The rapid development of artificial intelligence is drastically changing the operating landscape of data centers across Europe and pushes them toward a more environmentally friendly approach. As the demand for AI capabilities is increasing, developers are now feeling the heat to reduce the water temperatures of such energy-intensive facilities so as to accommodate the high-performance chips that leading tech companies, including Nvidia, are using.
And because AI is likely to lead to a whopping 160 percent increase in demand for these data centers in 2030 which is a crisis of epic proportion for Europe, Bloomberg stated.
Large language models are a very important part of AI technology and need high-powered chips such as graphics processing units for training and deployment. They produce high heat which requires water temperatures to be low for proper cooling. High-density computing is required to power these GPUs. According to Andrey Korolenko, chief product and infrastructure officer at Nebius, energy consumption of AI can reach 120 kilowatts per square meter, which can be as high as the energy usage of 15 to 25 households.
That’s a very backward step,” said Michael Winterson, chairman of the European Data Center Association (EUDCA), referring to the call to cool the water. The “space race” about AI development, he mentioned, has been led by the Americans; land rights and energy access and sustainability were usually placed in the “backseat” in that “racing” effort for market supremacy.
Herbert Radlinger, managing director at NDC-GARBE, believes the request from U.S. chip manufacturers to adapt cooling protocols from European suppliers is a cause for alarm since it comes at a time when initially, liquid cooling systems were expected to handle higher temperatures.
The European Commission aims to cut total energy consumption by 11.7% in the year 2030. However, the AI advance rate may increase the data center energy demand at a factor of two or three times more than the already projected 28%. According to Winterson, this trend of lowering the temperature for water does not support the recently passed directive of the EU on energy efficiency that has ensured transparency over energy consumption, especially for big data centers.
Schneider Electric is very deeply engaged with the EU in the sourcing of sustainable energy for AI data centers. Vice president Steven Carlini added that while energy use will increase, the efficiency metrics will not reflect the additional energy demand placed on cooling systems.
The AI demands will stretch the data centers in Europe, forcing them to adapt and take these demands forward. For a long time, data centers will remain concerned with matters of energy efficiency and sustainability. Cooling technologies and operations continue evolving and therefore shall be the backbone balancing Europe’s environmental commitments with burgeoning demands from the AI industry.
