Comment by ethbr1

How would that work, economically?

Wasn't a key enabler of early transitor work that required capital investment was modest?

SotA AI research seems to be well past that point.