The Pentagon Just Did Something It's Never Done to an American Company
The Pentagon has given military commanders 180 days to remove every trace of Anthropic's AI products from defense systems, officially designating the San Francisco startup as a "supply chain risk" — a label previously reserved for foreign adversaries like Huawei. The March 6 internal memo marks the first time the U.S. military has blacklisted a domestic tech firm under supply chain risk authorities, turning a public feud over AI safeguards into an unprecedented corporate death sentence.
The escalation follows a Tuesday meeting where Defense Secretary Pete Hegseth gave Anthropic CEO Dario Amodei until Friday evening to grant the military "unfettered access" to Claude or face "harsh penalties." According to a senior Defense official, the meeting was "not warm and fuzzy at all." Hegseth told Amodei the Pentagon won't let any company "dictate the terms under which the Pentagon makes operational decisions." Amodei denied the Pentagon's claim that Anthropic interfered during the Maduro raid operation, insisting the company's red lines have "never prevented the Pentagon from doing its work."
Why This Matters for Prediction Markets
The stakes are existential. Michael Mongan, Anthropic's attorney, warned a judge Tuesday that the supply chain risk designation could cost the firm "billions of dollars in revenue." Claude is currently the only AI model approved for the military's most sensitive classified work — a position that made Anthropic indispensable until this week. As one Defense official admitted to Axios: "The only reason we're still talking to these people is we need them and we need them now. The problem for these guys is they are that good."
The White House is preparing an executive order to formalize the purge across all federal agencies, sources told Axios. The Treasury Department has already begun offboarding Anthropic products, as @Polymarket flagged: "JUST IN: US Treasury terminating its use of all Anthropic products." Defense contractors are scrambling to assess their exposure — the Pentagon contacted Boeing and Lockheed Martin last week demanding inventories of Claude usage. Boeing confirmed it has no active Anthropic contracts; a Boeing executive said the company "sought their partnership [in the past] and ultimately could not come to an agreement. They were somewhat reluctant to work with the defense industry."
The Three Red Lines Anthropic Won't Cross
Anthropic has offered to adapt its usage policies for the Pentagon, but refuses to budge on three safeguards: no mass surveillance of Americans, no autonomous weapons that fire without human control, and no involvement in operations it deems ethically problematic. The company insists these are protected speech rights under the First Amendment. The Trump administration argues those safeguards constitute a national security threat when "industry intervenes during military operations."
Former defense officials and policy experts are now circling the wagons. A letter to Congress from defense and intelligence veterans slammed the Pentagon for setting a "dangerous precedent." Microsoft and Google threw their weight behind Anthropic — Microsoft asked a federal court to temporarily block the designation, while both cloud giants reassured customers that Claude remains available for non-defense work. The Information Technology Industry Council sent a letter to Hegseth expressing "concern" over the precedent, though it carefully avoided naming Anthropic.
What Happens Next
Trump boasted to Politico on March 5: "I fired Anthropic like dogs, because they shouldn't have done that." The same day, multiple outlets reported negotiations between the Pentagon and Anthropic had quietly restarted — though the 180-day removal order and pending executive order suggest those talks aren't going well. Anthropic filed suit Monday, calling the designation "unprecedented and unlawful" and arguing Congress never gave the administration authority to blacklist a U.S. company over protected speech.
The legal battle will test the outer limits of presidential procurement power. Trump used executive orders against foreign tech firms like Huawei in his first term, but never explicitly named a U.S. company outside standard contracting processes. For Anthropic, the clock is ticking: defense tech companies reliant on government contracts are already dropping Claude to stay compliant, and the firm is hemorrhaging commercial credibility as agencies race to comply with what one source called the administration's push to remove "woke" AI. The irony: Claude was good enough that the Pentagon built its most sensitive operations around it — right up until the company refused to let the military use it however it wanted.
