Rare Earths — A Complete Market Guide (2026)

Data as of 9 August 2026. Prices are quoted as full-year averages of assessed benchmark prices, not a single day’s snapshot, so this report stays useful over time. Reserves, production splits and historical series are estimates from agency data, rounded for clarity. This report is for information only and was prepared with AI assistance — see the disclaimer at the end.

Rare earths are the seventeen metals that make modern technology small, light and efficient — above all the neodymium magnets inside every EV motor, wind turbine, smartphone and guided missile. No commodity carries more geopolitics per kilogram: the rare earth market is tiny in dollar terms, yet China’s grip on it — roughly 70% of mining and 90% of refining and magnet-making — has made these obscure elements the sharpest pressure point in the US-China trade conflict and the object of billion-dollar Western industrial policy. This report is the free, big-picture primer on how the market actually works — what the seventeen elements do, how the price is set without an exchange, who mines and separates them, and what actually moves the sector. For the company-level data behind the charts — every producer screened by production, resources and role in the chain — go to Metal Pilot .

TL;DR & Key Takeaways

• What they are: 17 chemically similar metals (the 15 lanthanides plus scandium and yttrium) that are not geologically rare — but are rarely found in economic concentrations and are fiendishly hard to separate from each other. The value is concentrated in a handful of magnet elements: neodymium and praseodymium (NdPr), plus the heavy rare earths dysprosium and terbium.
• Market structure: the most concentrated supply chain in commodities. China mines ~69% of the world’s rare earths and controls ~90% of separation, refining and magnet production; supply inside China is run by state quotas allocated to just two state groups, and exports now require licenses under the controls introduced in 2025.
• Demand story: permanent magnets are ~48% of demand by volume and over 80% by value, pulled by EVs, wind turbines, electronics and defense. The rest of the basket — cerium, lanthanum for catalysts, polishing, glass — is a low-value by-product of producing the magnet elements.
• Price regime: no exchange, no futures — prices are assessed (Argus and peers), opaque and violently cyclical. The 2010–11 China embargo mania (prices up more than tenfold, then −90%) and the 2021–22 EV boom-bust are the pattern; NdPr oxide averaged ~$69/kg in 2025, well below the $110/kg floor the US government now guarantees its domestic champion.
• Biggest swing factor: Chinese policy — quotas, export controls and stockpiling — against the pace of Western supply-chain building (the MP Materials-Pentagon model of price floors and magnet plants).

Numbers to remember (rare earths at a glance)

Figure 1. Rare earths at a glance

Figure data: USGS Mineral Commodity Summaries 2026 , IEA and MP Materials ; see Sections 1.3–2.7.

Why it matters now: rare earths sit where the energy transition, the AI-and-electronics economy and great-power rivalry intersect. Export controls imposed by China in 2025 turned a sleepy specialty market into front-page leverage, and the West’s response — Pentagon equity stakes, guaranteed prices, new refineries — is creating a second, politically priced market alongside the Chinese one. For investors, the sector is small, volatile and policy-driven — and for the first time since 2011, structurally interesting. The big-picture case is below.

How to read this guide: new to rare earths? Read straight through from Section 1. Here for the investment case? Jump to Section 4 — how to own the theme, how to screen the producers, and what actually drives returns. Every technical term is defined in the Vocabulary (Section 7).

1. Rare earths & the market basics

1.1 What rare earths are — physical basics & quality

The rare earth elements (REEs) are the 15 lanthanides — lanthanum through lutetium — plus scandium and yttrium, which occur in the same ores and behave chemically alike. That chemical similarity is the industry’s defining curse: rare earths are always found together, in every deposit, and separating them into individual high-purity oxides requires hundreds of solvent-extraction stages, tuned per deposit — the hardest step in the chain and the one China spent four decades mastering. Despite the name, they are not rare: cerium is more abundant in the crust than copper. What is rare is a deposit concentrated enough to mine and a plant capable of splitting it.

The industry divides the basket in two, defined here once and used throughout:

• Light rare earths (LREEs) — lanthanum, cerium, praseodymium, neodymium, samarium. The workhorses: neodymium and praseodymium (NdPr) are the heart of the NdFeB permanent magnet, the strongest magnet known, while cerium and lanthanum go into catalysts, polishing and glass at a few dollars per kilogram. Light-rich ores — bastnaesite (Bayan Obo, Mountain Pass) and monazite (mineral sands) — supply most tonnage.
• Heavy rare earths (HREEs) — europium through lutetium, plus yttrium. Scarcer, harder to source, and strategically hotter: dysprosium and terbium are added to NdFeB magnets so they keep working at high temperature — no EV traction motor or fighter jet works without them. Heavies come overwhelmingly from ionic adsorption clays in southern China and Kachin State, Myanmar, where the metal is literally washed out of weathered soil with ammonium-salt solutions.

The demand side has a matching hierarchy. A NdFeB magnet is ~30% rare earth by weight, and magnets are what the market is really about: they convert electricity to motion (and back) more efficiently and compactly than anything else, which puts them in EV motors alongside copper windings, in direct-drive wind turbines, hard drives, speakers, robotics and precision-guided munitions. Everything else — fluid-cracking catalysts for refineries, autocatalysts, glass polishing, phosphors, nickel-metal-hydride battery alloys — consumes the cheap majority of the basket at a fraction of the value. This mismatch creates the sector’s basket problem: every tonne of NdPr produced drags along several tonnes of near-worthless cerium and lanthanum, so a project’s economics live or die on its magnet-element share, not its headline tonnage.

The value chain — from ore to magnet. Every route converges on the same middle: mine → concentrate → “crack” (acid/alkali attack) → separate (solvent extraction into individual oxides) → reduce to metal → alloy and manufacture magnets. Ionic clays skip the hard-rock steps (the clay is leached directly) but still need separation. The chain’s chokepoints sit mid-stream and downstream: China performs ~90% of separation and refining and ~90% of magnet manufacture; even ore mined in California or Australia has, until recently, needed a Chinese (or Chinese-adjacent) plant to become a usable product. One more quirk matters upstream: monazite carries thorium, a mildly radioactive by-product whose handling and disposal regulations shape where processing can happen at all.

