Hypersonic missiles are defined first by speed: at least Mach 5, five times the speed of sound, fast enough to cross the United States from coast to coast in under 45 minutes. Unlike ballistic missiles, hypersonics are maneuverable and difficult to track, able to fly low, change course mid-flight, and skirt radar. That mix of velocity and unpredictability is why the weapons have become a benchmark of military modernization—and so hard to defend against.

The second defining fact is the strategic gap: China and Russia have fielded hypersonic weapons, and the United States has not—not fully. The shortfall is not for lack of effort. Despite years of research costing billions and a series of successful tests, the U.S. has yet to fully field a hypersonic missile of its own, held back by testing complications and production issues. The technology has existed for some time, but it has proven expensive and difficult to produce at scale.

Closing that chasm is the bet behind Castelion, a startup valued at $13 billion that believes it can succeed where the defense giants have so far fallen short. "Humans have done it, just not Americans," says Bryon Hargis, the company's CEO and cofounder. "I firmly believe it's doable."

The Torrance, Calif.-based company has become a standout of the defense tech boom, in part because it embodies the movement's core premise: the U.S. must rebuild its industrial base faster, cheaper, and with an eye to scale. The war in Iran has crystallized that stark reality: the United States is firing missiles faster than it can replace them, by an order of years.

Castelion was founded in 2022 by three former SpaceX executives: Hargis, Andrew Kreitz, and Sean Pitt. Hargis started Starshield, SpaceX's military-grade satellite network; Kreitz ran classified finance; and Pitt led sales of Falcon 9 rockets in Europe.

The enterprise got off to an almost comically inauspicious start. In an early Zoom pitch to the Pentagon's innovation outfit, the founders announced that they planned to compete with Lockheed Martin in building affordable hypersonics. The room laughed, Pitt recalls.

The tables have since turned. In less than two and a half years, Castelion took its low-cost hypersonic missile, Blackbeard, from a drawing to an advancing reality—and secured a "program of record" designation, the elusive and coveted government budget line that makes, and often breaks, defense sector entrants. In practice, the designation means a weapon is slated for funding and fielding across future budget cycles rather than remaining a prototype or experiment.

That status places Castelion in direct competition with Lockheed, maker of Dark Eagle, a hypersonic race favorite that has been in development since 2017 and is reportedly three years behind schedule. ("Dark Eagle has successfully deployed in support of Indo-Pacific exercises and completed multiple successful flight-test events," an Army spokesperson said via email. "The Army continues planned operational assessment activities.")

In September, the U.S. Navy awarded Castelion a contract worth up to $200 million for Blackbeard, which, according to documents, can be made for less than $500,000 each. By comparison, Tomahawk missiles can cost millions apiece. That spread matters for stockpiles: in a fight where munitions are expended faster than they can be replaced, per-round cost shapes how many rounds a service can afford to buy and keep. Blackbeards have recorded successful test flights on fighter jets including the F/A-18 Super Hornet, and the company is racing to open its 1,000-acre New Mexico site, which could eventually turn out thousands of the missiles each year.

"This capability is needed now, and you have to sprint directly at what the country needs," says Kreitz, Castelion's CFO.

Kreitz argues that hypersonics matter not only as weapons but, more fundamentally, as deterrents. "It's really important we have a credible deterrent that we pass down for generations to come," adds Hargis. He gestures toward a Mad Men-esque poster in the company's mid-century office. "Cheesy as it is, this was the first poster we ever made for the company," he explains. "You can see the tagline was: 'Win before it begins.'"

Deterrence, of course, only has a chance if the weapons exist. Recent conflicts have created an unmissable urgency, Kreitz says. "This is the capability that will define warfare in the 2030s and beyond," he said. "The campaign was basically large, sustained volleys of back-and-forth long-range fires. That's what the future of warfare will look like."

Build, Fly, Test, Repeat

When Hargis, Kreitz, and Pitt were getting started in 2022, they presented to the Defense Innovation Unit, the Pentagon's tech acceleration division. It did not go well. "It was the tailish end of COVID," said Pitt. "So, it was a video presentation, and we explained 'we're here to make low-cost missiles.' They asked: 'Are you going to compete with Lockheed?' We said yes, and it was not a little laugh. There were a good number of people laughing."

The skepticism reflected the moment. In venture capital, 2022 was the bizarre era of peak metaverse enthusiasm. Defense tech was a novelty, largely viewed as controversial or ignored. Anduril, defense tech's greatest hardware success, had a skyrocketing 2022, but there was little sense that its breakthrough was imitable.

The three founders are candid about how precarious the very beginning was. Hargis, who has a forceful warmth and natural deadpan—his father, he says, worked at a nuclear power plant "like Homer Simpson"—does not hold back. "I was using my retirement to fund my living," he said. "I was messaging these guys: 'At what point would they have to call it quits if we didn't make anything?'"

