When Western officials talk about rare earths, they often begin with mines and maps. China digs a large share of the world’s rare earth ore. It has the Bayan Obo deposit in Inner Mongolia, the Baotou processing complex that grew around it, and ionic adsorption clays in the south. Those facts are true. They are also incomplete. The more consequential story sits further down the chain: separation of individual oxides, refining and metallisation, and the factories that turn neodymium-iron-boron (NdFeB) alloys into permanent magnets for electric cars, wind turbines and a long list of military systems.
Several 2025 studies describe the same pattern. The Atlantic Council’s mapping of China’s rare earth strategy put neodymium-iron-boron magnet shares roughly as follows: about 58 percent of mine production, 89 percent of separation, 90 percent of refining and metallisation, and 92 percent of magnet production. The International Energy Agency’s 2025 work on magnet rare earths—counting neodymium, praseodymium, dysprosium and terbium—put China at about 60 percent of mining and 91 percent of refining in 2024, with sintered permanent magnet production near 94 percent. CSIS, writing in October 2025, sketched a similar whole-chain picture of roughly 70 percent mining, 90 percent separation and processing, and 93 percent magnet making. The studies differ in detail. They agree on the shape. Chinese control tightens as you move from pit to finished magnet.
That is the subject of this essay. China did not merely sit on ore. Over decades it concentrated the middle and end of the magnet supply chain at home. From 2023 onwards it also began to treat process technology and, in 2025, export licences as instruments of industrial leverage—tools that look, in outline, like the semiconductor export controls Washington has used, but aimed at metals and magnets rather than chips.
What the supply chain looks like
Rare earths are seventeen elements. They are not scarce in the Earth’s crust so much as awkward to separate and refine at industrial scale. Light rare earths such as lanthanum, cerium, neodymium and praseodymium dominate Bayan Obo–style deposits and the Baotou complex. Heavy rare earths such as dysprosium and terbium are harder to find in convenient deposits. South China’s ionic adsorption clays, and cross-border feed from Myanmar that Chinese processors handle, matter because high-temperature magnets need those heavy elements. They raise coercivity so magnets keep their strength when motors run hot.
A permanent magnet for an electric-vehicle traction motor or a modern wind generator is not a lump of ore. Ore becomes concentrate. Concentrate is cracked and separated into individual oxides. Oxides become metals. Metals become alloys and powders. Powders become sintered or bonded magnets, then cut and coated for the job. Each step needs chemistry, capital, waste handling and experienced operators. For decades many Western mines could dig and ship concentrate. Chinese plants did most of the dirty middle work. Chinese magnet makers then made most of the finished goods.
Market reporting for 2024 puts Chinese NdFeB output on the order of about 280,000 to 300,000 tonnes, depending on whether the count is blanks or finished magnets. A round 310,000-tonne figure that circulates in some trade notes is better treated as an upper market estimate than as a single audited official total. Adamas Intelligence noted record Chinese rare-earth magnet exports in 2024, with Chinese customs-style tallies of magnet shipments in the high tens of thousands of tonnes. Volume is only part of the story. Grade matters. Magnets that must survive heat and harsh duty cycles take dysprosium or terbium. Those grades sat closest to the licensing wall Beijing raised in 2025.
Baotou remains the clearest emblem of light rare-earth industrialisation: pit, mill, separation trains, alloy shops, and a city economy built around these elements. Further south, the ionic-clay belt and Myanmar-adjacent flows have long supplied heavy rare-earth feed that Chinese separators know how to handle at scale. Consolidation under larger state-guided groups, including China Rare Earth Group, was meant to curb small illegal diggers and give Beijing clearer control of quotas, prices and environmental enforcement. Whether every tonne is cleanly booked is a separate question. The industrial geography is not.
How China gained control of the middle steps
Price did much of the early work. Through the 1990s and 2000s, Chinese producers—backed by state planning, cheap power in places, weak early environmental enforcement, and a home market that wanted magnets for hard disks, speakers and much else—undercut Western separators and magnet shops. Plants in the United States, Europe and Japan closed or shrank. Mountain Pass in California, once a Western flagship, spent years as a cautionary case: concentrate still needed Chinese processing. Know-how followed the work. Chemists, line managers and equipment vendors clustered where the tonnes were.
