Wrong, Uber and Lyft will be the operators. You think Google really wants to get into fleet management?

1) That's at least 3-4 years from now before it will be practical in most situations, which is a lot of revenue for Uber and Lyft.

2) Uber is already investing heavily in its own self-driving cars and has stated plans to buy as many self-driving cars as it can if someone else beats them to the punch. So really the choice will be between owning your own car (very high upfront cost and maintenance costs) and using a self-driving car service. I think many people will opt for the latter.

The network is important because we're not going to go from manual driving to self-driving overnight. It's a transition that will take years. For most Uber users, the change will be transparent; the only practical difference would be that there isn't someone in the front seat of the car that comes to pick you up. Growing the userbase and making the investments in becoming the primary method of ride acquisition is useful because that will carry over into the self-driving era.

What's the point of a self driving Tesla? A huge portion of the Tesla appeal is the high performance. Self driving cars are going to be the opposite of that: gentle acceleration and strictly following traffic laws and speed limits.

How do you hail that car? Using an app? What price should you pay? Which Tesla should come to you? These are all problems the middle man is solving.

No middleman and no competition? No business is really altruistic, despite their public image of being the good guys.

Have you tried Ubereats? Good experience with potentially interesting network effects for their business.