The rare-earth elements are seventeen chemically similar metals whose oxides and alloys underpin high-energy permanent magnets, and whose commercial chokepoint is not mining but separation and magnet-making, where China holds roughly 90% and 94% of world capacity respectively ref.
Summary
Rare-earth elements (REEs) are the lanthanide series plus scandium and yttrium. Despite the name they are not geologically scarce; the difficulty is that they occur intermixed in the same minerals and, because their chemistry is nearly identical, must be separated by long cascades of solvent-extraction stages before they are useful. The commercially decisive uses are the light REEs neodymium and praseodymium in NdFeB permanent magnets, and the heavy REEs dysprosium and terbium, which are added to those magnets to retain coercivity at motor operating temperatures ref. Those magnets are the torque-dense core of EV traction motors, direct-drive wind generators, robotic and defence actuators, and hard drives.
The supply structure is the whole story. In 2025 China mined about 270,000 t REO-equivalent against the United States’ ~51,000 t (Mountain Pass plus a Georgia operation), Australia’s ~29,000 t and Myanmar’s ~22,000 t. That is roughly 60% of mining, but China processes and separates about 90% of global REEs and makes about 94% of NdFeB magnets ref. A separate 2026 assessment frames the same asymmetry as G7 countries being able to mine but losing 85% of the chain before magnets, with non-China magnet capacity at about 6% ref. Heavy REEs are the most concentrated and highest-leverage segment of all.
The parameters that decide outcomes are therefore downstream, not geological: separation-cascade capacity and the metallisation and alloying steps between oxide and magnet; how far dysprosium loading can be cut by grain-boundary diffusion (cited at 30-50% reduction) and how fast magnet-free motor architectures scale ref; and the export-control regime, which since April 2025 has covered seven rare earths under a dual-use licensing system ref and was consolidated in March 2026 under State Council Order No. 834, China’s first dedicated supply-chain security framework ref.
Viability (5/5)
As a materials class this is fully industrialised. 2025 mine production of roughly 270,000 t REO-equivalent in China alone, with ~90% of separation and ~94% of NdFeB magnet output, is a demonstration of operating capacity rather than a promise ref. There is no technical uncertainty about whether rare-earth magnets work.
The live viability question is narrower: whether the same chain can be operated outside China at commercial scale and cost. Here the sources show capability, not yet capacity. G7 nations can mine, but about 85% of the chain vanishes before magnets, leaving non-China magnet capacity near 6% ref. The Perth assessment cites Western additions on the order of MP Materials’ +10 kt by 2028 as insufficient, in time, to offset re-imposed structural controls ref. The sources give no cost or yield data for Western separation plants, so their unit economics cannot be assessed here.
TLDR: Not a question: REE separation and magnet manufacture run at hundreds of thousands of tonnes a year, overwhelmingly in China.
Drivers (4/5)
Demand: the sources tie heavy REEs directly to high-temperature coercivity in NdFeB magnets for EV and wind motors and generators, robotics and defence actuators, plus semiconductor and photonics materials ref. The supplied evidence does not quantify tonnage growth by end market, so the demand curve’s slope is asserted rather than measured here.
Supply: the driver is political as much as geological. April 2025 restrictions on seven rare earths and a dual-use licensing regime continued to constrain exports to US and allied military end-users through 2026 ref; China restricted exports to the United States and Japan in March 2026 ref; and Order No. 834 of 31 March 2026 folded export controls, countermeasures and investment screening into one national-security framework ref. On the response side, Washington has been willing to redirect semiconductor money into minerals: a reported $2bn CHIPS Act reallocation toward critical minerals in August 2025 ref ref, and a December 2025 CHIPS incentive award to Korea Zinc subsidiary Crucible Metals for a US smelter and critical-minerals processing facility ref. Note also the Perth report’s warning that a falling dysprosium price is not evidence of loosening supply ref.
TLDR: Demand pull from motors, robotics and defence meets a deliberately constrained, policy-shaped supply side.
Novelty (2/5)
On the substitution side, the only quantified figure is a 30-50% dysprosium reduction from grain-boundary diffusion, which the source argues, together with magnet-free motor scaling, still cannot offset re-imposed controls ref. Rare-earth-free tetrataenite magnets appear as a 2022 manufacturing advance framed as reducing reliance, with no performance or scale figures given ref. So the sources support the judgement that no alternative displaces NdFeB on the evidence presented, but they do not let us quantify how far behind the alternatives are.
TLDR: The material class is the century-old incumbent; what is new in these sources is measurement and substitution research, both early.
Diffusion (4/5)
Adoption of rare-earth magnets is not the barrier. The barrier is geographic replication of the process chain, and the sources are consistent that this is the hard part: mining concentration (~60%) is far lower than separation (~90%) and magnet manufacture (~94%), so the bottleneck is refining and magnet-making, not ore ref. Non-China magnet capacity of roughly 6% quantifies how little of the chain has diffused ref.
