[Epistemic Status: Opinion]
The following is just one vision, but it is an enticing one, and something like it will probably come to bear.
Owning an autonomous car will not make sense, because as long as they’re sitting in your garage they are being wasted. Anyone who did own one would be under economic pressure to lease it out to a fleet. They would ultimately be convinced to sell it to the fleet, because it is worth much more to the fleet than it is to them. If our markets did not completely suffice this expropriation of commons resources, the people would have to run in torch-and-pitchfork and liberate the cars themselves. It is both evil and stupid to keep these noble service robots pent in a garage.
Transportation will become a utility.
When the cars- algorithms so extensively honed and tested- effectively never crash, they wont need a strong, heavy frame. Since their users are not their purchasers and most trips- laid out in advance- are guaranteed to be short, they will not need large batteries either: Neither a purchasing department nor a routing algorithm has to contend with range anxiety. With the relatively simple electric motor parts and mass production on the dime of the new taxi companies (which will hopefully not be Uber. I see no reason why Uber would really have an advantage here.), they will be cheap as hell relative to the old human-driven cars. They will flood the streets.
A car running a sufficiently sophisticated coordinated driving algorithm will not necessarily ever have to stop moving. At intersections, they’ll be able to weave through each other like schools of fish. If the public don’t believe this they will be invited to download a simulator and watch it for a while, see how it works, and how tolerant it is to interference. The code will be public, of course, as the autonomous network is a public resource, and cars from competing manufacturers would like to be able to coordinate.
Gliding along on low-friction bearings, these lightweight electric cars that never have to stop moving will be gratuitously efficient, while anything carrying more than 10 people must start and stop constantly to let them on and off. To satisfy their conflicting needs, a bus would have to take a highly indirect route. This is the reason busses are slower than driving and there is no reason to perpetuate their waste once you have autonomous electrics littering every curb. Unless you have the same use case as long-distance rail, there is simply nothing to be batched here. You don’t benefit from batching people onto the same trip, you only waste their time.
From the perspective of an engineer, this future may seem very important.
Refutations are welcome.