Just driving around cars that aren't in use gets brought up with self-driving cars all the time, but it's just a terrible idea, even ignoring the traffic considerations. You want them evenly distributed so they can be summoned quickly, sure, but wear and tear on all the expensive parts of a car is proportional to road miles. You want to park them
somewhere when they're not in use, and as long as you have peak hours you're going to have times where you need more or less cars.
Re: venture capital use, there are currently 65,000 Uber-affiliated cars in NYC alone, according to a random search. If you were to replace all of those with self-driving cars at $50,000 a pop, that would be $3 billion right there. You can argue they'll only need a tenth that number because of self-driving efficiencies and because most Uber drivers aren't constantly driving (for comparison, NYC allows ~13k taxies) and claim they'll be able to manufacture a self-driving car for half that, but you're still talking hundreds of millions of dollars for a single market.
To replace their current fleet of sub-contracted drivers with self-driving cars is going to cost billions and billions of dollars. Uber currently has something like $7B on hand and is burning through nearly a billion a quarter.
So 2021 comes around, and self-driving technology has been perfected by Uber, Google, Tesla and Ford.
You are a VC -- or a regular investor if Uber is public -- looking to invest in a self-driving taxi company.
Two companies are seeking $10B to build out their world-wide fleet of self-driving taxis. Would you rather invest in Uber, which has a successful ride-hailing App and an internally built self-driving car and is $20B in the hole on VC investment already, or would you rather invest in a newly formed corporation, Bob's Friendly Taxi Company, which has nothing, and is going to use its VC funds to buy a bunch of self-driving vehicles and develop a new ride-hailing app from scratch?