The Court Ruled Against Anthropic

The D.C. Circuit Court of Appeals denied Anthropic's motion for a stay on Wednesday, meaning the Pentagon's blacklisting remains in full force while the case works its way through the courts.

The court's language was blunt. "In our view, the equitable balance here cuts in favor of the government," the three-judge panel wrote. The judges acknowledged Anthropic "will likely suffer some degree of irreparable harm," but called the company's interests "primarily financial in nature."

Courts don't grant stays lightly. Framing Anthropic's constitutional retaliation argument as primarily a money complaint is a strong signal about how difficult this case may be to win.

What This Actually Means on the Ground

Two courts, two different outcomes, running simultaneously.

The D.C. appeals court said the Pentagon blacklist stands. Defense contractors must certify they are NOT using Anthropic's Claude models in military work.

But a separate federal judge in San Francisco — in a related but distinct case — issued a preliminary injunction blocking the Trump administration from enforcing a broader ban on Claude usage across government. That ruling means Anthropic can still work with civilian federal agencies.

The result: Anthropic is frozen out of DOD contracts specifically, but remains operational everywhere else in the federal government. Defense contractors can still use Claude for non-military work.

Tuesday's Oral Arguments Are Still Coming

None of this settles the main event. The D.C. Circuit is still scheduled to hear 15 minutes per side from the DOJ (representing the Pentagon) and Anthropic on the merits of the blacklisting itself.

Judges Karen Henderson, Gregory Katsas, and Neomi Rao will issue a written opinion after arguments. That opinion will determine whether the Pentagon's supply chain risk designation — a label historically reserved for foreign adversaries like Huawei — can legally be slapped on a U.S. company.

If the DOD can blacklist a domestic AI firm without clear statutory process, it can do it to any tech company.

The Context CNBC's Coverage Missed

The mainstream coverage focuses heavily on Anthropic as the sympathetic underdog fighting Pentagon overreach. That framing isn't wrong, but it's incomplete.

According to CNBC's own reporting, negotiations collapsed because the DOD wanted unfettered access to Claude across all lawful military purposes, while Anthropic drew hard lines against use in fully autonomous weapons and domestic mass surveillance.

Those are reasonable positions for any company with a safety mandate. But it also means Anthropic walked away from a deal — and the Pentagon's reaction, however heavy-handed, didn't come from nowhere.

Defense Secretary Pete Hegseth then went further than necessary — publicly bashing the company on social media and dropping the supply chain risk label. That's where retaliation looks most apparent. The appeals court's language about "judicial management of how, and through whom, the Department of Defense secures vital AI technology during an active military conflict" suggests the judges see DOD discretion as broad.

The court referenced the DOD's continued use of Anthropic's models to support military operations in Iran. The Pentagon is simultaneously blacklisting Anthropic AND using its technology.

President Trump told CNBC last month a deal is "possible."

The Business Reality

Anthropics's legal problems are landing at an unusual juncture.

CNBC just ranked Anthropic No. 1 on its 2026 Disruptor 50 list, ahead of OpenAI. CEO Dario Amodei claims revenue grew 80 times in Q1 alone. The company is reportedly in talks to raise capital at a valuation of up to $900 billion.

Co-founder Daniela Amodei told CNBC the acceleration over the past three to six months reflects "models getting smarter, products getting better" — driving massive enterprise value. Claude Code is being credited with reshaping software development workflows.

A company valued near a trillion dollars, locked out of the Pentagon.

What This Means

If the Pentagon can designate a U.S. technology company a national security threat — without clear legal standards, without transparent process, and arguably in retaliation for refusing unlimited government access — that sets a significant precedent.

Every AI company now knows what happens when they say no to the DOD.

Anthropic's refusal to allow autonomous weapons deployment or domestic mass surveillance wasn't reckless. It was a line any responsible company should hold. The government's response was to call them an enemy.

The merits hearing will determine whether that kind of executive power has legal limits. Right now, the Pentagon is winning.