Uber and Wayve launch London robotaxi service with safety drivers still in place

Londoners can now hail a self-driving car through the Uber app, provided they are willing to accept a catch: someone will still be sitting behind the wheel, ready to grab it. The ride-hailing giant and the British autonomous vehicle company Wayve began operating the capital's first robotaxi service on Thursday, using fewer than 20 Ford Mustang Mach-E electric cars fitted with Wayve's driving software. The launch is modest by the standards of San Francisco or Shanghai, but it represents the most tangible step Britain has taken toward putting autonomous vehicles on its public roads.

A deliberately small start

The fleet size tells its own story. Fewer than 20 cars is a trial, not a service. Uber customers in London cannot specifically request a robotaxi; they may simply be matched with one when they book a ride, and they can accept or decline. The price will be the same as a regular Uber trip. There will be no prompt to tip, which is a small signal that the company does not yet consider this a full commercial product so much as a live demonstration.

Sarfraz Maredia, Uber's head of autonomous mobility, said the deployment would start with under 20 vehicles and expand from there. He gave no timeline for expansion, nor for when the safety drivers might be removed. The British government's pilot framework requires operators to secure specific approval before running fully driverless services. Official UK statistics show London's population has grown past nine million, making it by some distance the largest city in Western Europe and a market any autonomous vehicle operator would want to crack eventually.

Why London is a hard problem to solve

London is not Phoenix. The Arizona city, where Waymo began its early deployments, features wide, gridded streets, predictable weather, and relatively light traffic by big-city standards. London is the opposite. Its road network grew organically over centuries, producing narrow lanes, irregular junctions, and one-way systems that confound even experienced human drivers. Congestion is routine. Cyclists, buses, motorcycles and electric rental bikes compete for the same tarmac.

Then there is the pedestrian question. Jaywalking is not a criminal offence in Britain, unlike in many US cities. People cross the road wherever and whenever they judge it safe to do so, and autonomous vehicles must be able to anticipate this behaviour without relying on the legal deterrents that simplify the problem elsewhere. A city where pedestrians routinely step into the road is a more demanding test environment than one where they wait at designated crossings.

Wayve's technology has been tested on London streets for some years, and the company has chosen to tackle a difficult market first rather than starting somewhere easier and moving in later. That approach has commercial logic: if the software can handle London, it can handle most European cities. But it also means the early results will be closely watched by competitors and regulators alike.

The regulatory holding pattern

Britain left the European Union in 2020, and one consequence is that it can write its own autonomous vehicle regulations without waiting for Brussels. The UK government has been keen to present Britain as a place where self-driving technology can develop, and has introduced legislation intended to create a framework for commercial deployment. But legislation and operational reality are different things. The pilot framework under which Uber and Wayve are operating still mandates a human safety driver for every trip. Removing that requirement needs separate approval.

Maredia was blunt about the timeline for going fully driverless. "That process is still playing out," he said, and it is not clear how long it will take. That uncertainty matters. Every month a safety driver sits in the car is a month of wage costs that undermine the economic case for autonomy. The whole point of a robotaxi is to remove the driver and cut the fare. With a driver present, the service is essentially a conventional ride-hailing trip with extra sensors and a public relations story attached.

Europe trails the US and China

The contrast with the United States and China is sharp. Waymo, owned by Alphabet, now operates fully autonomous taxi services in 14 American cities, with no human driver present at all. Amazon's Zoox and Tesla are expanding their own autonomous offerings. In China, Baidu, WeRide and Pony.AI run robotaxi services across multiple cities and have begun pushing into the Middle East, partnering with Uber or local fleet operators in Dubai, Abu Dhabi and Saudi Arabia.

