VAST Data says SK Telecom is using its storage for the GPU-as-a-Service (GPUaaS) element of its sovereign AI infrastructure offering.
SK Telecom (SKT) has launched a sovereign AI Infrastructure, providing GPUaaS based on more than 1,000 Nvidia Blackwell GPUs integrated into a single cluster named “Haein” (해인, 海印). This is the latest part of an ongoing overall AI infrastructure project to become a global AI company. SK Telecom, along with partners, has been developing a full-stack AI based on its own technologies spanning semiconductors, foundation models, data, and services, and open-sourcing the developed models to various companies within the domestic, South Korean, AI ecosystem. Here’s a diagram:
There are four AI DC business models: (1) Hyperscale, (2) Dedicated single customers, (3) Small and modular, and (4) GPUaaS for subscription-based AI cloud. The Haein Cluster has secured a role in South Korea’s Ministry of Science and ICT’s program for enhancing the foundation of AI computing resource utilization as part of the Proprietary AI Foundation Model project. SKT has prepared for this sovereign AI infrastructure launch with strategic global partnerships, which include Penguin Solutions for integrated AI DataCenter (AIDC) systems and Supermicro for globally sourced AI server procurement.
Kim Myoung Gook, head of the GPUaaS Business Office at SK Telecom, stated: “Haein Cluster’s GPUaaS will serve as a catalyst for enhancing both customer and national AI competitiveness. As an AI infrastructure provider, we are committed to building the nation’s AI superhighway.”
According to SKT, Petasus provides an enterprise-ready, high-performance, low-power cloud management platform for AI and general-purpose workloads. It is a sovereign GPU service within South Korea’s territory and features self-service resource management, multi-CPU architecture support, and high-performance networking with a distributed firewall. SKT developed its Petasus AI Cloud from an earlier 5G Multi-Access Edge Computing (MEC) platform.
STK carried over MEC requirements such as automation, orchestration, and low-latency networking into the design of the GPUaaS service. SKT can instantly partition and reconfigure the GPU cluster as customer needs change. Petasus has an “AI Cloud Manager” – an AIOps platform (AI for IT Operations) that manages the entire AI service lifecycle from development and training to deployment.
VAST has issued an announcement with accompanying blog, podcast, and forthcoming webinar, all making a big deal out of the fact that SKT is using its storage for Petasus. The company says SKT is deploying VAST’s AI Operating System on Nvidia Blackwell GPUs to accelerate virtualization, simplify AI data pipelines, and support national-scale AI training and inference.
DK Lee, Vice President and Head of AI DC Lab at SKT, said in a canned statement: “VAST Data’s unified architecture has been instrumental in helping us move from legacy bare-metal deployments to a fully virtualized, production-grade AI cloud. The VAST AI OS powers the performance, simplicity, and flexibility needed to support the next generation of sovereign AI workloads, and gives us the confidence to scale fast and securely. With VAST, we’re enabling a GPUaaS platform that meets the exacting needs of government, research, and enterprise AI customers in South Korea.”
