Earlier quoted context omitted.
The difference isn't that big: https://www.youtube.com/watch?v=WKN9nvXTGHE And for 5nm, I'm quite sure AMD already has contracts for that with TSMC. So the gap should be even smaller.
His “as tested” was 11 hours for AMD and 13 for M1. His “max time use” for AMD was 14.5 and 20 for M1. That’s a 38% greater time for the M1 in a best possible outcome and 18% on “average”. Substantial differences that users would perceive noticeably.
It’s not that it doesn’t matter at all, it’s just that it matters a lot less once you’ve checked boxes like that. The effective experience of both would be “you only need to charge it overnight”