LG Electronics announced that its 600kW Coolant Distribution Unit has secured official NVIDIA certification after meeting more than 100 technical evaluation criteria for AI Factory standardization. The approval qualifies LG as a verified infrastructure partner for NVIDIA AI Factory deployments and is intended to simplify supplier qualification for hyperscalers and data center operators.
The CDU uses Direct-to-Chip liquid cooling to deliver coolant to heat-generating components, including GPUs and CPUs. It operates as part of a hybrid cooling configuration alongside existing air-cooling infrastructure and is designed for the thermal loads produced by high-density AI servers. Temperature control precision is rated at ±0.25 degrees Celsius, although performance may vary with operating conditions and server loads.
Virtual sensor technology provides real-time monitoring and leak detection. The unit can also be connected to LG’s Data Center Cooling Control Manager for predictive control and centralized monitoring, with standard communication protocols supporting integration with data center management systems. NVIDIA’s evaluation covered cooling performance, reliability and failover capabilities.
System-level testing takes place at LG’s Chip-to-Chiller validation site in Pyeongtaek, South Korea. The facility is used to evaluate transient responses, temperature changes and low-load stability across the cooling chain, from AI chip loads and cold plates to the CDU and chiller. The 600kW unit was also tested against NVIDIA’s failover requirements for maintaining operational stability under demanding conditions.
The certified CDU forms part of LG’s Chip-to-Chiller portfolio, which includes chillers, CDUs and cold plates for AI data center cooling. LG intends to seek NVIDIA certification for additional CDU capacities of 1MW, 2.5MW and 4MW. The manufacturer has also secured cooling projects for hyperscale facilities in North America and the SM+ Data Center, an AI-ready facility in Jakarta, Indonesia.
