The Bottleneck In Rare Earth Supply

AThe seventeen metals grouped under the name rare earths are not rare. Cerium is about as abundant in the crust as copper, and even the scarcest member of the group is commoner than gold. The name was fixed in the eighteenth century, when the elements were first prised apart from a handful of Scandinavian minerals at enormous trouble, and it has long outlived the conditions that produced it. What is genuinely limited is not the ore but the ability to divide one of these metals from another. Chemically they are almost indistinguishable, and the industrial answer to that likeness is solvent extraction, in which a solution passes through hundreds of successive stages, each stage carrying the mixture a little nearer to purity. Plants of that kind are slow to build and are unevenly distributed: in 2020 a single country carried out roughly 85 per cent of the world's separation. Whether such concentration amounts to a strategic weakness, or to an ordinary bottleneck that prices will dissolve, is what divides the field.

BImani Andersen, a mineral economist at the Nordvik Institute of Resource Economics, argues that the bottleneck dissolves itself. Her case rests on what followed 2010, when export restrictions drove the price of dysprosium, the metal that allows a magnet to keep its strength when hot, from about 250 dollars a kilogram to more than 2,000. Her group examined 38 designs of traction motor built between 2009 and 2016 and traced the dysprosium in their magnets, which fell from an average of 6.5 per cent by weight to 1.4 per cent. Consumption of the metal in that sector fell by 40 per cent across those years even though the number of motors leaving the factories tripled. No new deposit was opened and no separation plant was completed in the interval, yet the shortage went away. Price, Andersen maintains, does quietly and at no public expense what stockpiles do slowly and at great expense.

CThe claim carries a boundary, and Andersen has drawn it herself. If designers are answering a price, the answer should appear only where the metal can be surrendered without surrendering the property that makes the component worth buying. Her second study followed 61 applications, among them catalysts, phosphors, polishing powders and magnets, from 2011 to 2016, and found reductions in 44 of them. The reductions were confined to applications in which a cheaper material could do the same work at some loss of efficiency; where the metal itself set the temperature or the magnetic field that the component could withstand, use per unit in 2016 was unchanged or slightly higher than in 2011. Andersen is careful to describe the pattern as the range within which her explanation is meant to hold, rather than as an assurance that any shortage will cure itself in time.

DTomas Berenguer, who studies industrial policy at the Aurelio Vega Centre in Lisbon, regards that episode as the least suitable precedent available. A change of design, he points out, is made inside a factory that already stands, by engineers already employed; separation capacity has instead to be permitted, financed, built and then taught to run at commercial yield, and the four stages cannot be exchanged for one another. He assembled the 22 separation projects announced outside the dominant supplier between 2010 and 2019 and followed each of them to 2024. Among those that reached first commercial output, the median interval from announcement was 9.4 years, and only four of the twenty-two reached it at all. Two limits attach to the figure, and he states them himself. His list includes only projects whose promoters published a timetable, and the costs he compared were taken from company statements, which are not audited. He presents the number as a measure of how long capacity takes to arrive, not as a claim that it cannot be made to arrive.

EThe argument left the journals long ago. A directive of 2023 obliges member states to obtain a fifth of their consumption from domestic or allied sources by 2032; several governments have bought stockpiles at prices they decline to publish; and recycling is promoted as the obvious remedy, although under 3 per cent of the neodymium in end-of-life magnets is at present recovered, the magnets being glued inside sealed assemblies never designed to be opened. Andersen accepts, awkwardly for her own case, that nothing in her analysis reaches buyers for whom cost is not the deciding consideration, since a price signal cannot discipline a purchaser who is indifferent to price. Berenguer answers that the two kinds of buyer cannot be kept apart in practice, because ministries read civilian substitution as evidence that military supply is safe. Both treat the habit of quoting reserves rather than separation capacity as the source of most confusion in public debate.

FA settlement of a kind is emerging in which the two accounts are assigned to different classes of use rather than ranked against each other. Where a cheaper substitute can be had at some cost in efficiency, price does the work and the shortage is a temporary matter of accounting; where the metal fixes what the component is physically able to do, no adjustment of the sort Andersen recorded has yet appeared, and supply becomes a question of plants, permits and years. What remains unresolved is whether the stockpiles bought after 2010 secured the interval in which the redesign took place or merely deferred a reckoning, and no method now available can separate the two, because the same years contain both. What both researchers reject is the reflex of treating a metal's abundance in the crust as a statement about its availability inside a factory.