I think often the problem is not knowing how to implement the vision. Say you want to start an electric car company in the mid 2000ies, how would you go about it without having gotten rich from selling some startup first?
As you break down the problems into further smaller problems you might encounter a problem which is possible to tackle with the currently available resources.
One of the possible outcomes here might be identifying a team within a comp / university who is working on the same problem and then joining them. That can be the first step in the long marathon of innovation.