Seventeen Elements and the Price of Every Magnet on Earth

Not scarce, just unrefined

Rare-earth elements are a group of seventeen metals, the fifteen lanthanides plus scandium and yttrium, prized for the magnetic, luminescent, and catalytic properties that make them irreplaceable in permanent magnets, precision optics, and a long list of industrial catalysts. Despite the name, they are not especially rare in the earth's crust. Deposits worth mining exist across Australia, the United States, Brazil, Vietnam, India, and several African nations, among others. The name describes how thinly they are usually dispersed, not how little of them the planet holds.

What is genuinely scarce is the capacity to do something useful with the ore once it comes out of the ground. Rare-earth minerals arrive bound together in complex mixtures, and separating one element from its close chemical relatives requires hundreds of repeated extraction cycles, strong acids, and waste streams that carry real environmental liability. China mines at least 60 percent and processes approximately 90 percent of the world's rare-earth metals, according to 2024 industry data cited in a 2025 legal analysis of China's export-control regime. The ten-point gap between the mining share and the processing share is the entire story: China's advantage sits far more heavily in what happens after the ore leaves the mine than in the mine itself.

That gap exists because the West largely walked away from the refining step over the past three decades, judging it too costly and too environmentally exposed to compete against a state-backed buildout, while China treated separation and alloying capacity as a strategic asset worth subsidizing for decades. The mining step is a commodity business; almost any country with deposits and capital can dig ore out of the ground. The refining step is where a shapeless mixture of oxides becomes a neodymium-iron-boron magnet that a motor manufacturer can actually buy, and that is the step where nearly all of the economic value, and nearly all of the strategic power, now sits.

This is the shape of the problem that would matter through 2025: not a mining monopoly, which other countries could in principle challenge simply by opening new mines, but a processing monopoly, which took China decades of deliberate investment to build and cannot be replicated by announcement.

The end products of that refining step are everywhere, well beyond headline defense hardware. The same neodymium-iron-boron chemistry sits inside the voice coil motor that positions the read head in a hard disk drive, the traction motor in an electric vehicle, the generator that turns a wind turbine's rotation into current, and the actuators that steer a precision-guided missile. A single refining bottleneck therefore reaches consumer electronics, the clean-energy transition, and national defense procurement at once, which is a large part of why a regulatory change to that one step can move so many unrelated markets on the same day.

The lever

The current tightened-control regime has a specific start date. A rare-earth export licensing system took effect in China on April 4, 2025, according to a November 2025 analysis by the Foundation for Defense of Democracies, marking the formal beginning of a period in which Beijing moved from treating rare-earth exports as routine trade to treating them as a governed, license-by-license instrument of policy.

The regime widened sharply six months later. On October 9, 2025, the Industry Safety and Import/Export Control Bureau inside China's Ministry of Commerce issued new export-control notifications covering rare-earth-related products, equipment, and technologies, according to a legal analysis published by Mayer Brown. The notifications did not just add more items to a restricted list. They changed the geography of the rule itself.

The choice to name equipment and technologies alongside the elements themselves is worth pausing on. Restricting the export of a finished metal is one kind of pressure; restricting the export of the separation equipment and process know-how needed to build a competing plant is a different and longer-lasting kind, because it slows the one activity, replicating the refining step, that would eventually erode China's advantage. A country can accumulate stockpiled metal against a shortage. It cannot as easily accumulate the specialized machinery and engineering expertise required to separate the next batch itself.

The centerpiece is a threshold: China's Ministry of Commerce designed the October 2025 controls to restrict the export of any product anywhere in the world containing more than 0.1 percent rare-earth content of Chinese origin, an extraterritorial reach that a White & Case legal analysis, citing the underlying MOFCOM Notification No. 61/2025, explicitly compares to the Foreign Direct Product Rule used in American export-control law. A regulation that started as domestic mining policy had become a claim of jurisdiction over manufacturing decisions made in Germany, Japan, or the United States, on the strength of a trace ingredient sourced from China.

The same notification introduced a second, related mechanism specific to this sector. Notification No. 61/2025 established what practitioners have taken to calling the "50% Rule," applying to rare-earth items and to certain semiconductor and AI-related items, restricting the export of foreign-made products that incorporate a defined threshold share of controlled Chinese-origin content, per the White & Case analysis. The rule did not stay confined to magnets and motors. It was written, from the start, to reach the chips and AI hardware built downstream of the same materials.

A blacklist riding alongside the minerals

China did not rely on the mineral controls alone. Its October 2025 export-control expansion added 14 additional entities to its own Unreliable Entity List at the same time it issued the new rare-earth rules, according to an analysis by CM Trade Law, combining mineral-supply pressure with a parallel corporate-blacklist mechanism in a single announcement.

The pairing matters more than either piece alone. A mineral export control restricts a category of goods; an entity listing restricts a named company's access to Chinese supply chains and, in some readings, to Chinese customers and partners. Deploying both instruments in the same notification let Beijing apply pressure on the material flowing into a targeted firm's products and on the firm's standing inside China's own commercial system, without needing a second announcement or a second round of internal approvals.

It is the kind of layering that turns a supply-chain regulation into an instrument closer to economic statecraft. A government that can restrict both a mineral and a named counterparty, in the same document, has built a lever with more than one point of application, and the October notifications used both at once.

