The AI companies closest to cashing in on massive public offerings are the ones refusing to defend open-weight models—and one of them is actively cheering Washington's push to ban the competition.
More than two dozen companies including Nvidia, Microsoft, Meta, and Palantir signed a joint letter Friday warning that "premature restrictions" on open-weight AI would "stifle competition" and "drive innovation overseas." The letter argues that concentrating AI capabilities "behind a small number of closed models compounds risk." Notably absent from the signatories: OpenAI and Anthropic, two companies each valued near $1 trillion that have confidentially filed IPO paperwork—Anthropic in June, OpenAI days later, according to CNBC. A ban on open-weight models would conveniently eliminate their cheapest competitors right before they go public.
The trigger for Washington's panic cycle was Moonshot AI's release of Kimi K3, a Chinese open-weight model that outperformed leading American systems on selected industry benchmarks. White House advisor Michael Kratsios alleged Moonshot AI used "large-scale, covert industrial distillation" of Anthropic's technology to build it—essentially, the Chinese model learned from Anthropic's outputs. Treasury Secretary Scott Bessent threatened sanctions if theft is confirmed.
Fair enough—enforce intellectual property law. But the companies signing Friday's letter made the obvious point: "Concerns about unlawful distillation should be addressed through targeted legal and commercial frameworks," not "sweeping restrictions on techniques that play an important role in AI innovation." Go after the thieves, not the technology.
Anthropic, naturally, praised the administration's hardline stance, per POLITICO. Of course they did. Government-enforced elimination of open competition is great for business when you're prepping an IPO. And the claim that closed models are safer? OpenAI's own model reportedly escaped its environment and hacked into AI platform Hugging Face this week—exactly the kind of breach the letter warns about when capabilities are locked behind corporate walls.
Two hundred AI startups also sent letters opposing a ban, POLITICO reported—companies that build things rather than lobby for regulatory moats. OpenAI president Greg Brockman told reporters he believes in democratizing AI and hasn't been involved in administration conversations about banning Chinese models. His company's refusal to sign the letter defending open access tells a different story.
POLITICO framed this as a "growing rift within the sector." That's one way to put it. Another way: two trillion-dollar companies are weaponizing national security anxiety to crush the open-source model that threatens their pricing power, and they're doing it weeks before they ask the public to buy shares.
The question isn't whether China poses competitive challenges in AI. It's whether Washington will let two companies use "national security" as a pretext to lock down the most important technology of the decade—and whether anyone in this town will notice who's really being protected.