Figure 2. The rare earth value chain — ore to magnet

Source: industry value-chain primers; conceptual diagram.

1.2 Units & measurement conventions

This report uses the conventions of the USGS and the trade press, stated here so every number is unambiguous. Quantities are metric tonnes (t) and thousand tonnes (kt) of REO — rare-earth-oxide equivalent — the standard way to add up different products (ores, carbonates, oxides, metals) on one basis. Mine production counts REO contained in concentrates; a deposit’s quality is its TREO grade (total rare earth oxides, % of rock) and — more important — its basket composition (the % of TREO that is NdPr, Dy, Tb). Prices are quoted in US dollars per kilogram (USD/kg) of individual separated oxide — there is no meaningful “rare earth price,” only element prices, and they differ by three orders of magnitude: cerium oxide ~$1.7/kg, NdPr oxide ~$69/kg, terbium oxide several hundred dollars per kilogram (all 2025 averages). Oxide and metal are related but distinct products: reducing oxide to metal adds cost, and ~1 kg of metal requires ~1.2 kg of oxide.

Flow vs. stock, flagged throughout: production and demand are flows per year; reserves (>85 Mt REO globally) and government stockpiles (China’s strategic reserves, the US Defense Logistics Agency inventory) are levels — and unlike exchange-traded metals, there are no public exchange inventories to watch, one of the market’s defining opacities.

Table 1. Rare earth units and conventions

Source: USGS Rare Earths Statistics and Information , 2026; price magnitudes from USGS MCS 2026 (Argus assessments), 2025 averages.

Numbers intuition: an EV traction motor uses roughly 1–2 kg of NdPr (in ~2–4 kg of magnets); a direct-drive wind turbine ~600 kg of NdFeB per MW; the entire world market for separated oxides is only on the order of $10–15 billion a year — smaller than a single large copper mine’s revenue — yet more than 80% of that value sits in the magnet elements. Small market, outsized leverage.

1.3 Pricing & benchmarks — no exchange

There is no LME contract, no futures curve and no closing bell in rare earths. Prices are assessed by price-reporting agencies — Argus (the USGS’s source), Fastmarkets, Asian Metal, SMM — which survey actual transactions, mostly inside China, and publish reference prices per element: oxide and metal, min 99%–99.99% purity, FOB China or CIF. The benchmark that matters most is NdPr oxide (or neodymium oxide, its close twin), because it prices the magnet chain; dysprosium and terbium oxide are the heavy benchmarks; cerium and lanthanum barely move and barely matter. Two structural quirks follow from assessed pricing: the market is opaque (volumes and inventories are unpublished; Chinese state buying is invisible), and since 2025 it has been bifurcating — Chinese domestic prices, ex-China prices for licensed material, and now a US government-guaranteed floor of $110/kg for MP Materials’ NdPr, nearly double the 2025 market average. A parallel, politically priced Western market is being built on top of the Chinese one, and the spread between them is the sector’s newest signal.

Table 2. Key rare earth price benchmarks

Source: USGS MCS 2026 (Argus Non-Ferrous Markets); MP Materials-DoD agreement , July 2025.

The long price cycle is the wildest in this blog’s series, and it is a policy cycle, not a macro cycle. Through the 2000s neodymium oxide traded in the $10–30/kg range as Chinese supply swamped the world (Nd oxide was ~$27/kg in 2008). Then came the mania: China cut export quotas in 2010 amid a diplomatic clash with Japan, and by mid-2011 rare earth prices had risen 10–26-fold, with Nd oxide briefly above $300/kg — before demand destruction, substitution and new supply collapsed the market just as fast (Nd oxide entered 2012 at $195/kg and ended it at $80). A long deflation followed to ~$40/kg by 2015–16, then stability, then the 2021–22 EV boom (annual average $134/kg in 2022) and a bust back to $56 in 2024 as Chinese quotas expanded — before the export-control era began lifting prices again in 2025 ($73 average). The reliable public annual series starts only in 2012; the table below carries it, and the landmarks above carry the earlier cycle.

Table 3. Neodymium oxide price, 2012–2025 (USD/kg, Argus/USGS)

Source: USGS Mineral Commodity Summaries , Argus assessments — 2017 edition (2012–2016), 2021 and 2026 editions (2016–2025). *2012–2015 are year-end assessments (midpoints of published ranges); 2016 onward are annual averages. NdPr oxide, the blended benchmark, runs a few percent below Nd oxide (2025: $69 vs $73). Pre-2012: Nd oxide ~$27/kg in 2008 and briefly >$300/kg in mid-2011 — reliable annual averages are not published for those years.

Figure 3. Neodymium oxide price, 2012–2025 (USD/kg)

Figure data: Table 3; DoD floor per the MP Materials-DoD agreement , 2025.

2. Supply, demand & the market balance

2.1 Where rare earths are mined — deposits & geology

Rare earth supply comes from three geological settings, each with its own economics. Carbonatite hard-rock deposits — igneous bodies exceptionally enriched in light rare earths — host the two giants: Bayan Obo in Inner Mongolia, a vast iron-ore deposit whose rare earth by-product has anchored Chinese dominance since the 1990s, and Mountain Pass in California, the highest-grade major deposit outside China (~7–8% TREO) and the source of nearly all US output. Mineral-sand monazite is recovered as a by-product of titanium-zircon mining (Australia, the US Southeast, India, Brazil) — cheap tonnage, but thorium-bearing. And ionic adsorption clays — southern China, Kachin State in Myanmar, and newer discoveries in Brazil — are the world’s critical source of heavy rare earths: trivially low grade (0.05–0.3% TREO) but leachable at surface with simple chemistry, and rich in dysprosium and terbium.

