Uber says AI agents booked just four rides in Q3 2026

The robots are coming for your commute. Eventually. Maybe.

Uber executives said on the company’s third-quarter 2026 earnings call that third-party AI agents arranged exactly four rides during the quarter. That is not four million or four thousand. It is four, a number you could count on one hand with a finger to spare.

Four rides, 500 burritos, 320 million trips

The disclosure came during Uber’s earnings call on November 4, 2026. Executives walked investors through how outside AI assistants are performing as a booking channel for the platform.

Food did slightly better than rides. Uber said AI agents fulfilled “around 500” Uber Eats orders during the week of Halloween.

For scale, Uber handled roughly 320 million rides and Uber Eats deliveries over that same stretch. That works out to about one agent-placed order for every 640,000 trips and deliveries.

Uber leadership didn’t spin the figures. Executives characterized the results this way:

“very, very early and very limited traction”

Executives also reaffirmed that Uber intends to keep building relationships with multiple AI companies, framing the strategy as a long game rather than a quarterly revenue driver.

The rest of the quarter looked a lot healthier

If the AI agent numbers were a rounding error, the core business was anything but. Uber reported gross bookings of $59.8 billion for Q3 2026.

That figure represents growth of more than 20% year over year. It also marks the fifth consecutive quarter Uber has cleared that 20%-plus growth bar.

Profitability moved in the same direction. Non-GAAP earnings per share came in at $0.88, up 35% from the year-earlier period.

Uber also updated investors on its autonomous vehicle plans. The company is targeting around 700 fully driverless vehicles on its network by the end of 2026. It is aiming for more than 2,000 by the end of 2027.

Two very different AI bets

It helps to separate the two AI stories Uber is telling, because they are easy to blur together.

The first is the autonomous vehicle push. That is AI doing the driving. The car itself replaces the human behind the wheel, and Uber acts as the marketplace connecting riders to those vehicles.

The second is the agent experiment. That is AI doing the ordering. A chatbot or assistant built by another company books the ride or the meal for you, so you never open the Uber app yourself.

Q3 suggests the second idea is far further from mainstream adoption than the first. Uber has concrete fleet targets for driverless cars. For agents, it has four rides and a cautious tone.

What to watch

What to watch from here: whether Uber keeps disclosing agent-driven volumes in future quarters, and whether those numbers start climbing off their very low base. Four rides makes for an easy comparison. Even modest growth would look impressive in percentage terms, so the raw figures will matter more than the growth rates.

Also worth tracking is how the driverless rollout progresses against the 700-vehicle target for the end of 2026. That program has firm milestones attached, which makes it the AI story investors can actually measure. The agent channel, for now, remains more of a science project than a sales channel.