Cette section n'est pas encore traduite. Le contenu ci-dessous est affiché en anglais, tel qu'il a été publié. Lire la version anglaise.
KT Cloud tests direct-to-chip and immersion cooling for next-generation GPUs and targets 1 GW of AI capacity by 2031
What changed
At its Mokdong 2 data centre in Seoul, KT Cloud is testing direct-to-chip liquid cooling on NVIDIA B200 racks with an in-house load bank that recreates server heat to check coolant flow, pressure and temperature, and runs an operational immersion cooling system. It plans modular, factory-built infrastructure starting with its Gunsan AI data centre, working with LS ELECTRIC on power and Vertiv on liquid cooling and modular design, and aims to supply 1 GW of AI data centre capacity by 2031, The Asia Business Daily reports.
View original source (opens in a new tab)
Read the original: The Asia Business Daily: KT Cloud tests direct-to-chip and immersion cooling for next-generation GPUs and targets 1 GW of AI capacity by 2031 (opens in a new tab)
We summarise in our own words and link the source. Figures are as reported by the source and not independently verified. Report a correction.
Related index data
A supplier link means the company is named in the story or offers this kind of product. It does not mean a contract unless the story says so.
Related on CoolPower Index: Chip power and cooling, AI news.
More in CDUs & Direct Liquid Cooling
- ToneCooling publishes measured datasheets for GB300 and GB200 direct-to-chip cold plate kits: about 1,067 W per GPU plate at 1.7 L/min 11 Nov 2026
- ZTE launches a 1.5 MW cold plate liquid cooling CDU, expandable to 2.4 MW, alongside 800 V DC power for AI data centres 10 Nov 2026
- India's National Stock Exchange installs direct-to-chip liquid cooling for its own AI servers, planned to scale to 224 GPUs 9 Nov 2026