3 ms·
Yet everyone is okay with a de-facto single supplier of CUDA capable AI GPUs.
Yet everyone is okay with a de-facto single supplier of CUDA capable AI GPUs.
Perhaps not. But (unlike in this hypothetical situation) nobody besides AMD and Intel can do much about that...
Outside of Nvidia, nobody is okay with that.
This is only true if “everyone” excludes all of the hyperscalers and the client device manufacturers.