So Anthropic makes a new model, Qwen or whoever learns how it works and makes their model nearly (but not quite) as good and then sells it back to Anthropic?
nope, never gonna happen. DC treats American AI supremacy as national security, IPO fortunes depend on the "America leads" story and the trump sycophants/toadies/roadies stand to gain millions from the IPO, and everyone thinks Chinese open weights are dumping meant to kill US labs, so Ant and OI will not want to wade into that.
Why pay more for a middleman if you can go directly to the source?
The problem for Anthropic and OpenAI is to devise some political scheme to stop them.