China first established the Unreliable Entity List in 2020, in the period following the United States' own use of its Commerce Department Entity List against Chinese technology firms, so the mechanism itself was not new by October 2025. What changed was where China chose to use it: pairing the corporate blacklist with a mineral chokepoint that reaches essentially every advanced manufacturer on earth gave a tool built for individual companies a scale it had not carried before.

The pause that was not a retreat

In November 2025, China introduced a one-year delay on the start of the newly expanded October 2025 controls, while explicitly retaining the underlying April 2025 licensing system, according to the FDD analysis. The headline read as a climbdown. The substance was narrower than that.

The FDD analysis describes the delay as widely interpreted as a way of preserving future options rather than a genuine policy reversal, and the structure of the move supports that reading. Beijing did not repeal the extraterritorial threshold, the 50% Rule, or the entity-list additions. It postponed their activation while keeping the licensing regime that predated the expansion fully in place, preserving the option to reactivate the broader architecture on its own timeline rather than under pressure to negotiate it away.

A government does not need to enforce a rule continuously for the rule to do its work. The credible threat that Beijing could switch the extraterritorial controls back on, at a date of its own choosing, is itself a form of pressure on every manufacturer now deciding whether to keep sourcing rare-earth content from Chinese-processed material or to pay more for a supply chain that carries less exposure to a single ministry's discretion.

The distinction matters for anyone actually buying the material. A rule that has been paused on a fixed schedule is not the same thing as a rule that has been withdrawn. Every manufacturer with a rare-earth magnet in its bill of materials now has to plan around a regime that could revert to full extraterritorial enforcement at the end of the delay, which is a different and more expensive planning problem than either a stable rule or a genuinely repealed one would be.

Why a mineral chokepoint belongs in a history of information infrastructure

The connection between rare-earth refining and the information economy is not metaphorical. Permanent magnets built from rare-earth alloys sit inside the electric motors, generators, actuators, and hard disk drives that make up the mechanical substrate of computing, communications, and clean-energy infrastructure, the physical layer that has to work before any digital layer built on top of it can function at all. A wind turbine, an electric vehicle motor, a precision-guided missile, and a data center's backup generator all draw on the same narrow refining capacity.

Beyond the physical link, the connection is written directly into the regulation itself. The same "50% Rule" that governs rare-earth items in Notification No. 61/2025 also applies to certain semiconductor and AI-related items, per the White & Case analysis, which means China's own export-control architecture already treats the rare-earth chokepoint and the AI-hardware chokepoint as a single administrative category, enforced by the same bureau under the same notification. The mineral behind the magnet and the chip behind the model are, in Beijing's own regulatory text, one lever.

The International Energy Agency assessed that the October 2025 controls turned longstanding theoretical concerns about critical-mineral supply concentration into a realized geopolitical risk, given the near-total dependence of clean-energy and defense supply chains on Chinese rare-earth processing, according to IEA commentary published in 2025. What had been a slide in an industry presentation for a decade became, in a single notification, a live constraint that treasury officials and defense planners had to price into their decisions immediately.

This publication studies the infrastructure of information across history, on the premise that whoever controls the dominant medium of an age, the road that carried the message, the cable that carried the signal, sets the economic and political terms for everyone downstream of it. The current chapter of that history runs through an AI system deciding which business gets named in an answer. The rare-earth story is the same mechanism working one layer down, in metal rather than in code: whoever controls the refining plant that makes the magnet, and now the same notification that governs the chip, sets terms for the manufacturers, and increasingly the AI systems, built on top of it.

The pattern is consistent across every chapter of this history. A message could only travel as fast as the road built to carry it; a signal could only cross an ocean once someone laid a cable able to survive the seabed; and an AI system can only name a business it can find, read, and verify. In each case, the constraint was never the information itself, it was the infrastructure standing between the information and the person who needed it. Rare earths are that constraint restated in physical form: the magnet has to exist before the motor turns, the motor has to turn before the server spins, and the refining plant has to run before any of it is possible.

What both sides of the argument get right

The optimistic reading has real substance. Rare-earth deposits are not confined to China, and a government willing to fund exploration, permitting, and new mines in Australia, the United States, Brazil, Vietnam, or elsewhere can in theory diversify the extraction side of the supply chain within a few years. China's October 2025 controls forced the United States, the European Union, Japan, and Australia into exactly that scramble, funding alternative processing capacity in response to a bottleneck that had gone from theoretical to urgent overnight. Mining, unlike refining, is a problem that capital and geology can solve.

The concentrated reading has equally real substance, and it is the harder one to argue away. Refining is not a capital problem alone. It requires decades of accumulated process engineering, a workforce trained in a genuinely hazardous chemistry, and an environmental permitting regime willing to tolerate the waste streams that separation produces, none of which a government can simply purchase into existence. China built that combination over thirty years while competitors were exiting the business for the same reasons that now make reentry slow. Replicating it is a matter of years, not months, and probably longer than the optimistic funding announcements from Washington, Brussels, Tokyo, and Canberra currently imply.

Both readings can be true at once, and the honest position is to hold them as scenarios rather than as a settled forecast. In one scenario, allied investment closes the processing gap within the decade and the extraterritorial threshold becomes a historical curiosity, a rule Beijing wrote at the peak of its advantage and lived to see erode. In another, the environmental and technical barriers prove as durable as the West's own three-decade retreat from the business suggests they will, and the 2025 controls turn out to be the opening move in a much longer period of dependence. The data available now supports either scenario being plausible. It does not support calling either one inevitable.