All reasonable responses in this thread to GP, but parent hits the nail on the head. For non-hobbyist problems where you really need a microcontroller, it is s

All reasonable responses in this thread to GP, but parent hits the nail on the head. For non-hobbyist problems where you really need a microcontroller, it is selected by spec (in terms of performance, operational predictability, physical/material properties, power consumption), and then purchased in volume (~ millions). Employing a general-purpose CPU for said problem, while potentially alright when prototyping, would not be satisfactory.