A significant portion of any real-world AI system is dedicated to extracting relevant features from raw sensor information. Doing that digitally has proven to have a very high cost rated in both watts and operating temperature.

Most machine learning algorithms are by design resilient to random noise. They can even be learned to be resilient to systematic noise (due to say.. variable hardware performance in mass production).

In those cases; a low power analog device with lower reliability guarantees is 'ok'.

I suspect the "decision engines" (the computation enforcing complex logic based on sensor readings) will probably continue to be on CPU's for some time.