Full Direct-To-Chip Cooling Lineup for AI Workloads

At DCW 2026, LG is featuring its Direct-to-Chip (DTC) cooling lineup for data centers that uses liquid cooling technologies to meet high thermal density and power demands. Its cold plate integrates a skived fin structure to optimize coolant flow and help manage heat from high-performance chips. The 1.4 MW Coolant Distribution Unit (CDU) combines a compact design with LG’s control and sensing technologies, supporting stable operation and improved energy efficiency through inverter-driven pumps.

Also on display is LG’s Computer Room Air Handler (CRAH) system, which incorporates high-efficiency EC fans and motors, along with an Air-cooled Centrifugal Chiller (ACC), extending coverage from server-level cooling to overall facility infrastructure.

Partnerships in Immersion Cooling Solutions

LG is expanding its portfolio with immersion cooling solutions through collaborations that respond to the growing cooling demands of high-density AI environments. The lineup includes immersion cooling tank systems developed with U.S.-based Green Revolution Cooling (GRC), as well as cooling fluids jointly developed with SK Enmove, a provider of premium base oils and lubricants. These solutions immerse IT equipment directly in dielectric liquid for more stable cooling performance even in high-heat environments.

Advanced Control and Operation Capabilities

To complement its hardware, LG is strengthening its control and operation capabilities through Data Center Cooling Management (DCCM). The software enables integrated monitoring and data-driven control across complex cooling infrastructure, including CDU, CRAH and ACC.

DCCM supports continuous operation through early anomaly detection and virtual sensor-based diagnostics. It contributes to stronger risk management with predictive maintenance, while real-time optimization based on IT workloads can improve operational efficiency. Its 3D visualization capabilities allow operators to monitor system status and manage data center operations more effectively.

Maximizing Compute per Megawatt™

Beyond cooling technologies, LG highlights software and power infrastructure solutions aimed at improving data center efficiency and increasing overall processing capacity. Among them is an AI-based workload orchestration platform developed by PADO, a company that emerged from LG North America Innovation Center (LG NOVA).

Serving as an intelligent “energy-aware compute orchestrator” for data centers, the platform analyzes IT, cooling and power systems, leveraging a multi-physics digital twin and reinforcement learning to deliver real-time operational changes. For example, it can shift power from idle servers to systems managing more demanding AI workloads, supporting 25 percent more utilization.

Jointly-Developed DC Grid Solution to Enhance AIDC Power Efficiency

In parallel, LG is also presenting a Direct Current (DC) Grid solution for data center operations, developed in collaboration with LS ELECTRIC, LS Cable & System and LG Energy Solution. The DC Grid solution reduces energy loss by minimizing the power conversion stages common in traditional Alternating Current (AC) systems, where around 25 percent of energy can be lost as heat.

The solution enables major data center equipment, including chillers, to operate directly on DC power, lowering initial power loss to approximately 15 percent.* When paired with renewable energy sources such as solar power, this figure can be reduced to as low as 10 percent. The result is improved power efficiency that can contribute to lower total operating costs.

“AI data centers require advanced cooling technologies, but they also need power and operations to work together in a more integrated way,” said James Lee, president of the LG ES Company. “LG will continue to expand its AIDC cooling capabilities, offering solutions that improve performance and support greater energy efficiency.”

Visitors to DCW 2026 can experience LG’s latest AIDC solutions, including the DTC cooling lineup, immersion cooling tank systems, DCCM integrated software, PADO’s AI-based workload orchestration platform and the DC Grid solution, at the company’s booth (#416, Walter E. Washington Convention Center) from April 20–23.