How End-to-End Learning Created Autonomous Driving 2.0: Wayve CEO Alex Kendall

Why It Matters

Wayve’s end‑to‑end AI model could dramatically lower the cost and complexity of deploying autonomous vehicles, speeding up industry adoption and reshaping the future of robotics.

Summary

In a recent interview, Wayve CEO Alex Kendall outlined the company’s vision of moving from the traditional, hand‑engineered autonomous‑driving stack (AV 1.0) to an end‑to‑end neural‑network architecture he calls AV 2.0. Wayve’s ambition is to become an embodied AI foundation model that can power the fleets of any OEM, eliminating the need for separate, bespoke networks for each vehicle or use case.

Kendall emphasized that the new approach hinges on three pillars: generalization across sensor suites, safety‑by‑design within a data‑driven framework, and massive, diverse training data augmented by synthetic world models. He described how Wayve’s early 100‑thousand‑parameter world model evolved into the current Gaia generative model, enabling the system to simulate complex traffic scenarios and develop emergent reasoning behaviors such as nudging through unprotected turns or adapting to foggy conditions.

Concrete examples underscored the progress: after only a few months of data collection, Wayve’s AI drove a Nissan vehicle in Tokyo, a city it had never seen before, and later demonstrated the stack in London, New York, and across Europe and Japan. Kendall highlighted the company’s partnership strategy that aggregates petabytes of dash‑cam and fleet data, and unsupervised clustering techniques that surface rare edge cases for targeted curriculum learning.

The implications are profound. By consolidating perception, planning, and control into a single, scalable neural engine, Wayve promises to slash development costs, accelerate time‑to‑market for OEMs, and push the broader automotive industry toward software‑defined vehicles. If the safety and interpretability challenges can be met, this foundation‑model approach could become the new standard for physical AI across robotics and autonomous systems.

Original Description

Alex Kendall founded Wayve in 2017 with a contrarian vision: replace the hand-engineered autonomous vehicle stack with end-to-end deep learning. While AV 1.0 companies relied on HD maps, LiDAR retrofits, and city-by-city deployments, Wayve built a generalization-first approach that can adapt to new vehicles and cities in weeks. Alex explains how world models enable reasoning in complex scenarios, why partnering with automotive OEMs creates a path to scale beyond robo-taxis, and how language integration opens up new product possibilities. From driving in 500 cities to deploying with manufacturers like Nissan, Wayve demonstrates how the same AI breakthroughs powering LLMs are transforming the physical economy.

Hosted by: Pat Grady and Sonya Huang