Gradient Dissent: Conversations on AI Uber, Nissan, and Mercedes Chose This Self-Driving Startup | Alex Kendall, Wayve
Apr 15, 2026
Alex Kendall, CEO and co-founder of Wayve, built an end-to-end deep learning driver that scaled from a garage demo to global automotive partnerships. He discusses avoiding HD maps, zero-shot driving across 500+ cities, camera-first generalization, shifting training approaches, human-like driving for acceptance, and commercial strategies for integrating autonomy into consumer vehicles and robotaxis.
AI Snips
Chapters
Transcript
Episode notes
Humble Garage Origins Sparked Wayve's First Breakthrough
- Alex Kendall started Wayve with $1.5M, friends, a rented house and a car in the garage to prototype end-to-end learning for driving.
- That garage origin led to early breakthroughs and memorable team culture like house jam sessions and server rooms at 40°C.
Ten Interventions Taught A Car To Lane Follow
- Early on Wayve used on-policy reinforcement learning and got lane-following after about ten human interventions on a front-camera-only car.
- Alex recalls cheering in the car the first time it lane-followed, and there's a video of that milestone.
AV 2.0 Scales By Avoiding HD Maps And Retrofits
- Wayve positions AV 2.0 as an AI-centric, mass-market approach avoiding expensive retrofits, HD maps and infrastructure.
- That enables integration into consumer vehicles built at scale with in‑chain hardware and OTA updates.