Divide more than 140,000 opted-in riders [4] by the fewer than 20 cars that actually went live [3] and each vehicle notionally carries a queue of over 7,000 people [29]. Riders have already shown up in force; getting enough cars on the road is the part still to be solved. Run the same denominator against the $8.6 billion valuation Wayve set in February's $1.2 billion Series D [15] and every Mustang Mach-E on a London street stands under more than $430 million of paper [30], a figure that only makes sense once you realize the money was paying for the software, not the cars.
What they bought is the licensing layer. Wayve does not want to run a fleet; it sells an AI driver that automakers wrap in vehicles and platforms wrap in riders [21], indifferent to vehicle platform and sensor configuration [34]. That is why Uber's investment arrived attached to 12 markets beginning with London [19], and why the same layer is meant to reappear in a Nissan Leaf in Tokyo this year, pending sign-off [23], with Stellantis signing a June memorandum on Level 4 vehicles globally [22].
The supervised phase cannot pay for itself, or rather the more interesting version of the question is whether it was ever meant to: every car carries a Transport for London-licensed private hire driver who can take over [5], the fare is the same as a regular Uber and shown upfront [6], and the rider is not prompted to tip [7]. A trip therefore absorbs a human wage and the capex of a sensor-laden EV while collecting less revenue than the all-human version of itself. Wayve raised roughly $240 million across its entire pre-Series D life and then $1.26 billion inside two months of 2026 [31], and in July let staff sell $85 million at the February mark, about one percent of the company [17][32]. That is a balance sheet built to fund a wait.
The next year or so could unfold three ways. British approval to drop the safety driver comes quickly, London's genuinely awful road geometry and its legal jaywalking [24] turn into the best training set anyone has, and the 12 markets convert. Or approval drags, Uber's Sarfraz Maredia is right that the process is simply still playing out with no known duration [12], and Waymo arrives in London [13] carrying an operating record from 14 fully driverless US cities [14] that a supervised British fleet cannot match on paper. Or the race resolves in Uber's favour regardless: a platform targeting as many as 15 AV cities by the end of 2026 and the largest share of AV trips by 2029 [20] is a platform that can bid its suppliers against each other.
My read is the second, with the third as the tax. The evidence that would break it is narrow and checkable: if the supervised fleet scales into the hundreds while fares stay at parity, then the driver in the seat was never the binding cost, and I have mispriced the option by pricing it as one.
