Waymo develops custom chips for its fleet of robotaxis

Waymo, the autonomous driving specialist division of Alphabet Inc., has revealed that it has designed custom chips for its driverless vehicles. This application-specific integrated circuit (ASIC) was manufactured with Taiwan Semiconductor Manufacturing Co.'s (TSMC) 5-nanometer process and represents a significant step in optimizing the hardware platform of robotaxis.

Quick Answer

  • Waymo has developed a 5nm ASIC to process sensor data in real time
  • The chip handles 13 cameras, 4 lidars, and radar with more than 1,000 trillion operations per second
  • The system ensures operational redundancy to instantly replace faulty components
  • The hybrid architecture combines custom processors with third-party CPUs/GPUs
  • The solution is already in production in the Ojai vehicle, developed with Zeekr

Hybrid architecture for optimized performance

Waymo has adopted a hybrid computing architecture that combines its custom circuits with third-party components. The company has developed an ML-primary architecture to run advanced neural networks with minimal latency. To handle non-ML critical tasks such as orchestration, data movement, and logging, Waymo paired its ML technologies with the best CPUs, GPUs, and accelerators available on the market. This approach allowed the creation of a heterogeneous and balanced system.

Responsibility, robustness, and redundancy

Waymo's computing system is designed around three fundamental requirements: responsibility, robustness, and redundancy. The system processes on-board driving information in milliseconds, with a computing capacity increased by 20 times in eight years. Additionally, it must withstand vibrations, impacts, and extreme temperatures, using the vehicle's liquid cooling system to maintain performance.

Safety adds another level of complexity. Waymo's computers operate as two independent engines that run workloads in parallel. If one engine malfunctions, the other can take control immediately, as there is no human driver available to intervene.

The impact on the technology sector

For technology leaders in companies, Waymo's move represents another signal that specialized AI workloads are pushing companies to reconsider how much control they want to exercise over their own computing stack. Instead of relying entirely on off-the-shelf processors, Waymo can optimize its own silicon based on the specific needs of autonomous driving, including latency, power consumption, redundancy, and sensor processing.

This approach may not be suitable for all companies, but the general lesson is clear: as AI workloads become more specialized, hardware choices increasingly influence performance, costs, and reliability. Waymo is applying this equation to vehicles that travel on public roads, where milliseconds count and system failures have consequences far beyond a slow application.

Implications for the future of robotaxis

As Waymo expands its fleet of robotaxis, its own custom silicon could become a crucial element for improving efficiency and reliability at scale. For IT leaders observing the growth of custom AI infrastructures, the trunk of a driverless car is becoming another battleground in the race for custom chips.

Future challenges and scalability

As Waymo increases the computing power in its vehicles, the biggest challenge remains ensuring that the technology can handle the unpredictable situations of the real world that emerge as robotaxi fleets expand. The ability to handle edge cases and exceptional situations will be crucial for the long-term success of robotaxi technology.

At this point, the integration of Waymo's custom chip represents a significant step toward optimizing the performance and reliability of autonomous vehicles. The combination of custom silicon and hybrid architectures could set new standards for the industry, influencing the future development of autonomous driving systems.

The industrial context and strategic collaborations

Waymo's decision to develop custom chips does not occur in an industrial vacuum but fits into an ecosystem of strategic collaborations. The production of the Ojai vehicle, in which these chips are already implemented, is the result of a partnership with Zeekr, a company controlled by the Chinese giant Geely. This collaboration demonstrates how innovation in autonomous vehicles is becoming increasingly a global affair, with companies sharing resources and expertise to accelerate technological development.

In particular, the alliance with Zeekr allows Waymo to benefit from the automotive engineering expertise and production capacity of the Chinese partner, while Zeekr gains access to Waymo's cutting-edge autonomous driving technology. This type of synergy could become a model for other companies in the sector, showing how vertical specialization (in Waymo's case, the development of custom chips) can be integrated with large-scale production by specialized partners.

The evolution of the chip market for autonomous vehicles

The innovation in custom chips represents only part of a broader ecosystem of enabling technologies for autonomous vehicles. Artificial intelligence, in particular, will play a crucial role in the development of increasingly sophisticated autonomous driving systems. Advanced neural networks, reinforcement learning, and other machine learning algorithms will be fundamental to improving the ability of vehicles to understand and navigate the surrounding environment.

Additionally, the integration of advanced sensors such as lidars and high-resolution cameras, combined with sensor fusion algorithms, will allow autonomous vehicles to achieve a more accurate perception of the environment. This will be essential to address the unpredictable challenges of the real world, such as adverse weather conditions, sudden obstacles, and complex traffic situations.

A new paradigm for autonomous vehicles

Waymo's adoption of custom chips represents a turning point for the autonomous vehicle industry. This move not only improves the performance and reliability of autonomous driving systems but also opens new possibilities for technological innovation and industrial collaboration. However, to fully realize the potential of autonomous vehicles, a series of technical, economic, and regulatory challenges will need to be addressed.

As the industry continues to evolve, it is likely that we will see further differentiation of hardware and software solutions, with companies developing increasingly specialized technologies to meet the specific needs of autonomous vehicles. This new paradigm could lead to a future where autonomous vehicles are not only safer and more efficient but also more accessible and integrated into our daily lives.

Editorial Note and Disclaimer

The guides and content published on GoYou are the result of independent research and analysis activities, for informational, educational, and in-depth purposes.

GoYou does not constitute a journalistic publication or an editorial product pursuant to Law No. 62/2001 and does not engage in real-time information activities.

The GoYou project does not provide professional, technical, legal, or financial advice and disclaims all responsibility for the improper use of the information published.

In the Crypto sector, every investment involves risks: readers are invited to always inform themselves independently before making any decisions.