In a late August earnings call, Uber CEO Dara Khosrowshahi sounded optimistic about his company’s “very strong” partnership with Waymo, the global leader in autonomous robotaxis. The two companies had been working together for years in several cities — including exclusively in Austin and Atlanta — and things were going swimmingly, according to Khosrowshahi.

“We believe we’ll continue to operate next year in those marketplaces,” he gushed. “It’s a terrific product, and the on-the-ground partnership continues to be very strong.”

The partnership would turn out to be more brittle than Khosrowshahi let on. Several weeks later, Waymo co-CEO Tekedra Mawakana revealed in an interview that her company planned to end its exclusive robotaxi partnership with Uber, calling the mixed-fleet experience a “crapshoot” for its own customers. Starting in January 2028, riders in Austin and Atlanta will be able to use the Waymo app to hail driverless rides directly. The unlikely partnership, it would seem, was coming to an end.

Waymo isn’t alone in souring on its experience in the Uber ecosystem. Serve Robotics, which operates thousands of sidewalk delivery robots in dozens of US cities, recently announced it would not renew its partnership with Uber Eats when it expires in 2027.

AI chatbots may be getting all the attention these days, but robotaxis and sidewalk bots are the ones putting in the physical work

AI chatbots may be getting all the attention these days, but robotaxis and sidewalk bots are the ones putting in the physical work. Autonomy is now a true commercial business — albeit an unprofitable one — and the companies building the robots no longer want an intermediary standing between them and their customers. That’s bad news for Uber, which aspires to be the premier robot-powered ridehail and delivery app on the planet.

After Waymo and Serve realized their respective visions for how to operate and bring autonomy to the world differed from Uber’s, the mega platform’s quest looked to be in doubt.

“This thing is real. It used to be a dream,” Serve Robotics CEO Ali Kashani told me recently. “And now you see Waymos everywhere, you see Serve robots everywhere. So some of the tensions about how that future is going to shake out [are] now actually coming to light.”

Waymo won’t be exclusive to Uber in Austin, Texas, and Atlanta after January 2028.
Photo: Getty Images

At first, Uber’s pitch to robotaxi and sidewalk robot operators was simple: Let us help you get off the ground by connecting your vehicles to our massive customer base and encouraging a skeptical public to give them a try. Initially, it seemed like a pretty good deal. If you’re a cash-strapped autonomous vehicle startup, why not piggyback on Uber’s platform, where hundreds of millions of people summon cars or order something delivered every day?

“I call this the Trojan horse,” said Harry Campbell, who writes about the autonomous vehicle industry in his newsletter, The Driverless Digest. “[Uber] introduces the Waymo product to a bunch of actual regular users and riders of Uber who would have never considered AVs.”

It wasn’t just Waymo and Serve that saw the benefit — more than 30 companies around the world have deals with Uber to feature their autonomous vehicles on its app. Amazon’s Zoox just launched with Uber in Las Vegas. China’s WeRide has vehicles operating on the Uber app in Abu Dhabi, Dubai, Riyadh, and Madrid. Avride, the tech startup that spun out of Russia’s Yandex, has both autonomous vehicles and sidewalk robots active on Uber and Uber Eats in Dallas.

This is all part of Uber’s big play to become the default robotaxi and robot delivery app in the world, Khosrowshahi said on that August earnings call. Not just one or two companies — dozens. By 2028, Uber says it expects to have “tens of thousands” of unsupervised AVs, across more than 50 cities. And citing a recent analysis by Jeffries, Uber notes that in cities where robotaxis have been operating for a while (like San Francisco, Phoenix, and Los Angeles), the company’s business is actually growing faster — and capturing a bigger slice of the ride-hailing market — than in the rest of the country.

“[Uber] introduces the Waymo product to a bunch of actual regular users and riders of Uber who would have never considered AVs.”

— Harry Campbell

“We want to make sure that we’re not dependent on one partner,” Khosrowshahi said in August. “We’re absolutely seeing a plethora of newer players in the AV ecosystem, just like you see in the foundation model space.”

The only problem with that strategy is that partners eventually get antsy and want to split off on their own, Campbell said.

“Waymo, I think, kind of held the cards and tried this experiment with Uber,” he said. “And within about six to 12 months, I think they knew it wasn’t working that well.”

One of Waymo’s biggest problems was consistency. Uber’s theory was that customers could hail a ride and get paired with either a robotaxi or a human driver, whichever was closer. But as Waymo discovered, many riders preferred to know exactly what they were getting.

“Early in adoption, predictability matters because many consumers are still forming their first impressions of autonomous vehicles,” TC Eley, an assistant professor of management information systems at the Villanova School of Business, told me in an email. “If riders specifically choose a robotaxi but receive a human driver, they lose an opportunity to learn about the technology on their own terms.”

The flip side can also be true. “If [riders] do not want to ride in a robotaxi but are given one, this first impression may lead to the customer having a negative experience and not wanting to ride in an AV in the future,” Eley added.

Uber has since pushed back on the idea that it wasn’t willing to adapt to Waymo’s needs. According to Bloomberg, the ridehail company pitched the idea of a dedicated robotaxi button in its app for Austin and Atlanta — but Waymo declined the offer. (A Waymo spokesperson didn’t reply to a request for comment.)

Serve, which operates more than 2,000 sidewalk robots as the second-largest autonomous fleet in the US behind Waymo, illustrates a similar dynamic. Rather than renew its partnership with Uber Eats, the company launched a direct developer API allowing any small business to build its own robot delivery service using Serve’s fleet.

