Armed with Nvidia GPUs, Samsung, SK, Hyundai, LG, Naver push into next-gen AI factories, robotics, smart manufacturing
GYEONGJU, North Gyeongsang Province — US chip giant Nvidia will supply 260,000 of its cutting-edge GPUs to the South Korean government and leading conglomerates — Samsung Electronics, SK Group, Hyundai Motor Group and Naver — in one of the largest AI infrastructure rollouts outside the US.
The announcement came Friday during a meeting between President Lee Jae Myung and Nvidia CEO Jensen Huang on the sidelines of the APEC summit in Gyeongju, North Gyeongsang Province. Also present were Samsung Electronics Chair Lee Jae-yong, SK Group Chair Chey Tae-won, Hyundai Motor Group Executive Chair Chung Euisun and Naver founder Lee Hae-jin.
“South Korea’s goal is to become the epicenter of AI in the Asia-Pacific region,” President Lee said. “We look forward to Nvidia’s partnership in building an AI ecosystem where infrastructure, technology and investment move in a virtuous cycle.”
The plan follows Thursday’s much-talked-about “chicken and beer” meetup among Huang, Lee and Chung, symbolizing deepening ties between the US chipmaker and Korea’s top industrial leaders.
"Korea has a chance to be one of the world's major AI hubs and today's announcement, along with President Lee's passion and enthusiasm and drive, and all of my CEO friends' who are dedicated to creating this journey for Korea,... this is a perfect example of sovereign AI," Huang said during a press briefing in the afternoon.
"There's no question in my mind that Korea will build many more AI factories here," he added, answering the question on whether he plans to collaborate with other companies in Korea as well.
Under the initiative, the Korean government will receive 50,000 Nvidia Blackwell GPUs, while Samsung, SK and Hyundai will each get their hands on 50,000 units. Naver Cloud will operate 60,000 GPUs to power its hyperscale AI services.
According to Nvidia, the GPUs will serve as the backbone of next-generation AI factories — massive data centers designed to integrate artificial intelligence across manufacturing, logistics, robotics and mobility.
Samsung will use Nvidia’s Nemotron, Cuda X and Omniverse platforms to build digital twins for chip production, enhancing yield and process precision. It is also developing home service robots using Cosmos and Isaac GR00T platforms, bringing “AI from the data center to daily life.”
SK Group plans to leverage the GPUs for AI-driven semiconductor manufacturing and cloud systems, while its telecom arm SK Telecom will use Nvidia’s RTX Pro 6000 Blackwell servers to power sovereign AI networks for local industries.
Hyundai Motor Group, focusing on autonomous driving and robotics, will co-invest about $3 billion with the Korean government to advance “physical AI” — technologies that apply AI reasoning directly to real-world systems such as vehicles and factories. The company also signed a separate three-way deal with the Ministry of Science and ICT and Nvidia to expand Korea’s physical AI ecosystem.
Beyond the industrial deployments, Nvidia said it will collaborate with the government and five Korean firms — LG AI Research, Naver Cloud, NC AI, SK Telecom and Upstage — to develop sovereign large language models using its open-source Nemotron datasets.
Officials said the collaborations will lay the groundwork for Korea’s next industrial leap — integrating AI across physical infrastructure to enhance efficiency, safety and productivity.
Samsung Electronics will establish a next-generation AI megafactory in partnership with Nvidia, embedding artificial intelligence across all stages of its semiconductor production chain — from design and wafer processing to equipment control and quality assurance.
The AI platform will run on more than 50,000 Nvidia GPUs, connecting and analyzing immense data streams to enable real-time prediction, diagnosis and optimization of manufacturing conditions.
“The Samsung AI Factory goes beyond traditional automation,” a company official said. “It connects and interprets vast data generated across chip design, production and equipment operations.”
Leveraging Nvidia’s Omniverse and Cuda-X platforms, Samsung will create digital twins of its fabs — virtual replicas that simulate and optimize production scenarios before physical implementation. The company will also apply cuLitho, Nvidia’s advanced computational lithography software, to accelerate chip patterning by up to 20 times, significantly improving yield and efficiency.
The project builds on a 25-year partnership between Samsung and Nvidia that began with DRAM supply for early GPUs. Their current focus centers on HBM4, a high-bandwidth memory solution integrating Samsung’s sixth-generation 10-nanometer DRAM with a 4-nanometer logic base die.
Samsung is also deploying Nvidia’s RTX Pro 6000 Blackwell and Jetson Thor processors in its robotics and smart manufacturing systems, using AI for predictive maintenance and autonomous operations.
The AI factory model will be rolled out across Samsung’s global network, including its new semiconductor plant in Taylor, Texas, underscoring its ambition to dominate every key semiconductor segment — memory, logic, foundry and advanced packaging.
“This marks a milestone in our journey toward AI-driven manufacturing leadership,” Samsung said.
