I think this is confirmed by the fact software vendors are still not taking advantage of ARM chips maximum performance.

Where this might shift is as we start using more applications that are powered by locally running LLMs.