Wayve and Uber put 15 supervised cars on the road in September, while Waymo, Baidu and Lyft wait for permission to take the driver out. Who controls that decision may matter more than who writes the best software, and where the money ends up is still an open question.
Since 3rd September, a Londoner who opens the Uber app and asks for an UberX,or an Uber Comfort or an Uber Electric may be matched with a Ford Mustang Mach-E that drives itself- yes, you heard that right.
The fare is the usual one, and a trained driver sits in the front seat, ready to take over, the first British journalist to ride as a paying customer described a mostly smooth trip through central London with a single intervention.
It is the first robotaxi London has offered the public, and it is a small one, with 15 licensed cars in a city with more than 100,000 conventional private hire vehicles.
The label needs a closer look, because Transport for London licensed those cars as private hire vehicles, not as automated vehicles, under a private hire vehicle (PHV) licence, the safety driver in the front seat remains legally responsible for the journey, which is why the service could start in September at all. In practice a robotaxi London launch happened through the existing door.
We think that detail says more about the state of the market than any demo ride does, because it shows which legal route Wayve and Uber took and which one everyone else is still waiting to use.
The Driverless Taxis London Was Promised for Spring
The route that would let a car carry paying passengers with nobody at the wheel is the Automated Passenger Services permitting scheme, created under the Automated Vehicles Act 2024 and brought into force on 15 May 2026.
The government had pulled its timetable forward in 2025 so that small commercial pilots could begin in spring 2026, and spring has come and gone without a driverless permit in London.
The scheme asks an operator to clear three gates in order. The Vehicle Certification Agency has to list the vehicle as capable of driving itself safely, the Driver and Vehicle Standards Agency has to grant the permit, and Transport for London, which the government treats as a consenting authority rather than a licensing authority, then has six weeks to give or withhold consent.
If it says nothing, consent is deemed given.
By the end of September, no operator had completed the vehicle-registration stage or applied for a driverless permit. TfL had not published the guidance operators need to prepare, and its director of strategy told the London Assembly's Transport Committee in July that a great deal of work was under way behind the scenes.
"I do not think we are in a rush to get the guidelines out." — Christina Calderato, Director of Strategy, Transport for London
There is a fair case for caution.
A TfL spokesperson has said any new vehicle licensed to carry passengers must align with the authority's Vision Zero goal, and the permit chain exists to test exactly that. Waymo's UK policy chief has said only that the company is working through the safety approval process and waiting on the same TfL guidance. The wider framework under the Automated Vehicles Act 2024 is not due to take full effect until the second half of 2027.
What nobody can yet say is how long the wait will run for SAE Level 4 operations, the kind with no human driver at all.
Read next: Would You Ride in Alex Kendall's Self-Driving Taxi?
Three Camps, One Bottleneck
The robotaxi London market, specifically, is usually described as a race, but the companies are not all running against each other.
Uber and Wayve are partners, and Uber invested in Wayve's $1.2bn Series D in February, which valued the company at $8.6bn, and committed a further tranche of up to $300m tied to milestones.
Wayve's cars run on cameras and radar, learning to drive from experience without the lidar and HD maps that Waymo's sensor-heavy system depends on, an end-to-end approach the company calls AV2.0.
Uber is also testing Baidu's cars on its network, which makes it a distributor with more than one bet. Baidu began testing its Apollo Go RT6 vehicles at the end of July, with safety operators on board, and they will also be offered through Freenow by Lyft. It expects to open to the public in 2027, after continued talks with TfL and the Centre for Connected and Autonomous Vehicles.
Waymo is the third camp and the only one going alone, the Waymo London operation runs around 100 Jaguar I-Pace cars across 19 boroughs, with Moove as its fleet operator, and the company has said it will not launch with a human behind the wheel.
In January, it set a September target for a commercial launch, and by late summer it was promising a fully autonomous service by the end of the year without saying whether the public would be paying.
Waymo raised $16bn at a $126bn valuation in February, more than ten times Wayve's. So we think that gap in firepower is real, but no amount of capital shortens a six-week consent window, and every camp now has to pass through the same three gates.
What Zagreb Knows That London Is Still Working Out
The nearest comparison is not American, it's Verne, the Croatian company spun out of Rimac, which opened what it called Europe's first commercial robotaxi service in Zagreb on 8 April.
It started with nine cars running Pony.ai's driving system, a promotional fare of €1.99 and a trained operator on board each one, and by 10 September, about five months later, the company and Pony.ai were reported to be carrying invited passengers in cars with no operator at all.
"After launching first commercial service in Europe, we are now going driverless." — Marko Pejković, CEO, Verne
The comparison only goes so far, Zagreb is a city of under a million people, it also operates under a different national regime, and the driving system belongs to a Chinese company, so what worked there will not transfer directly.