The 2025 map: China mined ~270 kt REO — about 69% of the world’s 390 kt — under a state quota system. The United States was a distant second at 51 kt (Mountain Pass plus monazite), almost all of it light-rich concentrate. Australia (Mount Weld, one of the richest carbonatites anywhere) produced 29 kt, and Myanmar — the wild card — an estimated 22 kt of overwhelmingly heavy-rich clay concentrate, essentially all of it trucked to Chinese separation plants. Myanmar deserves its own line in any risk model: its unregulated Kachin mines supply more than 60% of China’s heavy rare earth imports, and their seizure by the Kachin Independence Army in October 2024 cut shipments by ~89% year-on-year within months, spiking terbium prices — a live demonstration of how thin the heavy supply chain is.

Table 4. Leading rare earth mining countries, 2025 (estimated)

Source: USGS Mineral Commodity Summaries 2026: Rare Earths , February 2026. Figures rounded; Burma (Myanmar), Thailand and Madagascar estimated from Chinese import data.

Figure 4. Leading rare earth mining countries, 2025 (kt REO)

Figure data: Table 4.

At the asset level, supply is a story of one giant, one comeback and a scatter of newcomers. Bayan Obo alone, with the associated Baotou processing complex, underpins the majority of Chinese light-REE output; the southern Chinese clay districts (Jiangxi, Guangdong) hold the official heavy quota; and the Kachin clay mines — several hundred unregulated sites — are their offshore shadow. Outside China, Mountain Pass returned from bankruptcy to record output under MP Materials, Mount Weld feeds Lynas’s separation plants in Malaysia and Kalgoorlie, and Serra Verde in Brazil became the first commercial ionic-clay producer outside Asia in 2024.

Table 5. Major rare earth mines & districts (approximate recent annual output)

Source: USGS MCS 2026 , company reports (MP Materials , Lynas ) and reporting on Myanmar (MINING.COM/Reuters ), 2024–2025. Output approximate and on differing bases (concentrate vs separated products); Chinese figures inferred from quota allocations. Screen producers and projects on Metal Pilot .

2.2 Demand & consumption

Rare earth demand is really two markets wearing one name. The market that matters — permanent magnets — took about 48% of volume in 2024 and, because NdPr, dysprosium and terbium are the expensive elements, more than 80% of the sector’s value. Magnet demand is structural and compounding: every EV adds magnet content a combustion car never had, direct-drive wind turbines use hundreds of kilograms per megawatt, and electronics, robotics, drones, industrial motors and defense systems (a single F-35 contains ~400 kg of rare earths) stack on top. The second market — everything else — is mature and low-value: catalysts (~16%: fluid-cracking catalysts in refineries, autocatalysts), polishing powders (~10%: cerium oxide for glass and semiconductors), metallurgy (~6%), glass and ceramics (~9% combined), battery alloys (~2%, the lanthanum in nickel-metal-hydride batteries), phosphors and pigments. These uses absorb the cerium-lanthanum bulk of the basket at prices near $1–2/kg — economically, they are how the industry disposes of what it must co-produce to get NdPr.

Table 6. Global rare earth demand by end use, 2024 (share of volume)

Source: Natural Resources Canada rare earth facts (2024 shares, after Adamas Intelligence); value concentration per McKinsey (magnet REEs ~30% of volume, >80% of value). Shares approximate, rounded.

Figure 5. Global rare earth demand by end use, 2024

Figure data: Table 6.

The volume-value mismatch is the sector’s central economic fact, worth one chart of its own: the magnet elements are barely a third of the tonnes but essentially all of the money, while cerium and lanthanum — roughly half of every basket dug up — sell for about the price of gravel with a chemistry degree.

Figure 6. The basket problem — share of volume vs share of value, 2024

Source: McKinsey and NRCan , 2024; shares approximate.

Geographically, consumption follows the factories, and the factories are in Asia. China consumes roughly three-quarters of the world’s separated rare earths, because it makes ~90% of the magnets (and most of the catalysts, phosphors and polishing compounds); Japan is the second manufacturing hub (~5–10% of magnets, the high-performance end); the US and Europe consume mostly indirectly — importing magnets and finished goods rather than oxides. Direct US consumption of compounds and metals was just ~27 kt REO in 2025, with catalysts the leading domestic use — a reminder that the West’s exposure is embedded in supply chains rather than visible in trade statistics. That embedded dependence is precisely what China’s 2025 export controls monetised.

Table 7. Where rare earths are consumed, 2024–25 (approximate)

Source: consumption shares approximated from magnet-production geography (IEA : China ~90% of magnets; DOE via Fastmarkets ) and USGS US apparent consumption; figures approximate — most Western consumption is embedded in imported components and finished goods.

2.3 Supply: producing countries & processing

Mine share understates China’s control, because the chain narrows at every step downstream. In 2024–25 China accounted for roughly 69% of mining — but ~90% of separation and refining, a similar share of metal-making, and ~90% of NdFeB magnet production (94% for finished sintered magnets by some estimates). For the heavy rare earths the mid-chain share approaches ~99%: until Lynas began separating dysprosium and terbium in Malaysia in 2025, effectively every kilogram on Earth passed through a Chinese plant. The rest-of-world processing layer is short: Lynas (Malaysia + the new Kalgoorlie cracking plant), MP Materials (Mountain Pass separation restarted 2023–25), Neo Performance Materials (Estonia), Solvay (La Rochelle, France — NdPr restart 2025), plus refineries under construction (Iluka’s Eneabba in Australia, backed by a government loan) and paper pipelines elsewhere.

World mine production has nearly tripled in a decade — from ~130 kt REO in 2015–16 to 390 kt in 2025 — with China’s quota expansion and Myanmar’s rise supplying most of the growth, and the US the largest addition outside Asia.

Table 8. World rare earth mine production, selected years (kt REO)

Source: USGS Mineral Commodity Summaries , 2017 , 2021 and 2026 editions. Figures rounded; Chinese figures are quotas and exclude undocumented production.

