Artificial intelligence and data centers are transforming the global energy landscape, with their massive electricity consumption reshaping power supply systems, according to Gokberk Bilgin, secretary-general of the Energy Digitalization Association (EDIDER).
According to the International Energy Agency (IEA), electricity demand from data centers is surging due to AI. Their global electricity consumption has grown at an average annual rate of 12% since 2017.
In 2024, they consumed about 415 terawatt-hours (TWh) of electricity. The IEA projects this could more than double by 2030, reaching roughly 945 TWh—a figure that surpasses Japan's current total electricity consumption.
Bilgin told Anadolu that the race to develop larger AI models demands immense computational power, significantly driving up energy needs.
"Training a single large model today requires hundreds of high-performance GPUs running simultaneously at full load," he said. "Furthermore, using these models creates a continuous energy demand, keeping data centers active around the clock."
He noted that managing the heat has become a major burden: "Where traditional air cooling fails, there's a rapid shift to liquid cooling. Today, about 30-40% of energy in many data centers goes solely to cooling."
Adding the load from backup systems turns data centers into an almost constant, full-capacity operation, he added.
Bilgin also pointed to grid challenges, noting that high-density GPU clusters are demanding tens of megawatts of additional load in single locations, straining local grids in places like Dublin, Frankfurt, and Tokyo.
- Diverging strategies against energy demand
Bilgin outlined the different strategic approaches, saying that China's advantage is a power generation infrastructure that can expand rapidly and maintain lower costs, aligning state energy policy with AI as a geopolitical priority.
The US, meanwhile, is focusing on controlling the total energy load through efficiency gains and advanced GPUs, specialized cooling, and pairing data centers with renewables form the backbone of the US strategy, according to the expert.
- 'AI energy demand large enough to transform global supply structure'
"In conclusion, while AI is becoming an indispensable technology for managing renewable energy systems, the energy demand it creates is large enough to single-handedly transform the global supply structure," Bilgin said.
He explained that the high, constant load from data centers is difficult to meet with solar and wind alone in the short term.
"This situation maintains the importance of natural gas's flexible capacity and nuclear energy's baseload nature for the time being," he said.
Bilgin concluded that the countries that will shape the future will be those that prepare the necessary political, regulatory, and financial frameworks for this new era today.
By Fuat Kabakci
Anadolu Agency
energy@aa.com.tr