SK Group is partnering with Nvidia to establish a large-scale AI factory equipped with over 50,000 GPUs, designed to form the backbone of a national “manufacturing AI cloud.”
The platform will strengthen Korea’s industrial competitiveness while offering AI-computing resources to startups and small manufacturers. The initiative aligns with Nvidia’s broader mission to democratize access to high-performance GPU computing.
At the forefront of the project is SK hynix, which will integrate digital-twin technology across its semiconductor lines. By creating virtual replicas of production systems, SK hynix aims to improve yield rates, enable predictive maintenance and reduce operational costs.
The group’s telecom affiliate SK Telecom will also work with Nvidia to develop next-generation AI-based wireless communication technologies that can handle the explosion of data from AI-powered devices, autonomous systems and industrial robots.
SK hynix, which supplies most of Nvidia’s HBM3 and HBM3E memory chips, plans to leverage the collaboration to consolidate its position as the world’s leading AI memory provider.
“This collaboration will strengthen SK’s AI manufacturing ecosystem and accelerate innovation across Korea’s industrial base,” the company said, noting that the infrastructure will serve as a foundation for AI-native supply chains and sustainable data processing.
Hyundai Motor Group and Nvidia will jointly invest $3 billion to advance Korea’s physical AI — a next-generation domain where AI interacts directly with the physical world through robotics, mobility and smart factories.
The two companies will establish an Nvidia AI Technology Center, a Hyundai Physical AI Application Center and a dedicated data hub to cultivate engineers and expand joint research. Hyundai will deploy 50,000 Nvidia Blackwell GPUs to develop integrated AI models that connect autonomous driving, robotics and production systems.
“We are laying the foundation for an AI ecosystem that fosters innovation and talent,” said Executive Chair Chung Euisun, emphasizing the role of AI in transforming mobility and manufacturing.
Nvidia CEO Huang said AI “will revolutionize every facet of transportation,” from vehicle design to logistics. Backed by the Ministry of Science and ICT, the partnership combines Hyundai’s industrial data with Nvidia’s GPU infrastructure to expand AI integration across Korea’s mobility and energy sectors.
Using Nvidia’s GPU-powered AI factory technologies, Hyundai aims to link in-vehicle intelligence, factory automation and robotics into a single interconnected ecosystem.
“Together with Hyundai — one of the world’s top mobility providers — we’re building intelligent cars and factories that will shape the future of the multitrillion-dollar mobility industry,” Huang said.
LG Electronics is expanding its alliance with Nvidia to strengthen physical AI, digital-twin systems and smart-factory innovation. The company has joined Nvidia’s comprehensive AI platform ecosystem and is developing its own humanoid AI model based on Isaac GR00T, Nvidia’s general-purpose robotics inference engine.
LG is applying reinforcement learning and synthetic data to train robots using its vast datasets drawn from appliance manufacturing, automotive components and industrial systems.
In manufacturing, LG is creating real-time digital twins of its global factories through Nvidia Omniverse and RTX Pro 6000 Blackwell GPUs. The systems enable virtual testing and optimization of equipment configurations before deployment, while AI-based monitoring identifies bottlenecks and defects in real time.
LG is also pursuing Nvidia certification for its AI data-center cooling units, essential for managing the heat output of high-performance computing clusters. The company is working to integrate its direct-current power solutions and heat-recovery technologies into sustainable data-center ecosystems, reducing energy consumption and emissions.
“We aim to accelerate innovation in future technologies through strategic cooperation with Nvidia,” said Yoo Eu-gene, head of open innovation at LG Electronics.
Meanwhile, LG AI Research, the conglomerate’s AI research hub, is expanding access to its Exaone large language model through collaboration with Nvidia, offering Korean firms and academic institutions greater AI computing capability.
Naver: Bridging digital, physical worlds
Naver Cloud, the cloud and AI subsidiary of Naver Corp., has also joined forces with Nvidia to develop a physical AI platform that bridges real-world industries and digital systems through AI, cloud computing and robotics.
During a meeting with President Lee and Nvidia CEO Huang, Naver founder Lee Hae-jin said the partnership signals “the opening of the physical AI era,” where intelligent systems directly support industrial operations.
“Naver will help businesses better use data and level up their industries with AI and cloud technologies,” Lee said.
Under the memorandum of understanding, Naver Cloud will build AI infrastructure tailored to Korea’s key sectors — including semiconductors, shipbuilding and energy — using Nvidia’s Omniverse and Isaac Sim alongside Naver’s digital-twin and robotics platforms.
The project forms the first step of Naver Cloud’s “Sovereign AI 2.0” vision, which aims to embed AI into national industries and daily life, expanding beyond linguistic and cultural applications.
“This partnership marks the start of the physical AI era, where AI meaningfully enhances productivity, safety and efficiency on the factory floor,” said Naver Cloud CEO Kim Yoo-won.
