A look at obscure defense startup Rebellion Defense, which works to create AI and infosec applications for the US government and is backed by Eric Schmidt
Jonathan Guyer / American Prospect :
Context & Ripple Effects
This American Prospect profile lands between two bookends in Rebellion Defense's arc: Eric Schmidt's Pentagon advisory board had just published AI principles for the US military the year before, and eight months later the company would raise $150M led by Insight and Venrock at a $1B pre-money valuation. The piece is the first sustained look at how Schmidt operates on both sides of the defense-AI table — advising the Pentagon while bankrolling vendors like Rebellion that sell to it.
The profile also captures the sector's founding premise: AI and infosec applications pitched directly at US government buyers, with founder prestige and Schmidt's name doing the early credibility work that contracts would later have to validate — a bet that Rebellion's later struggles would test hard.
First-order effects
- The profile puts Rebellion's dual identity under scrutiny: a Schmidt-backed startup selling AI and infosec to the US government while Schmidt chairs the Pentagon advisory board shaping military AI policy — an immediate optics and conflict-of-interest problem for both Rebellion and the Defense Innovation Board.
- For Rebellion itself, the coverage functions as early due diligence in public: its government-first go-to-market and its dependence on high-profile backing become the story investors and DOD buyers read before the $150M raise.
Second-order effects
- Rebellion's subsequent collapse — failing to win a US DOD contract, losing its founders, and shuttering its UK arm — shows what happens when the credibility-from-backers model meets a customer that doesn't buy: the same investor prestige that powered the $1B valuation could not substitute for a contract.
- Schmidt's response is portfolio diversification rather than retreat: he backs adjacent bets like Ishtari, which uses machine learning to assemble and test war machines virtually, spreading defense-AI exposure across multiple vehicles instead of concentrating on one vendor.
Third-order effects
- The pattern points toward a defense-AI market structured around single-buyer risk: startups that cannot convert advisory-board influence or venture backing into actual DOD contracts fail regardless of valuation, making procurement outcomes the sector's real sorting mechanism.
- Schmidt's arc — from advisory-board principles to a portfolio of defense-AI investments — models the individual technocrat-investor as the connective tissue between Pentagon policy and the vendor ecosystem, a structure later players like Arkenstone Defense's back-office software for defense startups are now building around.
The trend: Defense AI is consolidating into a Schmidt-style model where Silicon Valley capital and Pentagon advisory influence set the agenda, but DOD procurement — not investor enthusiasm — decides which startups survive.