Full Breakdown
Waymo Deploys Custom ASIC Chip in New Ojai Robotaxi to Cut Costs and Boost Performance
8/20/2026, 7:43:50 PM
Core Development: Custom ASIC for Autonomous Driving
Waymo announced that its latest generation of robotaxis now incorporates an in-house application-specific integrated circuit (ASIC). The silicon, fabricated on Taiwan Semiconductor Manufacturing Co.’s 5-nanometer process, delivers more than 1,000 trillions of operations per second (TOPS) for front-end sensor processing and machine-learning inference. By locating this processing in the trunk, the system reduces latency between sensor capture and actuation, which Waymo describes as “pixels-to-actuation.”
Background & Context: From Third-Party Chips to In-House Silicon
Historically, Waymo’s autonomous-driving computers have relied on off-the-shelf processors from Nvidia and AMD, integrating them into custom-designed boards. The new ASIC marks a strategic shift toward vertical integration, mirroring Alphabet’s broader effort to design its own AI processors for data-center workloads. Waymo’s spokesperson said the move helps “diversify chip supply” and “control soaring AI-computing costs.”
Data & Statistics: Performance, Fleet Size, and Cost Estimates
- Processing power: >1,000 TOPS for front-end ML tasks.
- Manufacturing node: 5 nm (TSMC).
- Current fleet: ~4,000 Waymo-operated vehicles in more than 10 U.S. cities, completing roughly 500,000 paid trips per week.
- Ojai deployment: ~300 Ojai robotaxis are in commercial service; expansion to Denver, Las Vegas and San Diego is planned for later this year.
- Future volume: MoffettNathanson projects 5,000 Ojai vehicles in the United States by the end of 2026.
- Cost per vehicle: Industry estimates place the sixth-generation hardware at $20,000-$25,000 per Ojai, a steep drop from the fifth-generation estimate of $100,000-$125,000.
Official Statements & Responses
Bloomberg reported that the chip’s TOPS rating puts it on par with Nvidia’s latest autonomous-driving systems, though Waymo has not announced dropping Nvidia or AMD from its supplier list. The company noted that the ASIC is only one component of a “balanced, heterogeneous system” that pairs its own silicon with CPUs, GPUs and accelerators from other vendors.
Why It Matters: Cost Reduction and Competitive Position
By moving critical sensor-fusion processing onto a purpose-built ASIC, Waymo aims to lower per-vehicle hardware expenses while maintaining safety-critical latency. The reduced hardware cost—combined with the Ojai’s chassis from Zeekr—supports Waymo’s goal of scaling to millions of weekly rides, a target it cites as 1 million rides per week. The move also narrows the performance gap with rivals such as Tesla.
Conflicting Reports & Gaps
Cost estimates for the sixth-generation hardware vary: The Verge cites $20,000-$25,000 per vehicle, while Bloomberg’s analysis focuses on performance metrics without providing a price figure. No independent regulator has audited Waymo’s safety claims that its technology “significantly reduces crashes and injuries” compared with human drivers; the data come solely from Waymo’s internal reporting. The exact number of ASICs already installed in the Ojai fleet remains undisclosed.
What’s Next: Expansion Plans and Future Hardware
Waymo plans to roll out the Ojai robotaxi to Denver, Las Vegas and San Diego later this year, expanding the fleet beyond the current three markets (Los Angeles, Phoenix, San Francisco). Upcoming presentations at industry events such as Hot Chips are expected to provide further technical insight.
