The demand for energy US data centers could nearly triple over the next three years and consume up to 12% of the country’s electricity, according to a study backed by the Department of Energy (DOE) first reported by Reuters on Friday.
The increase would occur as the sector undergoes a transformation due to the artificial intelligence.
Lawrence Berkeley National Laboratory produced the report at a time when the energy sector and the US government are trying to understand how demand from Big Tech data centers will affect power grids, electricity bills and the climate.
By 2028, annual data center energy consumption could reach between 74 and 132 gigawatts, or between 6.7% and 12% of total US electricity consumption, according to the Berkeley Laboratory report.
The study includes ranges that depend in part on availability and demand of a type of AI chip known as a GPU. Currently, data centers represent just over 4% of the country’s electrical load.
According to Avi Shultz, director of the DOE’s Office of Industrial Efficiency and Decarbonization, “this really tells us where the frontier is in terms of growing energy demand in the United States.”
“What this report highlights is what is growing the fastest, and spearheading demand growth in the United States is the new growth of AI data centers,” Shultz said.
According to Shultz, the conclusions can serve as based on DOE efforts to increase grid flexibility and resilience, including building long-lasting batteries in data centers and commercializing new technologies such as small nuclear reactors and advanced geothermal energy.
The AI, which requires increasingly powerful chips and intense cooling systems, is the main driver of the projected growth of data centers.
