Waymo built a custom chip to power its robotaxi expansion

Waymo revealed a custom 5 nm ASIC chip that preprocesses sensor data before it reaches its self-driving system's core "brain," delivering more than 1,000 TOPS. TechCrunch Mobility reports the custom silicon powers the sixth-generation Ojai robotaxi as Waymo scales service across Los Angeles, Phoenix, and San Francisco.

Waymo gets frequent attention for its rapid robotaxi expansion, and the company's sixth-generation self-driving system — which debuted in its next-generation Ojai robotaxi — has been central to those ambitions. What became clearer this week is just how vertically integrated Waymo has become in pursuit of profitability.

Key Takeaways

• Waymo built a custom 5 nm ASIC chip to handle raw sensor data before it reaches the self-driving system's core processor.
• The chip delivers more than 1,000 TOPS, roughly comparable to Nvidia's DRIVE AGX Thor automotive processor.
• The Ojai robotaxi uses 13 high-fidelity cameras and is designed to be cheaper to build, operate, and maintain.
• Waymo partners with AMD, Micron, Nvidia, Samsung, Sandisk, Socionext, and TSMC on compute — but owns the custom silicon design.
• The company recently opened Ojai robotaxi service to all riders in Los Angeles, Phoenix, and San Francisco.

Why did Waymo build its own custom chip?

Until this week, it was less clear how hands-on Waymo had become in designing its own hardware. The company shared that it built a custom silicon chip — specifically a 5 nm ASIC — designed to handle the massive influx of raw data before it reaches the core "brain" of the self-driving system.

Waymo has repeatedly said its next-generation Ojai robotaxi is cheaper to build, operate, and maintain — ingredients required if the company has any hopes of someday turning a profit. The custom chip is part of that cost-and-efficiency strategy, giving Waymo tighter control over how sensor data is processed at scale.

How does the chip compare to industry alternatives?

To get a sense of how much data the system handles, consider that the Waymo Ojai has 13 high-fidelity cameras feeding information into the pipeline. Waymo said the chip delivers more than 1,000 TOPS (trillions of operations per second) of computing performance.

That statistic puts the chip roughly in the same performance range as Nvidia's latest DRIVE AGX Thor automotive processor, a powerful computer designed for automated driving applications. Waymo contends the result is a system with "unmatched efficiency and performance" that can react fast and safely in complex, high-density urban environments.

Who is Waymo working with on compute?

Waymo isn't going it entirely alone on compute. The company listed a slew of partners, some for the first time, including AMD, Micron, Nvidia, Samsung, Sandisk, Socionext, and TSMC. The custom ASIC represents Waymo's own vertical integration, while these partners supply supporting components and manufacturing expertise.

For more on how custom silicon is reshaping Future Tech & AI Wonders, this chip marks a notable shift toward in-house hardware design among autonomous vehicle leaders.

What does this mean for Waymo's expansion plans?

Waymo just opened its Ojai robotaxi to all riders in Los Angeles, Phoenix, and San Francisco — cities where the sixth-generation system and its custom chip are now in active commercial service. The company seems to add new cities or expand service areas on a near-weekly basis.

Separately, Waymo, Tesla, and Uber all received permits from Nevada regulators allowing commercial robotaxi services in Clark County, home to Las Vegas — permits that could deploy up to 8,000 robotaxis across the county over the next 12 months. Waymo has also responded to questions from the National Highway Traffic Safety Administration as part of an investigation into a crash in which a robotaxi struck a child at low speed, though released responses have been redacted.

The custom chip story, as reported by TechCrunch Mobility, underscores a broader industry trend: scaling robotaxis profitably may require owning more of the technology stack, not just the software.