of an NdFeB magnet) + Dy + Tb (high-temp dopants); Sm for SmCo magnets.
Why it matters for energy
NdFeB permanent magnets are the torque-density core of EV traction motors and direct-drive wind-turbine generators, plus heat-pump compressors. The energy transition is, materially, a magnet build-out — and the magnets need exactly the Nd/Pr/Dy/Tb that China controls. China’s April-2025 export licensing on Dy/Tb-bearing magnets (see 2026 06 13 Iea Critical Minerals Export Controls 2025) directly forced Western/EU carmakers to cut utilisation or idle lines in 2025.
So a rare-earth supply shock hits the semiconductor/photonics stack from several directions at once, not just via motors — which is why it sits beside Eu Semiconductor Sovereignty and Sovereignty Semi Bifurcation.
Supply concentration (the structural fact)
Per USGS (2026 06 13 Usgs Mcs Rare Earths Supply Concentration): China mines ~60% of global REE but separates ~90% and makes ~94% of NdFeB magnets — the bottleneck is downstream (separation + magnet-making), not the ore. The US (Mountain Pass), Australia (Lynas), and Myanmar supply ore; almost everyone still sends it to China to separate. This is the same “controls the chokepoint, not the commodity” shape as the semiconductor toll-booths, one tier up the periodic table.
The fundable wedges (vs the un-fundable layers)
Mining and magnet high-volume-manufacturing are capital-heavy and increasingly state-funded (USA Rare Earth $1.6bn; Vulcan Elements $1.3bn, 2025 11 07 Observations 004). The deep-tech, venture-scale layers are:
- Recovery / recycling of REE from e-waste, swarf, and end-of-life magnets — EU recycling is <1% today (Rare Labs, Cyclic, Noveon/Hypromag, Momentum).
- Novel separation chemistry (replacing solvent extraction) — the actual China chokepoint (Reetec, Less Common Metals).
- Rare-earth-free / reduced magnets — iron-nitride / tetrataenite class (Niron-class); the long-shot that removes the dependency.
Full investment treatment, cohort, and predictions: Rare Earth Supply Chain.
Frontier (open questions)
See frontmatter frontier:.