Rare is the wrong word. The scarcity is in the separation.

The name is a historical accident. The rare earth elements are not rare in the earth's crust; cerium is more abundant than copper, and neodymium is more abundant than gold by a factor of thousands. They were called rare because in the eighteenth and nineteenth centuries they were difficult to isolate from the minerals that carried them, and that part has not changed much.

They occur together, they are chemically similar, and they are separated by repetition. A modern separation plant runs a cascade of solvent-extraction mixer-settlers — hundreds of stages in series — where an organic phase and an aqueous phase are contacted, allowed to settle, and split, over and over, each stage nudging one element a little further from its neighbors. The plant is not exotic. It is enormous, finicky and expensive to tune.

That is the entire reason a country can hold a world-class deposit and still export its ore. Mountain Pass mined for decades and, for part of that time, shipped concentrate to China for the step that actually produces a saleable oxide. China now produces roughly 270,000 tonnes of rare earth oxides a year against about 60,000 from the rest of the world combined, and the gap is separation capacity, metal-making and magnet-making — not ore.

The buildout in the United States is aimed squarely at those middle steps: a heavy separation circuit at Mountain Pass financed with a $150 million federal loan, a magnet plant in Fort Worth designed to reach 10,000 metric tons a year by 2028, and a portfolio of coal-ash and tailings projects aimed at producing a mixed concentrate from material already above ground.

The binding constraint that gets the least attention is people. Metallurgists who have commissioned a separation cascade, mill operators who can read a solvent-extraction bank, instrument techs who understand radiation controls on a monazite feed — those skills were largely allowed to lapse in this country, and a price floor does not create them. Training a workforce is the slowest part of any of these projects, and it is the part most often left out of the announcement.