Reserves are large and widely distributed — the constraint is chemistry and capital, not geology. China holds ~44 Mt of the world’s >85 Mt REO reserves, with Brazil (~21 Mt), Australia (~6.3 Mt), India, Russia and Vietnam holding multi-million-tonne endowments; at 390 kt/yr of output, global reserves imply over two centuries of supply. Greenland’s undeveloped deposits (Kvanefjeld, Tanbreez) are a recurring geopolitical football for the same reason: the rocks are common, the working chain is not.

Table 9. Rare earth reserves by country, 2025 (Mt REO)

Source: USGS MCS 2026 , February 2026 (world total >85 Mt). Reserves have been revised heavily in recent editions — Vietnam downgraded sharply, while Brazil’s ~21 Mt is now second only to China; Australian JORC-compliant reserves ~3.3 Mt within the 6.3 Mt figure.

Figure 7. China’s share of the rare earth chain, 2024–25

Source: USGS MCS 2026 (mining); IEA Global Critical Minerals Outlook 2025 and IEA magnet-production data (downstream shares); heavy-separation share per industry reporting, pre-2025 Lynas start-up. Approximate.

2.4 The supply–demand balance

Rare earths have no INSG-style statistical body, no exchange stocks and no published global balance series — the market’s balance must be read through China’s quota system and the segment spreads. Read that way, the mid-2020s picture is a split market. The light/NdPr segment is structurally oversupplied by policy: China expanded its mining quota from 105 kt in 2017 to 270 kt in 2024 — a deliberate, twelvefold-cheaper-than-shortage strategy that crushed NdPr from its 2022 peak ($134/kg Nd oxide average) to $56 in 2024, took Western projects below break-even, and (Western agencies argue) weaponised oversupply just as effectively as the 2010 embargo weaponised shortage. From 2025 the quota itself went dark: allocations were issued secretly, to just two state groups, removing even the market’s main supply signal. The heavy segment is the opposite — structurally tight: official Chinese clay quotas have been flat for years, Myanmar (the swing supplier) is in armed conflict, and the April 2025 export controls sit precisely on Dy, Tb and the other heavies. Prices tell the same story from both sides: NdPr below the US floor price, terbium spiking on every Kachin headline.

Table 10. China’s rare earth mining quota — the market’s de facto balance lever (kt REO)

Source: China MIIT/NDRC quota announcements as compiled by USGS (2020: 140 kt) and Reuters/Mining Technology (2024: 270 kt mining, 254 kt separation; 2025 issued without publication, restricted to two state groups). Separation quotas track mining quotas a few percent lower.

Figure 8. China’s rare earth mining quota, 2017–2024 (kt REO)

Figure data: Table 10.

Table 11. The two-speed balance, 2024–2026

Source: USGS MCS 2026 ; Myanmar figures per MINING.COM/Reuters , 2025; price series Table 3. Qualitative balance assessment — no audited global balance series exists for rare earths.

Without exchange inventories, the stock buffer is state stockpiles — China’s strategic reserve purchases (unpublished, widely believed to absorb surplus NdPr in weak years) and the US Defense Logistics Agency’s modest but growing holdings (FY2025 planned acquisitions: 300 t NdPr oxide, 450 t of NdFeB magnet blocks, 60 t of samarium-cobalt alloy). Both stockpiles are strategic, not price-smoothing, and neither publishes levels — one more reason this market moves on policy announcements rather than data releases.

2.5 Supply structure & recycling

Almost no one mines “rare earths” alone — the supply structure is a nest of joint products. At the deposit level, every rare earth mine co-produces all seventeen elements in fixed geological ratios that rarely match demand: the basket problem of Section 2.2, which makes NdPr price recoveries self-limiting (more NdPr means more surplus cerium) and heavy supply stubbornly inelastic (you cannot mine more terbium without mining everything else). At the mine level, much of the world’s output is itself a by-product: Bayan Obo’s rare earths ride on an iron-ore mine, monazite rides on titanium-zircon mineral sands (Chemours, Iluka, Energy Fuels), and loparite on Russia’s titanium-niobium chain. By-product supply responds to the host commodity’s economics, not to rare earth prices — one more damper on price signals.

Recycling is small but strategically prized. Historically under ~1% of supply (magnets are dispersed in millions of small devices), it is growing from three directions: magnet-production swarf (up to 30% of magnet material is machined away and increasingly looped back), end-of-life magnets (EV motors and wind turbines create, for the first time, large recoverable magnet masses on predictable schedules), and policy — Apple’s $500 million deal with MP Materials commits to magnets made from 100% recycled feedstock for hundreds of millions of devices, the highest-profile bet yet that urban mining can supply a meaningful share of Western magnet demand. The EU’s CRMA sets a 25% recycled-content aspiration for 2030. Recycling’s appeal is obvious: it produces exactly the demanded elements, with no cerium ballast, no thorium and no Chinese mid-chain.

Table 12. How rare earth supply is actually produced

Source: structure estimated from USGS MCS 2026 production detail and industry reporting (Apple/MP recycled-magnet agreement , 2025); shares approximate and overlapping (Bayan Obo is both primary REE and iron-ore by-product).

2.6 Trade flows & the license era

The rare earth trade map is a hub-and-spoke drawn around China — but since 2025, every spoke passes through a licensing office. The physical flows: feedstock flows in (Myanmar clay concentrates — China’s largest external source of heavies; monazite from Africa and Southeast Asia; historically, half of Mountain Pass’s concentrate went to Chinese separators until MP halted China shipments in April 2025 amid the tariff exchange), and products flow out — separated oxides, metals and above all magnets, to Japan, Europe, the US and Korea. The parallel Western spoke is thin but real: Lynas ships Mount Weld concentrate to its Malaysian plant and sells separated products to Japan (under long-standing JOGMEC financing), Europe and the US; Neo runs the only separation plant in Europe (Estonia) on non-Chinese feed; MP now separates at Mountain Pass and makes magnets in Texas.

