Japan's stockpile policy and its diversification policy cannot both survive 2026

Two promises sit on Tokyo's desk right now, and they cancel each other. The first is the stockpile promise: the state holds reserves of critical metals so Japanese factories never stop when Beijing squeezes. The second is the diversification promise: within a few years, no single supplier can hold Japan hostage again. China has just made the squeeze big enough that honoring the first promise empties the very buffer the second one depends on. The reserve is being consumed faster than the replacement supply chain is being born, and 2026 is the year arithmetic catches up with policy.

The trigger arrived in stages. In April 2025, China required export licenses for seven medium and heavy rare earth items, dysprosium among them, then added five more elements such as holmium in October, then paused the regime until November 2026 after the US-China trade talks (Caixin Global, Aug 21). Early this year Beijing went further against Tokyo specifically, banning dual-use exports to entities tied to the Japanese military and imposing rare earth controls on forty Japanese defense, aviation and shipbuilding companies, with twenty more watchlisted (Caixin Global, Aug 21). The result on the ground is stark: China's dysprosium and terbium shipments to Japan fell 75 percent last year to 45 tons, and went to zero in the first half of 2026 (Caixin Global, Aug 21). Zero.

Underneath sits the slower pressure. Japan spent fifteen years proving it could cut light rare earth dependence after the 2010 embargo. Heavy rare earths stayed almost entirely Chinese anyway, because nobody else separates them at scale. China still refines nineteen of the twenty strategic minerals tracked by the International Energy Agency, and the agency estimates up to $6.5 trillion of economic output outside China rests directly on those concentrated processing nodes (IEA outlook via Skillings, Jul 28). Diversification was always going to be slow. The stockpile existed to absorb the gap. Nobody budgeted for a gap with no far side.

Japan's reserve was sized for the 2010 shock and is fighting the 2026 one, and no country can spend the same tonnage twice.

The gap is now measurable. Japanese manufacturers can currently secure only about two-thirds of the rare earth supply they need, and analysts at Argus expect the heavy rare earth shortage to persist until at least 2027, because projects backed by the state metals agency JOGMEC need one to two more years to reach large production (Argus, Aug 17). A survey of four hundred manufacturing executives by Resilire found over ninety percent lacked sufficient supply of at least one rare earth material, with an average sixty percent reliance on China for procurement, and costs up more than twenty percent where supply did arrive (Resilire survey via Caixin Global, Aug 21). Every month factories run on reserve tonnage is a month the buffer shrinks without replenishment.

So the two policies fight each other directly. The stockpile was designed as insurance against a shock measured in months. This shock is measured in years, since the alternative mines JOGMEC backs will not produce at scale until 2027 or 2028 (Argus, Aug 17). Drawing down the reserve keeps assembly lines moving today but guarantees Japan enters the next confrontation, or even the resumption of Chinese licensing friction in November, holding a thinner cushion than at any point since the program began. A policy of hoarding for emergencies cannot coexist with an emergency long enough to drain the hoard.

Reuters

Tokyo's answer is money and patience. In July, JOGMEC agreed to invest up to C$47.7 million, roughly 5.5 billion yen, in TJ Namibia Rare Earths, a special purpose company with Toyota Tsusho developing the Lofdal heavy rare earth deposit alongside Canada's Namibia Critical Metals, targeting exactly the metals now cut off (JOGMEC release, Jul 30). Lynas and Japan Australia Rare Earths signed a deal in March for 5,000 tons of NdPr alloy a year and started up a matching separation plant (Caixin Global, Aug 21). JAMSTEC plans a large-scale trial next year mining rare earth rich mud from the seabed near Minamitorishima (Skillings, Jul 28). All real. None fast enough to matter before the reserve runs low.

History offers one clean comparison and one warning against it. After China halted shipments during the 2010 Senkaku incident, Japan used state funding through JOGMEC to help build Lynas into a non-Chinese supplier and pushed dependence on Chinese rare earths down meaningfully over the following decade, proof that a stockpile plus subsidized diversification can work when the target is one metal family. The counterexample is the heavy end of the same story: reliance on those two heavy rare earths barely moved in fifteen years, because separating them outside China required chemistry, permits and customers no subsidy line could conjure quickly. The 2010 playbook works on the metals that are easy to replace and fails on the ones Japan actually needs most right now.

Follow the consequences outward and the bill lands on identifiable desks. Toyota's motor lines, Japan's defense electronics primes on Beijing's restricted list, and the magnet makers feeding them all pay in allocation and price while the reserve thins. The winners sit upstream: Lynas earns guaranteed offtake at premium terms, Lofdal's owners gain a state-backed buyer before producing a kilogram, and any non-Chinese separator with dysprosium capability can name its price through at least 2027 (Argus, Aug 17). Meanwhile the government pays twice, once to build the reserve and again in subsidies to escape the need for it.

What confirms the read is simple: a METI announcement releasing strategic metal stockpile tonnage this autumn, or official figures showing the reserve drawn below its stated target months ahead of schedule. What breaks it: Beijing quietly reissuing licenses to Japanese civilian buyers after the US-China truce holds, letting imports resume and the reserve rebuild, which would prove the whole episode was coercion theater rather than a lasting cutoff.

The judgment the numbers force: Japan's reserve was sized for the 2010 war and is fighting the 2026 one, and no country can spend the same tonnage twice. Until Lofdal ships or Minamitorishima mud yields, every magnet made from the stockpile is a loan against a mine that does not exist yet.