Brandon Tseng and the veteran founders rebuilding America's arsenal

Brandon Tseng spent part of his twenties as a Navy SEAL, doing one of the most dangerous jobs in modern warfare: going into buildings without knowing what was inside. Today he is the co-founder and president of Shield AI, a San Diego company that builds artificial intelligence to fly military aircraft without a human pilot. A funding round in March 2026 valued the company at $12.7 billion. Forbes estimates that Tseng and his brother Ryan each hold a stake worth about $400 million.

For Tseng, building a defense company was not a change of career. It was the same mission with different tools. He also stands near the front of a new class of American entrepreneur: veterans who came home from the wars that followed September 11, looked at the technology they had been issued and decided they could build something better.

The room nobody wants to enter

Tseng graduated from the U.S. Naval Academy with an engineering degree, served as a surface warfare officer, then made it through SEAL training and deployed overseas, including to Afghanistan. The experience that shaped him most was brutally simple. Troops were sent into buildings with no idea whether they would find fighters, explosives or a family eating dinner. People were hurt and killed for lack of information that a machine could have gathered.

Small drones existed, but they depended on GPS and on a steady radio link to a human operator. Inside a concrete building, or against an enemy with jamming equipment, they were of little use. What the military needed, he concluded, was a machine that could find its own way.

After leaving the Navy, Tseng went to Harvard Business School and kept turning the problem over. In 2015 he teamed up with his older brother Ryan, who had already founded and sold a wireless-charging startup, and the engineer Andrew Reiter. They founded Shield AI with a mission statement that sounded ambitious for three people in San Diego: to protect service members and civilians with intelligent systems.

Selling to the hardest customer on earth

The timing looked poor. Much of Silicon Valley wanted little to do with defense. In 2018 Google employees protested the company's work on a Pentagon project, and Google let the contract lapse. The Pentagon's procurement system, for its part, was built around giant contractors and programs that run for a decade, and had a reputation for starving startups before a contract arrived.

Shield AI's first product was Nova, a small quadcopter that could fly into a building and map it with no GPS, no pilot and no communications link, showing troops what was waiting before anyone stepped through the door. It was deployed with U.S. forces in the Middle East in 2018. The tool Tseng had wished for in Afghanistan was now going through the door first.

Hivemind

Nova was never really the product. The product was the software inside it, which the company calls Hivemind: an AI pilot designed to fly an aircraft and carry out a mission without GPS, communications or a human at the controls.

The ambition grew quickly. In 2021 Shield AI bought Martin UAV, maker of the V-BAT, a drone that takes off and lands vertically on its tail and needs no runway. The same year it bought Heron Systems, whose AI agent had beaten an experienced Air Force F-16 pilot five to zero in a simulated dogfight run by DARPA in 2020. Hivemind has since flown real aircraft, including a modified F-16 test jet.

Then came a proving ground nobody wanted. In Ukraine, electronic warfare has knocked cheap drones out of the sky in large numbers by jamming GPS and severing control links, the same failure Tseng had seen years earlier. Shield AI says its V-BATs have flown more than 130 sorties there alongside Ukrainian forces, and that the aircraft kept operating through both Russian and Ukrainian jamming. For militaries watching, it was evidence that autonomy is moving from a research project to a requirement.

Investors followed. In March 2025 the company raised $240 million at a $5.3 billion valuation, with the defense contractor L3Harris and South Korea's Hanwha Aerospace among the backers. The same month it brought in Gary Steele, a veteran software executive, as chief executive, with Brandon Tseng continuing as president. In October 2025 it unveiled its most audacious product yet: X-BAT, an AI-piloted, jet-powered aircraft that takes off vertically and needs no runway. The March 2026 round more than doubled the valuation again.

The new arsenal

Tseng is not alone. Anduril, founded by the Oculus creator Palmer Luckey and staffed heavily with veterans and alumni of Palantir, is valued in the tens of billions of dollars. Former fighter pilots, Army Rangers, Marine officers and intelligence analysts have founded companies building autonomous boats, counter-drone systems, battlefield software and satellites. Venture capital, which avoided the sector a decade ago, has put tens of billions of dollars into defense technology since Russia's invasion of Ukraine and rising tension with China ended the assumption that great-power conflict was history.

These founders share an edge that is hard to buy. They were the end user. They know what it means when equipment fails at the worst moment, and they know how the military actually makes decisions, which in this market is worth as much as any algorithm. Their pitch to Washington is blunt. The United States will struggle to out-build China in ships and missiles, they argue, but it can compete with software, and with large numbers of cheap, smart, expendable machines rather than a handful of exquisite and irreplaceable ones.

If they are right, the rewards will be large. The Pentagon's annual budget exceeds $800 billion, and even a small shift away from the traditional prime contractors would be one of the biggest market openings in the world.

The hard questions

The shift has critics. Arms-control advocates warn that machines able to find and strike targets on their own lower the threshold for war and blur accountability when something goes wrong. Tseng has said publicly that decisions about lethal force should stay with humans, and that the point of autonomy is to let those humans keep operating when communications are jammed. He also makes the deterrence case: the surest way to prevent a war with a rival power, in his view, is to make it unwinnable.

There are business risks too. Defense budgets are political, and a startup selling aircraft must eventually prove it can manufacture at scale and operate safely, not only demonstrate well. Shield AI has had setbacks. In 2024 a Navy service member was seriously injured in an accident involving a V-BAT, an episode Forbes later reported had cost the company momentum. In 2025 China added the company to its list of unreliable entities.

Mission, continued

What stands out about founders like Tseng is how little of the story they tell is about money. The stake is real, and so is the valuation, but the account he gives in public is about the room nobody wanted to enter, and about the conviction that the next generation of Americans in uniform should not be sent into a fight with worse technology than the enemy has.

The SEAL teams taught Brandon Tseng to work the problem in front of him, however unlikely the solution looked. He is still doing that. The problem has become larger, and so has what rides on solving it.

Sources

Forbes: Battlefield to boardroom, the veteran founders riding defense tech's boom (August 3, 2026)

TechCrunch: Shield AI raises $240M at a $5.3B valuation (March 6, 2025)

Shield AI: more than 130 V-BAT sorties in Ukraine

TIME: Brandon Tseng, the 100 most influential people in AI 2025

Forbes: How a grisly injury threw Shield AI off course (May 13, 2025)

Wikipedia: Shield AI