I think people in the Python world just don't yet have a good mental model of async either. I recently have had conversations with a number of people who have switched to FastAPI because "it's faster" even when they're using it to serve ML models which are compute bound and there's nothing you can await on.