I wonder if we’ll see more of a movement towards virtual threads Java Loom style where cheap threads in the language and cheap cores on the cpu will combine to create a programming style where you’ll just dispatch little worklets all over the place without worry about “is this a large enough amount of work to justify a thread” the way we do now. I also wonder if that will cause a trend towards optimizing startup latency in both language threads and the efficiency cores.

I think there’s a Smalltalk implementation that tries to do something like that.