Europe, by comparison, remains in the testing phase. Uber launched its first commercial robotaxi service on the continent only last month, in Zagreb, Croatia, in partnership with a local operator and Pony.ai. Even there, human safety drivers are required. The Croatian launch was a deliberate choice: a smaller, less congested city with simpler road conditions than London, Paris or Berlin. OECD work on autonomous vehicle policy has repeatedly highlighted the tension between encouraging innovation and managing public safety, and European governments have consistently erred on the side of caution.

The caution is not irrational. European cities are older, denser and more complex than the sunbelt American cities where autonomous vehicles first took hold. Regulators are also more responsive to public anxiety about safety. A single serious accident involving a robotaxi could set the entire European project back years. But caution has a cost: it slows the accumulation of real-world driving data that autonomous systems need to improve, and it gives American and Chinese competitors a head start that may prove difficult to close.

The competition gathering in London

Wayve is not the only autonomous vehicle company eyeing London. Waymo, the most experienced robotaxi operator in the Western world, has been testing its vehicles on the city's streets. Baidu, the Chinese technology group, is also preparing to offer commercial passenger service in the British capital. The arrival of multiple operators will intensify competition, but it will also create pressure on regulators. If one company can demonstrate a strong safety record, it becomes harder for the government to justify keeping restrictions in place for all operators.

Uber's strategy is to partner with multiple autonomous vehicle developers rather than build its own technology. The Wayve deal is one such partnership; the Zagreb launch involved Pony.ai. This approach gives Uber flexibility, but it also means the company is dependent on the regulatory progress of its partners. If Wayve cannot secure approval to remove safety drivers, Uber's London robotaxi service remains an expensive experiment rather than a scalable business.

What passengers will actually experience

For now, the passenger experience will be almost indistinguishable from a regular Uber ride. The car will arrive, the passenger will get in, and a safety driver will be present throughout the journey. The only visible difference will be the sensor equipment mounted on the vehicle and, perhaps, the sight of the steering wheel moving without the driver touching it on stretches of road where the autonomous system takes over.

Uber has said prices will not differ from standard fares, which removes one potential incentive for early adopters. In some American cities, robotaxi rides have been offered at a discount to build ridership. Uber appears to have decided that in London, the novelty value is sufficient, or that discounting a service with a safety driver would confuse the brand positioning.

The economics of a driver with a driver

The fundamental problem with the current London deployment is that it does not yet test the economics of autonomous ride-hailing. A robotaxi with a safety driver costs more to operate than a conventional taxi, because the vehicle carries expensive sensor and computing equipment in addition to paying a wage. The business case for autonomous vehicles rests on removing that wage. Until the safety driver is gone, the service is subsidised, whether Uber and Wayve acknowledge it or not.

This is why the regulatory timeline matters so much. If the UK government takes two years to approve driverless operations, the accumulated costs of running a subsidised fleet will mount. If it takes five, the economics become harder to justify. Wayve and its investors are betting that the data gathered during the supervised phase will be valuable enough to justify the cost, and that regulatory approval will come before the money runs out.

There is also a competitive dynamic at work. If Waymo or Baidu secures driverless approval in London before Wayve, the British company loses its home-market advantage. Wayve's position as a domestic operator with local knowledge counts for something, but it counts for less if a rival can offer a genuinely driverless ride while Wayve is still running supervised trips.

The data question

One underappreciated aspect of the London launch is data collection. Every supervised trip generates information about how the autonomous system handles real-world conditions, from double-parked delivery vans to pedestrians stepping off kerbs without warning. That data is the raw material for improving the software. London, with its complexity and unpredictability, is a rich environment for training. Wayve may be running a small fleet, but it is gathering disproportionately valuable data from each kilometre driven.

The data also has regulatory implications. If Wayve can show that its system has completed tens of thousands of kilometres in London without a safety driver needing to intervene, the case for removing the driver becomes much stronger. Conversely, if the safety driver is frequently taking over, regulators will want to understand why before granting more autonomy. The early weeks of the service will generate a stream of evidence that could determine how fast Britain moves from trial to commercial reality.