Most defence-tech companies sell you a machine. Shield AI is really selling you the pilot: an artificial one, in software, that can fly an aircraft when the GPS is jammed, the communications are cut, and no human is at the controls. That focus, autonomy for the hardest, most contested environments, has turned it into a $12.7 billion company and one of the most closely watched names in the field. This is the second entry in our Company Profiles library of ones to watch.
As with every profile in this section, this is a business and investment piece on a company in a serious, ethically contested industry. We cover the money, the record and the debate at the level already public, and we leave the operational detail of any weapon well alone.
The founders: a Navy SEAL and his brother
Shield AI was founded in 2015 by Brandon Tseng, a former US Navy SEAL officer, with his brother Ryan Tseng and Andrew Reiter. The origin story matters here more than most, because Brandon Tseng’s pitch comes from having been the person the technology is meant to protect. His argument, made repeatedly in interviews, is that autonomy keeps operators out of the deadliest situations, and that the West should build these systems before adversaries who will build them anyway. He has been candid that this is uncomfortable territory, framing the work around the “AI weapon no one wants” but many believe is coming regardless.
What Shield AI actually builds
The crown jewel is Hivemind, the company’s AI pilot software. It is designed to fly aircraft autonomously in GPS-denied and communications-denied environments, execute missions without a human operator, and coordinate multiple aircraft as a team. The flagship hardware it flies is the V-BAT, a vertical-take-off reconnaissance drone that needs no runway, and newer platforms like the Nova 2. Crucially, Hivemind is meant to be portable: in February 2026 an Anduril YFQ-44A Fury flew switching between Shield AI’s Hivemind and Anduril’s own Lattice within a single flight, a sign that the AI pilot is being positioned as a layer that can run on other people’s aircraft.
The battlefield record is a genuine differentiator. Shield AI says its Hivemind-powered V-BAT has logged more than 130 combat sorties in Ukraine since June 2024, in heavily contested electronic-warfare conditions, the exact environment where ordinary drones that rely on GPS and a radio link fall out of the sky. Real combat use, rather than a demo reel, is the strongest thing on Shield AI’s CV.
| Shield AI at a glance | Detail |
|---|---|
| Founders | Brandon Tseng (ex-Navy SEAL), Ryan Tseng, Andrew Reiter |
| Founded | 2015 |
| Latest valuation | $12.7bn (Series G, up ~140% from $5.3bn) |
| 2026 revenue (projected) | at least $540m, 80%+ growth |
| Core product | Hivemind AI pilot, V-BAT drone |
| Combat use | 130+ V-BAT sorties in Ukraine since June 2024 |
The ethics, said plainly
The same unavoidable question sits under Shield AI as under the whole sector: is it acceptable to hand flight, and potentially targeting, to software with reduced human oversight? Shield AI’s answer leans on the human-out-of-harm’s-way argument and on deterrence. The counter-argument is that autonomy in weapons, once normalised, is very hard to contain, and that “a human is still in the loop” is a promise that market pressure tends to erode. This profile does not resolve that for you; it flags it as the load-bearing debate the whole valuation assumes will be settled in the industry’s favour.
The investment case, even-handed
Shield AI is private, so this is about how investors and analysts frame it, and the frame is the same boom-or-bubble split that hangs over the sector. (None of this is investment advice.)
The bull case. Shield AI’s backers point to something rare in this hype cycle: a product with a real combat record, revenue projected above $540 million in 2026 growing more than 80% a year, and an AI pilot that can run on multiple airframes, which makes it a software play rather than a single-drone company. The $1.5 billion Series G plus $500 million in preferred equity, at a valuation up around 140% in a year, is the market betting Hivemind becomes the default autonomy layer for Western unmanned aircraft.
The bear case, deferred to the sceptics. The doubts are the sector’s doubts, voiced by the financial press. Fortune has argued the defence-tech boom is turning into a bubble, with startups fetching 17 to 50 times revenue; Shield AI at $12.7 billion on roughly $540 million of projected sales sits toward the friendlier end of that range at around 23 times, but that is still a rich multiple that assumes years of flawless execution. Analysts’ broader worry applies squarely: capital is arriving faster than governments can convert new technology into recurring production orders, and a company priced on future dominance is exposed if procurement is slower or bumpier than the deck implies.
The neutral read. Shield AI has the most convincing evidence of real-world usefulness of anyone in this section, which is a genuine point in its favour, and a valuation that still requires the growth story to keep delivering. Watch whether the revenue and combat-usage numbers keep climbing, whether Hivemind wins deals to fly aircraft it did not build, and whether the US and allied procurement machine turns interest into signed, recurring contracts. Evidence of use is the bull’s best card; the pace of government orders is the bear’s.
What this means
Shield AI is the purest bet in this section on the idea that the valuable thing in autonomous warfare is the software brain, not the airframe, and it has more real combat evidence behind that bet than its rivals. It is also, like all of them, priced for a future of dominance in a field whose ethics remain unresolved and whose government customers move slowly. Watch the sorties, the revenue and the contracts, and treat the valuation as a claim on the future rather than a description of the present. (None of this is investment advice; do your own research.)
Related on Top Tool Stack: Anduril: the $61bn arsenal · Helsing: Europe’s answer
Did you know: Shield AI’s V-BAT has reportedly flown more than 130 combat sorties in Ukraine in exactly the jammed, GPS-denied conditions that knock ordinary drones out of the sky. Real combat use, not a demo, is its strongest asset.