> With the inherent difference in size and power consumption between 8-bit and 32-bit processors being relatively
> small (considering memory, interface, peripheral, etc. costs), I would have guessed that 8-bit processors would
> be marginalized, surviving only from legacy use and better customer relations. I.e., economies of scale would
> eventually kill 8-bit processors. (From the little I have read, it looks like what is actually happening is that
> 16-bit processors are being squeezed from below and above rather than displacing 8-bit processors.)

Ah, but the power difference between 8-bit and 32-bit µC isn't small.
For comparable technology, it's often 3 orders of magnitude. It's mW vs sub-µW.
The technology is often not comparable, with many slow 8-bit µC operating directly at 3.3V or 5V, something that's not an option for the faster 32-bit µC.

>
> So you do not think that Cortex-M has a sufficient inherent disadvantage that it could not compete
> with 8-bit processors--that the absence is effectively a "historical accident"--, correct?

Cortex-M cores are far too complex and power hungry to be used where 8-bit µCs are.

These devices do not survive from legacy.
They're being actively used in new designs that have zero legacy concerns and some of them architecture variants are rather recent.