"To encourage the company's self-funded facility expansion, it was seeking authorizations and appropriations to buy more than 12,000 missiles over five years."
That line, buried in the Pentagon's May 13 framework agreement with Castelion, is the most important sentence written about defense tech this year. Read it again. The government wants 12,000 missiles. It has not appropriated the money to buy them. It is seeking the authority. And the reason it is telling Castelion all this — publicly — is to get Castelion to spend its own money building a factory first.
Start with the seduction, though, because the seduction is real.
A four-year-old startup founded by three SpaceX alumni just raised $1 billion at a $13 billion valuation. JPMorgan's Strategic Investment Group, Andreessen Horowitz, and Carlyle co-led the equity. T. Rowe Price came in new. Lightspeed, Altimeter, and General Catalyst all doubled down.
The product justifies the enthusiasm on paper. Blackbeard, Castelion's hypersonic strike missile, is expected to cost roughly $384,000 per unit, per Reuters. Lockheed's Dark Eagle reportedly runs about $41 million, per Tectonic. That is not a discount. That is a different category of thing.
Clean sheet to program of record in under four years, with fielding targeted for 2027. More than $500 million in military contracts over 18 months. A 1,000-acre manufacturing campus in New Mexico. This is the story venture wants to tell itself: that private capital and speed have finally cracked the defense-industrial base open.
Katherine Boyle put it as cleanly as anyone. "We backed Castelion when it was a small team that wanted to build what the department of War most needed faster and cheaper than the experts thought possible." Four years later, she said, "this is the clearest proof" that excellent teams and private capital can solve physical problems.
But look at what the Pentagon actually did, and the direction of the causation flips.
The framework agreement contemplates a two-year procurement contract for a minimum of 500 missiles annually — once Castelion completes testing and validation. The 12,000-missile figure is a five-year aspiration contingent on appropriations that don't exist. And as Aerotime reported reading the release closely: as of that announcement, "neither the procurement contract nor those appropriations were in place."
The Department was explicit about the design. Its new-entrant vendors, it said, would reach production scale "without direct Department investment." It described a partnership model that "rewards private sector capital."
That is the whole trick, stated in the government's own words. The Pentagon is not writing the check. It is dangling demand — a number, 12,000, large enough to underwrite a factory — so that private balance sheets build the capacity the government isn't willing to fund up front.
The Pentagon is not writing the check. It is dangling demand large enough to underwrite a factory, so that private balance sheets build the capacity the government isn't willing to fund.
Castelion has already committed more than $250 million of private money to Project Ranger. It just raised a billion more, in part, to "commit hundreds of millions more." The equity investors are, in effect, financing capacity against a purchase order that is still a legislative wish.
This inverts a pattern worth naming. When Nvidia reportedly invests in the startups that then buy Nvidia chips, a company manufactures demand for its own product to pull capital in. Here the government is doing the manufacturing — dangling 12,000 missiles it hasn't paid for — and private capital rushes to meet a demand signal the Treasury has not actually committed to.
The Pentagon has learned to underwrite private capital instead of the other way around. And once you see the mechanism at Castelion, you see it everywhere hard tech has gotten too capital-heavy for the market to carry alone.
Look at batteries, where the exact same move runs in reverse.
The One Big Beautiful Bill gutted battery and EV production incentives, undercutting a huge chunk of future demand. The domestic factories that sprang up under the Inflation Reduction Act suddenly had no market to serve. By pure market logic, a wave of battery startups should have died.
Then the Department of Energy announced $500 million in grants — explicitly justified to "reduce reliance on foreign sources, bolster national security, and advance American energy dominance." Much of it went to startups.
Coreshell got $50 million to expand manufacturing for its metallurgical silicon anode material. Lilac Solutions got $100 million to build a lithium-carbonate facility on the Great Salt Lake. Nth Cycle got $100 million to refine recycled black mass.
An administration with open disdain for EVs kept the battery supply chain alive by relabeling it. Nth Cycle's CEO Megan O'Connor told TechCrunch there are "clear demand drivers from the defense sector." Coreshell said defense applications are "absolutely playing out in discussions" — and brought on a strategic VC whose parent company is a defense supplier.
This is the tell. The consumer and industrial demand that markets recognize was legislated away. National-security framing was the mechanism that put the money back.
Same trick, opposite direction. At Castelion, national security manufactures demand to pull private equity into a factory. In batteries, national security manufactures a justification to keep public money flowing after the market rationale was killed. Either way, the label is doing the financing.
And the reason the label matters is that capital-intensive hard tech has stopped penciling out on its own. The Defense Logistics Agency was buying $200 million of batteries a year even in 2021 — real, but a rounding error against the roughly $18 billion, per Mordor Intelligence, the auto industry is expected to spend on U.S. battery manufacturing this year. Defense demand cannot replace consumer scale. It can only buy time and keep the supply chain from going dark.
The uncomfortable implication is that this is now the bipartisan financing path for anything too heavy for pure market logic. A Trump administration that hates EVs funds battery plants under a security flag. A Pentagon unwilling to appropriate 12,000 missiles gets a private factory built by promising them anyway. The national-security frame is the one reliable way to move real money into physical infrastructure the market has abandoned.
Which raises the question of what happens to the sectors that don't get the label in time.
The Information reported this month that China is moving rapidly toward commercializing semisolid-state batteries while U.S. startups remain far from large-scale production — and that a new report warns China's manufacturing lead could be difficult to close.
That is the shape of the risk. National-security financing keeps chosen technologies breathing, but it is triage, not industrial policy. The money flows to what gets framed, when it gets framed. The technologies still waiting for their security label — or arriving too late to one — lose the race while the frame is being drafted.
Castelion may well succeed. The team is credible, the price point is genuinely disruptive, and the demand for cheap hypersonics is real. But its $13 billion valuation does not prove that venture discovered defense.
It proves the government figured out how to make venture do the building. The check the Pentagon didn't write is the whole story — and everyone is celebrating the round that filled the gap.