Samsung SDS to Supply Blackwell GPUs for National AI Research

Launch of National AI Research GPU Computing Service
Project Worth 15.3 Billion Won; Deployment of Latest NVIDIA GPUs

Samsung SDS is providing graphics processing unit (GPU) resources to support artificial intelligence (AI) research in the South Korean academic community.

Samsung SDS announced on August 11 that it has been selected as a GPU resource provider for the 2026 AI Research Computing Support Project, sponsored by the Ministry of Science and ICT and the Institute for Information & Communications Technology Planning & Evaluation (IITP), and began providing its services starting July 20.

Samsung SDS logo. Samsung SDS

This project is a nationwide AI research support initiative that offers GPU resources for AI research to academic and research institute communities in Korea. It supplies computational resources so that researchers at universities and government-funded institutes, who have difficulty securing their own large-scale computing infrastructure, can pursue studies on super-scale AI models and large language models (LLMs).

The project is valued at approximately 15.3 billion won, with services available through March 2027.

The evaluation criteria for selecting resource providers included GPU supply plans and the scale of available resources, the quality of the research environment, operational support systems, and security capabilities. Among multiple candidates, Samsung SDS was selected as the top provider.

For this project, Samsung SDS is offering not only NVIDIA H100 (Hopper) GPUs, but also the latest B300 (Blackwell) GPU clusters. The B300 is a state-of-the-art GPU that exceeds the H100-level specifications outlined in the project's Request for Proposal (RFP). It delivers enhanced performance and higher memory capacity compared to the previous generation, particularly in the training and inference of large-parameter models.

Samsung SDS has deployed the latest GPUs directly in the National AI Research GPU Computing project, enabling domestic researchers to work in high-level computing environments. According to Samsung SDS, the major challenge in Korean research fields has been bridging the gap between actual available resources and the level of global infrastructure due to fierce competition to secure cutting-edge GPUs.

In addition, Samsung SDS provides large-scale cluster configurations by pooling multiple GPUs into a single resource platform and offers high-speed interconnect networks. This enables researchers to flexibly address both large-scale, long-term projects and short-term experimental studies.

Beyond simple GPU supply, Samsung SDS pre-configures and provides central processing units (CPUs), memory, storage, network, and development environments. This allows researchers to focus fully on their studies by relieving the burden of setting up infrastructure.

Earlier, in March, Samsung SDS became the first in Korea to launch GPU-as-a-Service (GPUaaS) based on NVIDIA's B300 GPU through the Samsung Cloud Platform (SCP). In July, it further expanded its research infrastructure by introducing NPU-as-a-Service (NPUaaS) based on FuriosaAI's second-generation NPU "Renegade (RNGD)", a domestically developed neural processing unit specialized for AI inference, also through SCP.

Hojoon Lee, Senior Vice President and Head of Cloud Services at Samsung SDS, stated, "Through this project, we will provide stable services so that Korean researchers can achieve innovative results on world-class AI infrastructure," adding, "We will continue to expand the foundation of national AI research infrastructure and contribute to strengthening Korea's competitiveness in AI."

This content was produced with the assistance of AI translation services.