Shield AI

Shield AI builds Hivemind, an AI pilot that lets drones and crewed aircraft fly, navigate, and make tactical decisions on their own when GPS and communications links are jammed or denied. Founded in 2015 by Brandon Tseng, Ryan Tseng, and Andrew Reiter, the company sells autonomy as software rather than a platform, deploying Hivemind on its own V-BAT vertical-takeoff UAS as well as on third-party drones and fighter jets. Hivemind has flown V-BAT in combat in Ukraine and been validated on an F-16 via the X-62A test jet, and in June 2026 the U.S. Air Force awarded Shield AI a production contract to provide Hivemind mission autonomy for the Collaborative Combat Aircraft program, flying on Anduril's YFQ-44A. Shield AI has raised roughly $2.49B total, most recently a Series G at a $12.7B valuation, announced alongside its acquisition of simulation specialist Aechelon Technology.

It is the clearest example of autonomy sold as software: an AI pilot that ports across aircraft and keeps flying when GPS and comms are denied, the hardest case for autonomous flight.

shield.ai Drones & UAS Autonomy

United Statesprivate

At a glance

Total raised

$2.5B

last round series g

Headcount

1k-5k

approximate · public range

Headquarters

San Diego

United States

Founded

2015

11 years active

Status

Private

Drones & UAS Autonomy

Latest valuation

For subscribers

Subscribe to see the full profile

Including Funding rounds, Bull / Bear thesis, Stock + earnings, Roster changes, Patents, News, and Open roles.

Already subscribed? Sign in →

AI bull / bear

Bull case

Shield AI is the purest expression of the sector's best thesis — autonomy sold as software — and the U.S. Air Force just validated it. In June 2026 the company won a production contract to supply Hivemind mission autonomy for the Collaborative Combat Aircraft program, with its AI pilot already flying aboard Anduril's YFQ-44A: the airframe prime builds the jet, Shield AI supplies the brain. That is the position every software company wants in a hardware war. Hivemind is combat-proven in the hardest conditions — V-BATs flying 130+ sorties in Ukraine's jammed, GPS-denied airspace, including autonomously locating a Russian Buk air-defense system — and has been validated on an F-16 via the X-62A dogfighting trials, so the porting-across-airframes claim is demonstrated, not aspirational. The March 2026 Series G — $12.7B valuation, up roughly 140% in a year, with ~$2.49B raised — gives it the balance sheet to prosecute the X-BAT, a runway-independent VTOL combat jet aimed at a CCA market the Pentagon plans to buy in the thousands. Defense autonomy budgets are the most durable spending line in Washington, and Shield AI is the only pure-play with production-contract proof that its software wins on other companies' aircraft.

Bear case

The valuation prices Shield AI as a software company, but the revenue looks like a drone maker's: Hivemind was roughly 30% of revenue as of early 2025, meaning the majority still comes from selling V-BAT aircraft — hardware, with hardware margins and lumpy program timing. The X-BAT ambition makes the structural tension worse: by building its own Group 5 combat jet, Shield AI turns Anduril, General Atomics, and every airframe prime — the very customers the Hivemind story depends on — into direct competitors, and those partners can commission or build alternative autonomy rather than arm a rival. Anduril in particular has every incentive and the engineering depth to replace Hivemind on the YFQ-44A over time. Execution risk is visible: X-BAT's first flight, once floated for 2025, is now targeted for late 2026, and a clean-sheet VTOL fighter is a multi-billion-dollar development that companies with far deeper aerospace benches have fumbled. Meanwhile a $12.7B mark — up 140% in twelve months on a company that has raised ~$2.5B and has not demonstrated profitability — leaves no room for a CCA increment loss, an X-BAT slip, or a Ukraine-driven procurement shift toward cheap attritable drones over premium autonomy. Defense revenue is real but budget-cyclical, and program cancellation is a single-customer risk equity markets persistently underprice.

Generated by editorial-sourced · July 1, 2026Refreshes quarterly · see disclosures