Western trade lawyers often called this dumping. Beijing preferred the language of industrial policy. Both descriptions point to the same market outcome. The world learned to treat Chinese separation as the default utility. Once that utility existed, building a rival looked expensive, slow and environmentally contentious. Permitting a separation plant in California or Europe is not like building another warehouse. Communities remember tailings. Investors remember Chinese prices that can fall whenever a new Western plant looks viable.
Technology control came next. On 21 December 2023 China banned the export of rare-earth extraction and separation technologies, and tightened the list around magnet-related production know-how. That was a quiet prelude. Digging ore abroad is one thing. Teaching engineers how Chinese plants separate and sinter at industrial yields is another. The December 2023 step showed that Beijing understood the moat as process skill, not only geology.
History had already offered a warning. In September 2010, after a collision near the Senkaku/Diaoyu islands and Japan’s detention of a Chinese fishing-boat captain, Japanese industry reported that rare-earth shipments from China slowed or stopped through customs delays and inspections. Beijing denied a formal embargo. Scholars still argue how cleanly the episode maps onto retaliation versus earlier quota cuts. What stuck in Tokyo, Washington and Brussels was simpler: dependence felt real, and China could turn paperwork into pressure.
The April 2025 export licensing rules
On 4 April 2025 the Ministry of Commerce and the General Administration of Customs issued Announcement No. 18. It placed export controls on a set of medium and heavy rare-earth related items: samarium, gadolinium, terbium, dysprosium, lutetium, scandium and yttrium, together with related metals, alloys, targets, oxides and compounds. Permanent magnet materials were explicitly in scope where they contained the controlled elements. That covered samarium-cobalt magnets and NdFeB magnets or magnetic powders containing terbium or dysprosium. Exporters had to apply to MOFCOM for a licence under China’s export-control and dual-use rules. Customs could hold cargo while they queried the paperwork.
This was not a total ban on every Chinese magnet. Light rare-earth NdFeB without the controlled heavy additives could still move more freely in principle. Commercial reality was harsher. High-performance grades for electric vehicles, wind and defence often need dysprosium or terbium. Once those grades needed a licence, buyers faced delay, uncertainty and the chance of a quiet denial. Adamas Intelligence, writing for buyers in May 2025, stressed that each order needed a licence and that suppliers were advising applications months ahead of desired ship dates.
Timing mattered. The announcement landed in the middle of a tariff war with the Trump administration. Analysts read it as retaliation and as leverage. Either reading still leaves the industrial fact: Beijing chose a licensing regime rather than a theatrical total embargo. Licences can be sped up, slowed down, approved for some end users and denied for military ones. That is how modern export-control states prefer to work.
How May and June 2025 showed the leverage in practice
Shipments fell sharply. Chinese customs data reported in Reuters and Bloomberg showed rare-earth magnet exports plunging in May 2025. One widely cited tally put May magnet shipments down about 74 percent year on year, to roughly 1,238 tonnes—about half the April level and the weakest reading in years. Shipments to the United States shrank to a trickle. Automakers felt it. Ford paused Explorer production in Chicago for a week in May over rare-earth shortages. European suppliers warned of lines going idle. Nissan and Suzuki reported disruption. Magnosphere’s leadership told the German press that factories could be quiet by mid-summer without fresh magnet flow.
Diplomacy chased the shortage. On 11 May 2025 United States and Chinese officials met in Switzerland and struck a ninety-day tariff truce meant to restore rare-earth access for American firms taken off Chinese blacklists. Licences still lagged. President Trump accused Beijing of reneging. CSIS’s Gracelin Baskaran and Meredith Schwartz, writing in June, estimated that only about a quarter of the hundreds of licence applications filed since April had been approved. Delay itself was the instrument.
