Tensor and Arm have formed a multi-year partnership to develop the computing architecture for Tensor’s planned autonomous passenger vehicle, the Robocar. The vehicle is designed as a privately owned car capable of Level 4 autonomy, meaning it can operate without human intervention under defined conditions.
The Robocar is built around artificial intelligence systems that control driving, sensing and decision-making functions. As part of the collaboration, Tensor is using Arm’s processor architecture as the foundation of the vehicle’s onboard computing platform. Each vehicle integrates 433 Arm-based processor cores that distribute computing tasks across the system, from central AI processing to smaller control units connected to sensors and safety systems.
The processors handle workloads including AI model execution, vehicle control, sensor processing and real-time safety monitoring. The configuration includes Neoverse AE cores for high-throughput AI processing, Cortex-X cores for system control and in-vehicle computing, Cortex-A cores for functions such as drive-by-wire systems and lidar processing, Cortex-R cores for safety-critical operations, and Cortex-M cores for low-power subsystem management. The system operates alongside Nvidia-accelerated hardware used to run Tensor’s autonomous driving software.
The vehicle’s autonomy system relies on a large set of sensors designed to monitor the surrounding environment and vehicle systems. The Robocar incorporates 37 cameras, five lidar sensors, 11 radar units, 22 microphones, 10 ultrasonic sensors and three inertial measurement units, along with collision detectors, water-level sensors, tire-pressure sensors and a smoke detector. The vehicle also uses global navigation satellite system receivers and triple-channel 5G connectivity to support navigation and communication.
Drew Henry, executive vice president of Arm’s Physical AI Business Unit, said autonomous vehicles require computing platforms capable of delivering high performance while meeting strict safety and power efficiency requirements. Tensor Chief Operating Officer Jewel Li said deploying personal autonomous vehicles at scale requires engineering focused on safety, redundancy, reliability and power efficiency.
Tensor is working with a range of automotive suppliers, semiconductor companies and cloud providers on the Robocar program, including Autoliv, ZF, Continental, Nvidia, AMD, Qualcomm, Samsung and Oracle. The company plans to introduce the Robocar in the United States, the European Union and the Middle East in 2026.
