LG Electronics is stepping up its push into the rapidly expanding AI data center market with a new large-capacity cooling system designed to address the rising power and thermal demands of AI infrastructure.
The company said September 20 that it has launched a new air-cooled, oil-free inverter turbo chiller. The product expands LG Electronics’ data center cooling lineup beyond its existing screw chillers, targeting high-capacity applications where AI servers generate substantially more heat than conventional computing systems.
The new chiller has been developed around three key requirements for AI data centers: reducing electricity consumption and total cost of ownership, responding to tightening environmental regulations, and shortening deployment times while allowing operators to scale capacity more flexibly.
At the heart of the system is LG Electronics’ proprietary oil-free magnetic bearing technology. The technology uses electromagnetic force to suspend the compressor shaft during operation, eliminating mechanical friction and the need for lubricating oil.
By removing oil from the refrigeration circuit, the system does not require oil circulation pumps or associated piping. It also avoids the formation of an oil film inside the heat exchanger, which can reduce cooling performance over time.
LG Electronics said the new turbo chiller achieves a non-standard part-load value (NPLV) more than 20% higher than inverter screw chillers commonly used in data centers.
The oil-free design is also intended to reduce maintenance requirements. Operators can avoid routine oil replacement and filter management while potentially extending compressor life and reducing the risk of downtime.
Another feature is a refrigerant-based free-cooling system designed for cooler seasons. When outdoor temperatures fall during transitional seasons or winter, the system can stop operating its compressor and allow refrigerant condensed by the outdoor air to circulate naturally.
Because the compressor accounts for more than 80% of a chiller’s power consumption during operation, shutting it down under suitable outdoor conditions can significantly reduce electricity use.
LG Electronics said the refrigerant free-cooling approach can reduce annual power consumption by approximately 30% compared with conventional water-based free-cooling systems. The technology is also designed to improve Power Usage Effectiveness (PUE), a key efficiency metric for data center operators.
The company has also selected R-513A refrigerant for the new system. Its global warming potential is more than 50% lower than that of R-134a, providing greater flexibility as environmental regulations governing refrigerants become stricter in major markets.
Deployment speed is another area LG Electronics is targeting.
The new chiller incorporates a “Hydronic Modular” system that integrates pumps, valves, emergency buffer tanks and other components inside an ISO-standard container before delivery to the data center site.
In conventional installations, numerous pumps and pipes connecting the chiller to server rooms must be assembled and welded on site, adding time and labor costs to construction.
With the modular approach, the major hydronic components are preassembled and delivered as a finished unit. At the site, operators can connect the module to the chiller rather than installing the individual components separately.
LG Electronics expects the approach to shorten construction and commissioning schedules while allowing data center operators to add cooling capacity in line with future expansion plans.
The launch comes as AI workloads are driving demand for increasingly dense computing infrastructure, making thermal management a critical factor in the design and operation of next-generation data centers.
By combining oil-free compression, seasonal free cooling, lower-GWP refrigerant and modular installation, LG Electronics aims to address both the operating efficiency and construction requirements of AI-focused data centers.
The company plans to continue expanding its customized cooling solutions as global demand for AI computing infrastructure accelerates.
