LG Secures NVIDIA AI Factory Certification for 600kW CDU
Demand for AI infrastructure is increasing scrutiny of the cooling technologies that keep high-density compute environments operating reliably.
Against that backdrop, LG Electronics has secured official NVIDIA certification for its 600kW Coolant Distribution Unit (CDU), a milestone that strengthens the company's position in the liquid cooling market for AI data centres.
The certification validates LG's thermal management technology against NVIDIA's AI Factory infrastructure standards and makes LG an officially verified partner for AI Factory deployments. The validation is intended to simplify supplier qualification for large-scale AI infrastructure while reducing procurement risk.
“As AI data centres continue to scale, the need for highly reliable liquid cooling solutions has never been greater,” says James Lee, President of the LG ES Company.
“Leveraging our advanced cooling technologies together with the LG Group's integrated infrastructure capabilities, we are expanding our AI infrastructure business with end-to-end solutions that help enable physical AI, where AI creates value in real-world environments."
Validation for AI infrastructure
LG says its 600kW CDU passed more than 100 technical evaluation criteria as part of NVIDIA's verification process.
Those tests are designed to assess cooling performance, reliability and failover capabilities required in high-density AI environments.
The certification comes as operators increasingly adopt liquid cooling to support AI clusters built around power-intensive GPUs.
Traditional air cooling alone is becoming less practical as rack densities continue to increase, leading many facilities to deploy direct liquid cooling technologies alongside existing air-based systems.
LG's CDU is designed around a direct-to-chip liquid cooling approach, delivering coolant directly to components including GPUs and CPUs.
The company says this creates a hybrid cooling architecture that works alongside conventional air cooling rather than replacing it entirely.
The system is engineered to maintain temperature control within ±0.25 degrees Celsius while incorporating virtual sensor technology for continuous monitoring and leak detection.
LG also integrates its Data Center Cooling Control Manager platform with the CDU, providing predictive control, centralised monitoring and compatibility with wider data centre management systems through standard communication protocols.
Building a chip-to-chiller portfolio
The certification forms part of LG's wider strategy to provide cooling infrastructure across the entire thermal chain inside AI data centres.
Rather than focusing on a single component, the company offers what it describes as a "chip-to-chiller" portfolio, covering chillers that generate chilled water, coolant distribution units that circulate coolant throughout the facility and cold plates that transfer heat directly away from AI processors.
To validate the complete cooling chain, LG tests its systems at a dedicated Chip-to-Chiller validation facility in Pyeongtaek.
The site evaluates transient system responses, temperature variation and stability under changing workloads, from the AI chip through the cold plate, CDU and chiller.
The company says the testing programme also ensures compliance with NVIDIA's failover requirements, allowing systems to continue operating under demanding conditions.
Looking ahead, LG plans to expand NVIDIA certification beyond its 600kW CDU by seeking validation for higher-capacity models, including 1MW, 2.5MW and 4MW systems.
As AI clusters continue to increase in scale, larger coolant distribution units are becoming an important part of supporting facilities that require megawatt-level cooling capacity.
Supporting hyperscale deployments
Alongside certification, LG is also expanding its presence in hyperscale data centre projects.
The company says it has recently secured cooling contracts covering hyperscale facilities in North America as well as the Sinar Mas data centre, SM+ Data Center's AI-ready campus in Jakarta, Indonesia.
LG attributes its position in the sector to its experience in heating, ventilation and air conditioning technologies, together with in-house production of components including inverter systems and magnetic bearing compressors used across its chiller portfolio.
The company views the NVIDIA certification as another step in strengthening relationships with hyperscale operators that are standardising infrastructure for AI workloads.