The decision, according to Kashani, was largely economic. Delivery apps like DoorDash and Uber Eats take a massive 15 percent to 30 percent cut on every order, largely because paying human couriers is expensive. By letting businesses connect directly to its fleet through software, Serve wants to let local shops — like bakeries, florists, and pharmacies — add robot delivery straight to their own websites and checkout registers without paying steep middleman fees.

“At least in our case, there was a different vision for how we wanted to operate and kind of bring this to the world versus how Uber wanted to,” Kashani said. “They’re doing what’s right for their business, and we’re going to do what’s right for ours.”

In a statement, Uber spokesperson Noah Edwardsen said the fault lies with Serve, not Uber.

“We believe autonomy will be a key part of the future of both delivery and mobility, and we continue to scale with many partners around the world,” Edwardsen said. “Beginning in early 2026, we repeatedly raised concerns to Serve about operational performance and reliability, and unfortunately performance did not improve to the level we required.”

Front angle view of a parked Uber self-driving car.

Uber once aspired to build its own self-driving cars — an effort that ended in tragedy.
Photo: Vjeran Pavic / The Verge

It wasn’t long ago that Waymo and Uber were locked in a bitter standoff over the future of autonomous vehicles.

In 2017, the Alphabet-owned company sued Uber over allegations of trade secret theft and patent infringement. The case went to trial a year later but ended abruptly when the two sides reached a surprise settlement after just a week of testimony. Uber later admitted that it misappropriated some of Waymo’s tech and agreed to license it for future use. Anthony Levandowski, a former Google engineer, was sentenced to 18 months in prison for stealing Waymo’s trade secrets. (He was later pardoned by President Donald Trump.)

Infamously, Uber once aspired to develop its own fleet of robotaxis with the intention of eventually replacing its human drivers, but the program was effectively dismantled after a woman was struck and killed by one of the company’s test vehicles in March 2018.

Uber’s exit gave Waymo an opening. The company launched its own ridehail app, Waymo One, in late 2018, aiming to own the entire stack: building the tech, acquiring the customers, and controlling the rider experience. Then, the covid-19 pandemic put the nascent robotaxi business on its back foot. Many autonomous startups had to lay off workers, while others shut down completely. That gave Uber its own opening: pitch itself as the de facto app for all the robots.

“At least in our case, there was a different vision for how we wanted to operate and kind of bring this to the world versus how Uber wanted to.”

— Serve Robotics CEO Ali Kashani

Yet Uber could never go all in on autonomy without threatening the millions of human drivers and couriers currently powering its platform. This conundrum has increasingly put Uber in the awkward position of needing to appear fully prepared for an autonomous future for investors while quietly trying to slow it down as much as possible.

In Washington, D.C., for example, Uber lobbied against a bill legalizing robotaxis in the nation’s capital, arguing it would displace human drivers. The company pushed to amend the legislation to require AV networks to include a “hybrid human network” — a regulatory maneuver designed to force Waymo back into its arms. Waymo pushed back, and the bill ultimately stalled.

Behind the scenes, Uber has also been hit by a wave of departures among its top AV executives, including global head of autonomous mobility and delivery Sarfraz Maredia, head of AV business partnerships Guy Peterson, VP of engineering and science Danny Guo, and head of legal for autonomous mobility and delivery Lando Juarez.

Even though Uber drivers worry that self-driving cars will put them out of work, competition from robotaxis has actually forced Uber to take a smaller cut of each fare in certain cities — meaning human drivers are making a bit more per ride for now. In the long run, however, drivers face a much bigger risk of losing customers completely to Waymo’s app, which is already happening in San Francisco.

“What we’re seeing right now — especially in cities like San Francisco, where you can only call Waymo directly on the Waymo app — is that it’s cannibalizing a lot of Uber and Lyft,” Campbell said. “They came in [and] took entry market share from Uber [and] some from Lyft. Anecdotally, talking to riders, they love Waymo… saying, ‘Yeah, I don’t take Uber anymore. I want to only take AVs.’”

Serve is not renewing its partnership with Uber Eats in 2027.

Waymo isn’t ready to completely put Uber in its rearview mirror. Even after January 2028, Uber says it hopes to keep hosting Waymo rides via non-exclusive agreements. Meanwhile, Waymo’s newer partnership with Lyft in Nashville, where Waymo runs its own app for core demand while using Lyft to soak up excess supply during off-peak hours, seems to point to a new dynamic in the industry.

“Many digital ecosystems evolve through what researchers call ‘co-opetition,’ where firms cooperate and compete simultaneously,” Villanova’s Eley said. “The challenge emerges when customer data, pricing power, and user relationships become strategic assets. As markets mature, participants often seek greater control over those assets.”

Still, if Uber continues to fight standalone robotaxi apps in major markets like Washington, D.C., and New York, the trench warfare could intensify. Another major test of Waymo’s independence is playing out behind the scenes: whether outside fleet operators — like Moove or Avis — are willing to step in and buy large numbers of Waymo vehicles. If third-party companies start spending hundreds of millions of dollars to own and manage these cars themselves, it will be the clearest sign yet that running a robotaxi business could become profitable.

Ultimately, the question isn’t whether driverless vehicles and delivery bots will take over city streets — it’s who will own the relationship with the person tapping “request” on their phone. For years, Uber pitched itself as the indispensable bridge between robotics startups and the mass market. Now that the robots have learned how to navigate the real world, they’re deciding they don’t need a chaperone.

Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.

Share.
Exit mobile version