LG CNS and Naver Cloud Deploy Direct-to-Chip Liquid Cooling at Samsong Data Centre to Power Next-Gen AI Services

LG CNS announced on Tuesday that it will collaborate with Naver Cloud to implement direct-to-chip (DTC) liquid cooling technology at the Samsong data centre, a move aimed at laying the groundwork for supporting high-performance AI services. The partnership marks a significant step in adapting existing data centre infrastructure to the escalating thermal demands of advanced AI workloads.

The Samsong facility, where LG CNS previously assumed outsourced operations and signed a colocation agreement with Naver Cloud, will be retrofitted with a liquid-cooling infrastructure designed to enable Naver Cloud to operate Nvidia's next-generation GPU, Vera Rubin, which is owned by the National IT Industry Promotion Agency under the Ministry of Science and ICT. According to LG CNS, Vera Rubin is expected to deliver approximately 3.3 times the inference performance of the existing Blackwell Ultra, underscoring the critical need for more efficient cooling solutions.

The DTC system to be deployed attaches a cooling plate carrying coolant directly to the GPU chip to absorb heat, leveraging liquid's superior heat-transfer efficiency compared with air. This approach is increasingly vital as power density in high-performance GPU server racks rises beyond 200 kW per rack, far exceeding the 20 to 30 kW per rack limit that conventional air-cooling can handle. Industry experts view liquid cooling technologies such as DTC as essential for building so-called AI factories, where dense computing clusters generate unprecedented thermal loads.

LG CNS will combine its data centre design, build, and operate (DBO) capabilities with AI infrastructure expertise to deliver an integrated implementation covering DTC liquid cooling, GPU server operating environments, power supply systems, and the operational foundation for AI platforms. Naver Cloud, in turn, plans to leverage its AI platform operations capabilities and extensive experience with large-scale GPU services to provide high-performance AI services to domestic public institutions and enterprises. The companies aim to complete construction and verification of the liquid-cooling infrastructure by next year.

Jo Heon-hyeok (조헌혁), an executive director overseeing LG CNS's data centre business, said data centre cooling technology is emerging as a core competitive edge for AI service quality and operational efficiency. He added that LG CNS will work with Naver Cloud to continue building key infrastructure that supports the growth of the domestic AI ecosystem, drawing on LG CNS's strengths in data centre DBO capabilities and next-generation cooling technology. The collaboration underscores a broader industry trend where advanced cooling is becoming a strategic differentiator in the AI era, as operators race to accommodate the power and thermal requirements of next-generation processors.