The chokepoint is not a strait — it is a license. China’s April 2025 export controls put seven elements (samarium, gadolinium, terbium, dysprosium, lutetium, scandium, yttrium) and related magnets under MOFCOM permits, with case-by-case approval, sub-25% approval rates for some Western applicants at the outset, and extraterritorial claims over foreign-made products containing Chinese inputs. The October 2025 expansion (five more elements, tighter tech rules) was suspended for a year in the November truce — the suspension runs to 10 November 2026 — but the April architecture, and the precedent, stands. Enforcement has since widened rather than eased: in June 2026 China placed US producers MP Materials and USA Rare Earth (with eight other US firms) on its dual-use export-control list, and on 24 July 2026 added fourteen European entities — among them defence group Rheinmetall — so the controls now reach Western producers and defence firms by name. Every Western re-shoring program is, at bottom, an attempt to route around that licensing office.

Table 13. Major rare earth trade roles

Source: USGS MCS 2026 (US import sources 2021–24: China 71%); EP Think Tank and CSIS on the 2025 license regime; Benchmark Mineral Intelligence (June 2026 US designations) and S&P Global (24 July 2026 EU additions).

2.7 Market organisations & supply coordination

Rare earths are the purest case of state coordination in commodities — no cartel is needed when one government owns the market. Inside China, supply is run like a utility: the MIIT/NDRC quota system sets annual mining and separation volumes, consolidated since 2021–24 into two state groups — China Northern Rare Earth (lights, Baotou) and China Rare Earth Group (heavies and the south) — with quota publication halted in 2025 for “security reasons.” Above them sits the export-control regime (MOFCOM licensing under the 2025 rules) and the State Reserve Bureau’s unpublished stockpiling. This machinery has been used in both directions: withholding supply (2010–11, 2025 controls) and flooding it (the 2022–24 quota surge that halved NdPr prices).

The Western counter-institutions are younger and smaller: the US DoD/DPA complex (the MP partnership’s price floor and offtake — a pre-agreed contract to buy set volumes of future output — plus DLA stockpiling and Section 232 tariff authority), the EU Critical Raw Materials Act (2030 targets: 10% domestic mining, 40% processing, 25% recycling for strategic minerals), Japan’s JOGMEC (whose financing kept Lynas alive after 2011 — the only successful Western supply-security intervention of the last cycle), and coordination forums like the Minerals Security Partnership. There is no OPEC and no study group; the closest thing to market governance is two governments’ industrial policies colliding.

Table 14. Who shapes the rare earth market

Source: USGS MCS 2026 , CSIS , MP Materials-DoD agreement , 2025–26.

3. The companies & the value chain

3.1 The largest rare earth companies

The corporate landscape mirrors the political one: two Chinese state giants, two Western champions, and a long tail of processors and developers. China Northern Rare Earth — the world’s largest producer by far, listed in Shanghai but state-controlled, feeding on Bayan Obo — and the unlisted China Rare Earth Group hold the entire Chinese quota between them. The investable Western core is a pair: MP Materials, the Las Vegas-based owner of Mountain Pass, which has vertically integrated from mine (50.7 kt REO in 2025, a record) through separation (2.6 kt NdPr, doubled year-on-year) to magnet-making in Texas — now with the Pentagon as its largest shareholder (~15%, plus the $110/kg floor and a 100% offtake of its second magnet plant) and Apple as an anchor customer; and Lynas Rare Earths, the largest non-Chinese separator (record 6.6 kt of NdPr in FY2025 from Mount Weld via Malaysia and Kalgoorlie, plus the first Western dysprosium-terbium separation, backed by Japanese offtake since 2011). Around them: Shenghe Resources (the Chinese-listed trader-processor that historically bought MP’s concentrate), Iluka Resources (building Australia’s government-backed Eneabba refinery), Energy Fuels (monazite processing at its uranium mill), Neo Performance Materials (Estonia separation + magnets in Estonia/Canada) and Solvay (the restarted La Rochelle plant). Sized, as always, on production, resources and chain position — never market capitalisation.

Table 15. Leading rare earth companies, 2024–25 (approximate)

Source: company reports and filings, 2024–25: MP Materials (2025 production), Lynas (FY2025), MP-DoD agreement ; Chinese company scale inferred from quota allocations (2025 quotas unpublished). Listing notes: Northern RE and Shenghe trade publicly under state/parent control; China RE Group is wholly state-owned. No market-capitalisation figures are shown by design.

3.2 Company archetypes along the value chain

Rare earth business models differ more than in any metal market, because the value chain — not the deposit — is the scarce asset. Concentrate producers dig and ship — historically to China, which means their realised prices carry payability discounts and, since 2025, political risk on both ends. Integrated producer-separators (MP, Lynas) capture the full oxide margin and the strategic premium; they are the only Western firms that matter to governments, which now shows up directly in their economics (price floors, offtakes, equity). Developers face the sector’s brutal build economics — separation plants are chemistry-intensive, deposit-specific and slow to commission (the 2011-era cohort mostly died; Ambatovy-style overruns are the norm, not the exception) — so government anchoring is close to a financing requirement. Processors and magnet-makers (Neo, Solvay, the Japanese magnet firms) earn conversion margins and benefit from re-shoring subsidies without mining risk. Recyclers are the newest archetype (element-exact supply, policy tailwind, feedstock still scarce). And the Chinese state groups are their own category: quota-fed, vertically complete, and priced off policy rather than markets.

Table 16. Rare earth company archetypes

Source: company filings; the Metal Pilot project-type taxonomy, 2025.

Figure 9. Rare earth company archetypes by price sensitivity

Source: company filings; conceptual, see Table 16.

