Add preferred source
South Korean cloud infrastructure company Innogrid has built an AI analytics environment for the Korea SMEs and Startups Agency, integrating the agency’s existing virtualization infrastructure and GPUs into a unified management framework. Based on its proprietary private cloud platform Openstackit and cloud management platform TabCloudit, the company converted the SME Big Data Platform (SIMS) analytics environment from a closed on-premises structure to a cloud-based system, enabling GPUs to be utilized as virtualized resources. GPUs are allocated at the virtual machine (VM) level, with real-time monitoring and dynamic scaling capabilities. Innogrid views this project as the starting point of a strategy linking foreign virtualization replacement and AI infrastructure expansion into a single trajectory. Under its “From xPU to AI Platform” vision, the company plans to expand its business toward integrating and operating diverse processor infrastructure under a unified system.
Key Elements
South Korean cloud infrastructure company Innogrid has built an AI analytics environment for the Korea SMEs and Startups Agency, integrating the agency’s existing virtualization environment and graphics processing units (GPUs) into a unified management framework. The strategy goes beyond simple GPU expansion, implementing a structure that extends existing IT infrastructure to AI workloads, directly targeting AI infrastructure transition demand in the corporate and public sectors.
Innogrid announced on the 1st that it has completed the “GPU-based AI Big Data Analytics Infrastructure” project for the Korea SMEs and Startups Agency, built on its proprietary private cloud platform “Openstackit” and cloud management platform “TabCloudit.”
The core of this project was converting the analytics environment of the SME Big Data Platform (SIMS), operated by the Korea SMEs and Startups Agency, from a closed on-premises, physical PC-centric structure to a cloud-based resource system. Notably, the project satisfied the requirement for adopting South Korean-made virtualization software certified by the National Intelligence Service’s security functionality verification, while also establishing the high-performance GPU computing environment needed for developing advanced analytics models.
Converting GPUs into Virtualized Resources
The most noteworthy aspect of this deployment is the design enabling GPUs to be utilized as virtualized resources. AI big data analytics can be performed on virtual machines (VMs) equipped with GPUs, and development VMs combining multiple GPUs can be leveraged for tasks requiring high-performance computing. This structure is designed to handle computationally intensive research such as policy impact analysis and forecasting simulations.
GPU resources are allocated at the VM level. The configuration allows flexible resource distribution based on the nature and scale of analytics tasks, and a monitoring system has been applied that enables real-time visibility into resource status at both the host and virtual server levels, with the ability to predict expansion timing based on usage trends. Snapshot, replication, and dynamic scaling capabilities were also added to enhance resource utilization flexibility.
From Foreign Virtualization Replacement to AI Infrastructure
Innogrid plans to use this deployment structure as a foundation for extending existing virtualization environments into AI infrastructure. The vision is to support customers seeking to migrate away from foreign virtualization environments such as VMware by moving their existing VM workloads to a private cloud, then operating multi-cloud, hybrid cloud, and GPU-based AI workloads all within a single framework.
This strategy is also driven by a shift in the AI infrastructure market, where attention is moving from the race to secure GPUs toward “how efficiently secured resources are utilized.” Rather than operating business virtualization environments and AI computing environments separately, the approach reduces the disconnect between existing IT infrastructure and AI infrastructure by allocating and controlling resources within a single management framework.
This approach connects to Innogrid’s “From xPU to AI Platform” technology vision. The core concept is linking and controlling diverse xPU infrastructure—not just GPUs but also neural processing units (NPUs), central processing units (CPUs), and quantum processing units (QPUs)—through a single control plane spanning AI development, training, deployment, and operations environments.
Kim Myung-jin, CEO of Innogrid, said, “Virtualization transition and GPU adoption must be addressed within a single framework, not as separate tasks, to efficiently scale AI infrastructure. This project is an example of implementing a structure that operates virtualization infrastructure and GPU resources within one system using a South Korean cloud platform.”
He added, “Virtualization transition is evolving beyond simply replacing foreign products toward leveraging GPU resources for AI. We will advance our technology framework that connects everything from existing infrastructure modernization to xPU resources and AI development and operations environments, enabling customers to expand their AI infrastructure on top of their existing foundation.”
Market Impact and Expansion Strategy
This project is significant as a case where a South Korean public and policy research institution adopted a domestic cloud platform in the process of introducing AI analytics infrastructure. The fact that an integrated operating structure combining South Korean-made virtualization software and GPU resources has been actually implemented in the public sector, where security requirements are stringent, can serve as a reference model for other institutions with similar needs.
Building on this, Innogrid plans to advance an expansion structure that progresses from existing virtualization infrastructure modernization to multi-cloud, hybrid cloud, and GPU-based AI workloads. The goal is to support customers in incrementally expanding AI infrastructure at the timing and scale they need, without requiring a complete replacement of existing IT infrastructure.
Particularly amid rising license cost burdens and supply chain uncertainties surrounding foreign virtualization solutions, demand for transitioning to South Korean cloud platforms is expected to continue growing. Innogrid is expected to use this deployment case as a springboard to fully pursue a market strategy that ties virtualization transition and AI infrastructure adoption into a single continuous trajectory.
Once added, BigGo Finance appears first in Google Search Top Stories, so you get the broadest, most up-to-the-minute, and most comprehensive global financial news first.
