Beijing's one-year pause on sweeping rare earth export controls lapses on November 10. What returns is a licensing machine that has already slowed Tesla's Optimus humanoid — and that sits upstream of every motor, turbine and missile the West makes.
Published 6 October 2026 · 06:00 GMT
On November 10, the pause ends. Beijing's one-year suspension of sweeping rare earth export controls lapses, and a licensing regime that has already slowed Tesla's Optimus humanoid could snap back in full — with extraterritorial rules that reach deep inside factories far from China's borders.
The problem arrived in two words. On Tesla's earnings call, chief executive Elon Musk told investors that production of Optimus, the company's humanoid robot, was being hit by what he called a «magnet issue» — the shorthand for rare earth permanent magnets that now require an export licence from Beijing before they can be shipped anywhere. Exporters face weeks of disruption navigating the new licensing requirements, Tesla among them, and the company has sought assurances from Chinese officials that the magnets are destined for robots, not weapons. Musk still aims to build thousands of Optimus units by year-end and scale toward one million by 2030 — but he conceded production «will move as fast as the slowest and least lucky component.»
That component is neodymium-iron-boron — NdFeB — the most powerful commercial magnet known, and the heart of the small, high-torque motors packed into a humanoid's joints. Published estimates put magnet content at between two and four and a half kilograms per robot. Public trade data compiled by the supply-chain intelligence firm Sayari shows Tesla has sourced its NdFeB from Chinese suppliers, including Beijing Zhongke Sanhuan. In a machine that must balance, lift and gesture inside the tight volume of a human frame, there is no practical substitute for the magnetic force that dysprosium and terbium harden into these alloys.
Capital can be wired overnight. Craft cannot — and the magnet is the proof.
Two suspensions lapse on the same day. The first is China's: MOFCOM Announcement No. 70 of 2025 paused the October 2025 export-control package — Announcements 55, 56, 57, 58, 61 and 62 — for precisely one year, and that year runs out on November 10. The truce was never a rollback — a point earlier reporting made clear — but a diplomatic instrument with a hard expiry, timed to last autumn's trade talks. The April 2025 controls on seven heavy rare earth elements were never suspended and continue to run in parallel. The second suspension is Washington's: the US Bureau of Industry and Security agreed to a one-year pause of its affiliates rule, which extends export-control jurisdiction to subsidiaries of blacklisted parties. Chip export controls, notably, were excluded from the official US-China agenda and remain fully in force. Analysts sketch three paths from here: an extension of the truce, a selective reimposition aimed at specific elements or end-users, or the full return of the October 2025 framework.
The sharpest edge belongs to Announcement 61. Its so-called de minimis rule would require a Chinese export licence for any foreign-made product containing as little as one-tenth of one percent Chinese-origin rare earth content by value — shipments that never touch Chinese territory. If the suspension lapses, manufacturers across Europe, North America and Southeast Asia could fall under Beijing's licensing jurisdiction overnight. The recent flow data already looks like rationing by licence: China shipped 4,735 metric tons of rare earths in August 2026, up more than twelve percent from July's depressed reading, but still eighteen percent below the same month last year and below the year's monthly average. A speed bump would become a wall.
Because the magnet is inside everything else. The International Energy Agency's 2026 Global Critical Minerals Outlook estimated that full enforcement of China's controls could put $6.5 trillion of downstream production at risk each year across the automotive, defence, high-tech and energy sectors outside China — with the United States and Europe holding nearly half that exposure, roughly $3.5 trillion, according to industry analysis. The damage does not require a formal embargo: even selective tightening through licensing denials can cascade through supply chains with far less warning than an announced ban. Automakers already know the feeling — Ford temporarily idled plants in the summer of 2025 for want of magnets, and General Motors has partnered with MP Materials to build a domestic magnet supply chain anchored at its Fort Worth plant. The hardest magnets, the ones hardened with dysprosium and terbium for heat resistance, are also the ones whose manufacturing knowledge is thinnest outside China.