We think the useful lesson here is narrower.
Where the approval route is open, the step from supervised to driverless can take months, and the driverless taxis in London are still waiting for depend less on whether the technology can do it than on when three bodies agree that it may.
Wayve's Kaity Fischer has argued that London is the harder test, with 20 times the roadworks and 10 times the vulnerable road users of San Francisco. A robotaxi London service will have to prove that claim in practice rather than in a launch announcement.
Self-Driving Taxis UK Startups Can Build, and Who Gets Paid
The government's forecast puts the self-driving industry at £42bn and 38,000 jobs by 2035, that’s a figure for the whole sector rather than for robotaxis alone. The harder question is which layer of the stack collects the money.
Software is the only sector where UK self-driving startups have their strongest claim; it includes names like Wayve, which has chosen to sell its driver to carmakers and fleet operators rather than run its own cars, and Oxa, which already has self-driving systems working at Heathrow. Distribution sits with Uber, which says it wants to be the world's largest facilitator of AV trips by 2029 and expects to carry them in as many as 15 cities by the end of this year.
Now, the third layer gets the least attention.
Moove, a Nigerian-founded company that began by financing cars for ride-hail drivers, says it now runs around 42,000 vehicles across 29 cities in 13 countries. It raised $250m at a $2.1bn valuation in August, is headquartered in Dubai, and is Waymo's fleet operator for London.
"Autonomy requires fleets, charging, maintenance, data systems and 24/7 operations in every city." — Ladi Delano, Co-Founder and Co-CEO, Moove
Uber's Sarfraz Maredia has been candid about why that layer matters, and it comes down to cost.
"Today, nobody's able to operate AVs cheaper than a traditional human-driven ride." — Sarfraz Maredia, Global Head of Autonomous Mobility, Uber
We think the margin will go to whoever that can fund and run a fleet at a lower cost per mile than a human-driven one, which is why the fleet layer deserves as much scrutiny as the software.
Based on the evidence so far, the UK's strongest position is in software, and the fleet and finance layers sit mostly elsewhere. So, each robotaxi London trip needs at least three companies behind it: one to drive the car, one to book the ride and one to own the vehicle, and out of those three, only the company that drives is British.
The People Who Drive for a Living
The robotaxi London debate also includes the people whose livelihoods depend on the current system. The Licensed Taxi Drivers' Association, which represents black-cab drivers, has been openly sceptical about the whole idea.
"a solution looking for a problem" — Steve McNamara, General Secretary, Licensed Taxi Drivers' Association
The GMB union, whose private hire members are more directly in the path of a service sold through Uber, has asked TfL and the government for a plan to protect drivers before any move to driverless operation. Its organiser David McMullen has warned of "unprecedented levels of social and economic disruption". GMB's congress was actually told that more than 300,000 jobs are at risk, a union estimate rather than an official forecast.
Londoners themselves are cautious of this.
A GLA and YouGov poll of 1,053 residents in June found 29% in favour of autonomous passenger vehicles and 42% opposed, with only 17% trusting the companies to collect and analyse their personal data, yet more than 100,000 people signed up to Uber's interest list.
We think both facts can be true at once, as curiosity costs nothing before the first ride, while trust has to be earned after it.
What Happens When the Driver Gets Out
The driverless taxis in London that were promised for spring 2026 have not arrived yet, and the Automated Passenger Services permitting scheme that would unlock them has not yet been used.
Will TfL publish its guidance, and will a first application reach the Driver and Vehicle Standards Agency before the Automated Vehicles Act 2024 comes into full force in the second half of 2027?
How often have the safety drivers in Wayve's 15 cars had to step in, which is a figure nobody has published?
If Zagreb could move from supervised to driverless in about five months, what would the next London robotaxi milestone need to look like on narrower streets and in heavier traffic?
And when the first London robotaxi passenger pays for a ride with nobody in the driver's seat, who gets the money? And will any of it stay with British companies?
Editorial Note
This piece is purely based on secondary research.
We welcome any additional context, corrections or first-hand perspectives from Wayve, Waymo, Baidu, Lyft, Uber, Moove, TfL, the DVSA or anyone with direct knowledge of the permitting process.
Sources: GOV.UK (Automated Passenger Services permitting scheme and pilot guidance; DfT, June 2025); London Assembly Transport Committee; Covington; Burges Salmon; Jones Day; Wayve, Uber, Waymo, Verne and Moove company announcements; Reuters; TechCrunch; Sky News; The Guardian; AP; Evening Standard and Financial Times (via Tech Times); The Next Web; Automotive World; Tech.eu; Electronics Weekly; E&T; EVwire; Unite.ai; Mobile World Live; TransportXtra; GLA/YouGov polling, June 2026; Press Association.
