Elice Group Builds B300-Based Modular AI Data Center for National AI Infrastructure
Elice Group has developed a modular AI data centre with B300 GPUs, advancing South Korea’s AI computing infrastructure under a government programme.
Elice Group announced on 13 November the completion of its B300 graphics processing unit (GPU) infrastructure at its modular AI data centre, initiating beta service on 30 October as part of South Korea’s Ministry of Science and ICT’s Advanced GPU Acquisition, Deployment and Operations Support Program. This development contributes to expanding the national AI computing resources and underpins efforts to strengthen high-performance GPU utilisation.
Quick Skim
Elice Group Builds Infrastructure Following Government Programme
Since selection as a provider in June, Elice Group has been deploying the necessary infrastructure to operate B300 GPUs within its proprietary modular AI data centre, branded AI PMDC. The project is aligned with the Ministry of Science and ICT’s initiative aimed at scaling Korea’s AI capabilities through enhanced computing hardware deployment.
AI PMDC Integrates Hybrid Cooling for Energy Efficiency
The AI PMDC employs a combined cooling system integrating hot-water cooling, utilising water heated above 40 degrees Celsius, and air-side cooling, which reduces coolant temperature by leveraging outside air. This hybrid approach optimises power consumption for cooling operations.
The centre is designed to achieve a power usage effectiveness (PUE) target of 1.1 or lower, a key metric indicating energy efficiency in data centre operations.
Programme’s Role in Expanding AI Computing Capacity
The Ministry of Science and ICT’s Advanced GPU Acquisition, Deployment and Operations Support Program aims to broaden access to high-performance GPU resources critical for AI workloads. Elice Group’s involvement in this government-backed programme positions it to support Korea’s national strategy to bolster AI research and industrial applications.
Implications for Operations and Infrastructure in Korea
The deployment of modular AI data centres operating B300 GPUs can accelerate large-scale AI computation, benefiting sectors reliant on advanced machine learning and data analytics.
Energy-efficient cooling methods reduce operational costs and enhance sustainability, key considerations for industrial-scale data centre operation.
Modularity in data centre design supports scalability and flexibility, which is important for adapting to fluctuating demand in AI workloads and facilitating staged infrastructure expansion across Asia-Pacific.
Future Service Launch and Development Plans
Elice Group intends to launch full services by the end of 2023, following the current beta phase. CEO Jaewon Kim indicated ongoing commitments to advance the AI PMDC and cloud platform technologies, contributing further to strengthening South Korea’s AI infrastructure competitiveness.
Subsequent developments will likely focus on increasing capacity, enhancing operational efficiency, and integrating with broader AI ecosystems supporting industrial and research applications.


