NVIDIA today announced NVIDIA Halos for Robotics, the industry’s first full-stack, comprehensive safety system for robotics and physical AI that unifies AI compute and safety. The announcement was made at Automate in Chicago. The system is designed to provide a standardized, unified safety architecture that connects AI compute, system software, sensor data, safety applications and inspection for robotic systems operating in dynamic environments alongside humans.

The new system extends NVIDIA Halos’ proven autonomous vehicle safety to robotics and physical AI. It draws on 18,600+ engineering years of autonomous vehicle safety development. The architecture enables companies to scale autonomous systems into factories, warehouses and logistics operations with greater confidence.

System Architecture and Components

NVIDIA Halos for Robotics spans three key layers needed for robot safety. The hardware layer includes NVIDIA IGX Thor and NVIDIA Holoscan Sensor Bridge, which provide industrial-grade AI compute, built-in safety and sensor connectivity for real-time workloads. The software layer features NVIDIA Halos OS, which includes Halos Core to support safety-related operating functions and safety applications built with the NVIDIA Halos Outside-In Safety Blueprint.

The inspection layer consists of the NVIDIA Halos AI Systems Inspection Lab. This lab is the world’s first ANSI National Accreditation Board (ANAB)-accredited program for functional and AI safety for physical AI. It helps partners prepare Halos integrations for third-party certification by leading certification bodies including TÜV Rheinland, UL Solutions, TÜV SÜD, exida, SGS and CertX.

Partnership with Agility

Agility, a humanoid robotics and physical AI company, is the first to team with NVIDIA to incorporate elements of Halos for Robotics into its proprietary safety system. Agility will integrate NVIDIA IGX Thor and Halos Core into its safe human detection system for its humanoid robot Digit. Digit is designed for industrial work in logistics, manufacturing and warehouse operations for customers including Amazon, GXO, Schaeffler and Toyota Motor Manufacturing Canada.

Agility will also participate in the NVIDIA Halos AI Systems Inspection Lab. The companies will use the lab to ensure Digit’s safety-related software, AI components and cybersecurity protections meet rigorous standards such as IEC 61508, ISO 13849 and ISO/IEC TR 5469 before final third-party certification.

Ecosystem and Availability

The NVIDIA Halos for Robotics ecosystem includes partners across software, systems, sensors and silicon. Software partners include Acontis, FreeRTOS and QNX. Embedded systems partners include Advantech and NexCobot. Sensors and silicon partners include Infineon, NXP Semiconductor, STMicroelectronics and Texas Instruments. Industrial applications partners include FORT Robotics, Inventec, KION Group, Lyte AI and Neurealm.

NVIDIA Halos Core for NVIDIA IGX is available in early access for registered developers in Linux and Linux plus QNX OS for Safety 8.0 configurations. The open source NVIDIA Halos Outside-In Safety Blueprint is now available in early access on GitHub.

NVIDIA announced the NVIDIA Vera Rubin platform at ISC High Performance 2026, designed to deliver world-class supercomputers for science by combining native double-precision (FP64) performance, NVIDIA CUDA-X libraries and the full-stack capabilities of the NVIDIA AI platform. The platform is built to unite high-precision simulation, AI and data analytics, specifically targeting the era of agents to advance scientific discovery in fields such as climate modeling, computational fluid dynamics, quantum chemistry and energy exploration.

The Vera Rubin system provides more than 7 exaflops of AI for science, 5 petaflops of native FP64 support and extreme memory bandwidth with up to 144 GPUs per rack. This configuration allows the system to deliver performance on par with the TOP500 list of the world’s most powerful supercomputers, enabling research centers and industrial enterprises to run larger models, improve fidelity and shorten time to discovery.

Platform Architecture

The NVIDIA Vera Rubin platform combines NVIDIA Rubin GPUs and NVIDIA Vera CPUs connected via high-speed NVIDIA NVLink-C2C, NVIDIA ConnectX-9 SuperNICs and NVIDIA BlueField-4 DPUs in a direct liquid-cooled architecture. For scientific computing, the platform provides native FP64 capabilities to accelerate simulations requiring the highest accuracy, alongside the AI performance needed for surrogate models, scientific foundation models and AI-assisted analysis.

Global Adoption and Systems

Leading supercomputing centers are adopting the platform to build next-generation systems. The Leibniz Supercomputing Centre (LRZ) will utilize the platform for its Blue Lion system, scheduled to come online in 2027, delivering approximately 30x the computing power of LRZ’s current system. The National Energy Research Scientific Computing Center will deploy Doudna, a Dell Technologies system powered by NVIDIA Vera Rubin, for large-scale HPC workloads and AI training. Los Alamos National Laboratory has selected NVIDIA Vera Rubin for its Mission, Vision and Veritas systems, with Veritas specifically designed for agents to advance scientific discovery.

Manufacturer Availability

Global system manufacturers including Bull, Dell Technologies, GIGABYTE, HPE and Supermicro are bringing NVIDIA Vera Rubin NVL4 to market through direct liquid-cooled AI and HPC racks. These systems are designed to help research institutions, national labs and enterprises deploy rack-scale accelerated computing. NVIDIA Vera Rubin NVL4-based systems are expected to be available from global system manufacturers in Q4 this year.