OpenAI secured a $950 million contract with the Department of Defense, stepping into a vacuum created when Anthropic was blacklisted as a "supply chain risk" by the Pentagon. The deal, signed in the opening days of the Iran conflict, is reshaping the emerging AI defense industrial base along political lines [1].

The contract covers AI prototypes for "warfighting and enterprise" applications, expanding on a smaller $200 million deal from July 2025 that also included Google and xAI. The timing was striking: OpenAI's deal was finalized hours after Anthropic's blacklisting, a sequence that Sam Altman himself acknowledged "looked opportunistic and sloppy".

Altman added that Fourth Amendment protections should apply to military AI systems — a notable caveat that did nothing to slow the contract signing. OpenAI, which began as a nonprofit committed to beneficial AI, is now the Pentagon's primary commercial AI partner during an active war.

Anthropic has responded by taking the Department of Defense to court, alleging the blacklisting constituted an improper "supply chain" designation designed to exclude the company from defense contracting. The lawsuit argues the Pentagon's decision was politically motivated rather than based on legitimate security concerns.

The broader implications extend beyond two companies. The AI defense industrial base is sorting itself into camps: companies willing to build military AI (OpenAI, Palantir, Anduril) and those being excluded or self-excluding (Anthropic, for now). The $950 million contract establishes the commercial scale of military AI procurement during wartime.

The contract terms remain partially classified. What is public: OpenAI is now building AI for the Pentagon while the Pentagon uses AI in Iran.

-- DAVID CHEN, San Francisco