Building a supply chain for rare earth elements

The first thing to understand is that “rare earths” is a category hiding a much smaller category that does all the work. There are 17 elements, and five or six are what the industry calls magnet rare earths: neodymium, praseodymium, dysprosium, terbium and samarium. Those take about 40% of rare earth volume but carry over 95% of the dollar value in the market, which means the other 60% of volume is worth under 5%. The reason is physical. Neodymium and praseodymium combined with iron and boron produce the strongest permanent magnets that can be made, with high energy density in a small volume, and tiny concentrations of dysprosium and terbium keep those magnets magnetic at elevated temperatures. That combination yields the most efficient electric motors and generators available, which is why they end up converting battery power into motion in an electric vehicle and motion into power in a wind turbine. If you have already spent heavily on a battery pack, he argues, it makes sense to pair it with a motor that uses the electricity efficiently. Demand for these elements therefore tracks the energy transition directly.

The supply picture is unusual even by critical minerals standards. Roughly 60% of mining happens in China, some of it a byproduct of iron oxide mining, with the remaining 40% elsewhere including the US. But most of what is mined outside China is shipped there anyway: over 90% of processing and over 95% of magnet manufacturing happen in China, which Ghahreman describes as acting like a vacuum. That is different from lithium or cobalt, where China dominates refining but not the mine. His explanation combines geology and decisions. The US and Canada led rare earth production in the 1960s and 1970s. Rare earth ores carry radioactivity at an awkward concentration, high enough to be a real environmental problem but too low to be worth recovering, and the chemistry that turns ore into metal is difficult and dirty. Over twenty or thirty years the West exported that burden and then, in his words, woke up to find the production capacity inside Chinese borders. He does not frame it as something done to the West: China published five-year plans for two decades saying it wanted these metals, and the West miscalculated how much they mattered. Rare earths were about a $20 billion market in 2024, while the industries they unlock run to trillions.

The processing chain shows why the position is hard to contest. Ore contains 1% to 2% rare earth element or less and is upgraded to a concentrate at 50% to 60% rare earth oxide. Heavy chemistry then turns concentrate into mixed rare earth oxide, a basket of all the elements together, and the basket is the problem: for California material shipped to China the output is roughly 70% to 80% lanthanum and cerium, and across deposits generally something like 70% to 85% of the basket is what Ghahreman calls value-negative, elements so overproduced that they are a disposal question rather than a product. Separating the valuable minority means solvent extraction, a chain of around a thousand unit operations, because all 17 elements behave almost identically in chemical processes, which makes each step inefficient. It is expensive and it generates waste. Metallization, converting purified oxides into metal, is about 95% Chinese, for the straightforward reason that the oxides are not available anywhere else. Only then do you get alloy, magnet and motor. So mining rare earths means deliberately overproducing material nobody wants in order to obtain the small fraction everybody does.

Against that, Ghahreman sorts the responses into the three Kann proposes. On substitution, rare-earth-free magnets already exist in the alnico and ferrite families and new ones are being developed, which he thinks is the right thing to attempt out of necessity. But he does not believe they are there yet, and expects the near-term applications to be speakers and similar products where strength per unit volume and high-temperature performance matter less than in a traction motor that runs hot. On Western mining and processing he is much more positive, citing MP Materials mining in California and working downstream toward neodymium-praseodymium production and magnets, Lynas building a Texas refinery with over $200 million from the Department of Defense, and Vacuumschmelze building magnets in South Carolina with over $100 million, alongside the IRA and equivalent programs elsewhere. He adds a demand-side reason it might stick: inside a carmaker, engineers love rare earth magnets and cannot get enough of them, while buyers dislike them precisely because the supply chain runs through one country. Multiple sources removes the buyer’s objection without touching the engineer’s preference. He still says plainly that in 2024 the dependence on China remains heavy. Recycling, the third path, sits under 1% globally for a mechanical rather than economic reason: magnets are magnetic, so at end of life they travel with the steel into steel recycling, where the rare earths report to slag, a thermodynamically stable glass that locks them away for good. You could recover them, but it would cost more than it returns and be environmentally harmful. Cyclic’s answer is hub and spoke. Spokes sit near population centers where end-of-life products accumulate and split something like a traction motor into copper, aluminum, steel and magnet; the commodity metals sell into existing local markets, while the magnets, under 5% of the incoming feed, ship to a single North American hub for hydrometallurgical processing into mixed rare earth oxide. Logistics dominate recycling economics, so the spokes stay close to feedstock, while one chemical plant holds down capital cost. Two consequences matter more than the corporate detail: recycled feedstock was already refined once, so the output is concentrated in magnet metals and skips the value-negative basket entirely, and no radioactivity enters the process because none comes in with the feed.