Shield AI

Shield AI was founded in San Diego in 2015 by brothers Brandon and Ryan Tseng along with Andrew Reiter, and its founding idea still runs through everything it builds: aircraft that keep flying and completing missions after GPS and radio links go dark. That idea lives in Hivemind, the company's core autonomy software stack. Hivemind is not a single drone, it is the pilot software that Shield AI now puts inside several different airframes, letting them navigate, avoid obstacles, and coordinate with each other without a human joystick or a satellite signal to lean on. The company markets Hivemind as proven in contested and communications-denied environments, which is the scenario its military customers actually worry about. The clearest expression of that software is the V-BAT, also designated the MQ-35A, a ducted-fan VTOL aircraft that takes off and lands without a runway or catapult and can fly for more than twelve hours on a single tank of heavy fuel.

It came into the company through the 2021 acquisition of Martin UAV, the small Texas firm that built the original airframe, and it has since become Shield AI's most fielded product. The Army picked it for the Future Tactical Unmanned Aircraft System competition to replace the aging RQ-7 Shadow, teaming Shield AI with Northrop Grumman on that bid. The Navy selected V-BAT to compete for up to 800 million dollars in contractor-owned, contractor-operated ISR task orders, and the Coast Guard awarded a separate 198 million dollar contract for maritime surveillance. Poland has bought V-BAT for its navy, and the aircraft has flown ISR missions for the US Navy during exercises like UNITAS. In October 2025 Shield AI unveiled a far more ambitious airframe, the X-BAT, an autonomous VTOL fighter jet designed to launch from ships or austere strips without a runway, carry weapons internally or externally, and fly more than 2,000 nautical miles.

GE Aerospace signed on in November 2025 to supply the F110 engine with thrust vectoring, and first flights are expected in 2026. Alongside the two larger aircraft, Shield AI also builds the Nova family of small, rucksack-portable quadcopters used for room clearing and close reconnaissance, born out of the Heron Systems acquisition that brought Shield AI deep reinforcement learning talent from the DARPA AlphaDogfight trials. The company has grown through acquisition on the sensing and simulation side too, buying Sentient Vision Systems for wide-area computer vision and, in early 2026, Aechelon Technology for visual simulation used in training its own autonomy models. Shield AI stayed private through all of this, raising a Series G in 2026 that pushed its valuation past 12 billion dollars, and it now counts US Special Operations Command, the Air Force, the Marine Corps, and a growing list of allied militaries as customers.