50 articles found
Robot Talk Episode 164 features Michelle Lu, CEO of Vsim Technology, discussing how simulation and artificial intelligence can accelerate robot learning. Lu previously led simulation technology at NVIDIA where she developed Isaac Gym, a GPU-accelerated reinforcement learning framework, and now focuses on addressing efficiency, scalability, and sim-to-real gap challenges in robotics AI.
NVIDIA teams use ChatGPT Work to reduce manual tasks, connect fast-moving signals, and scale successful workflows globally.
Filing comes after Elon Musk announced exclusive arrangement to kit out its data centers.
PlusAI announced updated commercial readiness metrics for H1 2026, showing progress toward deploying factory-built autonomous trucks at scale in 2027, including improvements in safety validation, autonomous miles percentage, and remote assistance-free trips. The company launched commercial trials with International and Ryder in Texas, expanded operations to Southern Europe, released SuperDrive 6.0 with enhanced capabilities including night driving, and expanded partnerships with NVIDIA and international logistics operators.
VicOne has released a free cybersecurity extension for NVIDIA Isaac Sim that enables robotics developers to test how attack scenarios could affect their robot models, based on research from DEF CON 34's Robotic Hacking Community. The tool translates real-world hacking challenges into repeatable simulation scenarios covering AI models, perception, cloud services, communications, and embedded systems to help developers validate cyber-safety earlier in development.
The article examines five foundational physical AI infrastructure platforms shaping robotics development in 2026, including NVIDIA's accelerated computing and simulation ecosystem (Isaac, Newton), and Applied Intuition's validation engineering for autonomous systems. These platforms represent critical layers in the emerging physical AI stack, addressing bottlenecks in robot perception, reasoning, action, and real-world deployment across industries like automotive, manufacturing, and autonomous vehicles.
AMD's new robotics platform and the broader shift toward custom silicon reveal a quiet power struggle in robotics: whoever controls the compute layer controls the industry. As giants from AMD to NVIDIA battle for the robot brain, the question isn't just about performance — it's about who gets to define what robots can do.
AMD unveiled its Ryzen AI Embedded X100 series and Kria AI Robotics platform designed to provide deterministic real-time control and unified CPU-GPU-NPU memory for robotics applications. The company claims the X100 outperforms NVIDIA's Orin and Thor on system-level robotics workloads while supporting industrial arms, autonomous mobile robots, and humanoids on a single scalable platform.
Medtronic unveiled Touch Surgery Aide, an AI-powered compute platform for operating rooms that leverages NVIDIA's accelerated computing infrastructure to deliver real-time surgical insights during procedures. The platform's first FDA-cleared application, Instrument Exit Point, uses computer vision to alert surgeons when robotic instruments move out of view, representing the initial step in Medtronic's broader effort to embed AI capabilities into its Hugo robotic-assisted surgery system.
At SIGGRAPH 2024, NVIDIA showcased advances in neural rendering, world models, and AI-powered simulation technologies applicable to robotics, autonomous systems, and digital twin design. Key announcements include breakthroughs in 3D-guided neural rendering, AI physics models for weather and aerodynamics, and the Cosmos world foundation model platform for developing physical AI systems.
NVIDIA's new Rubin generation of AI infrastructure uses innovative liquid cooling systems with 45-degree Celsius (113°F) coolant, which paradoxically cools more efficiently than traditional cold-air systems. This approach reduces data center cooling energy costs by an estimated $4 million annually for large facilities, eliminates water consumption, reduces noise pollution, and improves space efficiency compared to conventional cooling methods.
Erin Davis calls it the "SuperDuperPOD." That's two things in one name: pharmaceutical giant Bristol Myers Squibb (BMS) already runs one of the largest AI clusters in life sciences, with serious resul...
NVIDIA introduced Vera Rubin and Nemotron 3 Ultra, focusing on optimizing post-training workloads for agentic AI systems that continuously adapt in production environments. The company emphasizes 'intelligence per dollar' as a key metric for agentic AI, where post-training becomes a continuous process rather than a one-time step, and released open-source NeMo libraries to make post-training infrastructure more accessible and repeatable.
Fujitsu, Nvidia, and Japan's leading robotics manufacturers (Fanuc, Yaskawa, Kawasaki Heavy Industries) announced a major collaboration to develop "physical AI" — autonomous robots that can think independently and work alongside humans in factories, homes, and hospitals. The initiative aims to address Japan's labor shortage by combining Nvidia's AI technology with Japanese manufacturing excellence, with the first phase of collaboration expected later in 2026.
