Imagine hailing a ride in central London and the car that pulls up has nobody in the driver’s seat. That’s not happening yet. But it just got a step closer. Uber and Chinese self-driving company Pony.ai plan to start London robotaxi tests within weeks.
A robotaxi is a taxi that drives itself, using cameras, sensors and software to read the road. It’s physical AI in its plainest form: software that has to see, decide and move a real car through real traffic. If that idea is new to you, our guide to physical AI covers the basics.
So what actually changed?
On Thursday, October 8, the two companies said they’d start testing Pony.ai’s seventh-generation robotaxis in London. The timing, as TechCrunch reported, is “in the coming weeks” and nothing more precise. It’s part of a wider partnership that covers Europe.
Now for what they didn’t say. According to The Next Web, there’s no date for a paying service and no number of cars. Nobody has said who’ll own and run the London fleet either. So this is a test announcement, not a launch. Keep that in mind when the headlines get excited.
Still, it’s a real signal. Pony.ai says the trials will run “under the UK’s frontier plan” for self-driving vehicles. In other words, the company wants to work inside Britain’s official process. That matters if you’re watching how fast this moves.
How the London robotaxi setup will probably work
Here’s where it gets interesting. London will likely copy the model the partners already use in Zagreb, Croatia. That setup splits the job three ways, and honestly, it’s a clever one.
First, Pony.ai supplies the self-driving system and the car itself. That’s the Arcfox Alpha T5, which it developed with Chinese carmaker BAIC. Next, a startup called Verne owns and runs the fleet. Verne was set up by Mate Rimac, the founder of Rimac Group. Finally, Uber brings the part you actually see: the app and its ride-hailing network.
The Zagreb service went live in August 2026. Even there, though, the first European rides on Uber had a safety driver on board. That’s a person who sits behind the wheel and can take over if something goes wrong.
So why split things up like this? Because each company does what it’s best at. The tech firm builds the brain, a local operator handles the cars, and the app handles the customers. It also means no single company has to carry every cost.
Why Uber keeps adding robotaxi partners
Uber isn’t betting on one horse. It has deals with more than 30 autonomous vehicle companies, TechCrunch says. It also expects to offer self-driving trips in up to 15 cities worldwide by the end of 2026. That’s split roughly evenly between the US and other countries.
London already has an Uber robotaxi partner, too. In September, Uber and British startup Wayve launched the city’s first robotaxi service, with a safety driver in every car. Uber also invested in Wayve. Wayve raised $1.2 billion in February 2026. That round could grow to $1.5 billion with an extra $300 million from Uber. The catch is that Wayve first has to deploy robotaxis, starting in London.
Then there’s Dubai, where Uber has separate robotaxi deals with Baidu and WeRide. See the pattern? Uber wants to be the place you book the ride, whoever builds the car. Pony.ai, meanwhile, runs robotaxi services in four Chinese cities and is pushing into Europe and the Middle East, including Qatar. The two companies first teamed up in May 2025, and they’ve talked about putting 2,000 robotaxis on European roads.
The catch (because there’s always one)
Testing isn’t the same as a service you can book. And so far, the London rides that exist still have humans behind the wheel.
Rules are the other big question. Britain created a legal route for driverless taxis in May, but the Uber and Wayve service didn’t use it. Pony.ai’s tests will fall under the UK’s frontier plan. Still, nobody has given a timeline beyond the coming weeks.
There’s also the question nobody’s answered: who actually runs the London cars? In Zagreb, it’s Verne. In London, we don’t know yet. That gap matters to us, since someone has to keep the cars charged, clean and fixed.
Okay, but should you care about a London robotaxi?
Honestly, yes, even if you’ll never visit London. Robotaxis are one of the clearest tests of physical AI. A chatbot can make a mistake and you shrug. A car can’t afford that. So every city that allows these tests tells us how much trust regulators are ready to give machines.
We also think the three-partner model is worth watching from Armenia and other smaller markets. A global tech supplier, a local fleet operator and a ride-hailing app can share the risk. That’s a path where local companies get a real seat at the table, instead of waiting for a giant to arrive alone.
If you want to go deeper on the tech, our piece on how sensor fusion helps robots combine their senses explains how these cars read the street. So here’s our open question: when the first London robotaxi finally drops its safety driver, will riders trust it, or will they wait to see who goes first?