AI deployment is already hard thanks to anti-open source nature of Nvidia. Have you ever tried to deploy models using nvidia-docker? Why are GPUs not part of Li

AI deployment is already hard thanks to anti-open source nature of Nvidia. Have you ever tried to deploy models using nvidia-docker? Why are GPUs not part of Linux Kernel? Why are server grade GPUs so expensive? Why does PyTorch/Tensorflow not really support AMD GPUs. Nvidia is a monopoly. A monopoly is never good for the market and innovation in the long term.