--- title: "Nvidia Invests Billions in Physical AI Technologies" url: https://noti.group/nvidia-invests-billions-in-physical-ai-technologies/ language: en publisher: "Noti Group" section: "Technology" published: 2026-10-08T11:15:33.000Z updated: 2026-10-09T02:26:55.249Z id: 5c9e688e-b76f-4d05-a17b-aca1f1a0d3d3 source: "Ars Technica https://arstechnica.com/ai/2026/10/nvidias-big-bet-on-physical-ai-aims-for-safer-robotaxis-humanoid-robots/" attribution: "Link to https://noti.group/nvidia-invests-billions-in-physical-ai-technologies/ and name Noti Group when you quote or summarize this story." --- # Nvidia Invests Billions in Physical AI Technologies Nvidia has made significant investments in physical AI technologies, including robotics and self-driving cars, which is estimated to be worth billions of dollars. Nvidia has made a significant investment in physical AI technologies, including robotics and self-driving cars, which is estimated to be worth billions of dollars. The company's efforts to address safety concerns in autonomous vehicles led to the development of the Halos system, a full-stack safety system that provides hardware and software tools for developers. The system was launched in 2025 with the goal of reducing the risk of accidents involving self-driving cars and other autonomous vehicles. Nvidia has now expanded its safety architecture to include robotics, announcing Nvidia Halos for Robotics in June 2026. This move aims to enable safe deployments of autonomous mobile robots in various settings, such as warehouses and factories, as well as humanoid robots performing tasks inside these environments. According to Amit Goel, head of robotics ecosystem and edge computing at Nvidia, safety is becoming a critical bottleneck in the development of physical AI technologies. The company's goal with Halos for Robotics is to unlock the capabilities of these systems by providing a robust safety framework. Nvidia's expanded safety offering for robotics is part of its broader push into physical AI, a business segment that already generates nearly $10 billion in annual revenue for the company. This growth area was highlighted by Nvidia CEO Jensen Huang as early as 2025. At the heart of Nvidia's full-stack safety system is hardware designed specifically with safety in mind. The Nvidia IGX Thor computing module, for example, features an independent processor dedicated to handling safety-related workloads, ensuring that critical functions are isolated from other processes. The operating system at the center of this safety framework is Halos, which enables constant monitoring of every component and software library within a robotic system. This allows for swift identification and mitigation of potential failures or anomalies. Nvidia's Holoscan Sensor Bridge plays a crucial role in connecting sensor data with safety-critical processing, helping to detect corrupted information that could compromise the integrity of the system. This bridge can be integrated into individual hardware components, such as microcontrollers or Field Programmable Gate Arrays. The Halos package also includes tools for simulating robotic behavior in virtual environments, allowing developers to test and refine their systems without putting physical assets at risk. Additionally, an inspection lab program provides partners with rapid feedback on safety-related issues that arise during development. To ensure that Nvidia's AI technology is adaptable to various robotic applications, a different approach was needed compared to autonomous vehicles. Unlike autonomous driving, where safety standards are relatively consistent across industries and regions, robotics involves diverse scenarios with unique safety concerns. For instance, a robotic vacuum cleaner navigating a hallway poses distinct risks than a forklift handling heavy loads in a warehouse. Nvidia's engineers had to rework their foundational platform to provide developers with the flexibility to define custom safety functions without compromising the underlying infrastructure. This involved striking a balance between giving developers control and maintaining stability. Robots operating in unstructured environments, such as factories or hospitals, face challenges like detecting blind spots or anticipating potential hazards. In these situations, robots may need to slow down or even come to a halt to avoid collisions with unknown objects. Agility Robotics has successfully integrated Nvidia's Halos system into its latest Digit 5 humanoid robot, enabling it to operate safely in close proximity to humans without the need for isolated workstations or physical barriers. The integration of Halos has brought all essential safety sensors and hardware within the robot itself, whereas previous Digit robots relied on external sensors placed around their work cell for safety purposes. This shift allows Agility's robot to move freely in a factory or warehouse setting without requiring new infrastructure for each task. According to reports, the incorporation of Halos has given Agility Robotics its "robot unchained" - the safety features now accompany the robot wherever it goes. As a result, the company can deploy its robots across various locations without the need for extensive setup and reconfiguration. Boston Dynamics is another prominent robotics firm collaborating with Nvidia on the Halos safety accreditation program. The Massachusetts-based company aims to develop a comprehensive safety platform that encompasses its range of robots, including Spot robot dogs, wheeled Stretch robots, and Atlas humanoid robots. Additionally, other companies are working to integrate Nvidia's Halos system into their operations, including German robotics firm KION Group, which specializes in self-driving forklifts and autonomous mobile robots. South Korean company LG is also developing its own humanoid robot using Nvidia's Isaac GROOT model, a foundation for building human-like robots. The development of humanoid robots and robotaxis has taken a significant leap forward as governments and tech companies invest heavily in robotics research. These advancements are leading to robots being tested in various environments, moving beyond their traditional confines of performing single tasks in one location. According to experts, the need for robots to be able to handle multiple tasks in different settings has become essential for mass production, which can only be achieved if safety is improved. --- Source: [Ars Technica](https://arstechnica.com/ai/2026/10/nvidias-big-bet-on-physical-ai-aims-for-safer-robotaxis-humanoid-robots/) Published by Noti Group: https://noti.group/nvidia-invests-billions-in-physical-ai-technologies/