As a AMD GPU user, I feel as if trying to mandate CUDA only or some other way to increase NVIDIA lock in like that would be met with people leaving the platform or at lest simply working on the non NVIDIA formats away from hugging face. Though requiring a lot of space it's quite easy to spin up a competitor at least for hosting weights. And if they try some legal shenanigans then many countries outside the US will still be happy to host I am sure why not China?
The only thing I can see them being able to get away with is increasingly bending the hugging face python API and any other features of that sort they develop to NVIDIA only. I personally don't use that and don't see a reason to and I am not sure how many people do use the hugging face python library.
Hf’s python / transformers is a hot mess. It conflates so many different ideas into huge monoliths that it’s barely usable for anything other than the small snippets listed for trying out models.
They need to take a machete to all the cross coupling they’ve metastasized.
Totally not sure why any one would use it for anything other then if they stumbled across it and where completely green.