사진 확대 On the 13th (local time), Koo Kwang-mo, head of LG, and Jensen Huang, CEO of NVIDIA, pose for a commemorative photo holding a miniature humanoid robot at NVIDIA’s headquarters in Santa Clara, California. LG
The memorandum of understanding (MOU) signed on the 13th (local time) by LG and NVIDIA for a strategic business partnership at NVIDIA’s headquarters in Santa Clara, California, carries major significance for both companies.
For the LG Group, the deal gives its affiliates a clearer path forward after being relatively left out of the recent boom in artificial intelligence (AI) infrastructure investment. It also raises expectations for meaningful growth in AI, one of the visions Koo Kwang-mo has set for the group. For NVIDIA, the agreement secures LG as an important partner with key technologies in physical AI.
Expectations for the partnership were also reflected in comments from NVIDIA CEO Jensen Huang. In a statement that day, Huang explained, “The most important opportunity presented by physical AI is to give every machine the ability to understand and reason about the real world, and to act safely alongside people. This will transform our daily lives, from homes and factories to roads.”
He added, “Building on years of collaboration, LG and NVIDIA are combining LG’s leadership in product engineering and manufacturing with NVIDIA’s technology to accelerate a new era of robotics, AI factories and autonomous vehicles.”
사진 확대
◆LG speeds up humanoid development
Through this partnership, the LG Group expects to move ahead in the humanoid robot market. LG Electronics is developing a next-generation bipedal humanoid robot based on NVIDIA’s open humanoid foundation model Isaac Groot and its robotics safety system NVIDIA Halos, with a goal of unveiling it in the first quarter of next year. The robot will be equipped with NVIDIA Jetson Thor GPUs. LG Electronics’ strength is that key humanoid components other than Jetson Thor, including actuators, sensors and batteries, are made directly by LG Group affiliates. This vertical integration can improve the competitiveness of humanoid robot manufacturing.
LG already has plenty of industrial sites where the robots can be deployed in practice. Later this year, it will introduce LG Cloid, a wheeled robot, to the washing machine production line at LG Electronics’ U.S. Tennessee plant for a proof of concept in a real manufacturing environment. Based on the results, the company plans to improve the robot’s performance and usability, then expand its use to a wider range of settings, including factories, homes and commercial spaces. LG’s software capabilities will also be used. The company will pursue technical cooperation to build a robot data factory based on LG CNS’s robotics platform PhysicalWorks. Through PhysicalWorks, the partners will expand cooperation so that a physical AI data factory, which repeatedly collects and generates data, trains models and verifies results, can be implemented and operated in real-world settings. LG also expects the robot data and manufacturing-site experience gained through this partnership to strengthen the competitiveness of its own robot foundation model under development.
From NVIDIA’s perspective, LG Electronics can be an excellent partner in humanoid robotics. Many companies are already building humanoid robots based on NVIDIA’s technology. However, U.S. companies lack manufacturing competitiveness, while Chinese companies face rising risks from political tensions. NVIDIA has already experienced the loss of the AI accelerator market in China to domestic players such as Huawei amid U.S.-China conflict.
To improve execution of the partnership, the two companies will operate a joint task force made up of technology and business experts.
◆Strengthening AI factories with ‘One LG’
LG expects the partnership with NVIDIA to allow it to enter the AI factory market in earnest, as large-scale investments are being made worldwide.
NVIDIA is currently building the DSX AI factory ecosystem and certifying companies that provide power, cooling technology, servers and software for AI factories. Certified companies are more likely to supply AI factories built on NVIDIA GPUs.
LG plans to build a reference AI factory based on NVIDIA’s DSX AI factory architecture. It will create a One LG Integrated Solution by combining capabilities from the data centers of LG Electronics, LG Energy Solution, LG CNS and LG Uplus, as well as LS Group’s power solutions.
In the first half of 2027, the company plans to complete technical validation based on NVIDIA Vera Rubin GPUs. In the first half of 2028, it will expand the AI factory in Cheonan, South Chungcheong Province, to an 80 MW scale. In particular, the AI factory will use a prefabricated modular design that is expected to shorten construction time by more than 20%. The facility is expected to be used to advance physical AI and LG’s robot foundation models.
The National AI Computing Center being built at the Haenam Solaseado Data Center Park will also be an 80 MW facility, with about 50,000 GPUs to be deployed by 2030.
If the AI factory is successfully built with LG’s technology, the company expects to be able to supply proven AI factory solutions to global big tech customers as a One LG package.
◆Developing next-generation vehicle solutions
The partnership will also expand into mobility. LG plans to develop a next-generation in-vehicle high-performance computing (HPC) platform based on NVIDIA’s DRIVE Hyperion autonomous driving platform. NVIDIA has recently been supplying automakers with both automotive semiconductors and solutions based on the Hyperion platform and its autonomous driving technology Alpamayo. Mercedes-Benz is already working with NVIDIA, and Hyundai Motor Company is also developing autonomous vehicles using NVIDIA technology. NVIDIA is taking market share that once belonged to Qualcomm.
LG Electronics is currently growing in the automotive components market based on infotainment systems and telematics. The company plans to use this solution to create a next-generation vehicle platform that integrates autonomous driving functions and in-car AI services based on NVIDIA technology, and then supply it to global OEM customers.
[Lee Deok-ju/Park Yoon-ye]
This article has been translated by GripLabs Mingo AI.