목차
Executive summary 2
Key Messages 2
Potential Impacts of Artificial Intelligence on Data Center Load Growth 4
EPRI U.S. Data Center Load Projections 4
Data Center Power Demands Are Concentrated in a Few Regions 5
A Roadmap to Support Rapid Data Center Expansion 6
Introduction 7
Research Questions 7
Data Centers in the United States 7
Data Centers’ Primary Electricity-Consuming Hardware and Equipment 9
AI and Data Center Power Consumption Insights 10
Immense Volumes of Data are Being Processed Daily 10
History of Energy Efficiency in the Data Center Industry 11
Uneven Geographic Distribution Creates Imbalance in Data Center Load 12
AI Implications for Power Consumption 14
Chat GPT and Other Large Language Models (LLMs) 15
Forecasting Data Center Load Growth to 2030 17
Four Scenarios Based on Historical Data, Expert Insights, and Current Trends 17
Energy Efficiency, Load Management and Clean Electricity Supply 18
Energy-Efficient Training Algorithms 18
Energy-Efficient Hardware 19
Energy-Efficient Cooling Technologies 19
Scalable Clean Energy Use 20
Monitoring and Analytics 20
Reducing Data Centers’ Environmental Footprint 21
Actions to Support Rapid Data Center Expansion 21
Improve Data Center Operational Efficiency and Flexibility 22
Increase Collaboration through a Shared Energy Economy Model for Sustainable Data Centers 22
Better Anticipate Future Point Load Growth through Improved Forecasting and Modeling 23
Appendix A: State-Specific Scenarios 24
Projected Data Center Load Scenarios for Top 15 States 24
Regional Differences in Data Center Capacities by Metropolitan Area 27
Projections of Potential Power Consumption for 44 States 28
Appendix B: Insights Into the Energy Use of AI Models 29
References 31
