Shield AI Hits $2.8B Valuation in Defense Tech's Biggest Raise
Shield AI's $2B raise marks defense tech's largest funding round. While military AI dominates headlines, businesses seeking efficiency should explore the best AI tools for presentations to streamline their own operations.
Shield AI has closed a $500 million Series F that values the San Diego defense startup at $2.8 billion, according to two people familiar with the deal. The round, led by Riot Ventures and including participation from ARK Invest and Snowpoint Ventures, marks the largest single raise for an autonomous defense technology company this year.
The company isn't just selling drones. It's selling autonomous decision-making software that lets small quadcopters operate without GPS, without radio links, and without human pilots calling every shot. That capability is already in active use. Shield AI confirmed its Hivemind software has flown "thousands of combat missions" across multiple theaters since 2022, though it declined to specify locations for operational security.
From Garage Startup to Pentagon Prime
Founders Brandon Tseng and Ryan Tseng built their first prototype in 2015 after Brandon, a former Navy SEAL, watched a teammate die in Afghanistan because they lacked aerial surveillance. The problem wasn't drone availability—it was that every available system required trained pilots, secure communications, and hours of setup.
Hivemind flips that model. The software runs onboard, uses computer vision to navigate, and coordinates swarms through mesh networking. A single operator can task a group of drones with area surveillance; the machines handle collision avoidance, search patterns, and target identification independently.
The approach has attracted $1.2 billion in total funding and, more critically, $800 million in active contracts including a $200 million Air Force deal signed in late 2024 for V-BAT drone production. The V-BAT—a vertical-takeoff aircraft that transitions to winged flight—can launch from a pickup truck bed and stay aloft for 11 hours.
The valuation gap with Anduril reflects different bets. Anduril builds complete hardware-software systems for fixed installations. Shield AI's pitch is software-first: put Hivemind on any airframe, from quadcopters to F-16s. The company demonstrated autonomous F-16 dogfighting in 2023 through a DARPA program, with the AI winning every engagement against a human pilot.
What $500 Million Actually Buys
Defense hardware scales slowly. The new capital targets manufacturing velocity—specifically, producing 1,000 V-BATs annually by 2027, up from roughly 150 in 2024.
But the bigger play is software licensing. Shield AI wants Hivemind on competitors' drones, on allied nations' aircraft, eventually on autonomous ships and ground vehicles. Each deployment feeds training data back to improve the core models.
"The moat isn't the airframe. It's the millions of hours of real-world flight data no simulation can replicate," Brandon Tseng told reporters in January. "Every mission in contested comms environments makes the next version harder to jam, harder to spoof."
That data advantage is why investors accepted terms that include no new board seats for this round, according to one source—founders kept control because the proprietary dataset can't be replicated with capital alone.
The Ethics Question Nobody's Answering
Autonomous weapons systems operate in a legal gray zone. The Pentagon's 2023 directive requires "appropriate levels of human judgment" for lethal decisions, but doesn't define what "appropriate" means. Shield AI states its current systems only conduct surveillance and targeting assistance, with humans retaining fire authority.
Still, the technology is designed for denial environments—exactly where communications break down and human oversight becomes theoretical. If a drone loses its link and continues a pre-authorized mission, who bears responsibility for mistaken strikes?
The company points to its published AI ethics framework and participation in DOD's Responsible AI program. Critics note those commitments are voluntary and lack enforcement mechanisms.
Congress hasn't clarified the rules. The 2024 National Defense Authorization Act directed a study on autonomous lethal systems, with findings due in September 2026. Until then, deployment decisions rest with combatant commanders and company policies.
What This Means for Tech's Defense Pivot
Shield AI's raise follows a clear pattern: Anduril at $14 billion, Helsing at $5.4 billion, Saronic at $4 billion. Venture capital is pouring into defense tech at rates unseen since the Cold War, driven by Pentagon procurement reform and geopolitical demand.
But the economics differ from consumer AI. These companies burn capital on hardware, regulatory compliance, and security clearances. They can't iterate weekly or A/B test in production. A failed drone flight in Ukraine costs $200,000 and potentially lives.
The bet is that autonomy becomes the defining military advantage of the next decade—cheaper than manned systems, faster to deploy, and harder to counter. If that thesis holds, Shield AI's $2.8 billion valuation looks conservative. If autonomous systems face international bans or catastrophic failures, the downside is total.
For now, the company is hiring 400 engineers and expanding its Dallas manufacturing facility. The Series F includes a secondary component allowing early employees to sell shares—rare in defense, where liquidity events are scarce.
Watch for two signals: whether Hivemind licensing deals materialize with NATO allies, and whether the Pentagon accelerates its "loyal wingman" program pairing autonomous drones with manned fighter jets. Both would validate the software-first strategy—and likely trigger a pre-IPO round before 2027.
---
Related Reading
- Harvey Hits $11B Valuation in Bet on Legal AI Over Models - OpenAI Sweetens Private Equity Pitch Amid Anthropic Rivalry - The Next Class War Is About Who Can Actually Use AI - OpenAI Hires Ex-Meta Exec to Lead Its First Ad Business - OpenAI's Data Center Shift Spooks Investors Before IPO