NVIDIA deepens its AI footprint in South Korea: assisting Samsung, SK, and Hyundai in building AI factories, Samsung may supply its HBM4 chips

According to the Zhitong Finance APP, during NVIDIA (NVDA.US) CEO Jensen Huang's visit to South Korea on October 31, NVIDIA announced collaborations with Samsung Electronics, SK Group, and Hyundai Motor Group to promote the construction of AI factories.

Under the agreements reached, NVIDIA will supply over 260,000 accelerated chips to help South Korea launch AI projects.

Samsung Electronics

NVIDIA will collaborate with Samsung to build a new generation AI factory, marking a new era of integration between intelligent computing and chip manufacturing. Samsung's semiconductor AI factory will be equipped with over 50,000 NVIDIA graphics processing units (GPUs). Samsung stated that it is in "close discussions" to supply NVIDIA with the next-generation high-bandwidth memory chip HBM4. The Korean company, which plans to launch the new chip next year, did not disclose the specific shipping date for the HBM4 chips.

Samsung's HBM chips are competing with those of Korean peer SK Hynix, which is currently the main supplier of NVIDIA's HBM chips. American Micron Technology (MU.US) is also a strong competitor in this field. SK Hynix plans to start shipping the latest HBM4 chips in the fourth quarter of this year and expand sales next year.

In a joint statement regarding the AI factory, NVIDIA noted: "From our first collaboration using Samsung DRAM in NVIDIA's first graphics card NV1, to the launch of the industry's first commercial HBM chip, and now the key supply collaboration for HBM3E and HBM4, the solid alliance between our two companies over 25 years has laid the foundation for today's AI transformation."

Additionally, Samsung stated that it is optimizing chip manufacturing through NVIDIA's CUDA GPU-accelerated optical proximity correction platform, achieving a 20-fold performance improvement in computational lithography and technology computer-aided design simulation.

Samsung is also utilizing NVIDIA's Omniverse platform to build digital twin systems for global wafer fabs, shortening the cycle from design to operation, and achieving AI-driven predictive maintenance, operational optimization, and real-time decision-making. A digital twin is a technology that virtually simulates physical objects, processes, or systems using real-time data from sensors.

Samsung emphasized that it is leveraging NVIDIA GPUs, CUDA-X libraries, and solutions from Synopsys (SNPS.US), Cadence, and Siemens to achieve significant performance leaps in simulation, verification, and manufacturing analysis.

SK Group

NVIDIA stated that it is jointly building an AI factory with SK Group to promote semiconductor R&D manufacturing, cloud infrastructure development, and support for digital twins and AI agents. The AI factory being developed by SK Group will be equipped with over 50,000 NVIDIA GPUs, with the first phase of the project scheduled for completion by the end of 2027. The new factory will support SK Hynix, SK Telecom, and other group subsidiaries and external organizations through a GPU-as-a-service model.

The two parties are also collaborating to develop SK Hynix HBM and next-generation advanced storage solutions for NVIDIA GPUs, focusing on semiconductor manufacturing and communication infrastructure. In addition, NVIDIA's cloud service provider SK Telecom plans to deploy an industrial AI cloud platform equipped with NVIDIA RTX PRO 6000 Blackwell server GPUs in Asia to help startups and government agencies accelerate digital twin and robotics technology innovation The first phase will include over 2,000 NVIDIA RTX PRO 6000 Blackwell GPUs, running NVIDIA Omniverse workloads to support SK Hynix semiconductor manufacturing, wafer fab digital twins, and internal AI agent applications. SK Telecom is developing a foundational model named A.X., which is built on NVIDIA's NIM microservices and AI Enterprise software platform, empowering SK Hynix's AI agent system to help over 40,000 employees and production personnel collaborate and accelerate problem-solving, enhancing chip R&D and manufacturing efficiency.

Hyundai Motor

NVIDIA and Hyundai Motor Group will collaborate on the new generation Blackwell architecture AI factory to promote innovations in autonomous vehicles, smart factories, and robotics technology. The two parties will deploy 50,000 NVIDIA Blackwell GPUs to jointly develop AI capabilities for mobility solutions, build next-generation smart factories, and advance semiconductor technology breakthroughs at the device level. Achieving integrated AI model training, validation, and deployment through GPU clusters is the core goal of both parties.

In response to the South Korean government's initiative to build a national physical AI cluster, Hyundai Motor and NVIDIA will work with relevant government agencies to accelerate ecosystem development, with a total investment of approximately $3 billion to promote the development of physical AI in South Korea. The South Korean government has announced extensive cooperation with NVIDIA to procure over 260,000 state-of-the-art AI GPUs, covering deployments in both the public and private sectors, including collaborative projects with Samsung, SK Group, and Hyundai Motor.

NVIDIA and Hyundai Motor will jointly establish the NVIDIA AI Technology Center, Hyundai Motor Group Physical AI Application Center, and regional data centers. According to the cooperation plan announced earlier this year, Hyundai Motor will use the NVIDIA Omniverse Enterprise platform to create digital twins of factories. Currently, Hyundai Motor is exploring the development of digital twins of automotive factories and robotic systems using the Omniverse and Cosmos platforms on NVIDIA RTX PRO servers equipped with RTX PRO 6000 Blackwell server GPUs.

Additionally, Hyundai Motor is developing advanced AI models based on NVIDIA Nemotron open-source AI inference models and the NeMo software platform, aiming to achieve cloud-based online upgrade capabilities for vehicle functions and features