The rare earth supply chain is a five-to-seven-stage industrial sequence: mining, beneficiation, chemical cracking, solvent-extraction separation, oxide-to-metal reduction, alloying, and sintered magnet or phosphor production, with the midstream separation and reduction steps controlled by a narrow set of Chinese processors [S3][S5].
Mine output is only the visible tip: the bottleneck sits in separation, where mixed rare earth concentrates are split into individual oxides such as Nd2O3, Dy2O3 and Tb4O7, the precursors that downstream magnet and phosphor plants require at 99.0–99.999% purity [S3][S1].
Stage 1 to 2: mining and beneficiation
Carbonatite deposits (for example Bayan Obo, Mountain Pass, Mount Weld) and southern Chinese ion-adsorption clays dominate the resource base, with carbonatite type deposits "currently the main sources and most likely will dominate future targets for exploration and mining" [S3]. Beneficiation uses froth flotation, magnetic separation and gravity concentration to upgrade run-of-mine ore into a rare earth mineral concentrate, typically grading 30–60% total rare earth oxide (TREO) before chemical processing [S3][S1].
Ion-adsorption clays in Jiangxi, Fujian, Guangdong and Yunnan are unusual in that leaching can be applied directly to the ore, bypassing the cracking kiln, which gives them a structural cost advantage for the heavy rare earths Dy, Tb, Y and Er that the carbonatite bodies carry only in trace amounts [S3].
Stage 3: chemical cracking and impurity removal
Concentrate is roasted or acid-baked to break down the mineral lattice, then leached with ammonium sulfate or hydrochloric acid to pull rare earths into solution while rejecting thorium, uranium, iron, calcium and phosphate gangue [S3]. A common process "separates rare earth ore from mineral concentrates by using basic or acidic solutions in order to selectively dissolve" the target oxides [S1].
The output of this stage is a mixed rare earth chloride or sulfate solution, with thorium and uranium removed to levels set by each jurisdiction's radioactive residue rules, the reason that downstream separation capacity clusters in jurisdictions with established radioactive waste handling infrastructure [S1][S3].
Stage 4: solvent extraction separation
Multi-stage mixer-settler or centrifugal contactor trains, often running 50–80 cells in series, use organic extractants such as D2EHPA, PC88A (P507) and Cyanex 272 to separate the mixed solution into individual high-purity oxides, the technical and capital bottleneck of the entire chain [S3]. Each element pair (Nd/Pr, Dy/Tb, Sm/Eu/Gd) requires its own circuit and its own set of stripping acids, so a commercial-scale separation plant typically hosts 8–15 parallel element-specific trains [S3].
This is the stage where single-firm Chinese processors such as China Northern Rare Earth, China Minmetals and Shenghe Resources hold the dominant global position, a fact that makes any non-Chinese "mine to magnet" project dependent on toll separation in China until its own separation train is commissioned and qualified [S5].
Stage 5: metal reduction, alloying and magnet sintering
Individual oxides are calcined, then reduced to metal via molten salt electrolysis for the light rare earths (La, Ce, Pr, Nd) or via metallothermic reduction (calcium or lithium reduction of fluoride salts) for the heavy rare earths (Dy, Tb, Y) that cannot be electrowon economically [S3]. The freshly reduced rare earth metals are then alloyed in vacuum induction furnaces, typically into NdFeB melt-spun ribbon, which is jet-milled, pressed in a magnetic field, sintered, and aged into finished NdFeB magnets with energy products commonly in the N35–N52 (38–52 MGOe) range and Dy or Tb grain-boundary diffusion applied where high-temperature demagnetization resistance is required (EV traction motors, wind turbine generators) [S3][S5].
For comparison, the four primary sintered NdFeB grades specified by buyers diverge sharply on cost, temperature rating and dysprosium content, and that trade-off is what magnet buyers actually negotiate, not the oxide spot price:
Standard N-series magnets (N35–N48) use no heavy rare earths, deliver 33–43 MGOe, and suit motors below roughly 80°C. M-series (35M–48M) add 1–3% Dy for 100°C operation. H-series add roughly 3–5% Dy and are rated to 120°C. SH/UH/EH series push 5–9% Dy or 1–3% Tb and hold coercivity up to 200°C, the grades that go into EV traction motors, wind turbine generators and defence actuators [S3][S5].
