[cavisne]

This feels a bit overdone. OpenAI has had problems with every compute partner they've ever had. It's just not a solvable problem, who would they go to to allegedly get next-gen chips quicker?