Inside NVIDIA Halos for Robotics: A Full-Stack Functional Safety System for Physical AI

developer.nvidia.com Jun 23, 2026

Inside NVIDIA Halos for Robotics: A Full-Stack Functional Safety System for Physical AI

Curated by Shubham · TechCapsules · Jun 23, 2026

AI-summarised brief · reviewed before publication

NVIDIA has announced the launch of NVIDIA Halos for Robotics, a full-stack functional safety system designed for physical AI in unstructured environments. The platform combines powerful AI compute and safety in one platform, leveraging NVIDIA's years of experience in autonomous vehicle safety. Halos OS is a comprehensive safety system built on a proven foundation, extending the company's safety stack from AVs to robotics. Agility, a humanoid robot manufacturer, is already adopting NVIDIA Halos for its safe human detection system. This marks a significant milestone in the development of safe autonomous machines.

💡 Why It Matters

· The industry's adoption of a shared, standards-aligned foundation for robotics safety, like NVIDIA Halos, signals a shift away from ad hoc implementations and toward a more structured approach.
· This continuity in safety development processes and standards can accelerate the development of safe autonomous machines, ultimately paving the way for widespread adoption of physical AI in various industries.

Read full article on developer.nvidia.com

Europe Expands Quantum-GPU Computing as NVIDIA Announces 35 New AI Supercomputers

NVIDIA has announced the expansion of Europe's quantum-GPU computing infrastructure through 35 new systems across 23 countries. This move aims to strengthen the region's capabilities in quantum research, AI development, and scientific computing. The new systems will provide advanced computing resources to over 3 million researchers and will be integrated with quantum hardware and software platforms to support hybrid quantum-classical computing applications. The broader buildout includes major AI factory deployments, climate and energy research initiatives, and approximately 800 AI exaflops of computing capacity.

💡 The European effort to integrate quantum processors with traditional high-performance computing systems could accelerate breakthroughs in fields like materials science, climate modeling, and biotechnology. By combining quantum hardware with powerful GPU-accelerated supercomputers, researchers can tackle complex problems that were previously unsolvable.

us.headtopics.com · Jun 23

Nvidia seeks to make humanoid AI robots safer around humans

Nvidia Corp. is working to make humanoid robots safer around people by providing technology for robots to make their own decisions based on analysis of their surroundings. The company will offer software and semiconductors, including its Halos software, to give robots better awareness of their environment. This technology will allow humanoids to enter the workplace and interact with people, even making physical contact if necessary. Nvidia's goal is to enable robots to handle split-second decisions, increasing productivity and cooperative activities. The company is also offering a lab for robot makers to run safety tests and get necessary certifications.

💡 Humanoid robotics is expected to generate $200 billion in revenue by 2035, but safety concerns must be addressed to unlock this market. Nvidia's technology aims to overcome current safety hurdles that hinder robot productivity and cooperation.

techtimes.com · Jun 22

Automate 2026 Opens Monday: NVIDIA Pavilion Marks Humanoid Robot Shift to Production

Automate 2026 opens with a focus on humanoid robots, marking a shift to production with companies like Figure AI and Boston Dynamics achieving significant milestones. Figure AI's BotQ facility has increased its throughput by 24 times in under 120 days, delivering over 350 units. Boston Dynamics has begun commercial shipments of its electric Atlas, with its entire 2026 production run committed to Hyundai and Google DeepMind. The event features a NVIDIA-sponsored pavilion, highlighting the growth of humanoid robotics and its potential applications. Humanoid robots are generating interest across the automation ecosystem, with many attendees expected to learn about the technology's current state and future possibilities, driving innovation and development in the field.

💡 NVIDIA's investment in the pavilion signals its commitment to humanoid robots as a key part of its addressable inference market. The shift to physical AI enables manufacturers to deploy automation more effectively, overcoming traditional programming limitations and opening up new possibilities for robotics in manufacturing.

techtimes.com · Jun 22

Pharmicell Is the Sole Supplier of a Key Material Behind Doosan Nvidia…

Pharmicell, a Korean company, is the sole supplier of a critical ingredient in Nvidia's AI server circuit boards, providing materials to Doosan, whose end customer is Nvidia. Demand has driven Pharmicell's sales up, with first-quarter revenue from low-dielectric materials rising over 1.5 times from a year earlier. The company expects annual sales of these materials to roughly double this year. Pharmicell's position is difficult to dislodge due to lengthy customer-specific optimization and consecutive evaluations required for new suppliers. A third plant in Ulsan will roughly double capacity once online.

💡 Pharmicell's defensible position in the supply chain gives it a unique advantage, making it a quiet beneficiary of the AI buildout. Its high barrier to entry protects its market share, allowing the company to capitalize on growing demand for AI hardware.

thenewstack.io · Jun 22

“An agent is an LLM and a harness”: What Nvidia really thinks…

Nvidia's Director of Deep Learning Software, Bryan Catanzaro, has shed light on the company's stance regarding OpenClaw, a new open-source project that aims to create a more efficient and scalable deep learning framework. Catanzaro described OpenClaw as "an agent is an LLM and a harness," suggesting a symbiotic relationship between the framework and large language models (LLMs). This statement implies that OpenClaw is designed to work seamlessly with LLMs, potentially unlocking new levels of performance and efficiency in deep learning applications.

💡 Nvidia's endorsement of OpenClaw highlights the growing importance of open-source collaboration in the deep learning space. By embracing OpenClaw, Nvidia is signaling its commitment to advancing the field through community-driven innovation, rather than relying on proprietary solutions. This shift could have significant implications for the development of future AI applications, as open-source frameworks like OpenClaw become increasingly influential in shaping the direction of the industry.

memeburn.com · Jun 21

Amazon Trainium Chips Take Aim at Nvidia

Amazon is in discussions to sell its Trainium processors directly to third-party data centers, a move that could reshape the economics of artificial intelligence infrastructure. Trainium has reached a $20 billion annual revenue run rate, with major commitments from OpenAI, Anthropic, and Uber. Amazon claims Trainium3 can match Nvidia's Blackwell at a lower cost. The company's shift from a "captive silicon" model to direct sales would place it in competition with Nvidia's core business model. Amazon's chip business could be worth $50 billion if operated as a standalone company. The demand for Trainium is outpacing supply, with the third-generation chips largely sold out.

💡 Sovereign AI demand is driving the need for locally controlled computing resources, and Amazon's move addresses this rising international appetite. By selling chips externally, Amazon can tap into the growing demand for AI infrastructure independence from US cloud providers.