At an all-hands meeting made public on October 22, Figure AI announced that its fourth-generation humanoid robot, Figure 04, has completed design lock and entered the hardware verification and testing phase, and that a larger version of its in-house end-to-end vision-language-action model Helix 3.0 is expected to complete internal testing within a few weeks.
ByteDance's Seed Model team added ten robotics positions, covering body design, core components, low-level control, application software, and data evaluation, extending from the model layer toward full-robot development. Unitree Technology released the humanoid robot H2-X "Kung Fu Legend Guardian"; a demo video shows a maximum battery life of about 8 hours, the ability to lift a car bare-handed, and a landing from 4 meters high. Daxiao Robotics released the embodied model ACE-Brain-0.6, which surpassed GPT-6Astra in six of eight Grounding evaluations.
Teradyne Robotics and Sereact have formed a partnership under which Sereact's Cortex physical AI platform will drive Universal Robots robot arms for unstructured tasks such as returns handling and picking; the two parties have already delivered dual-arm systems made up of hundreds of UR robot arms to customer sites in Europe.
ByteDance's Seed team updated its robotics hiring this month, extending roles from algorithms and simulation to data collection and robot hardware. Embodied data positions offer monthly pay of 55,000 to 85,000 yuan, and the data collection lead role starts above 70,000 yuan. The hiring covers Beijing and Shanghai and requires candidates to have experience in robot data collection, teleoperation, or on-site project management. Industry rumors say Seed has been collecting Ego data in China for several months.
A research team from Harvard, NVIDIA, Berkeley, MIT and Stanford proposes RAX (Robot Agent eXploration), which connects a general coding agent to robot perception and control interfaces, uses online exploration to write successful decisions into executable programs and accumulate them as reusable skills, while keeping model parameters frozen.
Why it matters: RAX uses a frozen model plus online exploration to generate reusable programs, leading by an average of 37.4 percentage points on five simulation benchmarks with zero action demonstrations, offering a path for robot learning different from scaling data and models.
Humanoid robots are extensively reusing resources from the automotive and industrial automation supply chains. McKinsey data shows their BOM cost is about $30,000 to $150,000 per unit, with a long-term target of pushing it below $20,000. However, actuators, force and tactile sensing, and computing and control are listed as high-risk bottlenecks, and dexterous hands are difficult to develop by reusing adjacent industries because of the lack of a standardized, large-scale supply ecosystem. Tesla Optimus's two hands, integrated with the forearms, contain more than 50 actuators, while the rest of the body has only 25 to 30; Tesla has used the Chinese supply chain to cut costs to $20,000 to $30,000, and Unitree's R1 is officially priced from 29,900 yuan.
The US Federal Communications Commission (FCC) has granted Intel a one-year rule exemption, allowing it to continue using the relevant equipment under an import ban on certain foreign-made advanced robots. Intel says these automated guided vehicles (AGVs) are used to transport wafers and materials within semiconductor factories, have been in use for more than twenty years, and that the market lacks mature US domestic alternatives that meet its operational requirements. The ban was introduced on July 28 this year and places newly manufactured foreign-made humanoid robots, quadruped robots, and networked power inverters on the controlled list.
Europe has spent generations mastering science, engineering, and manufacturing. That knowledge and expertise can now power AI of its own. "Let's build Europe's AI economy together and put AI into practical use." — @JensenHuang at GTC Berlin
NVIDIA released an illustrated tutorial that uses a fictional startup building a warehouse humanoid robot as an example, connecting tools including Isaac Sim, Isaac Lab, GR00T, Cosmos, and Omniverse, and covering virtual environment setup, demonstration teaching, policy training and testing, handling the Sim2Real gap, and deploying AI to real hardware. The tutorial also explains the steps between an AI model deciding to move an arm and the motor actually executing it, and the author hopes to exchange with developers who have actually used these tools about how real workflows differ from the examples.
Sereact this week reached a partnership with Teradyne Robotics (including Universal Robots and Mobile Industrial Robots) to have the Cortex physical AI platform drive UR robotic arms in handling unstructured tasks such as returns sorting and picking; the two sides say they have already delivered hundreds of UR dual-arm robotic systems at customer sites in Europe.
Dexory announced a collaboration with NVIDIA to advance physical AI in warehousing: DexoryView Storage Health integrates NVIDIA Cosmos, using foundation model reasoning to identify storage anomalies such as opened boxes, misaligned pallets, and misplaced goods.
