It's hard to believe the tiny microprocessors can be fast and flexible enough to effectively emulate or replace application-specific chips on something modern - is this some hack unlikely to be seen outside the PocketBeagle or do you think this will become more common elsewhere?
It's a bit of a hack, but it's also somewhat a cry of frustration.
For example, the cheapest I2C ADC I can cough up on Digikey has fewer bits of accuracy, fewer channels, is slower, lacks an internal reference, lacks an internal clock, is larger and is more expensive than these full microcontrollers.
Using a microcontroller means that my interface isn't limited to to one or two I2C addresses. Using a microcontroller means that my interface can be I2C, SPI, UART, CAN, Sent or whatever I can bit bang out of these.
The downside, of course, is that you now have two pieces of firmware you need to keep in sync and to keep track of.
Generic components will always be higher volume, therefore cheaper, than previous generations specialised components. In the retrocomputing scene there are lots of little ARM boards that emulate larger chips. It’s a matter of time for even the more complex ones.
OTOH, we now have to manage the complexity of software even in components we would never imagine would run programs.
