Many of the most important ML toolchains run natively on Apple Silicon, including PyTorch [0] and TensorFlow. For example the PyTorch folks have this to say about it [0]:
> Every Apple silicon Mac has a unified memory architecture, providing the GPU with direct access to the full memory store. This makes Mac a great platform for machine learning, enabling users to train larger networks or batch sizes locally. This reduces costs associated with cloud-based development or the need for additional local GPUs. The Unified Memory architecture also reduces data retrieval latency, improving end-to-end performance.
Plenty of other reports out there [1]:
>We ran a sweep of 8 different configurations of our training script and show that the Apple M1 offers impressive performance within reach of much more expensive and less energy efficient accelerators such as the Nvidia V100 for smaller architectures and datasets.
If one is looking for a great bang for the buck and a big savings in energy use, the GPUs in Apple Silicon are a compelling option. Plenty of Apple haters (presumably yourself) like to ignore what has been achieved with this technology, but that doesn't make it any less real.
Everyone targets Nvidia or AMD GPUs for both gaming and compute like ML.
Mac GPUs are mostly used by YouTubers to accelerate video editing.