On March 10, Anthropic filed suit to block the Pentagon. The company refuses to remove the safeguards in its AI against autonomous weapons and domestic surveillance. Defense Secretary Pete Hegseth had officially notified this designation on March 3, after months of tense negotiations.
Before the federal court in California, Anthropic denounces actions “unprecedented and illegal.” It also cites violations of its constitutional rights to freedom of expression and to due process, according to Reuters. In parallel, a second lawsuit targets a risk designation in the civilian supply chain, before the Court of Appeals for the District of Columbia.
Billions at stake
Anthropic’s executives present an alarming financial picture in their filings. Krishna Rao, the chief financial officer, estimates that the government could cut 2026 revenue “by several billions of dollars.” He adds that the impact would be “nearly impossible to reverse” if these actions persisted.
The company’s reliance on the Department of Defense already threatens hundreds of millions of dollars in 2026 revenue. Additionally, defense contractors could reduce their commitments by 50% to 100%. Finally, the company anticipates a loss of investor confidence, which would raise its financing costs.
The public sector: a pillar under threat
Thiyagu Ramasamy, head of the public sector, confirms damage that is “immediate and irreparable.” He estimates the direct loss at more than $150 million in annual recurring revenue. This amount concerns existing and anticipated contracts with the Department of Defense.
Growth in this segment had been spectacular. Between December 2025 and January 2026, the annualized growth rate of recurring public sector revenue had quadrupled. Moreover, projections pointed to several billions of dollars over five years. Thus, if defense contractors severed ties, the expected revenues for 2026—more than half a billion dollars—could “disappear entirely.”
Private clients also affected
Paul Smith, the chief commercial officer, also outlines the damage in the private market. One partner has abandoned Claude in favor of a competitor for a deployment with the Food and Drug Administration (FDA). This switch wipes out a projected revenue pipeline of more than $100 million.
Other contracts are also suffering. Negotiations with financial institutions totaling $180 million combined have been disrupted. A $15 million contract is on hold. In addition, a fintech client has reduced its commitment from $10 million to $5 million, citing directly the “situation” with the Pentagon.
More than 100 enterprise customers subsequently contacted Anthropic. They voiced “deep concern, confusion, and doubt” about the risks associated with partnering with the company.
On the AWS side, clients are now being assisted to migrate their DoD-related workloads to other models. However, AWS notes that Claude remains available for all workloads not connected to the Pentagon.
A clash with sector-wide repercussions
This dispute goes beyond Anthropic alone. It raises fundamental questions for the entire AI sector. Thus, 37 researchers and engineers from OpenAI and Google, including Jeff Dean, Google’s chief scientist, filed an amicus brief in support of Anthropic. They argue that the case could discourage AI experts from openly debating the risks and benefits of their technologies.
