LG Electronics CEO Ryu Jae-cheol said cooling technology is central to the competitiveness of AI data centers because reducing cooling power consumption frees more electricity for AI computing. According to Sina Finance, Ryu wrote on LinkedIn on the 20th that cooling can reduce its own energy use while making more power available for AI computing; on the 19th local time, he published an article outlining an AI data center cooling strategy.

Ryu identified power, water and space as constraints facing data centers, saying the key challenge is how much computing capacity can be delivered with limited resources. Power usage effectiveness (PUE), calculated by dividing a data center’s total power consumption by the power used by its IT equipment, is increasingly viewed in terms of how much available electricity can be allocated to AI computing. A PUE closer to 1 indicates less power is used by cooling and other auxiliary equipment.

Ryu said adaptability is essential because AI hardware evolves faster than the facilities supporting it, requiring cooling equipment to change when new chips arrive. He also said efficiency assessments should extend beyond PUE to consider how easily cooling systems can be adapted and revalidated for new hardware, the energy used to transfer heat from chips to coolers, and reliable operation under local power, water and climate conditions.

He called for IT, power and cooling systems to be integrated around AI computing performance rather than optimized separately. LG Electronics is addressing data center energy use through cooling-efficiency improvements, waste-heat recovery and energy management. Its products include chillers, cooling distribution units (CDUs), fan wall units (FWUs), heat pumps and energy management solutions. Ryu said the company aims to improve overall data center energy management by integrating these technologies rather than optimizing individual components.