Data centers consume 1-2% of the world’s electricity, and concerns about their energy use are growing as the world becomes increasingly interconnected and digitalized. Few data center operators publicly report electricity use, necessitating the use of mathematical models for generating the sector-level energy use estimates that guide public policy and shape public perceptions. This presentation will review the state of data center energy modeling, discuss how different methods can lead to (sometimes drastically) different results, and identify near-term priorities for improving data center energy models in light of emerging trends such as artificial intelligence, edge computing, and 5G that promise to increase societal demand for data center services considerably.
Eric Masanet is the Duncan and Suzanne Mellichamp Chair of Sustainability Science for Emerging Technologies at UC Santa Barbara. His research focuses on energy modeling of industrial and IT systems, with a longstanding focus on data centers. He led the analytical components of the landmark 2007 U.S. EPA Report to Congress on Server and Data Center Energy Efficiency, and has since published numerous sustainability analyses of data centers and ICT systems at regional and global levels. From 2015-2017, he led the Energy Demand Technology Unit at the International Energy Agency in Paris, where he oversaw energy analyses for the global industrial, buildings, and transport sectors. He is currently a Lead Author of Chapter 5 (Demand) for Working Group III of the IPCC’s Sixth Assessment Report and a member of the Research Advisory Board at the American Council for an Energy Efficient Economy (ACEEE). He holds a PhD in mechanical engineering from UC Berkeley, with a specialization in sustainable manufacturing.