It's interesting ,but the abstract lacks any concrete details, like feature size in nm. Does anybody has access to the paper , and can share such details ?
There is a drawback to having really small computers, they are easier to lose.
I can imagine a whole new class of software that has to run on these things. Talk about a network problem.
So I'm guessing that if it's a nanoelectrical FSM, you have to hook it up to some kind of nanoelectrical RAM to get a "real" computer, in the sense of a limited-tape Turing Machine?