A lot of the stuff you’re referring to as “application specific” is already present in Intel Macs with the T2 chip. Except obviously slower and older. People aren’t discussing it because it isn’t really “new”
They have indeed moved the system controller etc from the T2 onto the M1 more or less wholesale. If you look at the TEM image I have added to my post of the M1's sister chip, the A14, you'll see that the classical CPU and GPU take up only a small fraction of the die. The rest isn't memory as a naive take would suggest. It's custom silicon handling a large basket of previously CPU tasks.