3.3 Infrastructure & balance-sheet assets

Reading a rare earth company’s filings means valuing chemistry plants as much as ore bodies. The deposit is described by TREO grade, tonnage and — decisively — basket composition: two deposits with identical headline grades can differ threefold in value per tonne depending on their NdPr and Dy/Tb content, so the reserve statement must be read next to an element-by-element assay. The heavy capital sits downstream: the cracking plant (acid-bake or caustic attack; Lynas’s Kalgoorlie plant is the reference Western build), the separation plant (hundreds of solvent-extraction cells in series, tuned to one feed — capacity in kt REO/yr, but the real metrics are element recoveries and time-to-qualification), metal-making (electrolysis, mostly Chinese), and the magnet plant (capacity in tonnes of NdFeB; customer qualification cycles of years — which is why offtakes with Apple or the Pentagon are balance-sheet events). Two rare earth specifics: thorium/uranium handling — monazite processors need radioactive-materials licensing, an asset in itself (Energy Fuels’ uranium mill moonlights as its REE edge) — and government contracts as quasi-assets: a $110/kg floor, a 100% offtake or a strategic-reserve purchase agreement can be worth more than the ore body it supports.

Table 17. Rare earth company asset types and metrics

Source: company disclosures and reserve statements (JORC / S-K 1300), 2024–25.

4. Investing in rare earths

4.1 Ways to own it — exposure vehicles

Rare earths are the rare commodity you cannot own as a commodity. There are no futures, no exchange contracts, no physically backed funds — and therefore, unusually for this series, no roll-yield problem to warn about: the standard futures-ETP caveat (see the oil guide for the canonical contango trap) simply has no product to apply to here. Physical ownership is equally impractical — separated oxides are industrial chemicals with no resale market. Exposure therefore means equities, and the menu is short: the two integrated Western producers (the liquid core of every rare earth thesis, now carrying explicit government support), a handful of processors and magnet-makers, a wide field of developers (binary, financing-dependent, best treated as options on Western industrial policy), Chinese A-share producers (the actual market leaders, but hard to access and policy-priced), and thematic ETFs — VanEck’s REMX (Rare Earth/Strategic Metals) being the structural example — which bundle the sector plus adjacent strategic-metal names. Because the sector is small, all of it is volatile: the liquid names routinely move double digits on a single export-control headline.

Table 18. Ways to own rare earths

Source: issuer product documentation and exchange listings, 2025; structural attributes only — no fees or prices, which change.

4.2 How to value & screen rare earth producers

Rare earth screening starts from an uncomfortable truth: at Chinese-set prices, almost no Western project is economic unaided — 2024’s $56/kg Nd oxide average sat far below most Western all-in costs, which is exactly why the US government now guarantees $110/kg. So the screen inverts the usual mining logic: policy position first, rocks second. The questions that sort winners from casualties: does the company have a government anchor (price floor, offtake, grant, strategic-reserve customer) or a path to one? What is the basket value per tonne — the NdPr and Dy/Tb share of TREO, valued at benchmark prices, which varies 3× between deposits of identical headline grade? Where does it sit on the chain — concentrate sellers surrender most of the value and all of the leverage; separation capability is the moat, and magnet qualification the prize? For developers: capex per tonne of separated capacity versus the (subsidised) Western benchmarks, the separation flowsheet’s maturity, thorium handling, and — bluntly — cash runway versus qualification timelines measured in years. These are exactly the screens — chain position, basket quality, resource base — you can run across every listed producer on Metal Pilot .

Table 19. Rare earth producer screening metrics

Source: company filings and feasibility studies, 2024–25; framework follows the sector’s post-2011 survivorship record.

Figure 10. Illustrative NdPr cost positions vs price (USD/kg)

Chart source: illustrative — cost tiers approximate the Chinese, integrated-Western and greenfield cost bands from company disclosures and industry commentary, 2024–25 (not company-level data); the dashed line is the 2025 NdPr oxide market average ($69/kg, Table 3), and the $110/kg US DoD floor noted in the label sits higher (Figure 3), per the MP–DoD agreement .

4.3 Macro regimes, rates & correlations

Rare earths are the least macro-driven market in this series — the cycle that matters is the policy cycle, layered on a structural demand trend. Demand is genuinely pro-cyclical (magnets ride autos, electronics and capex; catalysts ride refining throughput), so recessions hurt — but the historical record shows macro forces being routinely overwhelmed by supply politics: the 2011 mania happened into a slowing world economy, and the 2023–24 bust happened into an EV boom. (For how the broad commodity complex behaves across the same regimes, see the macro regime guide .) The dollar and interest rates matter mainly through the equities: rare earth stocks are long-duration, story-driven assets that suffered badly in the 2022 rate shock and rally on policy headlines regardless of the discount rate.

Table 20. Rare earths across regimes

Source: price record (Table 3) and policy timeline (USGS , CSIS ); regime characterisations are historical, not predictive.

On past performance, the sector’s defining episode remains 2010–2015, and it is a cautionary tale told in full: China’s export cuts drove rare earth prices up 10–26-fold in under three years; a generation of Western companies IPO’d into the mania; then substitution, thrifting and restored supply collapsed prices ~90% from the peak. Molycorp — the equity amplifier case study — rode Mountain Pass from a $14 IPO (July 2010) to $79.16 (May 2011), then to Chapter 11 bankruptcy (June 2015) and a $0.35 delisting: a −99.6% round trip in five years, while Lynas — the survivor — still fell from A$26.63 (April 2011) to A$0.30 (June 2015), −99%, and lived only thanks to Japanese rescue financing. The lesson is not that the thesis was wrong — magnet demand grew exactly as promised — but that in a policy-priced market, equities amplify the policy cycle, not the demand trend. The 2021→2024 mini-cycle repeated the shape (MP fell ~75% from its 2022 high to end-2023 as NdPr halved) before the 2025 policy escalation re-rated everything again. Past performance is not indicative of future results.

On correlations: measured monthly over 2012–2025, NdPr prices show only weak positive correlation with the industrial-metals complex (≈ +0.2–0.3 with copper), weak negative with the dollar, and essentially none with gold or equities — the series is dominated by China-specific supply events that ignore the macro calendar. The investable equities are another matter: MP, Lynas and the REMX basket trade with high beta to risk appetite (≈ +0.5–0.6 vs equity indices) plus violent idiosyncratic jumps on policy news. In practice: the commodity diversifies, the stocks don’t.