Researchers from Shanghai Jiao Tong University, UC Berkeley, Stanford University and other institutions proposed the Physical RSI framework in the paper "Who Improves the Robot? Toward Physical Recursive Self-Improvement," incorporating task capability updates and improvement process updates into a unified analysis and presenting an L0–L5 grading scale.
Kazakhstan is moving from AI infrastructure planning to deployment. NVIDIA founder and CEO Jensen Huang visited Astana on October 23 and announced an investment in Kazakhstan-founded startup Higgsfield AI.
The October 23 morning brief covers AI industry developments: Meta's third-quarter revenue was $64.6 billion, but free cash flow turned to -$2.12 billion, and the AI agent Muse gained over 10 million downloads in 40 days after launch; OpenAI raised up to $30 billion at a pre-money valuation of $1.25 trillion, and Anthropic plans to publicly file an S-1 with a target valuation of about $2 trillion; Nvidia released the modular AI factory reference design DSX POD, compressing on-site deployment time from about 6 months to as fast as 14 days.
Events including Nvidia's GTC Berlin conference and Tesla's Q3 earnings call converged in the same window, bringing multiple catalysts to the A-share AI supply chain; the Wind AI concept index has risen more than 2% cumulatively since October 19. Analysts believe Nvidia's system-level layout for AI factories and physical AI, combined with Tesla's commercialization progress on Robotaxi and humanoid robots, is expected to drive supporting segments such as liquid cooling, high-speed interconnect, and robot actuators. Whether the rally can shift from event-driven to fundamentals-driven still depends on order visibility and progress in delivering real-world use cases.
At GTC Berlin, Jensen Huang announced that the Vera Rubin platform has entered full-scale production and introduced DSX POD, a modular AI factory reference design. It consists of 7 liquid-cooled computer racks interconnected by about 17 kilometers of cabling and uses 800V DC power; assuming power and site readiness, deployment time can be shortened from about 6 months to as fast as 14 days, with deliveries expected to begin in early 2027.
Totally adorable. Jensen Huang brought four Microduck robots on stage during his NVIDIA GTC Berlin keynote.
In his GTC Berlin keynote, Jensen Huang launched several products, including the OpenShell reference architecture that constrains agent permissions outside the sandbox, the DSX POD that can be pre-integrated at the factory and cut deployment time from about six months to as little as 14 days, and the Agile ONE SN humanoid reference robot developed with Agile Robots.
On October 21, 2026, NVIDIA held its first Europe-hosted GTC conference in Berlin, where Jensen Huang delivered a keynote. Ahead of the keynote, NVIDIA announced that all seven chips in the Vera Rubin GPU family had completed tape-out, with mass production moved up to the third quarter of 2026; OpenShell, the open-source agent security platform released at the end of September, has already been announced as adopted by over 100 organizations; and the RTX Spark series of PCs was confirmed to launch this fall.
Jensen Huang came to work the booth. Walking around GTC Berlin, meeting the people and partners bringing AI to communities across Europe. 🧵
Schneider Electric, Nvidia and HPE have formed a strategic collaboration to launch AI agents that can monitor and optimize industrial operations inside factories and process plants. The first batch, Industrial AI Agents at the Edge, focuses on operational technology (OT) monitoring, cybersecurity and energy consumption optimization.
At GTC Berlin, Jensen Huang said PCs need to be reinvented as personal agents and showcased RTX Spark, a four-year collaboration with Microsoft that uses a chip design combining CPU and GPU, offers up to 128GB of unified memory and up to 1 petaflop of FP4 AI compute, and can run larger models locally.
LG Innotek, part of LG Group, is partnering with Nvidia to jointly develop perception solutions for physical artificial intelligence and robotics. LG Innotek will provide camera modules, lidar, and radar fusion perception technology, while Nvidia will contribute its Holoscan Sensor Bridge platform to enable direct data connections between sensors and GPUs.
Tesla and Nvidia will invest 1.1 trillion won and 2.6 trillion won respectively to build an FC-BGA (flip-chip ball grid array substrate) line for AI semiconductors with Samsung Electro-Mechanics; these substrates have become a new bottleneck in the AI supply chain. According to industry sources on the 22nd, Samsung Electro-Mechanics has signed a long-term FC-BGA supply agreement with Tesla and is building a dedicated line at its Sejong plant; Tesla's investment funds line expansion and equipment, in exchange for medium- to long-term stable supply guarantees, and these substrates will supply the AI chips used in Tesla's self-driving electric vehicles and its Optimus humanoid robot.
NVIDIA has expanded the Cosmos platform with new training tools, simulation checkpoints and Cosmos Data Factory blueprints for training, evaluating and deploying world models.