The obstacles visible in the sources are time and permitting-scale capital rather than customer acceptance: announced Western additions are described as too small and too late relative to control re-imposition ref, and public capital is being pulled from other programmes to fill the gap ref ref. Environmental permitting, separation know-how and metallisation capability are all plausible constraints but are not documented in the supplied sources, and the 2025 Latitude Media item on the “uphill battle” against Chinese dominance carries no extractable detail ref.
TLDR: The materials are already everywhere; what has not diffused is the ability to make them outside China.
Impact (5/5)
The impact case does not depend on forecasts. The sources state that whoever controls separation and magnet-making controls the input to EV and wind motors and generators, robotics and defence actuators, and a range of semiconductor and photonics materials, and identify the ~90% separation / ~94% magnet share as the structural fact under both semiconductor sovereignty and energy-transition exposure ref. Heavy REEs are singled out as the most concentrated and hardest to source ex-China.
The realised impact has already been demonstrated through policy: repeated 2025-2026 export restrictions aimed at US, Japanese and allied military end-users ref ref, and a unified national-security supply-chain framework in March 2026 ref. What the sources do not provide is a monetary quantification of disruption, so the magnitude is argued structurally rather than in dollars.
TLDR: A single-country chokepoint sitting under electrification, robotics, defence and parts of semiconductors.
Timing Now (0-2yr)
Controls are already in force and being tightened rather than anticipated: seven rare earths restricted since April 2025 with dual-use licensing continuing through 2026 ref, targeted restrictions on the United States and Japan in March 2026 ref, and Order No. 834 on 31 March 2026 ref.
The near-term hinge is the temporary suspension of dysprosium and terbium controls, which expires in November 2026; the Perth assessment argues that neither grain-boundary diffusion nor magnet-free motors can offset re-imposition on that timescale, and that Western capacity such as MP Materials’ +10 kt lands only by 2028 ref. That gap between the 2026 policy date and the 2028 capacity date is the whole timing story.
TLDR: The constraint is active today, with a defined decision point at the November 2026 expiry of the dysprosium and terbium control suspension.
Overrated or underrated? Fairly rated
Rare earths are, if anything, the best-understood chokepoint in advanced materials, and the headline risk is priced into policy on both sides. Where attention is still misallocated is within the chain: the sources are unambiguous that mining concentration (~60%) is the least binding link and that separation (~90%) and magnet manufacture (~94%) are where the leverage sits ref ref. Investment framed around new mines, or comfort taken from a falling dysprosium price, is reading the wrong variable ref.
On substitution, the supplied evidence supports scepticism rather than dismissal. A 30-50% dysprosium reduction via grain-boundary diffusion is real engineering progress but is described as insufficient against structural controls ref, and rare-earth-free magnet routes appear only as a 2022 manufacturing advance with no scale data ref. Sensing improvements such as dual-comb laser-ablation spectroscopy attack exploration and ore sorting, which is upstream of the actual bottleneck; useful, but not the fix.
Prediction
By 31 December 2027, non-China NdFeB permanent-magnet capacity will still be below 15% of global capacity, against roughly 6% reported in mid-2026 ref.
Evidence base
- 2025 mine production: China ~270,000 t REO-equivalent, United States ~51,000 t, Australia ~29,000 t, Myanmar ~22,000 t; China mines ~60% but separates ~90% and makes ~94% of NdFeB magnets ref.
- Non-China permanent-magnet capacity is about 6%, with roughly 85% of the G7 chain absent between mine and magnet, reported 30 June 2026 ref.
- April 2025 restrictions on seven rare earths plus dual-use licensing continued constraining exports to US and allied military end-users through 2026 ref.
- State Council Order No. 834 of 31 March 2026 created China’s first dedicated supply-chain security framework, unifying export controls, countermeasures and investment screening ref.
- China restricted rare-earth exports to the United States and Japan in March 2026 ref.
- The suspension of China’s dysprosium and terbium controls expires November 2026; grain-boundary diffusion (30-50% Dy reduction) and magnet-free motors are judged unable to offset re-imposition in time, with MP Materials’ +10 kt arriving by 2028 ref.
- Laser-ablation dual-comb spectroscopy demonstrated REE atomic, ionic and oxide absorption signatures in certified reference materials across three spectral windows, targeting standoff exploration and ore sorting, 6 May 2026.
Open questions
- Does China re-impose or extend dysprosium and terbium controls when the temporary suspension expires in November 2026, and at what licensing stringency?
- Can Western projects close the metallisation and alloying steps between separated oxide and finished magnet, which is where the sources say 85% of the G7 chain disappears?
- How far can dysprosium loading actually fall in production magnets beyond the cited 30-50% grain-boundary-diffusion reduction, and at what cost or performance penalty?
- Do rare-earth-free magnet routes such as tetrataenite reach any documented production volume or motor-grade performance figure, rather than laboratory manufacturing demonstrations?
Assessment drafted 2026-08-31 from up to 16 KB sources using the technology-scorecard framework; scores are a draft read pending review.