NVIDIA introduced the Jetson T3000 and T2000 modules based on its Thor architecture to enable mass-market robotics and edge AI deployment. The T3000 delivers 865 FP4 teraflops with a compact form factor and functional safety capabilities, while the T2000 provides an entry-level option with 400 teraflops for visual AI agents and autonomous systems. NVIDIA also released AI agent skills that automate memory optimization, helping companies like Agile Robots and Hitachi reduce memory usage by up to 15GB while maintaining performance.
NVIDIA founder Jensen Huang visited Japan to showcase the company's AI and robotics ecosystem, highlighting how Japan's manufacturing expertise combined with AI can drive innovation in robotics and intelligent manufacturing. The visit included a Build-a-Claw event where developers used NVIDIA's platform to build robotic systems, and meetings with Japan's supply chain and technology leaders to discuss AI's potential to address labor shortages and create new economic opportunities.
NVIDIA introduces methods for evaluating general-purpose robot foundation models and policies for real-world deployment, addressing critical gaps in current simulation benchmarks. The company identifies key problems including visual domain overlap between training and evaluation, benchmark saturation, lack of diagnostic insights into failures, and statistical trustworthiness issues, and proposes solutions like the Clopper-Pearson confidence interval method for more rigorous robot policy evaluation.
NVIDIA's Nemotron Labs blog post discusses how open AI models enable enterprises to customize, control, and trust AI systems for their specific needs, removing barriers imposed by closed models. The article highlights real-world examples of companies like Abridge, Glean, Harvey, and others specializing Nemotron models for healthcare, legal, search, and multilingual applications, demonstrating competitive advantages through domain-specific fine-tuning and reduced inference costs.
NVIDIA publishes technical analysis on performance-per-watt as the critical metric for AI infrastructure efficiency, arguing that power constraints determine profitability for AI factories. The company highlights how its Blackwell NVL72 platform delivers up to 25x better performance-per-watt compared to Hopper generation systems when serving mixture-of-experts models, achieved through full-stack codesign across silicon, networking, and software optimization layers.
NVIDIA launches trading cards that feature 'GeForce PC gaming's great moments.'
NVIDIA and Hugging Face have integrated the Isaac GR00T 1.7 vision-language-action model and Isaac Teleop framework into LeRobot, an open-source robotics library, to provide developers with standardized tools for humanoid robot development, data collection, and policy training. The collaboration aims to democratize physical AI by connecting NVIDIA's 3 million robotics developers with Hugging Face's 16 million AI builders, with plans to add NVIDIA Cosmos 3, a frontier world model for physical AI, to the platform soon.
NVIDIA's Nemotron 3 Ultra language model, optimized by LangChain for its Deep Agents platform, achieves benchmark-leading performance at 10x lower inference cost than closed-source models while maintaining parity on business tasks. Rather than retraining the model, LangChain engineered the harness around Nemotron 3 Ultra by tuning system prompts, tool descriptions, and middleware, making the optimized configuration immediately available to developers. The open-source stack, called NemoClaw, gives enterprises full control over their AI agent systems across infrastructure, with early adoption from companies like Abridge, Amdocs, Box, and EY.
Max single-threaded CPUs at scale are a new category of CPUs built for the agentic AI era. Across the creation and deployment of an agentic system, the CPU is on the critical path for reasoning, resp...
DeepSeek, a Chinese AI company, is reportedly developing its own inference chips to reduce reliance on third-party providers like NVIDIA and Huawei. The company is hiring engineers and engaging with manufacturing partners as part of its expansion strategy in the competitive Chinese AI market.
NVIDIA and Hugging Face are integrating NVIDIA's Isaac GR00T 1.7 vision-language-action model and Isaac Teleop framework into LeRobot, an open-source robotics library, to democratize robot development tools and datasets for developers. The collaboration also plans to add NVIDIA Cosmos 3, a frontier world model for physical AI, providing a standardized pipeline for training, evaluating, and deploying robot foundation models across the open robotics community.
Open AI models and infrastructure have become foundational to modern AI research, as evidenced by thousands of papers at ICML 2026 citing open models like NVIDIA Nemotron, Cosmos, and Isaac GR00T. Key research areas include robot world models (such as DreamDojo for learning physical reasoning), AI for life sciences, synthetic data generation, and autonomous vehicle development, with companies like Sakana AI and Merck leveraging these open foundations for their work.
This roundup covers June 2026's top 10 robotics developments, including NVIDIA's physical AI tools, AGIBOT's 15,000th robot production milestone, humanoid deployments by Figure at BMW, major funding rounds for Standard Bots ($200M), NEURA Robotics ($1.4B), and General Intuition ($320M), Agility Robotics' SPAC merger announcement, RealSense's new AI-native depth camera, and Collaborative Robotics' Proxie Gen 2 mobile manipulation robot.
