In a swift move that has rippled through the defense technology sector, several companies working on U.S. military contracts have begun phasing out Anthropic's Claude AI model following the Pentagon's decision to blacklist the firm. The Trump administration announced the designation on Friday, labeling Anthropic's technology as a supply chain risk, which has prompted contractors to instruct employees to switch to alternative AI systems. This development marks a dramatic turnaround for Anthropic, which had recently gained significant traction in government circles through partnerships and contracts.
The blacklist stems from tensions over how Anthropic's AI would be used, particularly after company executives refused to comply with certain government demands. According to reports, Anthropic sought assurances that its models would not support fully autonomous weapons or mass domestic surveillance of Americans. Defense Secretary Pete Hegseth declared on X that any contractor or supplier engaging with the U.S. military is now barred from commercial activity with Anthropic.
Alexander Harstrick, managing partner at J2 Ventures, which invests in defense startups, explained the rapid response in an email to CNBC. "Most of our companies are actively involved in large defense contracts and so are very strict in their interpretation of the requirements," Harstrick said. He added that 10 of his firm's portfolio companies working with the Department of Defense have "backed off of their use of Claude for defense use cases and are in active processes to replace the service with another one."
Harstrick, a former military intelligence officer who deployed to Afghanistan and Iraq in 2017, emphasized that the switch is precautionary. "This in no way reflected a perceived shortcoming of Claude with most commenting that the situation was lamentable as the product itself is excellent," he wrote. Multiple unnamed defense tech executives echoed this caution, with one telling CNBC that employees were directed last week to begin transitioning away from Claude, a process expected to take one or two weeks in preparation for Friday's deadline.
The CEO of another defense tech company said this week that staff were instructed on Monday to halt use of Claude pending further guidance, assuming a ban would take effect. Meanwhile, larger contractors like Lockheed Martin are anticipated to excise Anthropic's technology from their supply chains, as reported by Reuters late Tuesday. President Donald Trump posted on social media Friday that federal agencies have six months to phase out the technology, and directives have already gone out at the Defense Department, Treasury Department, State Department, and Health and Human Services.
Anthropic's entry into the Department of Defense ecosystem began in late 2024 via a partnership with Palantir, a software provider that derives nearly 60% of its U.S. revenue from government work. This led to a $200 million contract allowing Claude to become the first major AI model deployed in classified government networks. CEO Dario Amodei noted in a January CNBC interview that about 80% of Anthropic's revenue comes from enterprise customers, underscoring the potential financial hit from the blacklist.
Despite the announcement, Anthropic's models continue to support U.S. military operations in Iran, as previously reported by CNBC. The company has pushed back against the Pentagon's authority, stating in a Friday blog post that Defense Secretary Hegseth lacks the power under a federal statute to restrict contractors from using its technology. "Should the supply chain risk designation be made official, it would only apply to companies' use of Claude as part of defense contracts and cannot affect how contractors use Claude to serve other customers," Anthropic wrote.
Anthropic has not commented further beyond the blog post and has yet to pursue a legal appeal, as the designation remains informal, largely confined to social media statements. In a related development, OpenAI CEO Sam Altman posted on X hours after the Pentagon's Friday announcement that his company had reached an agreement with the DoD on AI model usage. Facing weekend criticism, Altman followed up Monday, admitting the timing was "sloppy" and that OpenAI "shouldn't have rushed" the deal. He shared an internal memo amending the contract to state that "the AI system shall not be intentionally used for domestic surveillance of U.S. persons and nationals."
Palantir declined to comment on its plans regarding the blacklist, but analysts at Piper Sandler highlighted potential disruptions in a Tuesday note to clients. They described Anthropic as "heavily embedded in the Military and the Intelligence community" and noted that transitioning away could cause short-term issues for Palantir's operations. "While re-establishing AI functions with a new vendor can and will happen if needed, Anthropic was [a] trailblazer in terms of operationalizing AI models for data-sensitive environments," the analysts wrote, maintaining a buy rating on Palantir's stock. They added that onboarding replacements would consume time and resources that could otherwise fuel growth.
Not all defense tech leaders see an immediate need for change. Tom Siebel, chairman of C3 AI and a former CEO who counts the DoD as a customer through a partnership with Booz Allen Hamilton, said in an interview that he doesn't view Claude as requiring mitigation "until it gets litigated." A partner at another defense-focused venture firm noted that his portfolio companies have minimal reliance on Claude and primarily use OpenAI's technology.
One venture investor in defense tech told CNBC that serious government contractors avoid over-dependence on any single supplier, suggesting the switch from Anthropic won't pose major challenges. The government maintains contracts with other AI providers, including Google for its Gemini model and Elon Musk's xAI for Grok, providing alternatives amid the controversy.
Tara Chklovski, CEO of Technovation, a global tech education nonprofit, warned that fully severing ties with Anthropic could prove risky. She praised the company for its deliberate approach to military AI development, arguing that alternatives might lack similar safeguards. "Once the dust settles, they'll realize that Anthropic is the only one that has this very unique set of skills in technology," Chklovski said. "Competition is so fierce that people think going fast and without the weight of these safeguards is the only way to succeed. Anthropic is showing that's probably not the way."
The blacklist highlights broader debates over AI ethics in national security, where companies balance innovation with concerns about misuse. Anthropic's refusal to yield on surveillance and autonomous weapons reflects a growing pushback from tech firms against unrestricted military applications. As the situation unfolds, defense contractors are navigating compliance while preserving operational efficiency, with the potential for legal challenges that could reshape AI procurement in government work.
Looking ahead, the six-month phase-out period for federal agencies offers a buffer, but the immediate actions by private contractors signal the blacklist's broad impact. If Anthropic appeals successfully, it could reinstate access; otherwise, the sector may accelerate diversification to models from OpenAI, Google, and xAI. For now, the defense tech world is adapting to a landscape where ethical stances can swiftly alter supply chains and partnerships.