LG ties its humanoid robot to Nvidia while TIER IV opens an autonomous-driving chip
LG is building its first bipedal robot on Nvidia's humanoid foundation model, while TIER IV plans to open its autonomous-driving chip logic, exposing two critical layers of physical AI.
Artificial Intelligence··Midday
LG's bipedal robot and Nvidia's foundation
LG Group chairman Koo Kwang-mo and Nvidia chief executive Jensen Huang signed a partnership agreement on Thursday at Nvidia's Santa Clara headquarters covering humanoid robots, AI factories and autonomous driving. LG aims to show its first bipedal humanoid in the first quarter of 2027. The robot is to run on Nvidia's Isaac GR00T humanoid foundation model with the Jetson Thor computing platform and the Halos safety architecture. Actuators come from LG Electronics, sensors from LG Innotek and batteries from LG Energy Solution; initial production has started at the Changwon plant. The partnership stretches beyond one product reveal: the two companies will also develop a computing platform for AI-defined vehicles on Nvidia's Drive Hyperion. Bringing model, compute and safety layers into the same agreement makes the software foundation of a physical robot an explicit industrial choice.[1]
Factory line and power plan
Before the bipedal reveal, LG plans to put its wheeled CLOiD robot on a washing machine line at its Tennessee plant by the end of 2026. Physical automation therefore enters the factory first on a wheeled platform, with the bipedal form scheduled later. A reference AI factory site using Nvidia's Vera Rubin platform is set for the first half of 2027, followed by an 80 megawatt facility in Cheonan in the first half of 2028. Modular construction is expected to build that facility more than 20 per cent faster. The timetable ties the robot's software foundation to factory infrastructure and power capacity on the same calendar: model choice alone is not enough when compute and energy ground the plan. The robot, factory and vehicle strands of the deal ask readers to treat physical AI as production and power investment, not only as a stage demonstration.[1]
An open autonomous-driving chip
TIER IV has joined the Next-Generation Edge AI Semiconductor Research and Development Program run by the Japan Science and Technology Agency. The company plans to open-source the design assets and toolchain of a system-on-chip built for Level 4 autonomous driving. Inside the programme a research team led by Professor Yoshihiro Kawahara of the Graduate School of Engineering at the University of Tokyo works on use-case-driven, functionally differentiated physical AI chip design, while TIER IV develops the logic design of a chip meant to run inference for end-to-end autonomous driving. The chip is to support Autoware, the open-source autonomous-driving stack TIER IV maintains, and the company says it will open the compiler and related toolchain alongside the logic design so that chipmakers can build on the platform. Founder and chief executive Shinpei Kato tied the plan to verification, saying the ability to understand how an AI model is transformed for execution and to check its correctness will matter more and more. Kawahara said the project aims to overcome GPU power consumption constraints through functionally differentiated design. No schedule for a first chip was given. LG's closed partnership foundation for a humanoid and TIER IV's planned open chip logic put the software layer and the silicon layer of physical AI on the same agenda at different degrees of openness.[1], [2]
