□ Selected as a supplier for the 2026 AI Research Computing Support Project by the Ministry of Science and ICT
- The project aims to provide researchers in academia and research institutes with GPU resources for AI research
- Samsung SDS selected as the top-ranked supplier in the bid
- Services launched in July 2026, to be provided through March 2027
□ Providing NVIDIA’s latest B300 (Blackwell Ultra)-based GPU clusters
- Providing NVIDIA B300 GPUs as well as NVIDIA H100 (Hopper) GPUs
- Established a foundation for researchers in Korea to conduct research in a world-class computing environment
□ Contributing to the expansion of AI research infrastructure
- Launched B300 GPUaaS on Samsung Cloud Platform (SCP) in March, a first in Korea
- Launched NPUaaS based on RNGD (FuriosaAI’s Korean-developed second-generation NPU) in July
Samsung SDS announced that it was selected as a GPU resource supplier for the 2026 AI Research Computing Support Project supported by the Ministry of Science and ICT and the Institute of Information & Communications Technology Planning & Evaluation (IITP). The service was launched on July 20th.
The project aims to support national AI research by providing researchers in academia and research institutes with GPU resources for AI research.
This project is designed to provide computing resources to research entities that cannot internally secure large-scale computing infrastructure -- such as universities and government-funded research institutes -- so they can conduct research on hyperscale AI models and large language models (LLMs). The project is valued at approximately KRW 15.3 billion, with services provided from July 2026 through March 2027.
During the project bidding process, suppliers were evaluated based on the following key criteria: GPU supply plans and the scale of available resources; research environment; operational support systems; and security capabilities. Samsung SDS was ranked first among the selected suppliers.
□ Providing NVIDIA’s Latest B300 (Blackwell Ultra)-Based GPU Cluster
For this project, Samsung SDS provides the latest B300 (Blackwell Ultra)-based GPU clusters, as well as H100 (Hopper) GPUs.
B300 is based on the latest architecture and exceeds the minimum requirements for Hopper-class GPUs or higher outlined in the request for proposal (RFP). B300 delivers significantly improved performance and greater memory capacity than previous generations for training and inference of large-parameter models.
Amid intense competition to secure the latest GPUs, closing the gap between the computing resources actually available in Korea and the level of global AI infrastructure has become an urgent challenge for Korea’s research community.
By deploying the latest GPUs for the national AI research GPU computing project, Samsung SDS has laid a foundation for researchers in Korea to conduct research in a world-class computing environment.
To establish an environment that can flexibly meet the needs of researchers – from long-term, large-scale projects to short-term experiments – Samsung SDS provides high-speed interconnect networks and large-scale clustering configurations that combine multiple GPUs into a single resource pool.
Additionally, moving beyond simply supplying GPUs, Samsung SDS pre-configures and provides CPUs, memory, storage, networks, and development environments. This relieves researchers of the burden of building infrastructure themselves, allowing them to focus on their research.
□ Contributing to the Expansion of AI Research Infrastructure
In March, Samsung SDS launched NVIDIA’s B300 GPU-based GPU as a Service (GPUaaS) via its Samsung Cloud Platform (SCP), a first in Korea.
In July, Samsung SDS launched an NPU-as-a-Service (NPUaaS) offering on SCP based on FuriosaAI's second-generation RNGD, a Korean-developed NPU specialized in AI inference, further expanding access to AI research infrastructure.
Ho-joon Lee, Executive Vice President and Leader of the Cloud Service Business Division at Samsung SDS, said, “Through this project, we will provide stable services so that researchers in Korea can achieve innovative outcomes based on world-class AI infrastructure. Going forward, we will contribute to expanding access to national AI research infrastructure and enhancing Korea’s AI competitiveness.”
