Abstract: This talk presents key findings from the "2024 United States Data Center Energy Usage Report," which reveals that U.S. data center energy use more than doubled between 2017 and 2023, primarily driven by the rapid adoption of GPU-accelerated servers for artificial intelligence. By 2023, data centers consumed 176 TWh, or 4.4% of total U.S. electricity. Looking ahead, we explore future scenarios where this demand could swell to between 325 and 580 TWh by 2028, potentially accounting for up to 12% of the nation's electricity. We will dissect the hardware trends and operational practices behind this surge, from the power draw of multi-GPU servers to the infrastructure supporting them. Beyond simply forecasting growth, this presentation will address the critical need for new strategies to manage this load.
Biography: Alex Hubbard is a Technology Researcher at Lawrence Berkeley National Laboratory in the Buildings and Industrial Applications Department. He is a member of the LBNL's Center of Expertise for Data Center Energy, and supports a variety of projects relating to data center energy consumption. He completed his Master's Degree in Data Science at UC Berkeley, and his Bachelors in Mechanical Engineering at the University of Maryland.