The last time anyone flew a fighter jet without a human in the cockpit was a test. For Shield AI, its AI pilot Hivemind has been doing it in combat since 2018 — the longest uninterrupted combat run of any autonomous flight system in the world. On Thursday, that track record attracted $2 billion in new capital: $1.5 billion in Series G funding at a $12.7 billion post-money valuation, led by Advent International with JPMorgan Chase co-leading, plus $500 million in fixed-return preferred equity from Blackstone with a $250 million delayed-draw facility committed on top, according to Reuters and Bloomberg. The San Diego-based defense startup is now valued at $12.7 billion, up from $5.3 billion a year ago. But the number that matters isn't the headline valuation — it's the Air Force flight test campaign currently running Hivemind aboard the Anduril YFQ-44A.
Shield AI was founded in 2015 by Ryan Tseng, his brother Brandon Tseng — a former Navy SEAL — and Andrew Reiter. The core product is Hivemind, an autonomy stack that runs on a range of aircraft and ground vehicles, navigating by onboard sensors rather than GPS or remote communications links. Unlike a conventional autopilot, Hivemind is designed for disconnected, degraded, intermittent, or low-bandwidth environments, the conditions that make drones useless when they lose contact with their operators. The company claims Hivemind has now piloted 26 classes of vehicles, including F-16s, jet-powered UAVs, helicopters, drone boats, and ground vehicles. The product has been used continuously in combat since 2018, according to Tech Funding News, making it what Shield AI calls the world's first combat-deployed AI pilot.
That deployment record is what drew the Air Force's attention. Shield AI was recently selected as a mission autonomy provider for the U.S. Air Force's Collaborative Combat Aircraft program — the service's effort to pair autonomous drones with crewed fighters — and is actively conducting a flight test campaign of Hivemind aboard Anduril's YFQ-44A CCA platform. It is a direct answer to a question the Pentagon has been asking for years: at what autonomy level will the military trust an AI to fly into combat alongside a human pilot? The answer, it turns out, is sooner than many expected.
The other half of Thursday's announcement closes a loop that makes that question answerable at scale. Shield AI is acquiring Aechelon Technology, a simulation software company from Sagewind Capital, for an undisclosed sum, according to a Shield AI press release. Aechelon builds the high-fidelity synthetic reality environments used to train pilots and test autonomous systems before they fly live — including the Pentagon's Joint Simulation Environment, the virtual combat range used to evaluate aircraft and tactics against realistic battlefield threats. CEO Gary Steele called it directly: the acquisition will accelerate Hivemind's development, particularly in simulation, and feed real-world operational data back into the model. Simulation trains the AI pilot. Combat refines it. The acquisition is expected to close pending regulatory approval, with Aechelon continuing to operate independently under co-founder Ignacio Sanz-Pastor.
Sacra estimates Shield AI hit approximately $300 million in revenue for the year ending March 2025, up from $267 million the prior year — a 64 percent year-over-year gain. The company is targeting 70 to 100 percent annual revenue growth to reach $1 billion by the year ending March 2028, with Hivemind software accounting for roughly 30 percent of revenue currently and a goal of 50 percent software revenue by 2028. That revenue trajectory is the real basis for the valuation step-up: $5.3 billion in March 2025 to $12.7 billion now is a contract momentum story, not a multiple-expansion story.
International sales are a significant part of that momentum. The company estimates over half of its 2025 business is international, with new V-BAT contracts in 2025 alone including the Romanian Navy at approximately $30 million, Japan's naval forces, Greece, Canada, the Netherlands, Ukraine, and Egypt. Ukraine's Ministry of Defence formally named Shield AI a verified business partner in August 2025, allowing it to compete for state procurement contracts. The V-BAT — Shield AI's flagship unmanned aircraft — has flown more than 130 sorties in Ukraine, making it, according to the company, the only long-endurance ISR and strike platform to penetrate both Ukrainian and Russian GPS and communications jammers.
The next chapter is the X-BAT, an autonomous VTOL fighter jet unveiled in October 2025. Shield AI has been developing the aircraft for roughly 18 months, targeting first vertical takeoff and landing trials in fall 2026, full-system flight testing and operational validation in 2028, and production beginning in 2029. Sacra estimates it will take roughly $1 billion to reach operational capability. The design requires no runway, can be deployed from ships, and is fully piloted by Hivemind. Range is cited at more than 2,000 nautical miles with full payload — putting contested airspace hundreds of miles from a carrier within reach without a crewed aircraft in sight.
The comparison to traditional defense contractors is instructive. Lockheed Martin and Boeing work with the Pentagon through cost-plus contracts, starting development only after a request for proposal. Shield AI front-loads its own R&D, betting on a product before a contract exists, then presenting a finished platform to the DOD and allied forces. It's a tech-company model applied to defense, and it's producing systems that are already flying in active combat zones while much of the traditional industrial base is still writing white papers about autonomy. The $2 billion infusion will fund X-BAT development, Aechelon integration, and international expansion — and it will test whether the model can scale from impressive deployments to a durable defense prime.
