I would too, but sadly that's Khronos' job to organize, and they've had trouble getting American vendors to work together.

It's likely that CUDA will continue dominating until they put aside their differences. The current MLX/MPS/ROCm ecosystems are too fractured to threaten Nvidia.