Table 21. Rare earth correlations (monthly, 2012–2025, approximate)

Source: author analysis of assessed-price series (Table 3) against FRED macro series, monthly, 2012–2025. Correlations are approximate, sample-dependent and repeatedly broken by policy shocks.

Figure 11. Rare earth correlations, monthly 2012–2025

Figure data: Table 21.

4.4 Price drivers & cycles

The durable drivers, read through concluded episodes only. On the demand side: magnet demand growth (EVs, wind, electronics — the compounding floor under the market), substitution and thrifting (the great shock absorber: after 2011, manufacturers cut rare earth content, redesigned motors and killed the phosphor market with LEDs — demand destruction that took years to reverse), and the ordinary industrial cycle. On the supply side, almost everything is policy: Chinese quotas (both directions), export controls, Myanmar’s conflict supply, and the slow, subsidy-dependent arrival of Western capacity. The settled case studies: the 2010–2015 embargo cycle (prices up 10–26×, then −90%; Molycorp dead, substitution permanent — proof that customers will engineer around the basket if pushed); the 2021–2024 quota cycle (EV demand met a 57% quota expansion; Nd oxide $134 average in 2022, $56 in 2024 — proof that Beijing can cap any rally it chooses to); and the 2024–25 Myanmar disruption (KIA seizure of the Kachin mines, Chinese heavy-REE imports −89% y/y, terbium spiking — proof of how thin the heavy chain is). The 2025-26 export-control confrontation — April controls, the October escalation and November truce, then the June 2026 US and 24 July 2026 EU export-control-list designations (Section 2.6) — is still unfolding and is noted here as a current situation, not settled history; its structural lesson so far matches the pattern above: every restriction accelerates the build-out of the parallel Western chain.

Table 22. Rare earth price drivers

Source: USGS , CSIS and the price record (Table 3). Case studies are concluded episodes; the 2025–26 control regime is flagged as unresolved.

4.5 Risks, controversies & ESG

Every risk in rare earths is a concentration risk wearing different clothes. Policy risk is total and two-sided: a Beijing decision can halve prices (2022–24) or double them (2011, 2025), and Western investors are now also exposed to Washington reversal risk — the sector’s economics lean on price floors, tariffs and stockpile budgets that a different administration could unwind. Demand risk is real despite the structural story: the 2011 episode proved substitution works (ferrite motors, rare-earth-free designs, thrifting), and several EV makers maintain magnet-free motor programs precisely as a hedge. Execution risk is the developers’ graveyard — separation chemistry has bankrupted more Western entrants than any other single cause. And the junior-sector froth risk is chronic: every geopolitical flare-up mints new “strategic” story stocks whose deposits will never see a separation plant.

The ESG ledger is heavy on both sides of the chokepoint. Chinese and Myanmar production carries a toxic legacy: ionic-clay leaching has poisoned watersheds in Jiangxi and turned parts of Kachin State into moonscapes of leach ponds — documented by Yale e360 and Global Witness among others — with unregulated militia-taxed mining funding conflict. Hard-rock processing produces thorium-bearing residues: Lynas’s Malaysian waste storage was a decade-long political controversy, and radioactive-waste licensing shapes every Western siting decision. Against this: rare earths are enablers of decarbonisation (EV motors, wind turbines), recycling is genuinely improving, and Western production is marketed explicitly on traceability and environmental standards — a real, if partly political, premium. These are contested questions, and reasonable analysts weigh them differently.

Figure 12. Rare earth risk map — likelihood vs. impact

Source: author’s qualitative assessment; see Section 4.5.

5. Future outlook & forecasts

The forward story is a race between a demand trend everyone agrees on and a supply-chain rebuild nobody has finished. Forecasts are scenarios, not measured facts — and in rare earths they carry an extra caveat: the 2011 cycle proved that both demand (substitution) and supply (quotas) respond to price and politics faster than models assume.

5.1 Demand

The IEA, in its Global Critical Minerals Outlook 2025, projects rare earth demand growing 50–60% by 2040 under its base scenarios, with magnet rare earths — pulled by EVs, wind power and electronics — supplying most of the growth on top of flat-to-declining traditional uses. Within the basket the growth is even more lopsided: NdPr demand from EV traction motors and wind turbines compounds with fleet electrification, while dysprosium and terbium demand grows with the high-temperature magnet share. The main demand-side uncertainties are the pace of EV adoption, the share of magnet-free motor designs (the substitution hedge several OEMs keep warm), and defense procurement — small in tonnes, decisive in politics.

5.2 Supply and the two chains

Supply forecasting in rare earths means forecasting two chains. The Chinese chain retains overwhelming advantages — cost, scale, four decades of process knowledge, and quota flexibility that lets it flood or starve the market at will; the IEA projects the top producers’ concentration in refining rising, not falling, this decade absent intervention. The Western chain is being assembled in public: MP’s 10X magnet facility (10 kt/yr US magnet capacity targeted by 2028, 100% offtake guaranteed), Lynas’s Dy/Tb expansion, Iluka’s Eneabba refinery (~2027), Solvay’s restart, Neo’s European magnet plant, Brazil’s ionic clays and a queue of anchored developers. The build-out is real but small: even optimistic Western pipelines sum to a modest share of world separation capacity by 2030, and every project’s economics assume policy support holds. The wildcards: Myanmar (whose conflict can tighten heavies overnight), recycling (the Apple-MP model scaling), and the export-control regime itself — every escalation raises the ex-China premium that finances the parallel chain.

Table 23. Rare earth outlook indicators (scenario figures)

Source: IEA Global Critical Minerals Outlook 2025 ; USGS MCS 2026 ; MP-DoD agreement . Scenario projections, not measured data.

