The most serious obstacle here is the fact that there is no such thing as a self-driving car.
I think it is the opposite... Uber's only real way to turn a profit while staying competitive is to get rid of the cost of drivers.
> What's stopping an upstart company to buy/lease 10,000 self-driving cars and let them loose on the street?
The fact that Uber is already working on this, has a platform, has tons of data?
Because deploying Robotaxi fleets will be so capital intensive, no single company is likely to dominate the way Uber has in rideshare. There is enough demand spread out across enough different cities that they won't need to compete directly for a long time.
If they can't maintain their market share in a driverless car world, it's difficult to imagine how they can justify their current valuation.
https://newsroom.uber.com/san-francisco-your-self-driving-ub...
https://www.theguardian.com/technology/2016/dec/16/uber-self...
You're right that self-driving cars would be a major challenge/opportunity. Re: what's stopping an upstart, letting self-driving cars loose on the street is inefficient. There is a huge amount learned/accrued intelligence involved in how those self driving cars are orchestrated for optimize the marketplace. So that's where a key capability is going to be important even in a world where self-driving is commoditized.