1) Intel process advantage vs TSMC at similar node is minimal in mobile.
So, why didn't Intel optimize the metal stack for density for its low power processes? Indeed, the company did talk about using more of the 80nm pitch layers in a density-focused mobile design and yet Bay Trail's density is still sub-par.
What's going on here? TSMC's 28nm has minimum metal pitch of, I believe, 96nm - not anywhere near as dense as Intel's 22nm M1.