Here is the contradiction most people miss about reshoring: the hard part was supposed to be the chemistry, and the chemistry just got solved.
In Marion, Indiana, a single chromatography column started producing commercial-scale germanium this month. ReElement's CFO Kirk Taylor put the pace bluntly: the company went "from lab scale to pilot scale to now commercial scale in less than six months," starting from a waste stream in March.
That is a genuinely fast retirement of technology risk. Solvent extraction — the incumbent process — is chemically nasty, capital-heavy, and slow to scale. ReElement swapped it for chromatography, the same purification approach pharma uses, and validated that the process holds at the largest column size it intends to deploy.
ReElement swapped the incumbent solvent extraction for chromatography, the same purification approach pharma uses, and validated the process at the largest column size it intends to deploy.
Last week this thread was about decree: an executive order making it harder for defense contractors to buy Chinese minerals, and a $2 billion federal push announced at the State Department. This week it's about physical output — one line, running, on time, under budget. The milestone is real.
Now zoom out to the number that reframes it.
Global germanium consumption is roughly 60 metric tons a year, and the US imports more than half of what it uses. So the honest question is what one validated column actually buys against that gap.
The answer is: proof, not tonnage. The germanium line is the first of four Phase I production lines at Marion — gallium, light rare earths, and heavy rare earths are still being commissioned over the next six months. Commercial-scale germanium demonstrates the platform. It does not yet displace imports.
And the marquee supply deals point at the same lag. NioCorp has a preliminary agreement to sell Lockheed Martin 15 metric tons of scandium a year — about a quarter of the ~60-ton global demand — from its Nebraska mine. That mine is reportedly "slated to open by 2028." The US hasn't mined scandium since 1969.
So the flagship contract is a commitment against capacity that does not physically exist yet, for a metal the country stopped producing during the Nixon administration. Rio Tinto, the only North American scandium producer, currently makes about nine tons a year.
This is the pattern across the reshoring story: announcements are compounding faster than plants. UNCTAD counted critical-minerals greenfield project announcements jumping more than 30% in 2025, even as actual project-finance value fell to roughly $8 billion. Pipelines are not operating mines.
The technology being de-risked doesn't change that arithmetic. ReElement's chromatography is a refining platform — it separates and purifies feedstock. It still needs feedstock. And feedstock comes from mines that take years to permit and build.
Which is where the second, slower constraint lives — and it's the one federal dollars struggle most to compress.
You can't run a mine, a separation plant, or a refining line without people who know how. On that front, the US has been quietly hollowing out for forty years.
The number of accredited mining engineering programs fell from 25 in 1982 to 14 today. Those 14 schools now graduate fewer than 170 mining engineers a year. Enrollment is reportedly down roughly 45% since 2015, per DOE.
Against that, the Department of Energy estimates the mining sector alone will need about 6,000 new engineers over the next decade. And China graduates more than 3,000 mining-related professionals a year — a pipeline more than seventeen times wider than America's annual output.
That is not a gap you close with a purchase order. It's the difference between two educational systems compounding in opposite directions for a generation.
The problem is worse than raw headcount, because the expertise that eroded fastest is the one reshoring needs most. Colorado School of Mines President Paul Johnson named it directly: "Our expertise has atrophied over time, particularly in the processing space."
Processing is exactly the link ReElement just proved out — and exactly where China built its decades-long lead. A country can import a chromatography column. It cannot import metallurgists, mineral processors, and materials scientists at the scale a new supply chain requires.
Washington sees this. Alongside the $2 billion in projects, Trump announced more than $180 million aimed at the workforce pipeline — the largest piece being DOE's planned $100 million PROSPECT initiative, which the department hopes will double mining-related graduates within two years.
There's also a $32.7 million Pentagon proposal for a Critical Minerals Innovation & Commercialization Hub at Colorado School of Mines — a 50,000-square-foot facility meant to bridge lab discoveries to commercial production quickly.
These are the right instruments. But look at the clocks they run on.
Doubling graduates in two years means those graduates finish degrees in, optimistically, the late 2020s — and reach senior competence years after that. The Nebraska scandium mine opens in 2028, if permitting holds. Mine timelines and academic timelines are both measured in years, and neither responds to urgency the way a wire transfer does.
Money can accelerate permitting, finance a plant, and underwrite a long-term purchase. It cannot manufacture a metallurgist with fifteen years of processing experience, and it cannot make an orebody produce before it's built.
That is the real asymmetry with China. It isn't that the US lacks capital or, anymore, the core refining technology. It's that China has 3,000 graduates a year and decades of accumulated processing depth, while the US has 170 graduates a year and a single validated column that just came online.
So the question this milestone forces isn't whether reshoring is real. Marion answers that — it is. Physical output exists, on schedule, and the technology scales.
The question is whether physical and human capacity can compound fast enough. The refining step now moves in six-month increments. The tonnage and the people move in five-to-ten-year increments — and they're the binding constraints.
Federal dollars operate on the fast clock. Mines and mining engineers operate on the slow one. The political window and funding cycle that opened this buildout run on the fast clock too.
The germanium column proves the idea works. What it can't prove is that the slowest inputs — refined tonnage and trained people — will arrive before the money, and the attention, move on.