Anthropic sued the Department of Defense on March 9, 2026, contesting a supply chain security designation that would restrict its business operations. The same day, Google announced plans to provide AI agents to the Pentagon, and OpenAI's hardware division leader resigned following the company's Pentagon deal.
The simultaneous announcements mark a fracturing point in how major AI labs approach government contracts. Anthropic CEO Dario Amodei defended the lawsuit, saying the company had no alternative but to contest the designation in court. Anthropic maintains strict limits on government use cases, refusing to allow its models for domestic surveillance of U.S. citizens.
The legal challenge appears to resonate with Anthropic's customer base. Monday, March 9 became the company's largest single day for sign-ups ever, following the lawsuit announcement. The surge suggests commercial customers value clear boundaries on military applications.
Google's Pentagon deal moves in the opposite direction, integrating AI agents directly into military operations. The contrast between Anthropic's lawsuit and Google's Pentagon partnership illustrates diverging strategies among frontier AI companies navigating national security demands.
OpenAI faces internal friction over military work. The resignation of its hardware leader immediately after announcing Pentagon contracts signals employee resistance to defense partnerships at some AI labs. The departure adds to evidence that military contracts create organizational tension.
National security concerns around AI supply chains are forcing labs to choose between government revenue and ethical positioning. The DOD supply chain designation that Anthropic contests stems from security reviews of AI development ecosystems. These reviews aim to prevent foreign influence but create compliance burdens that fragment the industry.
The split creates two paths for AI labs: embrace military contracts like Google, or resist government restrictions like Anthropic. Labs pursuing both commercial and defense revenue may need separate divisions with different policies, complicating unified product development.
Cross-border AI collaboration faces new barriers as supply chain security reviews expand. International research partnerships and deployment across borders become harder when labs must prove supply chain integrity to multiple governments with conflicting requirements.