Pentagon Signs Deals With Seven AI Companies to Deploy Models on Classified Military Networks

# Pentagon Signs Deals With Seven AI Companies to Deploy Models on Classified Military Networks

The Pentagon announced Friday that it has signed agreements with seven of the world's largest artificial intelligence companies to deploy frontier AI models on its most sensitive classified military networks, marking the most aggressive push yet to transform the U.S. military into what defense officials are calling an "AI-first fighting force." The companies -- OpenAI, Google, Microsoft, Amazon Web Services, Nvidia, SpaceX, and Reflection AI -- will integrate their models into Impact Level 6 and Impact Level 7 environments, the Department of Defense's highest security classifications for cloud services. Oracle was added to the list later in the day, bringing the total to eight.

The Deals

The agreements authorize the selected companies to deploy their AI capabilities for what the Pentagon described as "lawful operational use" across classified networks designed to streamline data synthesis, elevate situational understanding, and augment warfighter decision-making. Impact Level 6 networks handle information classified up to the secret level, while Impact Level 7 supports the processing of highly restricted data, including mission-critical intelligence.

Pentagon CTO Emil Michael framed the multi-vendor approach as a strategic imperative. "It's irresponsible to be reliant on any one partner," Michael told CNBC, emphasizing that the architecture is designed to prevent "vendor lock" and ensure long-term flexibility for the Joint Force. The move diversifies the Pentagon's AI supply chain at a moment when the technology is becoming central to everything from logistics planning to real-time battlefield analysis.

The roster itself is notable for both who is included and who is not. SpaceX and Reflection AI -- a relatively newer entrant in the frontier AI space -- join established defense contractors Microsoft and AWS, which have long held classified cloud contracts through programs like JWCC. OpenAI's inclusion signals a dramatic evolution for a company that was founded as a nonprofit research lab just a decade ago.

The Anthropic Exclusion

The most conspicuous absence from the list is Anthropic, the AI safety company behind the Claude model series. The exclusion caps a months-long dispute between the Pentagon and Anthropic over the terms under which the military could use its AI technology. At the center of the conflict: the Pentagon's insistence that all participating companies allow their models to be used for "any lawful purpose," a blanket authorization that Anthropic resisted.

Anthropic pushed for guardrails that would prohibit the use of Claude for fully autonomous weapons systems and domestic mass surveillance. The Trump administration responded by designating Anthropic a "supply chain risk" -- a classification typically reserved for foreign adversaries -- and began phasing out the company's tools across defense operations.

Defense Secretary Pete Hegseth made the administration's position unmistakable during a Senate hearing on Thursday. Hegseth compared Anthropic's stance to "Boeing giving us airplanes and telling us who we can shoot at," and called Anthropic CEO Dario Amodei an "ideological lunatic." Orders are now in place to phase out Anthropic's tools over the next six months, though reports indicate the White House has quietly reopened discussions with the company following recent technology breakthroughs.

Why This Matters

This announcement represents a watershed moment in the militarization of frontier AI. For years, the question of whether the world's most powerful AI models would be deployed in classified military settings was treated as hypothetical. That debate is now settled -- at least for the current administration. The deployment of large language models and other AI systems on IL-6 and IL-7 networks means these tools will be processing some of the nation's most sensitive intelligence data, with implications for everything from targeting decisions to strategic planning.

The Anthropic exclusion also sends a chilling signal to the broader AI industry about the cost of imposing ethical constraints on government customers. By publicly punishing a company that sought to limit military applications, the Pentagon is effectively establishing a precedent: full cooperation or no contract. That dynamic could reshape how every AI company approaches defense work going forward, particularly as the global AI arms race with China intensifies.

The multi-vendor strategy, while positioned as preventing lock-in, also creates a competitive marketplace among AI providers eager to prove their models can handle the unique demands of classified military operations. For companies like OpenAI and Google -- both of which have faced internal employee protests over military contracts in the past -- the deals represent a decisive bet that defense revenue outweighs reputational risk.

What to Watch Next

The next six months will be critical on multiple fronts. Watch for how quickly these AI models are actually deployed on classified networks versus how long security accreditation and integration take in practice -- historically, Pentagon IT modernization moves far slower than press releases suggest. The Anthropic situation remains fluid; the quiet reopening of White House discussions suggests the administration may ultimately prefer having every major AI lab at the table. And as these models begin processing classified data, expect intensified congressional scrutiny over AI safety guardrails, accountability frameworks, and the fundamental question of how much autonomy AI systems should have in military decision-making. The Pentagon has chosen its AI partners. The harder question -- what those partners will be asked to do -- is just beginning.