Arm and NVIDIA Enabling Intelligent Solutions For Roads and Robots

Across industries, intelligent machines are rapidly evolving. Vehicles are becoming AI-defined platforms capable of automated driving, AI-powered infotainment, and personalized in-vehicle experiences. At the same time, robotics is entering a new phase, where humanoid robots, autonomous mobile robots (AMRs), and smart industrial systems are expected to perceive, reason, and act in real time.

To realize the potential of intelligent machines requires compute platforms that prioritize safety while delivering a combination of datacenter-class performance, scalability, and efficiency. With Arm Neoverse as the CPU foundation, NVIDIA offers leading edge compute platforms with NVIDIA DRIVE AGX Thor and Jetson Thor.

Transforming Automotive with NVIDIA DRIVE AGX Thor

NVIDIA DRIVE AGX Thor Developer Kit is now available for pre-order, and will be shipping this September. NVIDIA DRIVE AGX Thor, built on Arm Neoverse V3AE CPUs, is a scalable and energy-efficient AI computing platform designed to process the complex workloads required for autonomous driving.

The Arm compute platform brings deterministic performance and ISO 26262 functional safety to NVIDIA DRIVE AGX Thor to ensure that advanced AI workloads run reliably in real-time. Delivering up to 2,000 teraflops of FP4 performance, NVIDIA DRIVE AGX Thor enables workloads such as predictive ADAS, real-time driving monitoring, and natural voice interaction to run simultaneously on one platform.

As the backbone of NVIDIA DRIVE AGX Thor’s compute fabric, Arm Neoverse V3AE CPU works together with NVIDIA’s Blackwell GPU  and dedicated accelerators to deliver a reliable and scalable foundation for the AI-defined vehicle era, where cars continuously learn, adapt, and improve over time.

Two developer kit versions are now available: A benchtop unit (SKU10) and an in-vehicle (SKU12), both designed to accelerate hardware and software development.  

Advancing Robotics with NVIDIA Jetson Thor

In parallel, introduction of NVIDIA Jetson Thor brings the same Arm Neoverse V3AE foundation into robotics and physical AI. Purpose-built for real-time, multi-model AI workflows, NVIDIA Jetson Thor enables robots to perceive, reason, and act simultaneously – a critical requirement for humanoid robots, AMRs, and industrial systems.

Arm Neoverse extends high-single performance and unmatched performance-per-watt efficiency to robotics to ensure that humanoid and AMRs run multiple AI models in parallel without compromising responsiveness.

Here’s what that looks like in practice:

• Humanoid robotics: NVIDIA Jetson Thor delivers the high-performance compute, low latency, and real-time reasoning needed for humanoid robots, where natural interaction requires multiple models such as vision, language, and motion planning to run in parallel.
• Industrial automation: In factories and warehouses, Jetson Thor powers autonomous machines and robotic arms that improve efficiency and safety.
• Healthcare systems: With local sensor fusion and advanced AI, Jetson Thor supports robotics in patient monitoring, rehabilitation, and surgical assistance.

Meanwhile, when paired with NVIDIA Blackwell, Jetson Thor delivers up to 2000 FP4 TFLOPS, with a 7.5x performance increase and 3.5x higher energy efficiency compared to Jetson Orin. These capabilities make Jetson Thor a versatile platform for deploying physical AI across diverse industries.

Along with a 3.1x higher performance CPU, NVIDIA Jetson Thor unifies a higher core count than its predecessor and more GPU performance in one platform, which helps enable real-time control for advanced robotics.

The role of Arm Neoverse V3AE

Today’s news is the perfect example of how Arm Neoverse V3AE is bringing datacenter-class performance to both automotive and robotics workloads, with efficiency and safety built in. It combines high single-thread performance with scalability to run multiple complex AI models in parallel. For safety-critical real-time tasks, Neoverse V3AE is built to meet the strictest automotive-grade standards, providing determinism, isolation, and reliability for safety-critical environments.

Arm’s performance-per-watt leadership is ensuring these systems can run advanced AI workloads sustainably, whether in vehicles or robots operating at the edge. This combination allows machines to fuse data from multiple sensors, run generative AI, and make safe, real-time decisions in environments where responsiveness is critical.

A Scalable Ecosystem for Developers

The scalability of any compute platform relies heavily on its ecosystem. Both DRIVE AGX Thor and Jetson Thor benefit from NVIDIA’s advanced software stacks, running seamlessly on Arm’s efficient compute architecture.

• NVIDIA DriveOS 7.0.3 for automotive, with ISO26262 certification readiness, secure boot, and TensorRT optimizations for AI and generative models.
• NVIDIA Isaac for robotics, Metropolis for visual AI, and Holoscan for sensor processing, enabling rapid development of intelligent robotic systems.

Paired with Arm’s robust development tools, and ecosystem support, developers can reduce complexity and accelerate time-to-market and focus on innovation instead of system integration.

Powering Intelligent Machines Across Industries

The availability of NVIDIA DRIVE AGX Thor developer kits and introduction of NVIDIA Jetson AGX Thor signal a new phase for intelligent machines with Arm at the core. In automotive, this means enabling AI-defined vehicles that adapt in real-time to improve safety, efficiency, and user experience. In robotics, it delivers safety-ready intelligence for humanoid robots, autonomous machines, and industrial systems.

Arm Neoverse V3AE is opening the door to scalable, safety-capable, and energy-efficient compute, ensuring fleets of vehicles and factory robots can run advanced AI workloads sustainably at the edge.

Together, with NVIDIA, we are laying a foundation for safe, scalable compute that powers every mile driven and every task performed.