[DigitalToday reporter Hyunwoo Choo (추현우)] The U.S. Department of Defense (Department of War) has signed contracts with Nvidia, Microsoft, Amazon Web Services and Reflection AI and plans to deploy their AI technology and models on classified networks. TechCrunch reported on Thursday, citing the Defense Department, that the purpose is lawful use of AI.

The contract recipients include SpaceX, OpenAI, Google, Nvidia, Reflection AI, Microsoft and AWS. Impact Levels 6 and 7 cover secret and top secret work zones, and the models will be placed in environments containing sensitive information and intelligence and operational data.

The Defense Department selected multiple U.S. suppliers to avoid dependence on a specific vendor. It aims to broaden options by keeping both closed models and open-source models on the table.

Roles differ by company. Nvidia will provide open-source Nemotron models, and Reflection AI will supply additional open-weight systems. Google will provide Gemini models tailored to lawful government purposes. SpaceX is expected to handle infrastructure linked to xAI's Grok model, while Microsoft and AWS will continue cloud and infrastructure roles.

Internal use is already growing. The Defense Department's generative AI platform, GenAI.mil, surpassed 1.3 million users within 5 months of launch, and recorded tens of millions of prompts.

Anthropic was left off the list. Defense Secretary Pete Hegseth (피트 헤그세스) said in February that Anthropic had not removed restrictions on autonomous lethal weapons and large-scale domestic surveillance, defining it as a supply-chain risk. A federal court later blocked enforcement of the ban, and legal disputes are continuing.

OpenAI said it is maintaining 3 of its own safety principles under this contract as well. Other companies, by contrast, accepted the phrase "all lawful purposes" without such public conditions.

The contracts also tie into the Defense Department's AI acceleration strategy disclosed earlier in 2026. The department set out a policy to expand a modular open-source architecture across combat, intelligence and administrative functions, and to prioritize U.S. suppliers, transparent open-weight options and rapid prototype development.

Separately, the Defense Department said more than 1.3 million people have used its secure generative AI platform, GenAI.mil, so far. The platform provides large language models and other AI tools in a government-authorized cloud environment and is mainly used to support unclassified tasks such as research, drafting documents and data analysis.