By treating the whole chain as the prize, not the mine. China digs up roughly seven in ten rare earths, but that was never the point of leverage: it refines about nine in ten and manufactures some ninety-four percent of sintered permanent magnets. Then, this year, Beijing consolidated the grip. State Council Order No. 839 of June 15 centralised control over mining, smelting, separation, quotas and traceability; a whistleblower channel for export-control violations opened on July 1; and in June and July Beijing restricted American and European firms — including MP Materials and USA Rare Earth — from access to controlled materials, then blocked dual-use shipments to fourteen EU companies. The response in Europe has been unusually direct: France and Germany are pressing the European Union for a rapid-response trade tool motivated in part by China's critical-mineral restrictions — a recognition that the licensing queue in Beijing has become a trade-policy front line. The deeper history is traced in our account of the metals monopoly.
Money, it turns out, is not the binding constraint — knowledge is. Non-Chinese volume magnet capacity has been described as «a 2028 story running into a January 2027 deadline,» the latter being the Pentagon's DFARS cut-off for defence supply chains. MP Materials, America's champion, only began manufacturing NdFeB magnets in December 2025, and its much larger 10X facility is not expected to be commissioned until 2028. Washington has guaranteed the company a price floor of about $110 per kilogram for neodymium-praseodymium oxide — a measure of how far the state will go to conjure a domestic chain. Sintered magnets demand multi-stage powder metallurgy in which decades of process knowledge decide yield and consistency; the grain-boundary diffusion of dysprosium and terbium that gives defence-grade magnets their coercive force is a step Western plants are only beginning to master. Capital can be wired overnight. Craft cannot.
Meanwhile the other side of the ledger is compounding. China deployed ten times more industrial robots than the United States last year, according to the International Federation of Robotics, and accounted for eighty-five percent of humanoids deployed, with AgiBot and Unitree building most of them, per Barclays data. AgiBot doubled shipments from ten thousand to twenty thousand units in six months. China's electric-vehicle giants are now pivoting into the field: Xpeng's robotics unit raised more than $900 million in August at a valuation above $6.3 billion — the largest single private financing in China's embodied-AI sector — and aims to put its IRON humanoid into mass production by year's end, with monthly output targeted at a thousand units. Chairman He Xiaopeng says building humanoids is roughly twenty times harder than building cars — and Chinese automakers can reuse up to eighty-five percent of their existing supply-chain components to do it.
Geopolitics first: this is leverage designed to be used, not merely held. Beijing timed the one-year suspension to the trade talks, bracketed it with a hard expiry, and layered entity-specific blacklists through the spring and summer — each move calibrated, reversible in form, durable in effect, aimed at the firms and materials where Western substitution is slowest. Macroeconomics: the cost of a rupture is measured in trillions because the magnet sits upstream of the entire electrified economy — cars, turbines, chips, weapons. Demographics: China's automation drive is also an answer to an aging workforce, and the robots it builds at scale are both the output of the chokepoint and its beneficiary.
History supplies the long arc. For decades Chinese planners treated rare earths not as a commodity but as strategic leverage — the downstream processing base that now exists was built through thirty years of state-guided accumulation while the West outsourced the chemistry. Structure is the deepest cut: the bottleneck is not ore but process knowledge, the grain-boundary diffusion and sintering craft that no subsidy can conjure in a quarter. That is the through-line of this story: industrial policy is the century's real contest, and it is fought not in summit communiques but in fabs, battery plants and refineries — the patient accumulation of advantage, made visible at last in a single licensing form.
The calendar is the first thing to watch: five weeks to November 10, and whether an extension is announced before or after the autumn's summits. The second is quieter and more dangerous — the licensing queue itself, where selective denials can throttle supply without any headline-grabbing embargo. Resource nationalism is spreading — Indonesia's nickel regime, Chile's lithium moves, the Congo's cobalt royalties — which narrows every alternative route at once. In Brussels, the French-German push for a rapid-response trade instrument will test whether Europe can convert anxiety into law. And in Washington, the January 2027 DFARS deadline approaches a magnet industry that, outside China, is still learning to sinter. The truce was a pause button. Someone has to decide what plays next.
There is an older pattern worth remembering. Export controls, unlike tariffs, cannot be arbitraged: if the licence is refused, the product does not move, at any price. That is what makes November 10 different from every tariff deadline of the trade war — and why the humble magnet, two to four kilograms of hardened alloy inside a walking machine, has become the clearest measure of who built the industrial century's real infrastructure. The contest was never only about who designs the future. It is about who can license its parts.
