[dlewis1788]

For training, yes, but no indications on inference workloads. Apple has said they would use their own silicon for inference in the cloud.

[walterbell]

Plus the Apple "Neural Engine" which has shipped on millions of iPhones for local inference.