Rare Earth Records Are Nice. MP Materials' Real Test Is the GM Magnet Shipment.

MP Materials has been telling two different stories over the past year, and only one of them is in the record books. Neodymium-praseodymium output — the pairing of elements that gives permanent magnets their strength — hit 917 metric tons in the first quarter, then 840 more in the second despite a month of scheduled maintenance. Management guided the third quarter past 1,000 tons. NdPr sales ran even hotter: 1,006 tons in Q2, up 127% from a year earlier.

The other story sits off the record sheet, and it's the one shareholders should be watching. On that same earnings call, the company said it had shipped finished magnets to General Motors for in-vehicle qualification testing and still expects the first commercial deliveries by the end of this year. This is the first time MP's entire integrated machine — mine to motor part — has to earn money from an actual paying customer rather than from a government floor. Everything before it has been structure. That year-end shipment is the test of whether the structure converts into a business.

What MP is actually building

MP is the only fully integrated U.S. rare earth producer. It mines ore at Mountain Pass, California, the country's sole commercial-scale rare earths mine, separates the individual elements, reduces the key ones to metal, and — at its Fort Worth "Independence" plant — presses them into sintered neodymium-iron-boron magnets. That last step is the point. China controls more than 90% of rare-earth separation and magnet manufacturing, and magnets are the component the West most depends on for EV motors, turbines, and drones. The strategic reason for MP to exist is plain.

Strategic reasons, though, don't pay the bills. The commodity stage of the business is already largely de-risked — and largely capped.

The government floor that also caps the upside

In mid-2025 the Pentagon put $400 million into MP for convertible preferred stock and warrants, roughly a 15% stake on a converted basis, wrapped in a decade of support. The centerpiece is a price floor: the Department of Defense guarantees $110 per kilogram of NdPr content — more than double the roughly $51 average MP realized in 2024 and well above the $50–60 market cost of NdPr oxide — and pays the difference if spot falls below it across concentrate, oxide, metal, and magnet alike. It also committed to buy 7,000 tons of magnets a year for 10 years. Banks supplied a further $1 billion in construction loans and Apple committed $500 million to recycling, and MP holds roughly $1.45 billion of cash, enough to fund the whole expansion plan without a near-term financing scare.

That is confirmed support. It is also what quietly caps the commodity upside. Under the profit share, once annual EBITDA passes $140 million the government takes the first $30 million above it, then half of everything beyond $170 million. Every ton sold into this structure earns a floor and shares its profits. The margin the market is ultimately rewarding has to come from downstream.

The bottleneck beneath the bottleneck

Follow the chain a layer down and the usual suspect stops being the constraint. At a targeted 60,000 tons of total rare-earth oxide, Mountain Pass could yield more than 6,500 tons of NdPr oxide a year — while the full 10,000-ton magnet program (the existing plant scaled toward ~3,000 tons, plus a new $1.25 billion "10X" campus in Northlake, Texas, commissioning in 2028) needs under 4,000. MP produces more of the "magnet metal" than its own factories can use.

The real choke is the heavy rare earths. High-performance EV magnets lean on dysprosium and terbium to hold their magnetism under motor heat and load, and Mountain Pass ore carries under 2% of them. So MP is doing two things at once: commissioning its first heavy-rare-earth separation circuit, meant to run on third-party feedstock and recycled material, and using its "grain boundary diffusion" process to shrink how much Dy/Tb each magnet needs in the first place. That is the genuine chokepoint in the domestic build-out — the node with no easy substitute and no installed U.S. supply.

Then the chain runs into the qualification gate, which is where this becomes a real business rather than a map. A magnet inside an EV motor has to survive warranty-grade duty cycles, and failure is expensive. GM is not buying a commodity; it is qualifying a supplier. MP says test magnets are delivered and commercial shipments land in the fourth quarter, initial volume modest and the ramp steady after. That part is confirmed: GM is a named long-term customer and the shipments are targeted, and qualification testing is happening under the hood. What remains undisclosed is the contract's size and price — how much of the future magnet output GM has actually committed to buy.

What the price already assumes

Put it together and the imbalance a retail holder has to weigh is sharp. The commodity business is booming and floor-protected but profit-shared, and the magnetics segment is where the revaluation must come from — yet it's tiny and currently shrinking, posting about $7.5 million of adjusted EBITDA for the quarter, down 7% year over year as start-up costs eat the ramp. The market is paying for the finished version: at roughly $50 a share, MP carries about a $9 billion market cap around 25 times trailing sales with negative earnings, and the stock sits near half its 52-week high of about $100. It is a multi-year execution story that has already given back a year's worth of gains.

None of that makes the structure wrong. It means the structure is known, and the market is now paying to see it executed at the one node that still has to prove itself. Watch the fourth quarter, not for a single shipment that changes the arithmetic, but because it is the first time the integrated engine earns real revenue from a real customer — and the clearest signal yet whether the dependency MP has spent years building is a business or just a better-subsidized map.