From Western capitals, November 10 looks like the day a vulnerability becomes a weapon. The reading in Washington and Brussels is that Beijing has converted a market position — painstakingly built through decades of subsidised processing — into a coercive instrument, timed to trade talks and aimed at the firms least able to substitute. The policy answer follows the diagnosis: price floors for domestic producers, defence supply-chain deadlines, and a French-German push for a rapid-response trade tool. The underlying bet is that dependence this deep is a security problem, and security problems get state money.
Yet the Western lens also carries its own blind spot: the assumption that the chokepoint can be bought away. The honest version of the argument admits that the gap is measured in process knowledge, not capital — that a magnet plant is not a fab that money can hurry. The sharper Western reading, then, is defensive rather than triumphal: build slowly, stockpile quietly, diversify where possible, and treat licensing queues as the new weather. The least honest version pretends the market will solve by 2027 what chemistry took thirty years to build.
From Beijing, the story reads in the opposite direction. Export controls on dual-use materials are presented as standard sovereign practice — every major power maintains them — and the licensing regime is framed as a security instrument, not a trade weapon. Chinese officials have stressed that licences are granted for civilian end-uses, which is precisely why the Tesla case matters to them: assurances that magnets go into robots, not weapons systems, are the stated condition of the regime. In this telling, the October 2025 measures and their suspension were always a negotiating instrument, calibrated and reversible, deployed in response to American tariffs and technology controls.
The deeper Eastern reading is about development, not just leverage. The processing base was built over thirty years of deliberate industrial policy, and Beijing views it as a legitimately earned advantage — one that Western buyers happily used while it was cheap. The suspension's hard expiry is, in this view, simply the other half of a bargain: relief granted in exchange for talks, leverage retained if talks stall. The risk Beijing accepts is that overuse teaches customers to leave; the bet is that leaving takes a decade, and a decade is a long time in geopolitics.
From the Global South, November 10 is a warning written in someone else's handwriting. Producer nations — Indonesia with nickel, Chile with lithium, the Congo with cobalt — are watching Beijing's licensing machine and drawing their own conclusions about resource nationalism: if minerals are leverage, they intend to hold some themselves. The risk for these countries is being drafted into a supply-chain war they did not start, forced to choose between the refiners who buy their ore and the buyers who promise processing plants that may never come.
The sharper Southern reading is opportunity rather than victimhood. Every Western diversification plan needs new mines, new refineries, new magnet plants — and the South holds the geology. The demand is straightforward: processing on Southern soil, technology transfer with real content, royalties that reflect the strategic value of what leaves the ground. The November deadline strengthens that negotiating hand. But the South also knows the oldest lesson in commodities: leverage without processing is just ore in the ground, and ore in the ground does not set the terms of trade.
They are licensing regimes run by China's Ministry of Commerce requiring exporters to obtain permission before shipping listed rare earth materials, magnets and related technology. The April 2025 controls cover seven heavy rare earth elements; the October 2025 package added five more elements plus an extraterritorial clause. Without a granted licence, the shipment cannot legally leave.
China's one-year suspension of its October 2025 export-control package lapses on that date unless extended. If it lapses, licensing requirements snap back — including the extraterritorial rule covering foreign-made products with as little as 0.1% Chinese-origin rare earth content by value. The US suspension of its affiliates rule expires the same day.
Tesla's Optimus humanoid uses small, high-torque electric motors in its joints, and those motors depend on neodymium-iron-boron (NdFeB) permanent magnets — the most powerful commercial magnets available. Estimates put magnet content at 2 to 4.5 kilograms per robot. China dominates NdFeB production, so Tesla must obtain Chinese export licences, a process that takes weeks and has delayed the rollout.
Not quickly. The binding constraint is manufacturing knowledge, not money: sintered NdFeB production requires decades of process expertise that Western plants are only beginning to develop. MP Materials only began making NdFeB magnets in December 2025, and its large 10X facility is due in 2028 — a year after the Pentagon's January 2027 supply-chain deadline.
