Uber wants to own the robotaxi era without building the cars. “Building a great autonomous driver is one part of building a successful autonomous business,” chief executive Dara Khosrowshahi told analysts on Wednesday. He then listed the rest: dispatch, fleet operations, charging, insurance and regulators. It is a bet the company is spending billions to defend, and one the market is not yet sold on.

On paper, the quarter was strong. Revenue rose 12% to $14.2bn and gross bookings jumped 24% to $58bn. Uber also generated more than $10bn of trailing free cash flow for the first time, it said, and signed up more first-time users than in any period in five years. Yet the stock fell as much as 6%, its worst day in six months, Bloomberg reported.

The immediate trigger was the outlook. Uber guided third-quarter bookings to a midpoint near $59.25bn, just shy of the $59.33bn analysts wanted. Its profit forecast landed below estimates too, CNBC reported. Trip growth also slowed, which Uber blamed “entirely” on Brazil, where the Chinese rivals Didi and Meituan are spending hard on deliveries.

The one question Uber keeps getting asked

None of that explains why the stock is down 12% this year. The real worry is robotaxis. Uber built its business on human drivers, and self-driving cars threaten to route around it. Its answer is to become the platform every robotaxi operator plugs into, rather than build the technology itself, and it is paying to hold that ground.

Uber said it will commit more than $10bn to autonomous vehicles in the coming years, mostly as equity stakes in partners and support for their fleets, Reuters reported. The Financial Times put the ambition at 120,000 driverless vehicles across 15 or more cities. Khosrowshahi said Uber is building “one of the most valuable positions in the AV ecosystem”.

Waymo is the nightmare scenario

The trouble is that Uber’s most important partner is proving the opposite case. Waymo, the US market leader and one of Uber’s first robotaxi allies, is ending their exclusive deals in Atlanta and Austin by early 2028. It increasingly sells rides on its own app. That shows a robotaxi firm can own both the car and the customer, and cut Uber out.

Khosrowshahi played it down. He called Waymo’s product “impressive”, but said Uber did not want to depend on any single partner. He argued Uber’s own ride-hailing stayed “really strong” even in cities such as San Francisco and Phoenix, where Waymo has run for years, with trip growth there accelerating. Uber also runs an AV Lab that feeds ride-hailing data back to the wider industry.

London is the thesis in action

Then came the counterpoint, on the same day. Uber and the British startup Wayve won licences from Transport for London to run supervised robotaxis in the capital, Bloomberg reported. That puts them ahead of planned rollouts by Waymo and China’s Baidu. It is the first time regular Londoners will be able to hail a self-driving cab.

The London cars keep a safety driver aboard, hands off unless they intervene, under a standard private-hire licence. True driverless operation needs further approval. Uber said more than 100,000 people have signed up as first riders. Wayve will run up to 15 electric Ford Mustangs on a one-year trial. It neatly demonstrates Uber’s pitch: own the marketplace, let someone else build the car.

That is the whole bet in miniature. In London, Uber is the aggregator and Wayve supplies the technology, exactly the model Khosrowshahi is selling. In Atlanta, Waymo is walking away to sell rides itself. Uber’s quarter was solid, and its $14.8bn Delivery Hero deal shows it still has room to spend. Which version of the future wins is the question the market keeps asking.

Get the TNW newsletter

Get the most important tech news in your inbox each week.