1) to address frontier model company stock valuations (openai for instance): deepseek is creating stuff months after frontier companies. In the AI arms race it

1) to address frontier model company stock valuations (openai for instance): deepseek is creating stuff months after frontier companies. In the AI arms race it might make sense to burn billions to be 3 months ahead (think about how you use that superintelligence to prevent anyone else acquiring it) 2) to address Nvidia valuation: there is no cap to demand on intelligence (or, we're not close). People will never be satisfied with the intelligence achieved and just stop asking for more. So Nvidia will still sell the hardware as the demand side is uncapped. Unrelated note that I was considering and would like an opinion on: Nvidia is the software play in AI, and TSMC the hardware play. Nvidia has competitors like broadcom/AMD/TPUs but beats out on software. TSMC will be frontier on manufacturing everyone's hardware.