This all sounds reasonable. I remember the initial friction with the transition to Apple Silicon/arm64, but those days are over now. Owning even a 6-year-old M1 today is a leg up to experiment with local LLMs, other architectures get bottlenecked more quickly.