London talks on 10 and 11 June 2025 produced another framework. Treasury Secretary Scott Bessent took part. Public accounts said China would resume rare-earth and magnet exports to the United States while tariff structures were reshaped. June shipments recovered in later customs tallies. The episode still taught a clear lesson. When China sits on magnet licences, assembly plants on other continents notice within weeks. Speeches about friend-shoring do not refill a parts bin.
The October 2025 rules that reach beyond Chinese ports
On 9 October 2025 MOFCOM Announcement No. 61 went further. CSIS described it as China applying a foreign-direct-product style rule to rare earths and magnets for the first time: the same broad idea Washington has used for semiconductors, now aimed the other way. Foreign-made magnets and related items that contain even small shares of Chinese-origin controlled rare earths, or that are made with Chinese mining, processing or magnet-making technology, would need Chinese approval. Reporting and law-firm notes put a 0.1 percent threshold in play for relevant Chinese-origin content. Overseas-product provisions were set to take effect from 1 December 2025, with Chinese-origin controls applying sooner.
Military end-use language hardened. Starting in December 2025, applications tied to foreign military users or military end uses were set for denial as a rule, not a discretionary maybe. Chinese nationals were restricted from supporting overseas rare-earth and magnet projects without authorisation. Case-by-case review stretched toward advanced semiconductor-related uses. Some later diplomatic suspensions and carve-outs were reported in trade coverage into 2026; treat those as moving parts. The architecture remains: Beijing wants a say not only over what leaves Chinese ports, but over magnets made elsewhere if Chinese material or Chinese know-how sits inside them.
In outline this mirrors American chip policy, drawn in metal. Washington tells the world that United States tools, software and technology create jurisdiction abroad. Beijing now says Chinese rare-earth content and Chinese process technology can do the same for magnets. The symmetry is both political theatre and industrial reality.
Why heavy rare earths and high-grade magnets matter
Neodymium and praseodymium carry the basic NdFeB story. Dysprosium and terbium carry the high-temperature story. Electric-vehicle motors that sit next to hot engines and inverters, wind-turbine generators that run for decades offshore, and many defence actuators and seekers need magnets that do not fade when the temperature rises. Heavy rare-earth additions raise coercivity. Without a reliable heavy rare-earth and high-grade magnet pipe, buyers redesign motors, accept lower performance, or wait.
Defence inventories make the dependence concrete. Open CSIS and defence-industrial writing lists rare-earth magnets across fighter jets, submarines, missiles, radar and smart munitions. Exact kilograms per platform are often classified or proprietary. The strategic claim does not need the exact bill of materials. If magnet licences stall, production schedules slip. China has spent years scaling munitions and platforms. Western defence ministries have spent years writing critical-mineral strategies. One side has the factories. The other has strategies on paper and a handful of plants.
Civil demand is larger still. Electric cars and wind are the volume pull. Consumer electronics and industrial motors fill the rest. IEA projections under current policies see magnet rare-earth demand rising hard into the 2030s, with electric vehicles as the big driver. Planned capacity outside China, even with subsidies, covers only a fraction of that path on today’s project lists. That gap works in Beijing’s favour.
Allied attempts to rebuild capacity, and their limits
Mountain Pass is again the American showcase. MP Materials mines in California. For years much of its concentrate still leaned on Chinese downstream capacity. In April 2025, after Chinese retaliatory tariffs and export controls, MP said it would stop China-bound concentrate shipments because sales under those tariffs made no commercial sense. It pushed domestic processing and non-China sales instead.
In July 2025 the United States Department of Defense—later referred to in some official speech as the Department of War—announced a large public-private package with MP Materials: about 400 million dollars of equity that made the United States government the largest shareholder on reported terms; a ten-year neodymium-praseodymium price floor around 110 dollars per kilogram to shield the firm from Chinese oversupply pricing; a 150 million dollar Office of Strategic Capital loan aimed at heavy rare-earth separation at Mountain Pass; and offtake arrangements tied to a planned 10X magnet-manufacturing campus near Fort Worth. The price floor is industrial policy by another name. It admits that Western magnet and oxide plants cannot survive a price war without a buyer of last resort.
