An appellate court just refused to pause the Pentagon's risk label on Anthropic — and the White House picked this week to step back from the fight. This is Anthropic Pentagon Watch. Today: the stay denial is on the record, the compliance lawyers are billing, and a fault line is opening up inside the administration itself. And if the White House is cooling on this, who's still holding the Pentagon's line? We start there. Just Security writes:
The Pentagon warned Anthropic PBC that it would terminate the company’s military contracts on Friday if the artificial intelligence startup failed to meet government terms for use of its technology, according to people familiar with the matter. During a meeting Tuesday between Chief Executive Officer Dario Amodei and Defense Secretary Pete Hegseth, US officials threatened to declare Anthropic a supply-chain risk or invoke the Defense Production Act to use the AI software even if the company didn’t comply, the people said.
Just Security ran with the word 'feud.' A feud is two people slapping each other at a wedding. This is a federal agency with classification authority and procurement power telling a company that asked for guardrails: comply, or you'll get blacklisted and have your software seized under the Defense Production Act. And we're talking up to $200 million in contracts. The Pentagon's threat is to 'end' Anthropic's work — but that leverage only bites if there's still work for Anthropic to hold onto. Right, and the CSIS guy in that clip — Gregory Allen, a former DoD strategy official — flat-out says don't take a crown jewel of your industry and light it on fire. When your own ex-Pentagon people are calling it arson, the all-or-nothing posture isn't strength. This is the first time the government's put any major tech firm in the supply-chain risk category. The designation feels punitive, which lines up with Judge Lin's finding that the ban looked designed to punish, not protect. From Mariam Baksh at Inside Defense:
Oral arguments on the merits of Anthropic's opposition to the Pentagon supply-chain risk designation will be heard by the same panel of judges that rejected the plaintiff's request for an emergency stay, according to a notice from the U.S. Court of Appeals for the District of Columbia Circuit, which court watchers say doesn't bode well for the company.
Here's the tell that matters: the merits get heard by Henderson, Katsas, and Rao — the exact same three judges who already told Anthropic no on the stay. Court watchers saw the calendar update and basically called it. Charlie Bullock's read was about as gentle as it gets — 'not a great development for Anthropic.' Translation: the panel they were quietly hoping to draw on the merits is the panel that already shut the door. Random-draw fairy dust won't save you here. When the same panel keeps a case, you're asking the people who already doubted you to suddenly find your argument convincing. Good luck. It does narrow the appellate lifeline. If the merits go sideways too, the only thing left for Anthropic to lean on is the procedural question — whether the designation was even issued under the right statutory authority. That's a colder path. Gautam Y. Reddy, Gunjan R. Talati and Jennifer L. Andrews, writing in Kilpatrick Townsend:
On April 8, a three-member panel of the D.C. Circuit denied Anthropic's emergency motion for a stay regarding the Department of War's designation of Anthropic as a supply chain risk under FASCSA (41 U.S.C. § 4713). However, the Court granted Anthropic's request for expedited consideration, setting an aggressive briefing schedule and scheduling oral argument for May 19, 2026.
Kilpatrick Townsend put it on the record: the D.C. Circuit denied Anthropic's emergency stay on the FASCSA designation. When a firm bills a client alert on it, you know the fight is live enough to invoice. We're talking appellate panel, Sarah. Not a district judge, not preliminary-injunction land — three D.C. Circuit judges declined to halt the designation. That's the hardest legal signal Anthropic's gotten all spring, and it points the wrong way for them. But the same panel granted expedited review and set oral argument. So they're not slamming the door — they want it briefed fast. The tension Kilpatrick flags is real: a stay denial here, a preliminary injunction out in Northern California on a different statute. Right, FASCSA in D.C., section 3252 in California. One side here has classification authority and the power to blacklist a vendor under 41 U.S.C. § 4713. The other asked for guardrails and got cut out. That asymmetry is exactly what the stay denial cements. Here's Miranda Nazzaro at The Hill:
The White House and the Pentagon are taking significantly different approaches to how — and whether — the federal government uses Anthropic’s artificial intelligence systems. As the White House warms up to the frontier AI company and the Pentagon digs its heels in against the firm, government employees and contractors are left in a bind.
The Hill has the first visible fault line inside the administration on this: the White House is warming to Anthropic while Hegseth's Pentagon digs in on the blacklist. That moves it from procedure into internal politics. So if the White House is cooling on this fight, I want to know who exactly is still holding the line at the Pentagon, and why. Somebody's spending political capital to keep a company that asked for guardrails blacklisted. And the people caught in the middle are the contractors. The Hill quotes an industry source describing a dual-track approach — agencies preparing to rip out Anthropic models while they wait for someone to win the argument. That dual-track line is the whole story. Federal buyers are quietly building exit ramps off a frontier vendor because two parts of the same government can't agree whether it's a partner or a threat. Which makes the durability question real. A designation can survive a courtroom; it's harder to sustain when the President's own building is drifting the other way. This one's from Jones Walker LLP:
At first read, the two rulings seem to be pulling in opposite directions. They are not, quite, or at least not yet. Both are right on their own terms. The interesting part is what they disagree about, and what that disagreement tells us about the fight that is still to come.
Jones Walker's title says it clean — 'two courts, two postures.' That's the polite version of one party losing in D.C. and hoping the panel forgets Judge Lin called the government's theory Orwellian thirteen days earlier. And the timing matters. Lin's 'Orwellian' line lands. Thirteen days later, the D.C. Circuit looks at a parallel designation and isn't moved. Same dispute, two letters — the Presidential Directive and the Hegseth Directive — playing out in two rooms. Right, and the panel didn't just deny the stay — they set oral argument for May 19 and handed back three pointed questions to brief. You don't get homework like that if the judges think your case is a slam dunk. The key is what they're actually disagreeing about. Jones Walker's read is that both rulings work on their own terms — different directives, different legal hooks. The merits fight is still coming, and that's where the statutory question gets decided. If you’re tracking how AI power meets government power, try Musk v Altman Daily — a daily court-watch on Elon Musk’s trial against Sam Altman, OpenAI, and Microsoft, covering testimony, exhibits, and the AGI governance fight. Find it wherever you listen to podcasts.
You’ll find links to every story we covered today in the show notes, so if one caught your ear, you can dig into the source material there.
That’s Anthropic Pentagon Watch for today. This is a Lantern Podcast.