Waymo, the autonomous driving development company under Google parent Alphabet, has revealed that it has designed and developed its own custom chips tailored for its robotaxi fleet. This move marks a strategic shift to gradually reduce its reliance on NVIDIA chips, which currently serve as its primary hardware.
The newly announced proprietary chip is optimized to process the massive amounts of sensor data generated while robotaxis are in motion and to execute real-time decision-making. By transitioning from general-purpose GPUs to custom hardware specialized for running autonomous driving algorithms, Waymo aims to improve power consumption efficiency and maximize computational performance.
Previously, Waymo's vehicles relied heavily on NVIDIA computing infrastructure. By bringing chip development in-house, the company enables a flexible system design unconstrained by specific third-party hardware. The goal is to enhance the practicality of its autonomous driving system by achieving both improved inference capabilities and optimized power consumption.
Waymo plans to progressively integrate these proprietary chips into its upcoming robotaxi vehicles. By deepening the vertical integration of hardware and software, the company aims to reduce medium-to-long-term operational costs and enhance system scalability.