The 600kW unit is engineered to handle the thermal load of high-performance AI GPUs through a Direct-to-Chip liquid cooling system. By delivering coolant directly to heat-generating components, the system achieves a temperature control precision of ±0.25 degrees Celsius. The hardware also features integrated virtual sensors for real-time monitoring and leak detection, designed to operate in tandem with existing air-cooling infrastructure.

To ensure reliability, LG subjects its cooling chain to stress tests at its Pyeongtaek validation site, measuring transient responses and low-load stability. This process is part of a broader strategy to scale LG's influence in the AI sector, with plans to certify 1MW, 2.5MW, and 4MW models in the future. The company is currently leveraging its 60-year HVAC legacy to supply major data center projects across North America and Southeast Asia, including the SM+ Data Center in Jakarta. By producing core components like magnetic bearing compressors in-house, LG aims to offer a unified 'Chip-to-Chiller' portfolio that spans from the cooling plant down to individual server processors.