I disagree with your analysis. The chips they acquired were not the top dog NVIDA chips. They figured out a way to train the models on these things which typically required bad ass top of the line chips. Then they sent out the source code to show others how to do this stuff on slower chips.So it seems like a Chinese company acquired 50,000 H100s on the black market, trained an AI model, and is now pretending they did it for next to no investment (relatively speaking).
If I traded individual stocks, I’d be long NVDA this morning.
The issue is this. People have a legacy software mindset on AI. Meaning they think there's going to be winners and losers like there is for enterprise software (e.g., SAP, Oracle, Workday, etc.). I think with the GenAI stuff it doesn't really matter. Once you're over some intelligence threshold on it there's litter difference between OpenAI, Google, or whatever other one you use. And if we now know how to produce these AI tools very cheaply, the value of OpenAI goes way down.
The move - and you can see it in the data - is to be in Industrials - the buyers of this tech.