As artificial intelligence (AI) workloads continue to grow, data centers are facing a new challenge: keeping increasingly powerful AI processors cool without sacrificing performance or reliability. To address this demand, LG Electronics has achieved NVIDIA validation for its 600-kilowatt (kW) Coolant Distribution Unit (CDU), a milestone that supports the company's expansion into the rapidly growing AI infrastructure market.
The validation confirms that LG's liquid cooling solution meets NVIDIA's technical requirements for AI Factory infrastructure, giving hyperscale data center operators greater confidence when selecting cooling systems for high-density AI environments.
What did LG achieve?
LG Electronics announced that its 600kW Coolant Distribution Unit (CDU) successfully passed NVIDIA's technical validation process after meeting more than 100 evaluation criteria covering cooling performance, system reliability, and failover capabilities.
The validation positions LG among a select group of technology providers whose cooling solutions have been verified for compatibility with NVIDIA's AI infrastructure standards.
For enterprise customers and hyperscale data center operators, this designation can help simplify supplier qualification and reduce deployment risks for large-scale AI facilities.
What is a Coolant Distribution Unit (CDU)?
A Coolant Distribution Unit (CDU) is a critical component in liquid-cooled data centers.
A CDU circulates coolant through servers to remove heat generated by high-performance processors such as graphics processing units (GPUs) and central processing units (CPUs). By maintaining stable operating temperatures, it helps AI systems run efficiently while reducing the risk of overheating.
As AI computing becomes more demanding, liquid cooling is increasingly being adopted because traditional air-cooling systems alone may no longer provide sufficient thermal management for dense AI server deployments.
How does LG's cooling system work?
LG's validated CDU uses a Direct-to-Chip (DTC) liquid cooling approach, which delivers coolant directly to heat-producing components inside AI servers.
Unlike conventional cooling methods that rely primarily on circulating cold air throughout a facility, Direct-to-Chip cooling removes heat at its source before it spreads through the system.
LG describes the solution as a hybrid cooling system, combining liquid cooling with existing air-cooling infrastructure to improve overall thermal efficiency.
According to the company, the system offers:
- Temperature control with ±0.25°C precision
- Real-time monitoring through virtual sensor technology
- Leak detection capabilities
- Predictive control using LG's Data Center Cooling Control Manager (DCCM)
- Integration with existing data center management systems through standard communication protocols
These capabilities are designed to help operators monitor cooling performance while supporting continuous AI server operations.
Why liquid cooling is becoming more important
Modern AI models require significantly greater computing power than traditional enterprise applications.
Training and operating large AI models often involve clusters of specialized GPUs that consume substantial amounts of electricity and generate large amounts of heat.
Liquid cooling is increasingly viewed as an effective solution because it can:
- Remove heat more efficiently than air alone
- Support higher server densities
- Improve energy efficiency
- Help maintain stable hardware performance
- Reduce the risk of thermal-related system interruptions
As organizations invest in AI infrastructure, efficient thermal management has become an important factor in maintaining system reliability and operational costs.
Supporting AI infrastructure from chip to chiller
LG says it is pursuing a "Chip-to-Chiller" strategy that covers multiple components of AI cooling infrastructure rather than focusing on a single product.
Its portfolio includes:
- Chillers - Produce chilled water for cooling systems
- Coolant Distribution Units (CDUs) - Circulate coolant throughout the cooling network
- Cold plates - Transfer heat directly away from AI processors
The company also operates a dedicated validation facility in Pyeongtaek, South Korea, where complete cooling systems are tested under varying operating conditions to evaluate temperature stability, transient responses, and failover performance.
Expanding beyond the 600kW system
LG plans to pursue NVIDIA validation for additional high-capacity CDUs, including:
- 1MW CDU
- 2.5MW CDU
- 4MW CDU
These larger systems are intended to support the increasing scale of next-generation AI data centers.
The company also noted that it has secured cooling projects for hyperscale facilities in North America and the SM+ Data Center in Jakarta, Indonesia, reflecting growing demand for AI-ready cooling infrastructure.
As AI adoption accelerates, data center infrastructure has become a strategic area of investment for technology companies.
While much public attention focuses on AI software and chips, supporting infrastructure—including networking, power management, and cooling—is equally important for ensuring reliable AI operations.
Industry analysts increasingly view liquid cooling as a key technology for future AI facilities because higher-performance processors require more advanced thermal management than conventional enterprise servers.
For LG, NVIDIA's validation strengthens its position in this evolving market by demonstrating compatibility with widely used AI computing platforms.
LG Electronics' NVIDIA validation for its 600kW Coolant Distribution Unit marks an important step in the company's expansion into AI infrastructure. By meeting NVIDIA's technical standards for AI Factory cooling, LG reinforces its strategy of providing end-to-end thermal management solutions for high-density AI data centers.
As enterprises and cloud providers continue investing in AI computing, technologies that improve cooling efficiency, reliability, and scalability are expected to play an increasingly important role in supporting the next generation of data center operations.


.jpg)







