Uber and Pony.ai Will Test London Robotaxis—A Launch Date Is Still Missing

Uber and Pony.ai disclosed plans on October 8, 2026, to test Pony.ai’s Gen-7 robotaxis in London in the coming weeks. The timetable covers road testing, not a ride that a customer can book: there is no announced date for a Pony.ai passenger service on Uber in the capital. That distinction matters because the partners already offer bookable autonomous rides together elsewhere in Europe, at a different stage of deployment.

Pony.ai said it looked forward to “working with local officials, and our partners” as plans for UK trials progress. The London announcement leaves the fleet owner and operator, vehicle count and safety-driver arrangements open. Its promised start is measured in weeks; the later step of carrying passengers has no published schedule.

The London timetable begins with testing

“In the coming weeks” establishes the next phase, but it does not identify a specific start day, road, neighbourhood or length of trial. The proposed vehicles are Pony.ai’s seventh-generation robotaxis, which have yet to be assigned a public booking area in London. A vehicle seen on the street during a trial would therefore say little on its own about when it might appear as a ride option in the Uber app.

Testing also leaves two different passenger-service questions unresolved. One is when riders could be matched with a Pony.ai vehicle through Uber; the other is whether that service would initially have a person ready to take control. The answers need not arrive together. Zagreb, where supervised autonomous rides became bookable while fully driverless operation remained a future goal, shows why a booking date and a driverless date should be treated separately.

A London fleet operator is another practical part of the schedule. Someone must own or fund the vehicles, maintain them and run the day-to-day service while Pony.ai supplies the driving technology and Uber connects riders to trips. The European agreement allows those responsibilities to vary by city. Until a local operator and the extent of the first fleet are identified, the regional deployment target cannot be translated into a London service size or coverage map.

Europe’s partnership has a working model in Zagreb

In an August 13, 2026, expansion agreement, Uber and Pony.ai set a target of more than 2,000 robotaxis across Europe, extending their Zagreb partnership to four additional cities. Pony.ai’s assigned role is Level 4 autonomous-driving technology, along with rider-experience and operational expertise. Uber’s role is access to customers, booking, payment and customer service. Local partners can take on daily fleet operations, and funding or ownership can sit with different participants in different markets.

The scale is European, not London-specific. The agreement offered no per-city allocation and said rollout details would be released in phases. Adding London to the companies’ plans thus brings a named market into view without establishing how many cars will run there or who will operate them. It also leaves open whether the first London fleet will resemble Zagreb’s in its ownership or supervision arrangements.

Zagreb offers a concrete example of how the split works. In Uber’s August 19, 2026, launch announcement, riders in the Croatian capital could book autonomous trips through the Uber app, with Pony.ai providing the driving system and Verne owning and operating the fleet. A licensed operator monitored from behind the wheel at launch. Riders requested UberX or Comfort and could be offered an autonomous vehicle when one was available, rather than selecting a dedicated Pony.ai ride.

That operational pattern illustrates what a commercial service announcement would add to the London testing plan: an operator, a defined availability area and an actual way for riders to obtain a trip. Verne’s Croatian role does not establish its role in Britain. Nor does the European vehicle total specify what London’s share will be; both depend on later local deployment decisions.

UK rules make the safety-driver question consequential

UK government pilot guidance draws a clear line between trials with a safety driver, who must be ready to resume control, and pilot deployments operating without one. A driverless passenger pilot also requires assessment of the vehicle and its operation, an automated passenger service permit and, for a taxi-like service, consent from the relevant local licensing authority. Those requirements concern the form of operation, rather than the marketing term “robotaxi.”

A supervised road trial, a supervised ride carrying paying customers and a driverless passenger pilot are separate milestones. The London plan specifies the first step, but its public description does not establish which supervision arrangement the initial vehicles will use. Even a subsequent announcement that trips can be booked would need to say whether a licensed person remains behind the wheel before those rides could accurately be called driverless.

For passengers, the decisive London update will be the one that names a service date and operating area, identifies who runs the fleet and explains how a Pony.ai vehicle becomes available in Uber’s app. The safety-driver arrangement will then show whether that launch offers supervised autonomous rides or a driverless service under the pilot scheme.