Most of the companies chasing the dream of the self-driving car have approached it like a mapping exercise. They send fleets of heavily instrumented vehicles to trace every kerb and traffic light of a city in painstaking detail, then let their cars follow that stored map like a train on invisible rails. Alex Kendall thinks this is the wrong way round. His company, Wayve, is built on the belief that a car should learn to drive the way a teenager does, by watching, practising and generalising, rather than by memorising a place it can never leave.
That contrarian idea has carried Kendall a long way. Wayve began in 2017 as a Cambridge project, testing prototypes on quiet English roads, and has grown into the most valuable artificial intelligence company Britain has produced. Along the way Kendall, still in his early thirties, has become the face of a national ambition, the man most often named when politicians talk of Britain leading the driverless age rather than watching it arrive from Silicon Valley or Shenzhen.
A GPT for driving
The technical heart of Wayve is an approach its founder calls embodied artificial intelligence. Instead of writing millions of explicit rules for every situation a car might meet, the company trains a single large model on vast amounts of driving footage and lets it learn the underlying skill. The system takes in what its cameras see and decides directly how to steer, brake and accelerate, end to end, in much the way a language model turns a prompt into a sentence. Kendall likes to describe it as a kind of GPT for driving, a foundation model for moving through the physical world.
The appeal of this method is flexibility. A car that has learned to drive rather than memorised a map is not confined to the handful of neighbourhoods its makers have surveyed. In principle it can be dropped into a city it has never seen and cope, reading a roundabout in Leeds or a rain slicked junction in London by drawing on everything it has learned elsewhere. That is a very different proposition from rivals whose robotaxis work beautifully but only within the invisible fences of their mapped zones.
The money follows
Investors have found the vision persuasive to a remarkable degree. Wayve raised more than a billion dollars in a single round led by SoftBank, with Nvidia and Microsoft joining in, a sum that ranked as the largest artificial intelligence funding round in European history and the biggest bet ever placed on a British AI firm. The backing values the company in the billions and, just as important, supplies the enormous computing power and data that its learning approach devours. In a field where progress is bought with scale, that war chest matters as much as any clever algorithm.






