One of the big reasons I'd say is that 90% of Python work doesn't actually benefit from async? Basically nothing in the data science / simulation / etc. world benefits at all. Web development it does, and Django has sprinkled it in (but not DRF), Flask has a fork that's async and FastAPI is async. At work people have suggested we should switch our Flask code to async to 'make it faster' but for us, we're largely using Flask to server ML models and so there's no benefit by being async at all since we're largely compute bound within the request cycle.

You'd save a bit of memory by letting 1 thread handle multiple connections, but that's about it.