Europe’s AI Boom Puts Strain on Green Goals and Data Centers
The rapid growth of artificial intelligence is challenging Europe to manage power-hungry data centers while staying aligned with strict environmental goals. As companies like Nvidia develop high-performance, energy-intensive chips, European data centers are grappling with the need to cool these devices efficiently, often requiring much lower water temperatures.
Steven Carlini of Schneider Electric emphasizes cooling as a major energy factor, second only to the IT load itself. “The energy use will rise” he explains, as Nvidia’s new Blackwell GB200 chips require water cooling between 20 and 24 degrees Celsius (68-75°F), while traditional temperatures are often higher. Ferhan Gunen from Equinix notes that their customers want higher-density servers, but implementing such measures isn’t straightforward: “It’s really an evolution discussion more than anything”.
The AI boom is putting additional demands on data center operators to increase density while also investing in sustainable cooling solutions. Liquid cooling, widely seen as a more efficient approach, requires reconfiguration, but Gunen believes it’s essential for handling the higher power requirements AI chips bring. The Nvidia Blackwell platform, designed to reduce AI’s energy and cost demands by up to 25 times, is already being adopted by companies like Nebius, which plans to invest over $1 billion in European AI infrastructure by 2025.
However, the push for more power conflicts with Europe’s decarbonization goals. The EU’s Energy Efficiency Directive aims to cut energy use by 11.7% by 2030, yet projections show that AI could drive data center demand up by 160% by the end of the decade. This dilemma has led organizations, like the European Data Center Association (EUDCA), to urge caution. Michael Winterson, EUDCA chair, warns that “lowering water temperatures will fundamentally drive us back to an unsustainable situation”.
As Europe navigates the high-stakes AI race, regional data centers face pressure to reduce energy consumption transparently. The EU now requires large data centers to report power usage to meet sustainability targets, with Schneider Electric working closely with regulators on sourcing cleaner energy. Sicco Boomsma of ING also highlights that U.S. operators are aligning with European green initiatives, further intensifying the market’s competitiveness.
With high-power GPUs consuming the energy of up to 25 homes in a single square meter, Europe’s data centers must strike a balance between AI’s energy demands and sustainability commitments, an increasingly complex task amid soaring demand and regulatory pressures.
Source: CNBC
Share:
The rapid growth of artificial intelligence is challenging Europe to manage power-hungry data centers while staying aligned with strict environmental goals. As companies like Nvidia develop high-performance, energy-intensive chips, European data centers are grappling with the need to cool these devices efficiently, often requiring much lower water temperatures.
Steven Carlini of Schneider Electric emphasizes cooling as a major energy factor, second only to the IT load itself. “The energy use will rise” he explains, as Nvidia’s new Blackwell GB200 chips require water cooling between 20 and 24 degrees Celsius (68-75°F), while traditional temperatures are often higher. Ferhan Gunen from Equinix notes that their customers want higher-density servers, but implementing such measures isn’t straightforward: “It’s really an evolution discussion more than anything”.
The AI boom is putting additional demands on data center operators to increase density while also investing in sustainable cooling solutions. Liquid cooling, widely seen as a more efficient approach, requires reconfiguration, but Gunen believes it’s essential for handling the higher power requirements AI chips bring. The Nvidia Blackwell platform, designed to reduce AI’s energy and cost demands by up to 25 times, is already being adopted by companies like Nebius, which plans to invest over $1 billion in European AI infrastructure by 2025.
However, the push for more power conflicts with Europe’s decarbonization goals. The EU’s Energy Efficiency Directive aims to cut energy use by 11.7% by 2030, yet projections show that AI could drive data center demand up by 160% by the end of the decade. This dilemma has led organizations, like the European Data Center Association (EUDCA), to urge caution. Michael Winterson, EUDCA chair, warns that “lowering water temperatures will fundamentally drive us back to an unsustainable situation”.
As Europe navigates the high-stakes AI race, regional data centers face pressure to reduce energy consumption transparently. The EU now requires large data centers to report power usage to meet sustainability targets, with Schneider Electric working closely with regulators on sourcing cleaner energy. Sicco Boomsma of ING also highlights that U.S. operators are aligning with European green initiatives, further intensifying the market’s competitiveness.
With high-power GPUs consuming the energy of up to 25 homes in a single square meter, Europe’s data centers must strike a balance between AI’s energy demands and sustainability commitments, an increasingly complex task amid soaring demand and regulatory pressures.
Source: CNBC