Why it matters: The original lists Cosmos's new training tools, simulation checkpoints and data blueprints, and gives specific data on users such as Porsche and ANYbotics, which can be used to assess changes in the world model development workflow.
NVIDIA announced several implementations of open models and physical intelligence in European retail: Shopify used NVIDIA Megatron Bridge and NeMo RL to post-train Nemotron 3.5 Lightning, cutting serving costs by 90% and delivering 7x faster responses on internal benchmarks.
Nvidia CEO Jensen Huang announced in Paris a new round of strategic investments in four French startups, covering health tech, chemical materials, business planning, and robotics. Genesis AI is the only robotics target, and it is negotiating a new funding round at a valuation of about $3 billion, roughly 20 billion yuan.
AgiBot announced that starting October 22, it will release seven full-stack technology achievements over seven consecutive working days, covering datasets, an operation platform, the heavy-duty robot G2 Max, the embodied Agent framework GenieMaster, a simulation platform, a foundation model, and a collective experience learning paradigm.
In his GTC Berlin speech, Jensen Huang said the ChatGPT moment for robotics is approaching rapidly rather than having already arrived, and on the same day released upgrades to the Isaac platform, a humanoid robot reference framework, and simulation and data tools to prepare the infrastructure for that moment.
NVIDIA CEO Jensen Huang said at the Build a Claw event in London that OpenClaw is one of the most important models to date, calling it "the first truly trustworthy agent runtime framework" and describing it as providing a real-time runtime framework for LLMs that can be used in robotics. He also stressed that autonomous systems must treat every object as untrusted by default and grant only minimum privileges, and introduced OpenShell, an isolation system NVIDIA developed for this purpose that continuously monitors for privilege violations and issues alerts.
Agile Robots unveiled the Agile ONE SN humanoid robot reference platform at Nvidia GTC Berlin, integrating Nvidia's Isaac GR00T robot development platform with the Jetson Thor edge computing system, with launch planned for early 2027.
In his GTC Berlin keynote, Jensen Huang announced that Nvidia and Agile Robots are launching the Agile ONE SN reference humanoid robot, standing 1.4 meters tall, weighing 40 kilograms, with 31 degrees of freedom, preloaded with the Isaac GR00T model and powered by Jetson Thor. Expected to launch in early 2027, it is intended to cut the time needed to set up robot research hardware and software from weeks to days.
At GTC Berlin 2026, NVIDIA expanded its Isaac robotics development platform. The now generally available Isaac Lab 3.0 supports multi-GPU training and connects to physics engines such as Newton and PhysX, while the open-source runtime framework Isaac OS connects AI models, sensors, control systems, and robot hardware.
UK AI and robotics company Humanoid announced an expansion of its collaboration with NVIDIA to jointly develop a high-performance robotic edge runtime for next-generation humanoid robots. Humanoid will integrate NVIDIA Jetson Thor, Isaac OS, Holoscan Sensor Bridge, and the Holoscan SDK to process large volumes of sensor data, run AI models onboard, and coordinate perception, reasoning, and control in real time.
At NVIDIA GTC, Agile Robots and NVIDIA unveiled Agile ONE SN, a compact humanoid reference platform for R&D, built on the NVIDIA Isaac GR00T development platform and integrating a complete workflow spanning demonstration capture, simulation, training, evaluation, deployment, and real-world testing.
Why it matters: The original article lists the specific configuration of the robot's whole-body degrees of freedom, compute power, and preconfigured software stack, which can be used to assess how this integrated solution for R&D labs connects with the industrial model.
NVIDIA released and open-sourced the data engine Flora, which converts raw robot video into training data such as timestamped action descriptions, process rewards, and outcome rewards; it is already used to train the GR00T series models. Flora consists of an action description model, a temporal segmentation model, and a reward model, fine-tuned from the Qwen3.5 series of vision-language models, and can label 200,000 hours of data per day on 512 GB300 GPUs.
Why it matters: NVIDIA's open-source data engine cuts robot video labeling costs to about 1/125 of a general-purpose large model and provides several benchmark comparisons, which can be used to assess the substitution space in the data labeling segment.
NVIDIA has released and open-sourced Cosmos3-H-Dreams, a real-time generative surgical world model that can generate surgical imagery in real time based on robot operation commands, simulating instrument movement, tissue traction, and deformation, with support for laparoscopic cholecystectomy scenarios.
1 petaflop.
Even Microducks need a nap. 😴
At GTC Berlin, NVIDIA showcased its full lineup of local AI hardware and announced multiple software and ecosystem updates for RTX Spark, DGX Spark, and DGX Station.