Pentagon Signs AI Deals With Seven Companies, Keeps Anthropic Out

The U.S. Defence Department has cleared SpaceX, OpenAI, Google, Nvidia, Microsoft, AWS, and Reflection AI for classified military networks, formalizing Anthropic’s exclusion as a court battle over its AI safety guardrails continues.

The U.S. Department of Defense announced on May 1 that it has reached agreements with seven technology companies to deploy their AI capabilities on classified and top-secret military networks.

The companies named are SpaceX, OpenAI, Google, Nvidia, Microsoft, Amazon Web Services, and Reflection AI. Oracle was added separately when the Pentagon posted confirmation on X the same day. Anthropic is not among them.

The Pentagon framed the agreements as part of a drive toward becoming “an AI-first fighting force.” The new terms permit what officials describe as “lawful operational use,” language broad enough to cover potential applications in targeting and other classified combat operations. Defence officials said the push is also designed to prevent overdependence on any single vendor.

ALSO READ: Pentagon Questions Anthropic on Claude’s Military Use

How it got here

Anthropic had been the first frontier AI lab to have its model, Claude, approved for use on the Pentagon’s classified networks, following a $200 million contract signed in July 2025.

Negotiations broke down when the Defence Department sought to renegotiate those original terms, demanding that Anthropic allow use of Claude for “all lawful purposes” without restriction. Anthropic refused, citing two specific limits in its acceptable use policy: it would not allow Claude to be used in fully autonomous weapons systems capable of selecting and engaging targets without human intervention, and it would not permit mass domestic surveillance of American citizens.

Defence Secretary Pete Hegseth designated Anthropic a “supply chain risk” on February 27, 2026, a label previously reserved for companies linked to foreign adversaries. President Trump simultaneously directed all federal agencies to stop using Anthropic technology, with some given a six-month window to transition off.

Anthropic filed two federal lawsuits in response, arguing the designation was retaliation for its public stance on AI safety and a violation of its First Amendment rights.

A federal judge in California, U.S. District Judge Rita Lin, granted Anthropic a preliminary injunction, blocking enforcement of the ban and calling the designation “an attempt to cripple” the company. A D.C. appeals court, however, denied Anthropic’s separate request to pause the blacklisting while litigation continues.

The result is a split: Anthropic can continue working with non-defence government agencies but remains excluded from Defence Department contracts.

What the new deals say

The agreements with the seven companies substantially expand the terms under which the Pentagon can use AI in classified settings.

Nvidia’s agreement, according to a person familiar with the terms, grants the Defence Department full operational use of its models, including for autonomous weapons development, with usage to remain within U.S. law and constitutional authority. No additional monitoring or evaluation mechanisms are stipulated beyond that.

Microsoft and AWS negotiated their terms with the Pentagon late into Thursday before the announcement. OpenAI had previously signed a similar agreement but has not yet deployed its models on classified networks; an OpenAI spokesperson confirmed implementation is underway.

One name on the list drew specific attention. Reflection AI, a lesser-known startup, raised $2 billion in October 2025 and is backed by 1789 Capital, a venture capital firm in which Donald Trump Jr. is a partner and investor.

On the ground

Since the Anthropic fallout, the Pentagon has sharply compressed its onboarding timeline for new AI companies onto classified networks. A process that previously took 18 months or more now takes under three months, according to newer AI entrants who have gone through it since February.

The Pentagon’s primary AI platform, GenAI.mil, has been used by over 1.3 million Defence Department personnel since launching five months ago. Claude is also deployed on Maven Smart System, the digital targeting platform used in U.S. military operations including actions related to Iran. Pentagon officials have given themselves six months to find and deploy replacements across these systems.

Pentagon staffers and IT contractors who work closely with the military have told Reuters they are reluctant to abandon Claude, which they regard as technically superior to the alternatives now being deployed.

Where Anthropic stands

The situation may not be permanent. Trump said last week that Anthropic was “shaping up” in his administration’s view. Anthropic CEO Dario Amodei visited the White House last month to meet with Chief of Staff Susie Wiles after the company unveiled Mythos, a new model with advanced cybersecurity capabilities.

Pentagon officials nonetheless maintained on Friday that Anthropic remains a supply chain risk. Speaking before Congress on Thursday, Hegseth called Anthropic’s leadership “ideological lunatics” and defended the department’s position: “We follow the law and humans make decisions. AI is not making lethal decisions.”

The legal dispute continues.

Disclaimer : This article is intended for informational purposes only, and is based on publicly available information. NervNow has not independently verified the claims.

MORE ON THE PENTAGON & ANTHROPIC DISPUTE
Federal Judge Blocks Trump’s Pentagon Ban on Anthropic’s AI Models
Banned by the Pentagon, Claude Hit No. 1 on App Store
Anthropic Refuses Pentagon Pressure to Remove AI Safeguards
AI Safety Clash Deepens as Pentagon Pressures Anthropic