CNBCยท2025-10-17 15:37Core Insights - Electricity companies in the U.S. are grappling with demand forecasts driven by the AI boom, as tech firms plan extensive data center builds that could consume electricity equivalent to entire cities [3][4][6] - The utility sector has seen a significant stock rally, gaining approximately 21% this year and nearly $500 billion in value over the past two years, amid concerns of an AI bubble [8][9] - Experts predict a historic increase in electricity consumption, estimating an additional 120 gigawatts of demand by 2030, with 60 gigawatts attributed to data centers [11][12] Demand Forecasting Challenges - Utilities face difficulties in accurately forecasting electricity demand due to competing requests for power connections from AI companies [5][6] - FERC Chairman David Rosner emphasized that small discrepancies in load forecasts can lead to substantial financial impacts on investments and customer bills [6] - Constellation Energy's CEO expressed concerns that current load projections may be overstated, suggesting a need for caution in planning [7] Infrastructure Constraints - The AI industry's rapid growth is straining existing electrical infrastructure, with competition for essential equipment driving up costs [14] - There is a lack of sufficient generation and transmission infrastructure to meet even modest demand targets, with natural gas turbines sold out through the end of the decade [14] - Renewable energy sources, particularly solar and battery storage, are positioned as the fastest means to expand capacity, with over 90% of current power projects awaiting grid connection being renewables [15][16] Strategic Responses - Some AI companies are exploring self-generated power solutions at data centers to bypass grid limitations, which could expedite energy availability [17][18] - Nvidia's CEO highlighted the importance of investing in diverse energy generation methods, advocating for faster deployment of self-generated power solutions [18]