Pentagon Expands AI Supplier List as Microsoft, Nvidia, and Amazon Gain Access to Classified Operations


Pentagon Expands AI Supplier List as Microsoft, Nvidia, and Amazon Gain Access to Classified Operations

The US government has expanded its roster of approved AI suppliers for classified military operations, adding Microsoft, Amazon, Nvidia, and Reflection AI to a growing list of companies authorized to provide AI technologies to the Pentagon.

The new agreements allow the Department of Defense to use these AI systems for “any lawful use” in classified environments, placing the companies alongside existing approved providers including OpenAI, xAI, and Google.

The phrase “any lawful use” has become controversial following a public dispute between the US administration and Anthropic. Anthropic CEO Dario Amodei argued that such wording could permit the government to use AI for mass civilian surveillance or autonomous weapons systems—applications the company wanted explicitly restricted.

The disagreement escalated after the Pentagon reportedly cancelled a $200 million contract with Anthropic, labeling the company a “supply chain risk.” Anthropic later challenged the decision in court, claiming significant financial damages and loss of future business opportunities tied to government influence.

Government officials reportedly criticized Anthropic as being overly ideological, marking the first known case of a US-based AI company receiving that type of designation.

Pentagon Pushes Toward an “AI-First Fighting Force”

In its official statement, the Pentagon said the expanded supplier network is intended to prevent dependence on any single AI vendor while ensuring long-term flexibility for military operations.

The Department of Defense plans to use the approved AI systems for highly classified “Impact Level 6” and “Impact Level 7” workloads—the latter involving some of the most sensitive national security materials handled by the US government.

Officials said the technologies will support data analysis, improve situational awareness, and assist military personnel in complex operational environments as part of the Pentagon’s broader push toward becoming an “AI-first fighting force.”

While generative AI is currently used mostly for administrative tasks such as document summarization and research inside defense agencies, the expanded partnerships suggest a broader role for AI in operational and intelligence functions.

Expanding Vendor Diversity

The Pentagon’s decision to work with multiple AI providers may also reduce risks tied to shifting corporate policies or executive decisions at individual companies.

In previous years, both Amazon and Google faced employee protests over military-related AI projects, with some workers objecting to the use of company technology in warfare or surveillance.

By diversifying suppliers, the government can reduce dependence on any single vendor whose policies or leadership could change unexpectedly.

Anthropic Still Active in National Security Work

Despite the ongoing dispute, Anthropic’s technology reportedly remains embedded in some government systems.

Its Claude AI model had previously been integrated into Palantir’s Maven platform for classified analysis tasks, and Anthropic’s Mythos model is reportedly still being evaluated for cyber defense and warfare applications by multiple intelligence organizations worldwide.

Reports suggest organizations such as the National Security Agency and the UK’s MI5 are among the agencies assessing the technology.

According to reports from Axios, officials inside the White House may now be attempting to repair relations with Anthropic despite the recent tensions. Sources claim the administration is exploring ways to restore cooperation, particularly as Anthropic’s coding models reportedly remain in use across parts of the US security infrastructure.

The White House stated that the US government will continue working closely with frontier AI companies to strengthen national security and protect American interests as AI adoption accelerates across military and intelligence operations.