The United States mines 13.1%. Its factories are the gap.

Mountain Pass gives the United States something many allied strategies still lack: a mine and concentrator running at scale. MP produced 50,692 tonnes of rare-earth oxide in concentrate during 2025. 4 The site also pushed toward a mine-output target of 60,000 tonnes a year, using the same orebody and much of the same infrastructure. 4

National reserves look more modest than the output rank suggests. USGS reports a national reserve base of 1.9 million tonnes of rare-earth oxide. 1 Concentration inside the country is more pressing: a serious outage at Mountain Pass would remove nearly all current US mine output, and a second American project cannot be switched on like a backup generator.

Mountain Pass's old dependency extended beyond mining because the company historically sold most concentrate through Shenghe into Chinese refining. All China sales stopped in July 2025, and the Shenghe purchase agreement expired the following January. 4 The break forced more material into domestic separation or non-Chinese markets, a commercial choice with strategic consequences.

Separated production is climbing, though it still trails concentrate by a wide margin. Mountain Pass made 2,599 tonnes of NdPr oxide in 2025. 4 Throughput remained below design capacity in parts of the separation circuit. America recovered the mine first and is debugging the refinery while ore production continues.

Nearly the entire US mine-output bar begins at one gate in California. 1 4

What the numbers say

Rare earth projects outside China have long faced a financing trap. A new refinery needs a high enough price to repay its capital, while Chinese integrated producers can sell into a lower domestic cost base. The US answer now includes a government-backed floor rather than another report asking private investors to absorb that gap.

Under its July 2025 agreement, the Defense Department bought US$400 million of MP preferred stock. 5 Full conversion with the attached warrant could leave the department holding about 15 percent of the company on the agreement's stated basis. 5 Government became a potential major shareholder and customer, an unusually direct form of industrial policy for an American mining company.

The support contract runs for ten years, a horizon long enough to cover more than an early trial production cycle. 5 Its NdPr floor is set at US$110 per kilogram under the same agreement. 5 When the benchmark falls below that level, the government covers the eligible shortfall. Above the threshold, later upside is shared under the contract. The price floor targets the problem that has killed Western projects before: a price slump arriving just as a new plant needs to service debt.

Defense Department procurement rules add demand pressure: restrictions for NdFeB magnets extend from mining and separation through subsequent production in covered nations on 1 January 2027, subject to exceptions and waivers. 6 The rule can create a protected customer base, though commercial vehicles and electronics remain much larger markets. If defence demand alone runs a line, that line may be secure without ever becoming globally cheap.

Mountain Pass commissioned domestic separation before the latest policy push, but reaching steady output has taken time. MP produced commercial NdPr oxide, then added samarium to the scope of a planned heavy-rare-earth facility. 4 A restart of the plant that supplies separation chemicals and new recycling lines sit inside the same program, because solvent extraction depends on more than a mine and a row of tanks.

White Mesa in Utah offers a second route through Energy Fuels' current circuit, which the company says can make between 850 and 1,000 tonnes of NdPr oxide a year from monazite. 8 The mill produced commercial material in 2024 and later had a Korean magnet maker approve a batch for automotive drive units. 8 That customer test is worth more than another purity claim because the oxide survived the rest of the chain.

Construction began on White Mesa's commercial heavy-rare-earth expansion in July 2026. 10 The larger phase-two concept would lift NdPr capacity above 6,000 tonnes a year. 9 It also targets 288 tonnes of dysprosium oxide a year. 9 The design figures are not output, and part of the raw material will come from monazite projects outside the continental United States.

Energy Department officials selected two demonstration projects in June 2026. 18 Potential awards across the selected projects total up to US$134 million. 18 Selection begins negotiations; it does not make either project commercial. Critical-minerals announcements often lose that distinction because an award headline can appear years before repeatable product reaches a buyer.

Commercial sintered NdFeB production now exists at more than one US site, a change from the near-empty map of the last decade. Noveon's San Marcos factory is operational and uses mined feed alongside recycled magnet material. 12 The harder test is in Noveon's customer book, where a binding order from Nidec gives the plant a five-year commitment. 13 Its potential volume exceeds 1,000 tonnes, evidence of a customer willing to take the material rather than a factory-capacity claim. 13 Actual deliveries still determine whether customers approve the magnets, the quiet barrier that plant-capacity tables skip.

MP's Independence facility in Fort Worth began manufacturing magnets on industrial equipment in December 2025. 4 It had already commissioned NdPr metal and strip-cast alloy, which gives the plant an unusually complete conversion path. Management is expanding Independence toward 3,000 tonnes of annual magnet capacity. 4

A second MP site called 10X is planned for Northlake, Texas. The design adds 7,000 tonnes of annual capacity beside MP's first Texas operation. 4 Trial production is planned for 2028, placing 10X behind the operating Independence line. 4 The Defense Department's purchase agreement covers the facility for a decade under agreed terms. 5 Equipment and trained operators still determine whether the plant can execute its design, even after the purchase agreement removes much of the demand risk. Customer tests then determine whether its magnets can sell at scale.

USA Rare Earth commissioned the first Stillwater line in March 2026. 14 The company is targeting 600 tonnes a year by the end of the year. 15 A larger South Carolina site is planned to begin trial production in 2028. 16 Meanwhile VAC's operating Sumter plant adds experienced production, although Energy Fuels' agreement to acquire the owner was still pending at the data cut-off. 11

US policy is moving from scattered grants toward a demand system. An April 2025 order opened a national-security investigation into processed critical minerals and derivative products. 7 The Energy Department had selected another US$500 million of critical-mineral and battery projects by August 2026. 19 Financing is no longer the only missing ingredient, because the new plants still have to deliver repeatable material.

Operational repetition is harder because a magnet line whose products pass customer tests needs stable oxide chemistry and alloy control. Its tooling must produce the same grain structure after maintenance. IEA lists strip casters and alignment presses among the specialized equipment bottlenecks outside China. 2 The United States is buying that kit while it rebuilds the engineers who know when a batch is drifting.

Heavy rare earths remain exposed even with White Mesa under construction and USA Rare Earth's demonstration-scale production of commercial-grade dysprosium oxide from recycled magnet-manufacturing scrap in July 2026. 17 That scrap could provide up to 30 percent of that company's future magnetic rare earth oxide feed if the process design scales as planned. 17 Recycling helps, but early scrap volumes cannot feed every factory now being announced.

The country may reach impressive design-capacity totals before it reaches reliable output. Restarted industries normally endure that gap, though it complicates political deadlines. The best signal through 2030 will be several quarters of shipments from more than one feed source. Repeat orders should come from customers who are free to buy elsewhere.