NVIDIA is introducing a new business model that provides AI startups, model builders, and enterprises with access to large-scale compute infrastructure through revenue-sharing partnerships with AI cloud companies. Sharon AI and Firmus are among the first partners deploying tens of thousands of NVIDIA GPUs through this initiative, with Firmus planning a 360-megawatt AI factory in Indonesia to serve customers globally.
NVIDIA announced advances in graphics, simulation, and agentic AI at SIGGRAPH, highlighting developments in real-time simulation and AI for robotics applications. The company introduced the Vera Rubin model for cost-efficient post-training workloads and emphasized the role of open models in enterprise AI development and research.
The Robot Report Podcast features an interview with Drew Henry, executive vice president of Arm Holdings' Physical AI Business Unit, discussing the company's strategy for computing and software technologies powering robotics, autonomous systems, and automotive applications. The episode covers industry news including Bear Robotics' acquisition of Kinisi Robotics, NVIDIA's release of Halos safety system for robotics, and a collaboration between Built Robotics and Penn xLAB on physical AI for construction.
NVIDIA announced its BioNeMo Agent Toolkit integration with Anthropic's Claude Science, enabling AI agents to access accelerated computational workflows for life sciences research through natural language interfaces. The toolkit allows scientists to run sophisticated analyses in genomics, proteomics, drug discovery, and protein engineering by connecting Claude Science agents to NVIDIA's accelerated models, libraries, and microservices.
Vention has expanded its AI-powered automation platform to support FANUC industrial and collaborative robots, enabling manufacturers to autonomously generate collision-free motion paths and deploy automation more quickly. The partnership leverages Vention's MachineMotion AI and goal-driven programming approach, combined with NVIDIA's Foundation Stereo model, to reduce commissioning complexity and accelerate adoption of robotic automation across factories.
NVIDIA and FORT Robotics both released comprehensive safety systems this week, signaling a fundamental shift in how the industry thinks about autonomous robot deployment. Safety is no longer a feature companies bolt on — it's becoming the foundation everything else builds upon.
FORT Robotics and NVIDIA launched an AI-driven Outside-In Safety blueprint that uses external infrastructure sensors and visual AI agents to enhance autonomous robot productivity and worker safety in industrial environments. The system, demonstrated at Automate 2024, extends robot perception beyond onboard sensors using NVIDIA IGX Thor and Holoscan Sensor Bridge, allowing robots to operate safely alongside workers at higher efficiency levels while dynamically adapting to complex warehouse and manufacturing settings.
NVIDIA launched Halos for Robotics, a comprehensive full-stack safety system designed to unify AI compute and safety for autonomous robots operating in factories, warehouses, and logistics. The system builds on NVIDIA's 18,600+ years of autonomous vehicle safety engineering and includes hardware (IGX Thor, Holoscan Sensor Bridge), software (Halos OS), and an accredited inspection lab, with early access available for developers and the open-source Halos Outside-In Safety Blueprint now on GitHub.
HPE and NVIDIA are expanding the HPE AI Factory to support agentic AI workloads, introducing the NVIDIA Vera CPU for agent-specific computing and the NVIDIA Agent Toolkit for HPE Private Cloud AI, alongside extended confidential computing capabilities across the entire portfolio. The Vera CPU, designed for tool calls and real-time data processing in agent loops, will be available in 2027, with early enterprise adoption from the New York Stock Exchange exploring the technology.
Richtech Robotics launched a round-the-clock interactive livestream platform featuring its ADAM AI-powered humanoid robot, allowing global audiences to interact with the robot in real time through chat and direct requests. The initiative demonstrates advances in embodied AI and natural human-robot interaction, with ADAM having previously served at the Kennedy Space Center and prepared noodles at a restaurant industry show, while the company expands its AI infrastructure with a new Las Vegas warehouse.
NVIDIA's Blackwell platform swept MLPerf Training 6.0 benchmarks, achieving the fastest training times across all seven benchmarks and scaling to 8,192 GPUs—the largest Blackwell-based submission to date. The results showcase performance improvements including up to 1.6x faster training with the GB300 NVL72 over GB200 NVL72, and demonstrate Blackwell's capability for large-scale mixture-of-experts model training including DeepSeek-V3 671B and Llama 3.1 405B.
Edge AI processors from companies like NVIDIA, AMD, and Qualcomm are enabling robots and automated systems to run AI models locally without internet connectivity, but their complexity limits adoption to experienced engineers. Numurus has developed NEPI, a software platform that abstracts away low-level complexity through plug-and-play drivers, built-in applications, and browser-based interfaces to democratize edge AI deployment in robotics, similar to how Windows made PCs accessible to non-experts.