Figure 13. Rare earth demand index, IEA scenarios (2024 = 100)

Source: IEA Global Critical Minerals Outlook 2025 demand scenarios, indexed to 2024 = 100 (intermediate years interpolated); the ~90% China refining share and the nascent Western chain are developed in Section 5.2 and Figure 7. Scenario projection, not measured data.

5.3 Catalysts to watch

The watch-list is a policy calendar with a mining calendar attached. Beijing: quota signals (now leaked rather than published), MOFCOM licensing practice, the 10 November 2026 expiry of the October-controls suspension, and export-control-list actions (the 24 July 2026 addition of fourteen EU entities, following June’s designation of MP Materials and USA Rare Earth, being the latest escalation). Washington: DoD/DLA stockpile budgets, further MP-style anchor deals, Section 232 outcomes. Projects: MP’s 10X facility progress toward 2028, Lynas’s heavy-REE ramp, Iluka’s Eneabba commissioning (~2027), Serra Verde’s expansion, and the first Western developer to reach a fully financed separation build. Physical: Kachin conflict developments and Chinese heavy-REE import flows; Dy/Tb price behaviour is the tightest real-time indicator the sector has. What would confirm the Western-rebuild thesis: magnet plants qualifying customers on schedule and surviving a Chinese price flood. What would break it: a durable US-China détente that removes the strategic premium, or a policy reversal that pulls the floors out from under Western economics.

Table 24. Rare earth catalyst calendar

Source: USGS , IEA , company guidance and the 2025–26 policy record.

6. Summary

Rare earths are seventeen chemically inseparable metals whose economic identity is one product: the NdFeB permanent magnet — ~48% of demand by volume, over 80% by value — inside every EV motor, wind turbine and precision weapon. There is no exchange and no futures market: prices are assessed per element, led by NdPr oxide, whose history is a policy cycle in price form — ~$27/kg in 2008, >$300 in the 2011 embargo mania, $40 at the 2015–16 bottom, $134 average in the 2022 EV boom, $56 in the 2024 quota-driven bust, $73 in 2025, against a new US-guaranteed $110 floor. Supply is the most concentrated in commodities: China mines ~69% and controls ~90% of separation and magnet-making through a two-group state quota system (unpublished since 2025) and, since April 2025, an export-license regime; Myanmar’s conflict-torn clay mines supply most of China’s heavy-REE feed, and the balance is best read as NdPr in policy-made surplus, dysprosium and terbium structurally tight. The companies divide into Chinese state giants (Northern Rare Earth, China RE Group), the two integrated Western champions — MP Materials (Pentagon-anchored, mine-to-magnet) and Lynas (Japan-anchored, the first Western heavy separator) — and a policy-dependent tail of processors, recyclers and developers, screened on chain position and basket quality, never market cap. The regime lesson is unique in this series: macro forces are routinely overwhelmed by the policy cycle, equities amplify that cycle (Molycorp’s $14 → $79 → zero remains the sector’s defining chart), and substitution punishes every overshoot. The single most important variable to watch is Chinese policy — quotas and export controls — against the Western rebuild, with the Kachin conflict as the heavy-REE wild card.

To go from this big-picture view to the actual companies — screening every rare earth producer by production, resources and chain position — explore Metal Pilot .

7. Vocabulary

Every specialised term used in this report, defined once at first use in the body, collected here for reference.

Table 25. Rare earth vocabulary

Source: definitions follow USGS terminology and industry usage, 2025–26.

8. Sources, methodology & disclaimer

8.1 Sources, methodology & data vintage

Agencies & official data: USGS Mineral Commodity Summaries 2026: Rare Earths (and the 2021 , 2017 and 2013 editions for historical prices and production); USGS Rare Earths Statistics and Information ; IEA Global Critical Minerals Outlook 2025 and IEA magnet-production data ; Natural Resources Canada rare earth facts ; FRED for macro series.

Policy & analysis: CSIS on the 2025 export controls and defense supply chains ; European Parliament Think Tank on China’s restrictions ; McKinsey on magnet REE value concentration ; the 2026 export-control-list additions via Benchmark Mineral Intelligence (June, US) and S&P Global (July, EU); Myanmar reporting via Reuters/MINING.COM and Yale e360 .

Company filings & statements: MP Materials-DoD partnership and Q3 2025 results ; Apple-MP recycled-magnet agreement ; Lynas FY2025 report ; China quota reporting (Reuters via Mining Technology).

Methodology: prices are full-year averages of assessed benchmark prices (Argus via USGS), never spot snapshots; 2012–2015 values are year-end assessments (range midpoints) as flagged in Table 3, because annual averages are not published for those years — a sanctioned adaptation for this assessed-price market. Mine production and reserves follow the USGS (Chinese figures are quotas; Myanmar and neighbours are inferred from Chinese import data). Demand shares follow NRCan/Adamas and the IEA; consumption geography is approximated from manufacturing shares because direct consumption statistics do not exist. There is no audited global supply-demand balance series for rare earths — Section 2.4’s balance assessment is qualitative, built from quotas, prices and trade flows, and labeled as such. Correlations use monthly data 2012–2025 and are approximate. Reserves, resources and forecasts are estimates, not measured facts.

Data as of: 9 August 2026. Intended update cadence: annually after the USGS Mineral Commodity Summaries (January/February) and the IEA Global Critical Minerals Outlook (May), plus ad hoc updates on major policy changes.

8.2 Disclaimer & disclosure

This report is for informational purposes only and is not investment advice, a recommendation, or an offer to buy or sell any security or commodity. Rare earth prices are opaque, assessed rather than exchange-traded, and exceptionally sensitive to government policy; the figures here are estimates as of the stated date that will change, and the sector’s history includes near-total losses in listed equities. Reserves, resources, correlations and regime observations are estimates and historical patterns that may not persist. Do your own research and consult a licensed financial adviser before acting. This report was prepared with the assistance of AI; its figures were sourced from the references above and reviewed, but readers should verify any number before relying on it. The author holds no position disclosed as a conflict in respect of the companies named.