LG Uplus said on the 24th that it will unveil a top-tier global AI data center (AIDC) strategy at MWC26, the world's largest mobile communications exhibition, based on synergies with LG Group. Centered on "ONE LG," which brings together capabilities across LG affiliates, the company will showcase a "Beyond AI-Ready" AI data center that covers power, cooling, and operations end to end, aiming to strengthen its position as a key infrastructure provider in the AI era.
At this year's MWC26, the latest technologies and next-generation operations strategy to be applied to the "Paju AI Data Center," now under construction as the largest in the Seoul metropolitan area, will be unveiled. Targeting completion in 2027, the Paju AI Data Center will mobilize groupwide technological strengths from LG Uplus, LG Electronics, and LG Energy Solution for its core components—operations, cooling, and power systems.
First, the heat generation challenge—the key bottleneck for AI data centers—will be addressed in collaboration with LG Electronics. The heat generated by high-performance graphics processing units (GPUs) is difficult to handle with conventional air cooling alone, requiring new solutions. In response, LG Uplus and LG Electronics are introducing a liquid cooling solution that directly removes heat (D2C, Direct to Chip) by attaching a dedicated metal cold plate to the GPU chip and circulating liquid through a coolant distribution unit (CDU). LG Uplus said its own tests showed this method improved energy efficiency by about 24% compared with traditional air cooling. The cooling water required for liquid cooling is produced by LG Electronics' air-cooled "free cooling chiller." Because it uses cold outside air to produce chilled water, the company said it can drastically cut power consumption to about 10% of conventional levels.
LG Energy Solution's technology will be applied to ensure stable power supply. The Paju AI Digital Center will be equipped with high-performance UPS batteries that immediately compensate for power during outages or voltage fluctuations, and it will support uninterrupted operations by minimizing fire and thermal runaway risks through a multi-layered safety design engineered in-house from battery cell to pack.
LG Uplus plans to gradually expand AI computing infrastructure with the Paju AI Digital Center as a starting point, and to that end, it is strengthening an integrated consulting business that designs, builds, and operates customers' data centers end to end by developing partnership models with asset managers and others. To efficiently handle the greater heat that will come as GPU performance improves, the company is also accelerating development and deployment of "immersion cooling" technology, which submerges servers directly in a nonconductive liquid.
On this foundation, LG Uplus will apply its integrated operational capabilities—combining technologies from group affiliates—across the data center. In particular, the company will directly handle not only design and construction but also operations. In this process, LG Uplus is applying its in-house–developed AI-based DCIM (data center infrastructure management) across all areas of the data center. DCIM serves as the "brain," monitoring in real time power usage, temperature and humidity, cooling conditions, and equipment anomalies, and optimally allocating resources through AI analysis. For example, AI can analyze power usage trends to predict capacity shortfalls in advance and detect overheating signs based on temperature data, immediately notifying administrators to prevent incidents. LG Uplus plans to raise data center infrastructure availability to 99.99% through this intelligent operating framework.
Also at MWC26, LG Uplus will showcase the "Sovereign AI Appliance," developed in collaboration with LG AI Research and FuriosaAI. It is a solution that enables immediate AI use for work by simply connecting power and network, without complex infrastructure buildouts.
An Hyeong-gyun, head of the AI Business Group (executive director) at LG Uplus, said, "We will make the Paju AI Data Center a top-tier global hub," and added, "With a strategy premised on AI running continuously in real-world environments, we will leap forward as an AI factory that operates and scales GPUs with the highest stability."