Need For AI Data Center Efficiency Drives Innovation

R. Pritchard

Summary Bullets:

• SKT is partnering to use targeted liquid cooling with thermal fluids to lower the temperature of the hottest server components in its AI data centers.

• Industry estimates indicate AI will accelerate data center power and water consumption to the equivalent of major developed countries – causing growing concern.

South Korea Telecom (SKT), the South Korean communications company that is positioning itself as ‘an AI company,’ has signed a memorandum of understanding (MoU) with Iceotope Technologies and SK Enmove to collaborate on the development of liquid cooling solutions to optimize the energy efficiency of its AI data centers because air conditioning and cooling systems consume the largest amount of energy in data centers. It has been calculated that AI will cause energy usage at data centers to double from their present levels: The International Energy Agency (IEA) sees electricity consumption from data centers, AI, and cryptocurrency doubling by 2026, having globally consumer an estimated 460 TeraWatt-hours (TWh) in 2022. If that happened, demand would be roughly equivalent to the electricity consumption of Japan.

Equally, it is estimated that global demand for water to cool data centers will be driven upwards by AI to between 4.2 billion and 6.6 billion cubic meters by 2027 – about half the amount consumed by the UK each year.

The technology being used is Iceotope’s precision liquid cooling, which delivers dielectric fluid to remove heat from the hottest components of servers to reduce energy consumption. The move is part of SKT’s plans to develop an AI-based autonomous cooling system that can analyze the temperature and power load in the AI data center and automatically control the supply temperature and flow rate of thermal fluids in cooling units to improve operational efficiency.

With an estimated 8,000 data centers globally (no one knows the definitive number), their consumption of power, water, and other resources is becoming a significant issue. For example, in Ireland, the granting of planning permission for new data centers is being scrutinized by the country’s environment minister, Eamon Ryan, who has said: “They cannot operate if they are putting energy security at home at risk.”

There is huge demand for data centers and is set to grow exponentially with the AI boom. Unfortunately, electricity grids and water supply systems were not built with this in mind and cannot be ramped up overnight. The need for smart solutions to reduce energy and water consumption is essential if the full potential of AI is to be realized.

Leave a Reply