Samsung SDS has been selected as the top-priority GPU resource supplier 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). The company began providing the service on the 20th of last month, with a project scale of approximately 15.3 billion KRW and a supply period lasting until March 2027.

Bridging the Gap in National AI Research Infrastructure

This national project supplies GPU resources to researchers at domestic universities and government-funded research institutes who face difficulties in securing large-scale computing infrastructure, thereby supporting research into hyper-scale AI models and Large Language Models (LLMs). The core objective is to help researchers conduct their studies using globally competitive infrastructure. Historically, the fierce competition to acquire the latest GPUs in the domestic research field has led to a gap between actually available resources and global infrastructure standards, which has been a persistent challenge.

Ranked No. 1 in Evaluation... Deploying B300 GPUs Exceeding H100-Level Specs

During the supplier selection process, key evaluation criteria included GPU supply plans, resource scale, research environment quality, operational support systems, and security capabilities. Among multiple bidders, Samsung SDS was chosen as the top priority. Samsung SDS will provide Nvidia's latest B300 (Blackwell)-based GPU clusters alongside the H100 (Hopper) GPUs. The B300 is the newest GPU that exceeds the H100-level or higher specifications outlined in the Request for Proposal (RFP), and it boasts superior performance and large memory capacity compared to previous generations when training and inferencing large-parameter models. The company explained that by directly deploying the latest B300 GPUs into the national AI research project, it has laid the foundation for domestic researchers to conduct studies in a top-tier computing environment.

Fostering Research Focus through Large-Scale Clustering and Pre-configured Environments

Samsung SDS provides large-scale clustering, which groups multiple GPUs into a single resource pool, along with high-speed interconnect networks. This enables researchers to flexibly handle everything from long-term, large-scale projects to short-term experimental tasks. Additionally, by offering pre-configured CPU, memory, storage, network, and development environments, the company has reduced the burden of infrastructure setup, allowing researchers to focus entirely on their work.

Continuous Expansion of Cloud-Based AI Infrastructure

In March, Samsung SDS became the first in South Korea to introduce a GPU-as-a-Service (GPUaaS) utilizing Nvidia's B300 GPUs on its Samsung Cloud Platform (SCP). In July, the company further expanded its AI research infrastructure domain by launching an NPU-as-a-Service (NPUaaS) based on the second-generation 'Renegade (RNGD)' by FuriosaAI, a domestic Neural Processing Unit (NPU).

Lee Ho-jun, Executive Vice President and Head of the Cloud Service Business Division at Samsung SDS, stated, "Through this project, we will provide stable services so that domestic researchers can achieve innovative outcomes based on the world's best AI infrastructure." He added, "Moving forward, we will continue to contribute to expanding the foundation of the national AI research infrastructure and enhancing South Korea's AI competitiveness."