Supplier strategy intact: AMD, Micron, Nvidia, Samsung, SanDisk, and Socionext remain collaborators alongside TSMC.
- Custom silicon milestone: Waymo built its first custom chip using TSMC's 5nm process, delivering more than 1,000 TOPS dedicated to raw sensor processing and machine-learning models.
- Sensor-heavy front end: The chip handles raw LiDAR, radar, and camera streams in real time, processing data from 13 high-resolution cameras simultaneously.
- Scale versus critics: Waymo runs 500,000 paid rides a week across 4,000 vehicles, countering Elon Musk's "sensor contention" criticism of multi-sensor systems.
Waymo has disclosed that it has built its own custom computer chip using TSMC's 5nm process technology — the first time the company has designed its own silicon for its robotaxi fleet.
A Sensor-First Architecture
The chip delivers more than 1,000 TOPS, and it is not a general-purpose vehicle computer. Waymo's silicon is dedicated specifically to processing raw sensor data and running machine-learning models at the front end of the system, extracting information from raw LiDAR, radar, and camera streams in real time. The company says the system can process high-fidelity data from 13 high-resolution cameras simultaneously.
That TOPS figure sits in the same territory as the DRIVE AGX Thor platform from Nvidia, though the two are not directly comparable — Waymo's chip handles only the sensor-processing front end rather than total vehicle compute.
Waymo frames the custom chip as additive, not as a break from its suppliers. Beyond TSMC, the company lists AMD, Micron, Nvidia, Samsung, SanDisk, and Socionext as ongoing collaborators.
Engineered for the Road, Not the Lab
The compute system is built around three principles: responsiveness, ruggedisation, and redundancy. Processing is handled onboard, hardware is engineered for vibration, shock, and temperature extremes, and independent compute units run in parallel so one can take over if the other fails — a necessity when no human driver is present. Waymo says it has scaled raw onboard compute power 20-fold over the past eight years.
The architecture is also a pointed answer to Tesla CEO Elon Musk, who has long argued the opposite approach.
Waymo is hardly alone on the other side. Ford's Jim Farley has called LiDAR "mission-critical" for his company's autonomous driving ambitions, and Robosense founder Steven Qiu has said SAE Level 3-4 autonomy is more or less impossible with camera sensors only.
Deployment at Scale — and the Cost Question
Waymo's multi-sensor fleet is operating at a scale its critics have not matched: more than 500,000 paid rides a week across 4,000 vehicles in more than a dozen US cities. Ojai launched in Los Angeles, Phoenix, and San Francisco, with approximately 300 Ojai vehicles already in the commercial fleet and further rollouts planned for Denver, Las Vegas, and San Diego in 2026.
Manufacturing, however, carries its own complications. Ojai is manufactured by Geely brand Zeekr in Ningbo, China, then fitted with Waymo's sixth-generation Driver system at a facility in Mesa, Arizona.
The number Waymo has not disclosed is cost. Outside estimates put sixth-generation hardware at US$20,000–$25,000 per vehicle — down sharply from the US$100,000-$125,000 of the previous generation. The engineering case for custom silicon is clear; whether the economics support profitable expansion at scale remains the open question.
Is this your company?
This article features your business. Claim it to add your logo, contact details, and a link to your website — or upgrade to reach more buyers.
Did you know 80% of Press Releases trigger AI content warnings? Reach out and the M4S team can assist.
