Waymo details custom silicon and hardware partners behind its robotaxi brains
Waymo disclosed the architecture and components of the high-performance computer that runs in the trunks of its driverless taxis, including its own 5-nm ASIC and a list of supplier partners.
Waymo revealed key specifications of the compute platform that powers its fleet of roughly 4,000 robotaxis operating in more than ten cities, describing a custom 5-nanometer ASIC designed to clean and fuse raw sensor inputs before they reach the central processor. The blog post emphasized three non-negotiable design principles—responsiveness, ruggedization and redundancy—to meet real-time driving demands while fitting within the vehicle trunk and staying quiet for passengers.
Executives Satish Jeyachandran and Daniel Rosenband noted that the system balances machine-learning workloads with traditional CPU, GPU and accelerator tasks. The company also listed its hardware supply chain, naming firms such as AMD, Micron, Nvidia, Samsung, SanDisk, Socionext and TSMC. Waymo argues that this integrated sensor-fusion approach enables safe, large-scale deployment of autonomous taxis, contrasting its strategy with critics of lidar-based systems.
Why it matters
Understanding Waymo's hardware reveals how autonomous vehicles achieve real-time perception and safety at scale.
In this story
