When Alex Kendall and Amar Shah founded Wayve, the autonomous-driving industry was dominated by a hardware-heavy worldview. Vehicles were expected to combine detailed maps, extensive sensor suites and carefully engineered rules. Wayve's thesis was closer to modern machine learning: show a system enough driving, let it learn the task, and build a model that can generalise to places it has never seen.
That sounds obvious after the rise of foundation models. In 2017 it was a contrarian technical bet. Kendall, who had worked on computer vision and robotics at Cambridge, became chief executive. Shah helped build the early company before later moving on from an operating role.
A different approach to autonomy
Wayve's system is built around end-to-end learning and embodied AI. Rather than maintaining a separate hand-coded rule for every driving scenario, the company trains models to interpret the environment and output driving behaviour across different vehicles and geographies.
The commercial promise is portability. If the same intelligence layer can operate across many vehicle platforms and cities, Wayve avoids part of the cost structure that made earlier autonomous-driving programmes expensive to scale. The technical risk is equally clear: generalisation is only useful if safety remains dependable in rare and difficult situations.
London became a global autonomy base
The company moved from Cambridge roots into a substantial London operation, turning the UK into one of the few places outside the United States and China hosting a frontier autonomous-driving company. That matters because the value chain around autonomy extends well beyond software into automakers, chips, simulation, safety validation and regulation.
Wayve's 2026 funding round showed how broad that coalition had become. Financial investors were joined by Microsoft and Nvidia as well as major automakers, while later investment from AMD, Arm and Qualcomm added another layer of strategic semiconductor support.
The founder bet is entering its hardest phase
Research credibility and fundraising are no longer Wayve's central tests. Commercial deployment is. The company has said supervised autonomy is expected in consumer vehicles from 2027 while robotaxi trials begin earlier. Moving from demonstration to production means satisfying automakers whose safety, liability and product cycles are far less forgiving than software release schedules.
If Wayve succeeds, the UK will have produced not simply another AI unicorn but a company embedded in one of the world's largest industrial markets. That would make Kendall's original technical contrarianism look less like a clever research thesis and more like an industrial platform strategy.