People constantly cite self-driving cars as Uber's salvation, but never really seem to think through the implication. Hypothetically, let's say that self-driving cars are invented next year. Necessarily, these will start in high-end vehicles (as nearly all new car features do) before getting rolled out to the broader ecosystem. Now if you're Uber you face one of two choices:
1) You enter into perhaps the most capital-intensive business of all time and buy a fleet of these new self-driving cars. This would be a humongous cash raise; NYC alone would likely need $1B in vehicles to match current coverage levels. You'd also need service and maintenance facilities (or pay for someone else to handle it)
2) You wait for enough of these vehicles to be bought by consumers to create a ride share network akin to the current one, except without the drivers
Option 1 is likely a non-starter. Uber-style businesses are popular with investors precisely because they avoid this business model. Buying the vehicles and facilities to maintain current US coverage would probably run $100B or more. That's not really feasible.
Option 2 saves Uber from that awful capital raise, but means they're waiting for self-driving cars to reach high enough saturation within the broader vehicle fleet that they can use as rideshares during owners idle periods. Based on what Google suggests for a typical car upgrade cycle, this would be 5 years or more. It also probably doesn't improve their margins enormously, because they're still licensing the car from an owner who will demand a cut of fares (whether it be directly or as an hourly lease fee). The assumption is that fee is materially lower than what drivers want for their car and time, but there's no reason to believe that's true (especially given, as many studies have suggested, drivers in most markets currently lose money on a per-mile basis). During the initial period of self-driving adoption, owners fees will be higher, as there are fewer of these vehicles available and owners will have better bargaining power via their scarcity.
But while Uber waits for adoption in scenario 2, their first-mover advantage in rideshare erodes. If you're GM or Mercedes, this looks like a pretty high-margin business, and it looks really tempting to use the self-driving tech you're already putting into these vehicles to spin-up a competitor, and right now there's nothing on the books saying you can't simply lock Uber out of it.
Long-term, that's my expectation: that self-driving cars will exist and Uber will die to competition from OEMs (or some kind of alliance between OEMs and Waymo).
[e]fixed some formatting.