>A self-driving taxi company is going to look a lot like a car dealership that doesn't sell cars, or a lot like a taxi company that doesn't have any drivers. They're going to have a huge number of cars, parking lots to store cars that aren't in use, and technicians to service, clean and fuel vehicles periodically.
No? You'll need charging stations to charge vehicles and will have enough to meet typical demand, when a large event is expected you have vehicles come in from other areas (Indy 500, Superbowl, etc). If vehicles aren't driving people or charging you use them as courier vehicles.
Yes you'll still maintain some in parking structures at some hours but with no driver needed you can pack considerably more into the same space as a vehicle that would require a human being to enter and exit the vehicle to move it.
Kids in their teens and early 20's are losing their desire to drive, individuals are getting their licenses much later. This is already a sign that a carless society is a very real possibility. These companies know that, they're working on taking the human driver out of the equation, they're building brand recognition, they're getting people comfortable with using a driving service (a large portion of the population has never used a taxi or even public transportation, yet within a few years hopping in a car with a stranger became normal) and they're collecting more data on driving and human movement via vehicle than ever before.
>Meanwhile, they're blowing through most of their venture capital selling rides below cost.
They are investing money in collecting data. The data alone they are collecting is worth the money they are burning through. You can't go buy that data from someone because it has never been collected and there's many ways to exploit that data to generate revenue both now and in the future as technologies become available like autonomous vehicles.