NVIDIA's Blackwell Ultra platform achieves leading performance on AgentPerf, the industry's first agentic AI benchmark, delivering 20x more concurrent agents per megawatt than the prior Hopper generation. AgentPerf benchmarks agentic AI workloads—where agents chain multiple LLM calls and tool invocations together—rather than single conversation turns, reflecting how AI agents actually operate in production environments like code generation tasks.
NVIDIA announced Halos OS, a comprehensive safety system for autonomous robotaxis that addresses regulatory requirements for level 4 autonomous vehicles through a certified operating system, standardized hardware/software interfaces, and verifiable AI guardrails. The article covers new robotaxi deployments by Uber, Foxconn, VinFast, and HUMAIN using NVIDIA's DRIVE Hyperion platform, alongside the technical architecture of Halos OS designed to ensure safe autonomous operation at scale.
German auto supplier Bosch is pivoting toward humanoid robotics and related sensor technology as its traditional auto parts business faces pressure from competition and weak demand. The company plans to expand production of MEMS sensors crucial for robotic dexterity and has partnered with robotics firm Neura, which recently raised $1.4 billion in funding from backers including Bosch, Nvidia, Amazon, and Tether to accelerate deployment of robots and physical AI systems.
Google DeepMind released DiffusionGemma, an experimental open-source language model that generates multiple tokens in parallel instead of sequentially, achieving up to 4x faster text generation speeds. NVIDIA has optimized DiffusionGemma to run efficiently on its GPU hardware including RTX, DGX Spark, and DGX Station systems, enabling low-latency local AI inference for developers and researchers.
NEURA Robotics announced a Series C funding round of up to $1.4 billion to accelerate development of physical AI and cognitive robots. The company is building an ecosystem called the Neuraverse that combines robotics, AI, sensors, and edge computing for manufacturing, logistics, and service applications, with strategic partnerships including Qualcomm, Amazon, NVIDIA, and Bosch.
This article discusses why robotics will not experience a clean open-source moment like Meta's Llama LLM, arguing that robot policies cannot simply be downloaded and deployed without significant integration work. The piece examines how models from DeepMind, NVIDIA, and startups like Physical Intelligence and Skild AI are becoming more accessible, but emphasizes that deploying robot policies requires extensive local adaptation, hardware-specific tuning, and handling of real-world site drift that differs from software model distribution.
NVIDIA's Confidential Computing technology is now being deployed in Apple's Private Cloud Compute to enable secure server-side inference for Apple Intelligence features, with NVIDIA Blackwell GPUs running on Google Cloud infrastructure. The technology uses trusted execution environments and cryptographic verification to protect user data during AI processing without compromising performance, representing a shift toward privacy-preserving cloud-based AI services.
Daimon Robotics and Galbot have launched RobOmni, the first omni-modal evaluation benchmark for tactile perception and dexterous manipulation in robots, debuting at ICRA 2026. Built on NVIDIA Isaac Sim, RobOmni provides a standardized framework to measure how tactile sensing improves robotic manipulation tasks like grasping, insertion, and assembly, addressing the industry's lack of unified benchmarks for evaluating physical interaction capabilities beyond vision alone.
The UK is advancing its sovereign AI capabilities through significant infrastructure investments and government backing, with NVIDIA playing a central role in deploying AI compute across the country. Key initiatives include the Isambard-AI supercomputer (powered by 5,400 NVIDIA GH200 chips), the Sovereign AI Fund supporting homegrown startups like Cosine and Cursive, and multiple cloud providers deploying advanced AI infrastructure across UK soil to reach 65 megawatts by 2027.
NVIDIA and LG Group are partnering to build an AI factory that combines NVIDIA's AI infrastructure with LG's manufacturing expertise to accelerate physical AI, robotics, and autonomous systems development. The collaboration includes integrating NVIDIA's Isaac simulation frameworks and GR00T foundation models into LG's home robot development (like CLoiD), creating a physical AI data factory for synthetic training data, and establishing an autonomous manufacturing ecosystem with real-time data connectivity. Additionally, LG will expand AI factory infrastructure capabilities aligned with NVIDIA's DSX platform, including thermal management and modular deployment technologies.
The Robot Report Podcast features Winston Leung from QNX discussing how deterministic real-time systems and functional safety are critical for robotics operating in human environments. The episode explores QNX's microkernel architecture approach to providing real-time control and cybersecurity for autonomous systems, with insights into partnerships with NVIDIA and Intel for optimizing high-performance computing in next-generation robots.
