Astra's success leaves no spare compute for the $200 plan, one user says
The author says Astra is a roaring success but OpenAI has no compute left for its $200 plan, and notes the company raised its compute spend projection from $600B to $750B.
We are still underappreciating how much compute is going to be required over the next few years, to the benefit of names like $CRWV and $NBIS, which are probably not going to $0. Astra is a roaring success, and in my time with it, I am truly impressed. As such, OpenAI has no more available compute to facilitate their $200 plan. They also announced they are increasing compute spend projections from $600b to $750b (though note this is still a far cry from the $1.4T number they mentioned last year).
This post quotes another post: x.com/thsottiaux/status/2098113585683808624