Aside from the fact that CUDA is proprietary & very obviously made to lock people into using NVidia products, the thing is also that it's not just about CUDA. I

Aside from the fact that CUDA is proprietary & very obviously made to lock people into using NVidia products, the thing is also that it's not just about CUDA. Indeed, deep learning primitives use CUDNN which have professionally written code hand tuned by PhDs in internal only assembly, getting performance not available to the public even if you did end up getting a CUDA compiler somehow.