After a public standoff with Anthropic over how its AI models can be used, the Pentagon has signed contracts with Nvidia, Microsoft, and Amazon Web Services to deploy artificial intelligence on classified networks.
The deals mark a deliberate push by the Department of Defense to spread its AI work across multiple vendors rather than anchor to any single provider. Nvidia supplies the GPU hardware that runs AI models; Microsoft and AWS provide the cloud infrastructure and software layers on top.
The Anthropic dispute is the key context. Anthropic placed restrictions on how Claude and its other models could be used in military contexts - reasonable for a safety-focused AI lab, but a direct problem for a government agency that needs full operational control over classified systems. Defense analysts cannot work with tools whose capabilities are constrained by a vendor's acceptable-use policies. So the DOD moved.
Nvidia, Microsoft, and AWS are infrastructure-first companies. They're far less likely to attach operational restrictions to compute contracts than a model-focused AI lab is to a software license. That distinction matters when the buyer is running operations it cannot describe to a vendor.
The broader signal for the AI industry: companies with strong ethical-use policies face a real structural tension when pursuing government work. Anthropic's position on military use may be entirely principled, but it opened a gap that three major competitors are now filling. AI labs that want defense contracts will face pressure to offer either policy flexibility or infrastructure-layer products where usage restrictions don't apply at all.
This isn't the first time national security requirements have reshaped how tech vendors structure their terms - and it won't be the last.