Waymo is taking a more direct role in the hardware behind its robotaxis. Alphabet's robotaxi subsidiary has developed a custom chip for its self-driving taxis, Bloomberg reports, and the component is already operating in the company's latest generation of vehicles.
The move shifts part of Waymo's computing stack away from outside chip suppliers. Until now, the company relied on chips from Nvidia and AMD for the systems that support its autonomous driving work.
What Waymo's custom chip is built to do
The chip is designed for the demanding work inside a robotaxi. Waymo says it processes sensor data faster and runs the AI models that help its vehicles understand and react to what is happening around them.
That matters because a self-driving taxi depends on a constant stream of information from its sensors. The vehicle must interpret that information, connect it to its AI models, and respond in a way that fits the road environment.
The source article does not describe every sensor or model involved, but it does make clear that the chip sits at the center of the vehicle's perception and response pipeline. In plain terms, Waymo is building dedicated computing hardware for the task its robotaxis perform every time they operate: reading the surroundings and deciding how to react.
The performance number Waymo is highlighting
Waymo says the custom chip delivers more than 1,000 TOPS (trillions of operations per second). According to the company, it is manufactured by TSMC on a 5-nanometer process.
Those details put the chip in the same performance conversation as Nvidia's current autonomous driving systems. The source article states that the chip is on par with Nvidia's current autonomous driving systems, which is a notable benchmark because Waymo previously depended on outside suppliers including Nvidia and AMD.
The performance claim is important, but so is the strategic direction behind it. By developing its own chip, Waymo is not simply swapping one component for another. It is bringing a key part of the robotaxi computing platform closer to its own vehicle design and operating needs.
Why reducing reliance on Nvidia matters
The stated goal behind the move is to bring costs down. For a robotaxi company, computing hardware is not a side detail; it is part of the vehicle platform itself. If Waymo can use a chip designed around its own AI models and sensor-processing requirements, it may have more room to tune the balance between performance and cost.
The source article does not say that Waymo is ending its relationship with any supplier. It says the company had relied on chips from Nvidia and AMD until now, and that the new custom chip cuts its reliance on Nvidia. That distinction is important: the development shows a change in dependency, not a full description of every future sourcing decision.
For Nvidia, the point is also straightforward. Autonomous driving systems are one of the areas where high-performance AI computing is valuable. Waymo building its own chip shows that at least one major robotaxi operator wants more control over that layer of the stack.
Where the chip is going next
Waymo is installing the chip in its new robotaxi, which it is building with Chinese manufacturer Zeekr. The article does not provide a launch schedule or production volume, so the clearest takeaway is that the hardware is already tied to Waymo's next vehicle direction.
That gives the chip more significance than a lab project. It is already running in Waymo's latest generation of vehicles, and it is being installed in the new robotaxi being built with Zeekr.
The broader implication is that Waymo wants its robotaxi platform to be less dependent on general supplier choices and more aligned with its own self-driving system. The chip processes sensor data, runs AI models, delivers more than 1,000 TOPS, and is built by TSMC on a 5-nanometer process. Those are the facts Waymo is putting forward as it moves more of the robotaxi computing stack in-house.