AMD's support for AI is just fine, you just have to choose a path - if you're on Linux, use their CUDA translation software (ROCm), if you're on Windows, use DirectML.
This is quite untrue, especially if you do actual research and not just run other peopleโs models. For example, ROCm is missing in many sparse autograd frameworks, e.g. pytorch_sparse, or having a viable alternative to Nvidias MinkowskiEngine. This is needed if you do any state-of-the-art convnets with attention-like sparsity.