Uber's CEO explains why your kids probably won't need a driver's license -- and how Uber plans to orchestrate a world of robot cars, flying taxis, and AI gig work.
Imagine you call a taxi, but nobody is driving it. The car drives itself using cameras and computers, like a really good video game AI -- except it's real life. That's a robotaxi.
Right now, Uber has about 10 million human drivers around the world. They do 40 million trips every single day. But robot cars are starting to join the fleet. In cities like Atlanta and Austin, when you open the Uber app, a self-driving car might show up instead of a person.
Uber's CEO, Dara Khosrowshahi, says these robot cars are still really expensive -- way more than a normal car. So for the next few years, you'll still see plenty of human drivers. But in about 10 years? You probably won't need to drive at all. And in 25 years, humans will be measurably worse drivers than robots.
Think of it like horses. People used to ride horses everywhere. Then cars came along. Now, very few people ride horses to get around -- it's just for fun. Driving might go the same way.
Uber isn't making its own self-driving cars. Instead, it partners with companies that do: Joby Aviation for flying taxis, BYD for affordable electric vehicles, and Chinese AV firms like Pony.ai, WeRide, and Baidu. Uber's real advantage is its dispatch brain -- the system that decides which car to send to which rider.
That system is shockingly sophisticated. When you request a ride, Uber doesn't just send the closest car. It predicts what will happen across the entire city in the next 5-6 seconds -- who else is about to request a ride, where demand is shifting -- and makes a globally optimal decision. As Dara Khosrowshahi put it, they might leave a closer driver free because they predict someone else nearby will need it.
This is why Dara says he's "psyched for machines." Human drivers can decline rides, behave unpredictably, cancel last-minute. Robots accept every dispatch and do exactly what the system says.
When Dara joined Uber, the company was losing $4 billion a year. Now it generates roughly $10 billion in annual cash flow. His argument: that scale means Uber should be taking bigger risks, not smaller ones. If a billion-dollar bet fails when you're losing $4B, that's fatal. If it fails when you're generating $10B, you're fine.
The expansion playbook follows a simple rule: "It's got to rhyme." Rides led to Eats (moving food). Eats led to Uber Freight (moving bulk). Ground transport led to Uber Elevate / Joby (air mobility). The newest branch is Uber AI Solution -- putting gig workers to work on AI tasks like data labeling and model testing. It doesn't involve movement, but it rhymes with Uber's core identity as a flexible work platform.
Uber currently has 10 million drivers. Rather than shrinking, Dara's target is 20 million platform workers by 2035 -- doing different kinds of tasks. He frames automation as augmentation: in his experience, companies automate 20-30% of work; the rest shifts to oversight, quality control, and new categories.
Peter Diamandis called Uber a "societal capacitor" -- absorbing and deploying human labor as the economy shifts. The old social contract of a job for life is gone; now it's a job for a week, then you move on. Uber's platform is built for exactly that world.
One striking statistic: roughly half of all US court cases are car-accident related. The insurance model for AVs is still being figured out -- it'll likely be multi-layered, with the AV manufacturer covering the vehicle's decisions and the platform providing an additional coverage layer. The actuarial challenge is volume: all AV trips worldwide last year were less than 1% of Uber's growth alone. The data doesn't exist yet to price risk properly.
Through its partnership with Joby, Uber is planning a network of vertiports -- landing pads designed for mass-market eVTOL (electric vertical takeoff and landing) aircraft. Uber's ride data tells them exactly where to place these for maximum impact: airports, city centers, and high-traffic corridors. Real estate near future vertiport sites could be a significant opportunity.