Anthropic may keep supplying Claude to the NSA despite being flagged as a supply chain risk by the Pentagon

The real story isn't the security designation itself but what it reveals about the leverage dynamic: Anthropic's position is strong precisely because the government has no adequate substitute that runs on available hardware, which means the Pentagon's own risk framework is being subordinated to a supply constraint it helped create by prioritizing Nvidia's newest chips across agencies.

This is largely disconnected from recent activity in our archive, as we have no prior coverage to anchor it to. It belongs, however, to a broader pattern visible across the industry: government AI procurement is increasingly driven by infrastructure bottlenecks rather than vendor evaluation on merit or security posture alone. The chip scarcity angle connects to ongoing coverage elsewhere of Nvidia's Grace Blackwell allocation constraints shaping which AI deployments are even feasible at the federal level.

Watch whether the final contract language restoring or permanently removing the 'any lawful use' clause gets disclosed through procurement records in the next 90 days. If it stays out, that signals Anthropic accepted a meaningful constraint; if it quietly returns, the security designation was effectively toothless.