Noveon Magnetics, described in October 2025 CSIS writing as the only United States rare-earth magnet manufacturer then in operation, signed a memorandum of understanding with Australia’s Lynas Rare Earths to link separated oxides to United States magnet making for defence and commercial customers. Lynas already runs separation outside China, including Malaysian capacity and Australian projects. Iluka’s Eneabba refinery plans and Arafura’s Nolans project sit in the same allied ledger, with heavy Australian public finance behind them. Europe talks of its own magnet and recycling schemes. Japan never forgot 2010 and has spent years on stockpiles, thrift and alternative supply.
None of this is yet a second China. MP’s early magnet volumes were measured in hundreds to low thousands of tonnes against a Chinese industry measured in the hundreds of thousands. Heavy rare-earth separation on United States soil was still being built, not finished, through 2025. Recycling helps but does not replace primary feed for a growing electric fleet. Friend-shoring runs into the same walls that built Chinese dominance: capital cost, permitting time, environmental opposition, scarce process talent, and the risk that Chinese prices drop whenever a Western plant approaches break-even.
What control of the middle of the chain allows
A mine is a place. Control of separation, refining and magnet making is a system. China can approve commercial magnet licences to calm carmakers while denying military end use. It can slow paperwork without announcing a crisis. It can threaten extraterritorial reach over foreign magnets that still contain Chinese heavy rare earths. It can keep process talent and equipment at home under the December 2023 technology rules. It can consolidate producers so quotas and environmental crackdowns also serve as market management.
Washington’s chip controls try to freeze China at certain process nodes. Beijing’s magnet and rare-earth controls try to remind everyone else that green industry and advanced weapons still run through Chinese separation halls and sintering furnaces. The tools rhyme. The materials differ. The practical message is about who can wait: interdependence is not a treaty; it is a balance of patience and spare parts.
Critics will say China risks pushing customers to build alternatives faster. That risk is real. The MP deal, Lynas–Noveon talks, Australian refineries and European schemes exist because April and May 2025 hurt. Supporters of Beijing’s approach will say the West spent decades outsourcing the dirty steps and should not be shocked when the landlord sets house rules. Both can be true. The industrial clock still favours the incumbent. Separation plants and magnet lines take years. Licence regimes can be announced in an afternoon.
Limits of the leverage
China is not omnipotent. Light rare-earth NdFeB without controlled heavies can still move under different rules. Smuggling, stock draws and redesigns blunt shocks. Diplomatic truces reopen taps. Announcement No. 61–style overseas rules invite legal fights and workarounds. Chinese producers also need export revenue. A permanent total cut-off would damage Chinese magnet exporters and their local champions. The preferred tool is calibrated friction, not scorched earth.
Western policy can fail in the opposite direction: announce resilience, fund a ribbon-cutting, then discover that the oxide still travels to China or that the magnet line needs Chinese powder. USGS and IEA-style scoreboards will keep showing Chinese shares falling only slowly even when headlines shout breakthrough. Treat every new Western plant as progress—and as a rounding error—until the tonnes show up year after year.
What follows from who owns separation and magnets
The rare-earth story is often told as a mining postcard: Baotou under a bright sky, trucks on the haul road, a map coloured for reserves. The operating reality is solvent-extraction trains, sintering furnaces, dysprosium additions, export-licence queues and a December calendar for military denials.
China built its position the long way: price, scale, process skill, consolidation, then law. April 2025’s Announcement No. 18 put heavy rare earths and high-grade magnets behind licences. May’s shipment collapse showed the leverage in customs data. June’s talks showed how quickly geopolitics can pivot on magnet paperwork. October’s Announcement No. 61 tried to project that leverage past the Chinese border the way American chip rules project past American ports.
The West is funding mines, price floors and magnet shops. It should. It will still live for years inside a system where Beijing can license or delay the middle of the chain. That is not a slogan about minerals. It is a statement about who owns separation, refining and NdFeB manufacture. Until those shares move in a serious way, China does not merely have rare earths. It holds a large share of the magnet gate—and it has learned to keep the key on a ministry desk.