
VAST Data has joined forces with SK Telecom to support South Korea’s largest sovereign AI infrastructure. Built on NVIDIA Blackwell GPUs, the system is designed for national-scale AI training and inference. The project, known as the Haein Cluster, falls under the Ministry of Science and ICT’s AI Computing Resource Utilization Enhancement Program.
Scalable sovereign AI infrastructure
The Haein Cluster integrates the VAST AI Operating System within SKT’s sovereign AI cloud. Consequently, the platform enables GPU-as-a-Service in under 10 minutes while maintaining near bare-metal performance. Moreover, it ensures data sovereignty by keeping AI workloads and information within national borders.
Advanced features for national AI growth
The infrastructure provides elastic, multi-tenant environments for research, government, and enterprise AI users. Additionally, it unifies training and inference pipelines, streamlining model development and deployment. This allows users to scale projects quickly while preserving security and high throughput.
Powering next-generation AI workloads
The Petasus AI Cloud leverages VAST’s disaggregated, shared-everything (DASE) architecture alongside NVIDIA HGX servers built by Supermicro. As a result, traditional provisioning bottlenecks are eliminated, delivering speed, resilience, and scale required for next-generation AI demands.