Google Signs Pentagon AI Deal Despite 600-Employee Revolt

Google has formally signed a classified agreement with the U.S. Department of Defense that gives the Pentagon access to its Gemini AI systems on classified networks, sources told NBC News and the Washington Post. Fortune pegs the contract’s value at roughly $200 million. The arrangement was announced Friday alongside similar pacts with six other tech giants, OpenAI, Microsoft, Amazon Web Services, Nvidia, SpaceX and Reflection, under a sweeping “War Department” initiative described by the Pentagon as indispensable to national security. Google kept the specific scope under wraps, citing classification, but employees quickly learned it includes a broad “all lawful use” clause that covers military applications.

More than 600 Googlers signed an open letter to CEO Sundar Pichai urging him to “refuse to allow the DoD to use its AI models for classified work.” The letter, first reported by the Washington Post and amplified by Semafor and Common Dreams, argues that Google technology will now be used for “coercive or violent purposes” and that human lives are already at risk from AI misuse. The protest echoes the 2018 Project Maven fight, when 4,000 employees petitioned and dozens resigned over a Pentagon drone-imaging contract Google ultimately abandoned. This time, however, leadership appears unmoved.

Business Insider and Fortune both report that Google’s internal power balance has tilted decisively toward management. Unlike the Maven era, executives now stress that competitors have already accepted identical Pentagon terms; refusing would expose Google to legal and business risks while ceding ground to rivals. Internal responses to employee questions assert the company “proudly” works with the U.S. military, a stance circulated on company message boards and leaked to Threads. The firm also sanded down dissent channels: Fortune notes that employee leverage has “waned,” while Business Insider characterizes the dynamic as an overt “war on employee anti-war demands.”

By joining the Pentagon’s approved-vendor list, Google gains privileged access to classified data streams and future budget lines that could dwarf the initial $200 million. OpenAI, xAI and others have already taken the same step, leaving Anthropic as the notable hold-out after it balked at domestic surveillance clauses. The collective shift signals that the industry’s self-imposed red lines against military AI are eroding fast. For developers inside Google, the message is blunt: refusal is no longer an option that scales.

Google’s move tightens the coupling between frontier models and defense applications, a precedent competitors will struggle to ignore. Enterprise clients in sensitive sectors, health care, finance, civic tech, now confront a clearer divide: buy from vendors that also serve the Pentagon, or seek smaller providers that haven’t. Open-source alternatives may gain traction among privacy-conscious users, yet they lack the classified-level security clearances that Pentagon contracts confer. The net effect is a two-tier market: defense-approved giants and everyone else.

Pichai is unlikely to reverse course; the contract is signed and the Pentagon has publicly named Google as a strategic partner. Employee organizers must decide whether to escalate, legal filings, coordinated resignations, or public whistle-blowing, or accept that internal protest has hit a ceiling. Congressional oversight hearings could surface if lawmakers probe how “all lawful use” squares with stated AI safety commitments, but no dates have been scheduled. For now, Google marches forward, and its critics inside the company watch from the rear.