That's a lot of imponderables based on unknown technologies, regulations, and economics, though. Right now Uber doesn't have to worry about fleet cap- and opex,

That's a lot of imponderables based on unknown technologies, regulations, and economics, though. Right now Uber doesn't have to worry about fleet cap- and opex, so the assumption is that buying and maintaining an autonomous vehicle fleet will provide more relative income than their current business practices. It also assumes that Uber's current loss-leader practices will not put the company under before commercially-available autonomous vehicles are in the market, that debt service will not constitute a major issue for company profitability, and that Uber's IP portfolio and marketshare will represent a significant barrier to entry for competitors with fleet management experience (such as rental car companies and existing taxi/limo companies). That's a lot of unproven assertions to make a bet on. If driverless cars get here fast enough, the cost profiles work, and Uber has enough IP to use against challengers, then it can work. But, personally, I'd rather have a business model that actually works today than one that might potentially work in the future.