VAST Data has formed a strategic alliance with SK Telecom (SKT) to support the development of South Korea’s largest sovereign AI infrastructure, powered by NVIDIA’s next-generation Blackwell supercomputing platform. As SKT advances its vision of becoming an AI-first company, VAST Data will play a key role by streamlining data pipelines, boosting performance, and ensuring efficient large-scale training and deployment of cutting-edge AI models through its AI Operating System.
The initiative is about the Haein Cluster, a next-generation AI computing system designed to support the nation’s growing demand for AI research and innovation. By virtualizing GPU resources and offering a GPU-as-a-Service (GPUaaS) model, the cluster will allow developers and organizations to access powerful computing capabilities securely and efficiently within Korea’s borders. Selected under the Ministry of Science and ICT’s national AI program, the Haein Cluster is set to become a key driver of the country’s AI foundation models and digital sovereignty efforts.
Unlike traditional bare-metal systems, which can take days or weeks to allocate new workloads, SK Telecom’s approach dramatically shortens this process. With the integration of VAST Data’s technology into the Petasus AI Cloud, GPU environments can now be provisioned in under 10 minutes, while retaining near-bare-metal performance. This improvement significantly reduces bottlenecks for developers and enterprises, allowing them to run experiments, train models, and deploy applications much faster.
The Petasus AI Cloud architecture is built on Supermicro’s NVIDIA HGX server platform, combined with VAST’s disaggregated, shared-everything (DASE) storage system. The infrastructure is designed to meet the needs of government agencies, research institutions, and commercial enterprises, offering flexibility while maintaining strict sovereignty controls.
A key strength of this platform is its ability to isolate workloads while ensuring data privacy and security. Its unified AI pipeline streamlines the entire lifecycle — from training to inference — and supports dynamic resource allocation. This means GPU and storage resources, along with networking capacity, can be assigned on demand to match each tenant’s requirements. Combined with carrier-grade reliability, the system promises enterprise-level uptime with minimal operational overhead.
Leaders from the companies involved emphasized the importance of this collaboration. DK Lee, Vice President and Head of the AI DC Lab at SK Telecom, noted that VAST Data’s unified architecture has been crucial in transitioning from legacy bare-metal environments to a production-ready, virtualized AI cloud. He highlighted that the VAST AI OS provides the performance and adaptability needed for sovereign AI workloads while giving SKT confidence to scale rapidly and securely.
Cenly Chen, Chief Growth Officer at Supermicro, added that Supermicro is proud to support SK Telecom’s national AI vision by delivering server architecture built for NVIDIA’s latest platforms. According to Chen, the joint work with VAST Data and SK Telecom demonstrates how leading hardware and software providers can come together to create cutting-edge sovereign AI infrastructure.
Sunil Chavan, Vice President for APAC at VAST Data, stated that the partnership represents a blueprint for how nations can establish their own sovereign AI systems. He explained that by removing traditional hurdles such as data movement inefficiencies, long provisioning times, and security risks, SKT and VAST are enabling an infrastructure that balances speed, flexibility, and compliance.
Image credits: VAST Data
