Korea Quantum Computing (KQC), a company specializing in quantum computing R&D, quantum security, and AI infrastructure operations, announced today that it has deployed a high-performance AI GPU Farm powered by NVIDIA H200 GPUs at the Digital Edge Bupyeong Data Center. The company will officially launch its cloud-based high-performance computing service (GPUaaS) on July 4.

In response to surging demand driven by large language models (LLMs), AI copilots, and multimodal AI applications—along with the need for larger memory and greater computational power—KQC enables enterprises and institutions to flexibly access high-performance computing resources via the cloud. Customers can achieve performance and stability comparable to direct GPU usage through dedicated bare-metal environments with full operating system control.

The AI GPU Farm is built on NVIDIA’s latest Hopper architecture and configured as a parallelized cluster optimized for compute-intensive workloads such as LLMs (including GPT-4, Claude, and Gemini), generative AI, complex simulations, and large-scale data analytics. Compared to the H100, the H200 GPU delivers significantly enhanced memory capacity and bandwidth, while its integrated Transformer Engine maximizes efficiency for LLM training and inference.

“KQC’s AI GPU Farm is expected to attract strong demand from big tech companies, AI startups, industries requiring complex simulations and data analytics, national research institutes, and HPC centers,” said John J.Y. Kim, CEO of KQC. “By offering stable, cost-effective AI and HPC infrastructure, we plan to introduce more advanced services in the future, including integration with quantum computing infrastructure.”

KQC’s GPUaaS accelerates large-scale data processing and AI model training while enabling efficient, concurrent GPU utilization through virtualization and resource-sharing capabilities. The service includes the NVIDIA AI Enterprise software stack, supporting seamless development and operations in cutting-edge AI environments. In addition, customers can elastically scale compute, storage, and network resources with a pay-as-you-go pricing model, reducing upfront investment. Built-in predictive analytics, infrastructure monitoring, automated deployment, and update features further enhance operational efficiency.

Changhoe Kim, Executive Director at KQC, who led the service launch, added, “This service offers strong competitiveness in both stability and cost. We will continue to roll out diverse systems aligned with evolving market needs.”

To support this launch, KQC signed a strategic MOU with the ITCEN Group in June to collaborate on the GPUaaS business.