In their first investment contracts, the founders promised to return the money if they could not raise $1 million. They ultimately cleared that bar through a SpaceX tender offer, in which Hargis and another employee sold shares and invested the proceeds in the company. Then, in April 2023, Lavrock, a Washington, D.C.-based venture firm, led a $5.6 million seed round—a shoestring sum for an aspiring missile-maker, but enough. The founders went into the Mojave Desert, near Edwards Air Force Base, to build their first rocket motor.

"The early, early days in the desert were threadbare," remembers Kreitz, a history buff keenly aware of the lines between past and present. "We all wore Tyvek suits and respirators, helped make propellant and assembled parts, and we were getting hardware built alongside the engineers and techs."

The work took months, and funds dwindled. Soon came a Hail Mary: the founders hired a high-speed cameraman to film a static-fire test—a ground test in which the motor fires while anchored in place. The resulting video—of a rumbling, firing motor—drew the attention of Andreessen Horowitz, which quickly led an $8.6 million seed extension in Castelion.

The race then truly began. Building a defense company is an amalgamation of tiny precisions, from the rigmarole of clearances to establishing processes for classified information. The endgame for any defense startup is selling to the Department of War, a bureaucratic behemoth with labyrinthine fiefdoms, and ordinary startup challenges—office setup, engineering plans—are magnified by the specifics of hardware and defense.

"What mattered most for us then was finding a customer who was willing to try something new and was willing to be different in how they approach the acquisition cycle," said Pitt, who is wryly persuasive and served as a SpaceX spokesperson in his 20s. "[The Navy] gave us that initial contract, $3 million. We did what we said we would for $3 million. That led to a $15 million contract, and it all went from there."

Castelion now holds a program of record with the U.S. Navy, and to its founders the rise has happened at a dizzying pace. Headcount grew from 159 employees at the end of 2025 to more than 400. The tougher truth is that, from here, the company must maintain that pace or go even faster. "Can you make enough? Are they cheap enough that the country can buy that many? And can you do that in a time frame of relevance?" said Pitt. "That's the bar."

Much of that scaling will take place at Castelion's Project Ranger site in Rio Rancho, N.M., where some residents have reportedly balked at a missile factory moving in. The company has held regular community meetings, worked extensively with local authorities, and its chief engineer has moved to town.

"This is going to be the defining production site for Castelion," said Kreitz. "We, the three of us, will be here all the time. All our senior leadership is going to be there all the time. These are going to be our friends and colleagues. They're going to be working there. They're going to be living there. We want this to be safe. We want this to be done correctly."

How quickly Project Ranger comes online, and whether output climbs toward the thousands of missiles a year the company envisions, is likely to be one of the clearest measures of whether Castelion can clear that bar.

There is, ultimately, a back-to-the-future quality to Castelion's efforts. The company is careful not to attack the modern defense primes—the Pentagon, after all, has expressed a great need for both high-end and low-cost hypersonics. But its reference points reach further into the past, to the Apollo era, which produced humanity's first moon landings in roughly a decade.

"In the Apollo era, the country used to build hardware like there was no tomorrow, test it, fly it, iterate and do it all again," said Hargis.

In his telling, that approach is partly how other nations moved ahead of the U.S. in the hypersonic race: "The way it was done was with an iterative approach and a lot of testing. America has moved away from that iterative approach over the past decades, and we're now reaping the expected benefits from that."

'We're in a Sprint'

Castelion could still fall short. The company currently holds more than $500 million in U.S. government contracts, but it operates in big-swing-or-bust territory. In some sense, it is running a reverse Anduril: the $61 billion defense tech giant started with sentry towers and built out a massive suite of products over time and successive deals, while Castelion barreled straight into one massive, intractable problem. If the company is right, it will be very right; if it is wrong, very wrong. To Pitt, the gamble is worth it.

"The pure existence of Castelion as a credible threat to the primes significantly benefits the country, because behavior has to change to deal with a competitive entrant that's offering services at a significantly lower cost," he said.

The founders are acutely aware of the harshest underlying reality of their work: they are building lethal weapons capable of destroying military targets and infrastructure, and of killing. "You have to think about it and confront the fact that one day it can be used for its intended purpose," said Kreitz. Still, he added, "When we say 'peace through deterrence,' when we say 'deterrence works,' we really do mean that."

Hargis pushed back against a common criticism—that building weapons inherently facilitates war. "If you want to deter someone's actions, they have to feel there's a consequence to those actions," he said. "Hypersonics give the country that ability and a level between economic sanctions and nuclear war, which no one wants to happen, ever."

As an uneasy post-Cold War equilibrium morphs into a more geopolitically volatile world, the question hanging over Castelion—and the entire U.S. defense industry—is utterly urgent: What will the next major conflict look like, and will the country be ready?

"We're in a sprint," Kreitz said. "We're going as fast as humanly possible. I want to know how much time we have… We have to act as if it's tomorrow."

This story was originally featured on Fortune.com.