Anthropic Warns Pentagon Supply Chain Label Could Cost Billions as Tech Rivals Rally Support

A high-stakes confrontation between the Pentagon and artificial intelligence startup Anthropic is escalating, with the company warning that a controversial “supply chain risk” designation could ultimately cost billions of dollars in lost business—and drawing support from unexpected quarters of Silicon Valley.

Six Anthropic executives submitted court filings Monday seeking a preliminary injunction that would preserve the San Francisco company’s ability to conduct business with the Department of Defense while litigation over the designation proceeds. The company says hundreds of millions of dollars in Pentagon-related revenue is already at risk this year, with potential total losses reaching into the billions.

The dispute stems from Defense Secretary Pete Hegseth’s decision to label Anthropic a supply chain risk—a move that bars the company from military contracts and forces other defense contractors to abandon its Claude AI models. Hegseth has accused Anthropic of “arrogance” and “duplicity,” framing the company’s safeguards against autonomous weapons and mass surveillance as incompatible with American principles.

Yet Anthropic’s technology was reportedly used in a U.S. Central Command operation targeting Iran’s Ayatollah Ali Khamenei just hours after the designation was publicly announced, according to defense officials who confirmed the AI system had undergone hundreds of hours of training under human oversight as part of “Operation Epic Fury.”

CFO Krishna Rao revealed in court filings that Anthropic has generated more than $5 billion in all-time sales since commercializing its technology in 2023, but remains deeply unprofitable after spending over $10 billion on computing infrastructure to train and deploy its models.

Chief Commercial Officer Paul Smith documented immediate fallout extending far beyond military contracts. A financial services customer paused negotiations on a $15 million deal, two major financial institutions demanded unilateral cancellation rights on $80 million in combined contracts, and a grocery chain canceled a sales meeting entirely.

“All have taken steps that reflect deep distrust and a growing fear of associating with Anthropic.”

— Paul Smith, Chief Commercial Officer, Anthropic

OpenAI and Google have now entered the fray, backing Anthropic’s legal challenge to the Pentagon designation. Legal experts have characterized the move as ideologically driven and potentially vulnerable under the Administrative Procedure Act, with precedents suggesting courts may view the designation as pretextual political theater.

The Treasury Department and General Services Administration have announced intentions to halt business with Anthropic following a directive from the Trump administration. Meanwhile, federal officials are preparing new AI procurement rules that would require companies to grant the government irrevocable licenses for “any lawful” use—a condition that could reshape the entire government AI marketplace.

The Pentagon has offered a six-month transition period for organizations using Anthropic services, though officials acknowledge replacement challenges given Claude’s demonstrated capabilities in generating software code and other advanced applications.

Anthropic CEO Dario Amodei has called on Congress to address AI privacy risks, including potential domestic surveillance concerns—a position that reportedly factored into the administration’s adversarial stance. The company maintains that safeguards prohibiting hypothetical autonomous weapons and population surveillance were never present in actual DoD contracts.