Wayve’s Embodied AI Learns Japan in Months, Enabling Faster Scaling for Automakers

LONDON, UK, October 22, 2025 — Wayve, a leader in Embodied AI for driving, has released new data showing how its AV2.0 embodied AI technology scales autonomy to new markets faster and more efficiently than traditional map-based systems.

In Japan, one of the world’s most complex driving environments, the Wayve AI Driver adapted within months, demonstrating how efficiently Wayve’s embodied AI technology can learn from small datasets and deliver measurable performance improvements across all of Wayve’s deployment markets.

Speed to market: Wayve AI Driver completed its first drives in Japan on Day 1 with no maps or re-engineering. Within one month, using just 1% locally collected data, it achieved a 2X improvement in performance compared to zero-shot performance. Within three months, after adding just 2% Japan data, Wayve developed a global baseline model that performed consistently in Japan.

Global efficiency: The same dataset delivered benefits well beyond Japan. Within five months, Wayve optimized its global baseline model, improving performance across all testing markets. By late September, results showed consistent gains over July: Japan +275%, Germany +190%, the UK +150%, and the US +210%. These outcomes highlight the scalability and data efficiency of Wayve’s AV2.0 approach, reducing the time and cost of global deployment.

Commercial readiness: The Wayve AI Driver has already been transferred from Wayve’s internal development fleet to an automaker platform in Japan, showing how quickly this AI software can integrate into vehicle platforms and adapt to different hardware configurations.

Read the full engineering story: Global Learning, Local Driving – Lessons from Japan