LG Uplus has officially declared its ambition to become a major player in the AI data center market, targeting KRW 5 trillion in cumulative orders by 2030.
The most critical bottleneck in today's AI infrastructure race is securing power and location, especially in the densely populated Seoul metropolitan area. Strict regulations and grid limitations mean that securing a large-scale power supply is a significant competitive advantage. This is why LG Uplus's announcement of having secured 200MW of power in Paju and already selling out the first 50MW building is so noteworthy; it signals both secured resources and confirmed market demand.
So, what's driving this urgency? The primary cause is a fundamental shift in AI demand. The market is moving from the 'training' phase, where models are built, to the 'inference' phase, where they are actively used to generate answers and services. This transition, accelerated by next-generation chips from companies like Nvidia, requires data centers that can operate 24/7 at high power densities, creating immense heat. Traditional data centers are ill-equipped for this challenge.
This leads to LG Uplus's strategic solution, which they call 'ACE on TRUST'. First, they are tackling the heat problem with a hybrid cooling system that combines conventional air cooling with advanced liquid cooling, developed in synergy with LG Electronics. Second, to meet tight deadlines, they are using a Prefabricated Modular Data Center (PMDC) approach, which speeds up construction. Finally, the entire operation is backed by LG affiliates, from LG Energy Solution's UPS batteries to LS Electric's power systems, creating a reliable, integrated package.
This strategy is also well-timed with government support. The Korean government's 'K-Moonshot' initiative, which includes large-scale GPU infrastructure deployment, provides a stable source of public and industrial demand. In a competitive market where rivals like SKT and Naver are also racing to secure power and build capacity, LG Uplus's move to offer a comprehensive solution—power, speed, cooling, and operational stability—positions it strongly. This AIDC business is set to become a crucial growth engine, complementing its stable telecommunications core.
- AI Inference: The process of using a trained AI model to make predictions or generate outputs based on new, real-world data. It's the 'live' or 'in-use' phase of an AI application.
- Hybrid Cooling: A data center cooling strategy that uses both traditional air-based cooling and more efficient liquid-based cooling methods to manage the high heat generated by densely packed servers, especially those running AI workloads.
- PMDC (Prefabricated Modular Data Center): A construction method where data center components are manufactured and pre-assembled in a factory as modules, which are then transported to and assembled on-site. This significantly shortens construction timelines.