If you're doing work in The Cloud and the servers are in America, you can still move all that to physically reside in whatever jurisdiction you live in. The tech is there.

But if you build stuff on top of Anthropic, OpenAI or Google AIs, their data resides and will always reside in the US. They have very specific laws that allow them to grab data from anywhere and nobody can say shit about it.

There's no realistic way for anyone to run "ChatGPT" at home. You can get pretty close with the 200B++ param models but they're still just a pale shadow of what an acre of industrially cooled compute will do in comparison.

But if your risk modelling doesn't flash a big red light about building a business on top of someone else's hardware and software that is physically in a currently very unstable and legally unreliable country, that's fine.

I've been using my VC subsidised LLM credits to build local-first systems I can still use when either the money runs out or the political craziness crashes down.