Software would be great... if it worked. It's going to take a very clear failure of all self-driving software for anyone to even possibly invest in smart roads.

Software would be great... if it worked. It's going to take a very clear failure of all self-driving software for anyone to even possibly invest in smart roads. Smart roads don't seem to be a startup thing to do - massive friction (like most infrastructure) and no promise of fantastic profits. Imagine the investors' horror of building smart roads a few years before someone figures out a pure software solution. It's hard to be sure that it won't happen.