LG and Nvidia signed a memorandum of understanding on August 13 to build LG's first bipedal humanoid robot, with Nvidia supplying the AI brain and LG supplying its own actuators, sensors and batteries.
The deal was signed at Nvidia's Santa Clara headquarters, with LG Corp chairman and CEO Kwang Mo Koo and Nvidia founder and CEO Jensen Huang both in the room. The robot will run on Nvidia's Isaac GR00T foundation model for reasoning and behavior, Jetson Thor for onboard compute, and Nvidia Halos, which Nvidia describes as the industry's first full-stack safety system built specifically for robots. LG is targeting a public unveiling in the first quarter of 2027.
What makes this deal different from most of Nvidia's other robotics partnerships is where the hardware comes from. Nvidia has been building humanoid reference designs with outside partners: it teamed with China's Unitree in June to unveil a GR00T-powered robot using Unitree's H2 chassis and hand hardware from Singapore's Sharpa. LG doesn't need an outside chassis supplier. Under what the company is calling a "One LG" strategy, the actuators come from LG Electronics, the sensors from LG Innotek, and the batteries from LG Energy Solution. Nvidia provides the brain. LG provides the entire body.
A quieter experiment first #
Before any humanoid shows up in public, LG is running a quieter experiment. The company will deploy CLOiD, its existing wheeled robot, on the washing machine assembly line at its Clarksville, Tennessee plant sometime this year. That factory opened in 2018 as a $360 million, one-million-square-foot facility that already uses hundreds of robots for material handling and assembly, and it produces more than a million washers and dryers annually. CLOiD's job there is to prove out navigation and safety on a real production line - coordination comes with it - before LG hands any of that responsibility to a two-legged robot with far more ways to go wrong.
The MOU goes beyond the humanoid, too. LG plans to build AI factory reference sites running on Nvidia's DSX and Vera Rubin platforms, and it's developing a next-generation AI-defined vehicle platform on Nvidia DRIVE Hyperion. That's a lot to sign in one afternoon. But it fits a pattern Nvidia has been running all year: sell the compute and the foundation model, let partners bring the metal.
Samsung Sold Out Every Gigabit of Its 2026 HBM4 Memory Chips Samsung has sold out its entire 2026 HBM4 memory supply, with CFO Park Soon-cheol confirming the sellout on an April earnings call. The chip is critical to Nvidia's Rubin and AMD's MI400 GPUs, and with SK Hynix slowing its own ramp and Micron locking in fixed contracts, prices are on track to roughly double by 2027.
The bigger bet is on Nvidia, not LG #
Frankly, the more interesting story here isn't the robot. It's what Nvidia is quietly becoming. Chips built the company's trillion-dollar valuation, but Jensen Huang has spent 2026 turning Isaac GR00T into something closer to an operating system for the entire physical robotics industry. Nvidia's list of partners building on that stack now includes Figure, Agility Robotics, KUKA, FANUC, ABB Robotics, Universal Robots and AGIBOT, alongside research labs at Stanford, ETH Zurich and UC San Diego, according to Nvidia's own newsroom disclosures. LG joining that list is notable less because LG is first and more because LG is huge, vertically integrated, and has zero robotics-first history to lean on.
That's the real bet LG is making. It isn't chasing Boston Dynamics or Figure on their own turf. It's wagering that a company already making the actuators, motors, batteries and sensors that go into washing machines and air conditioners - and into electric-vehicle components too - can repurpose that supply chain into a robot body, and let Nvidia handle the part LG has never built: the software that lets a machine walk, balance and decide what to do next.
Samsung has its own robotics push through its stake in Rainbow Robotics, and Tesla is racing to scale Optimus inside its own factories. Same bet, different toolbox. Both are also betting that whoever controls the physical supply chain, not just the AI model, wins the humanoid race longer term. LG just told the market it plans to be in that fight with parts it already knows how to make.
None of that guarantees LG ships a robot people actually buy. Jetson Thor and Isaac GR00T give LG a serious head start on the software side, the part that has sunk plenty of well-funded robotics startups. But turning a washing machine plant validation run into a robot that works in a real home or a hospital ward is a much longer road than a signing ceremony in Santa Clara. LG has eighteen months until its own deadline to prove it can walk that road.
Also read: Why AI Agent Approval Queues Are Replacing Full Autonomy for Founders • Palantir's USA Today Data Deal Triggers Revolt From 800 Journalists • Claude Opus Beat a Human NanoGPT Speedrun Record, But Not the Real Gap