Samsung SDS is accelerating AI transformation across businesses and public institutions by deploying AI agents and Generative AI with a full-stack approach that covers models, platforms, infrastructure, and solutions.

The B300 GPU delivers 12-layer HBM3E memory, 288 GB per GPU and 8 TB/s bandwidth, offering significantly higher memory and bandwidth than the H100 to boost AI inference performance.

Samsung SDS rolled out SCP B300 GPUaaS on the Samsung Cloud Platform on March 23, becoming the first in South Korea to provide Nvidia B300 Blackwell Ultra-based GPU services.

The service targets data bottlenecks in large language model inference by increasing memory capacity and reducing latency for demanding AI tasks, including agents and content generation.

Customers can use SCP B300 GPUaaS to run large-scale AI models with high-capacity memory and low latency, benefiting enterprises, mid-sized firms, and public institutions.

Samsung SDS expanded into the enterprise Generative AI market through a reseller and service partnership with OpenAI, delivering high-performance enterprise AI with easy integration via APIs.

At CES 2026, AI agents were showcased as a way to reduce government workloads by enabling scheduled briefings, multilingual video conferencing, on-the-go voice tasks, and automated image analysis, promising about a 67% reduction in working hours.

Samsung SDS has a track record of GPUaaS development, offering A100-based services in 2021 and H100-based services in 2023, establishing GPUs as core AI infrastructure across cloud environments.

Brity Works, the company’s Generative AI collaboration tool, facilitates practical AI model applications within real work environments, complemented by Brity Meeting with AI-enabled video conferencing.

Lee Ho-jun, EVP and head of cloud service business, stated Korea’s first B300 GPU service will improve resource efficiency and energy savings while advancing AI transformation.

A core strategy is a full-stack AI approach—encompassing infrastructure, platforms, and solutions—with cloud services via SCP and partnerships with AWS, Azure, GCP, OCI, plus Nvidia for B300 GPUs.

Public-sector expansion includes private cloud operations for the Daegu Center PPP, pan-government AI projects, an intelligent business management platform, and collaboration with the Gyeonggi Education Office on a digital platform.