Around 67% of electricity executives are facing difficulties in predicting electricity demand due to ‘phantom’ load requests from AI-driven data centers, with 19% of these requests never materializing, as per a report by Capgemini. This unpredictability is making it significantly harder to plan power systems, leading to distorted forecasts and increased risks of over- and under-investment.
The report, based on a survey of over 600 senior electricity executives at organizations with annual revenue exceeding $500 million, revealed that nearly 80% anticipate more extreme and volatile demand patterns. Additionally, 77% of respondents reported challenges in accurately forecasting future demand.
With the emergence of expected demand variability as a major system challenge, new planning and operational approaches are deemed necessary. While 60% of utilities foresee AI playing a crucial role in enhancing grid efficiency and operational gains, only a few have implemented advanced AI-driven strategies.
According to the report, utilities are tasked with determining not just the amount of capacity to invest in but also where and when to prioritize grid modernization investments to meet future demand without incurring stranded assets. Hyperscalers, facing similar challenges, must make significant infrastructure decisions amid uncertain demand forecasts, grid availability, and connection timelines.
The geographic concentration of data centers further complicates the situation, with over 50% of electricity executives identifying load concentration as a major hindrance to reliable service. The clustering of high-density facilities is creating local bottlenecks that impact system stability and investment planning.
Claire Gauthier, Global Head of Energy & Utilities at Capgemini, highlighted the transformative impact of AI on electricity systems, exposing structural constraints in grid capacity, planning, and power availability. She emphasized the pivotal role of utilities in leveraging AI insights to balance grid and customer-owned resources, accelerate deliverable capacity, and facilitate the growth of data centers.