Who actually controls each stage
Resource ownership is comparatively fragmented: at least 146 advanced and mining projects covering 303.4 million tons of rare earth oxide resource have been catalogued globally, with Asia the largest source but "Greenland, Africa, and Europe show important resource potentials if environmental issues and processing technologies are well addressed" [S3]. The United States was once self-reliant in domestic REE mining but "has become nearly fully reliant on REE imports in the past 20 years", with the gap driven by lower Chinese mining and processing costs and larger reserves [S1].
Mid- and downstream processing remains the choke point. China's rare earth export controls, introduced and tightened through 2024–2026, "sustain magnet exports, restrict key elements and enable geopolitical leverage", which is why allied governments are now co-funding toll separation, alloy and sintered magnet lines inside the US, Japan, the UK and the EU [S5]. The strategic assets already operating outside China are concentrated in a small set of names (Less Common Metals in the UK, HyProMag and Ionic Technologies for recycling, plus Mountain Pass in the US, Lynas in Malaysia/Australia, and a handful of Japanese alloy and sintered-magnet lines) [S5].
Failure modes and structural limits of diversification
A "mine to magnet" project typically needs 7–12 years from feasibility study to first qualified sintered magnet shipment, because each step (separation, electrolysis, alloy, sinter, grain-boundary diffusion) has to be qualified independently by the magnet buyer, and most end-users will not dual-source an automotive-grade magnet without two years of PPAP and field-trial data [S3][S5].
Heavy rare earths (Dy, Tb, Y) are the binding constraint inside any diversification plan: they are geologically scarce outside southern China and Myanmar, they carry the highest processing radioactivity burden, and they are the elements actually targeted by the most recent Chinese export licensing regime, so a project that recovers only light rare earths from a carbonatite orebody does not, by itself, break allied import dependence for high-temperature magnets [S3][S5]. The IEA's April 2026 review concluded that "new projects, partnerships and policies are needed to address supply chain risks" and called out substitution, recycling and demand-side efficiency as the only near-term levers that scale faster than new mines and separation trains [S2].
Decision criteria for buyers and project sponsors
For a magnet or motor buyer evaluating supply resilience, the relevant questions are not "is this deposit big" but: does the supplier control separation of all elements the spec requires (especially Dy and Tb), can the alloy/sinter step be toll-processed outside China, and is grain-boundary diffusion available to keep the heavy rare earth content low without losing coercivity [S3][S5]? A pure oxide offtake from a non-Chinese miner is not, by itself, a hedge against magnet-grade supply disruption, because the oxide still has to be toll-separated in most existing project pipelines [S5].
For project sponsors, the gating capital items, in order, are: a qualified separation circuit with 8–15 element trains, radioactive residue management infrastructure, molten salt electrolysis cells for light rare earths or metallothermic reduction furnaces for heavy rare earths, vacuum induction alloy furnaces, jet mills, and field-press/sinter/ageing lines matched to automotive PPAP cadence. A US Department of Commerce 2016 assessment of the four heavy and lighting rare earths (Dy, Er, Nd, Tb, Yb) flagged exactly these midstream gaps as the reason US REE manufacturing shrank even while mine production restarted at Mountain Pass [S1]. For buyers of finished magnets, pairing a sintered NdFeB specification with a written Dy/Tb traceability chain and a written second-source separation plan has become the practical de facto hedge through 2026, given the pace of new Chinese export licensing [S5].
Trackable signals to watch through 2026–2027: commissioning of non-Chinese separation trains (Lynas Kalgoorlie, Less Common Metals expansion, US DoD-funded demonstration lines), the next iteration of Chinese rare earth export licensing rules, and the IEA's next Critical Minerals Outlook update [S2][S5]. The wider European filter, fastener and industrial-equipment buyer base, including the supply chains covered in the Industrial Fastener Suppliers and Manufacturers 2026 Shortlist, now sits on the same rare earth feedstock curve as EV and wind buyers, so the same hedging logic applies to any motor or generator specification a procurement team signs in 2026.
Spec-level background on the components involved: earth ground tester, power supply, and dc power supply.