Why Japan’s Air‑Conditioner Recycling Can’t Bridge Its Rare‑Earth Gap

In early July, Mitsubishi Electric began dismantling discarded household air‑conditioners to extract neodymium‑based permanent magnets, which are then sent to Shin‑Etsu Chemical for separation and re‑refining before being remounted in new units. The plan targets about 230,000 units per year, covering roughly 35 % of the company’s own rare‑earth needs.

The initiative is not a new invention; “urban mining” was proposed in Japan in the 1980s and Mitsubishi’s recycling technology dates back to around 2012. The change in external conditions—China’s export licences for dysprosium and terbium starting April 2025 and tightened dual‑use controls on Japan in January 2026—has driven the scale‑up.

Model 1: Quantity – a drop in the bucket

Empirical measurement shows that dismantling 100 air‑conditioners yields less than 2 kg of high‑purity rare earths, i.e., about 20 g per unit. Scaling to 230,000 units gives roughly 0.2 t of recovered material, matching the official “several tonnes” claim but representing only a tiny fraction of Japan’s annual demand of 20 t. Even if every discarded air‑conditioner in Japan (millions of units) were processed, the total would stay in the hundred‑ton range, far below the required amount.

The time lag further limits impact: the magnets recovered today come from units sold a decade or more ago, implying a service‑life delay of 10–15 years. With demand for magnets rising due to electric vehicles, robotics and AI hardware, the recovery rate remains structurally constrained.

Model 2: Structural bottleneck – the wrong rare earths

Viewing high‑end production capacity as a “barrel” limited by its smallest stave, the shortage caused by the export ban concerns heavy rare earths (dysprosium, terbium). The recycled magnets from air‑conditioners consist mainly of light rare earths (neodymium, praseodymium) and contain little of the heavy elements. Thus the recycling effort supplements a resource that is not the bottleneck, while the true constraint—heavy rare earths—remains untouched.

Model 3: Depletion vs. substitution race

Representing the usable stock of heavy rare earths as a reservoir, the analysis compares two time scales: the depletion rate (months) and the build‑up rate of substitute capacity (years). Japan’s heavy‑rare‑earth inventories are already below six months, whereas alternative supplies—such as the Minamitorishima seabed project (35 t of slurry yielding ~70 kg of rare earths in February 2026) and Lynas’s dysprosium‑terbium output—are orders of magnitude smaller or slower to materialise. Consequently, the recovery from air‑conditioners, limited to a few tonnes per year, cannot bridge the multi‑year gap.

In quantity, the recycling is locked below the demand by the product‑life time lag; structurally, it addresses light rare earths while the blockade targets heavy ones; temporally, stock depletion occurs in months whereas substitution would take years, leaving a hard shortage.

Its practical effect is to signal “action” to the public and market, while the government simultaneously claims “no problem”. Nomura Research estimates that a one‑year continuation of the restrictions could cost ¥2.6 trillion, and Daiwa projects a 1.3 %–3.2 % GDP reduction.

The real moat in the rare‑earth industry lies not in the ore but in the refining and magnet‑making processes, which are dominated by Chinese capacity. For Japan, the most scarce resource is strategic clarity about time scales, not the rare earths themselves.