The US Department of Defense has signed agreements with seven artificial intelligence companies — including SpaceX, OpenAI, Google, Nvidia, Reflection, Microsoft and Amazon Web Services — to deploy their technology across classified military networks, excluding Anthropic amid an ongoing dispute over usage safeguards, according toAFP.

The Pentagon said the companies’ tools would be integrated into sensitive systems used for mission planning, weapons targeting and related operations. The move comes as the department reduces reliance on Anthropic, whose AI models had previously been authorised for classified use.

Anthropic is contesting restrictions imposed earlier this year after objecting to its technology being used for mass domestic surveillance or to directly control lethal autonomous weapons. The Pentagon subsequently designated the company a supply-chain risk, effectively barring its use by the military and contractors. In February, President Donald Trump ordered agencies to stop using its technology.

Shift to multiple vendors, open models

Emil Michael, the Pentagon’s de facto chief technology officer, said the agreements reflected a strategy to diversify suppliers. “What we’ve learned is that it’s irresponsible to be reliant on any one partner,” he toldCNBC.

The contracts include both proprietary and open-source AI models. A person familiar with the matter said Nvidia and Reflection would provide open-source systems, which can operate without ongoing licensing or vendor access. The Pentagon views these models as offering greater operational flexibility and reducing dependence on a single provider.

Anthropic’s Claude model — which is not open source — remains in use for now, and the company is challenging the Pentagon’s measures in court.

Military AI push and internal resistance

The Pentagon said the deployments form part of an effort to build an “AI-first fighting force”, aimed at improving data processing and decision-making. Its GenAI.mil platform has been used by more than 1.3 million personnel, generating tens of millions of prompts and deploying hundreds of thousands of agents within five months.

The agreements have drawn internal opposition at Google, where more than 600 employees have called on the company to reject the Pentagon contract. The backlash echoes a 2018 campaign that led Google to withdraw from Project Maven, a military AI initiative.

Project Maven is now led by Palantir and has evolved into a targeting and battlefield management system. During a recent operation against Iran, the system reportedly processed more than 1,000 targets within 24 hours.

An AWS spokesperson said the company would support the Pentagon’s modernisation efforts. Other companies did not immediately comment.