Capgemini analysis: AI infrastructure and electrical grid convergence reshaping power-procurement strategy and demand profile.

We are living in the Age of Electricity. While electrification across industry, transport, and buildings continues to push demand higher, one of the fastest-growing drivers is digital infrastructure—data centers and AI. In recent years, electricity demand has surged as data-center activity has expanded, a trend expected to continue over the next three to five years.

The Capgemini Research Institute's "AI meets the grid: Shaping the data center power play" examines how rapid growth in data-center operations, especially AI-driven workloads, is reshaping electricity demand and challenging energy organizations. The report explores the constraints power companies face, the growing role of on-site power generation, and how renewables, natural gas, and small modular reactors (SMRs) are expected to contribute to future solutions. It also highlights how AI, generative AI, and AI agents can create value across grid operations and system performance. Key insights are drawn from a survey of energy and data-center executives across 21 countries, complemented by qualitative insights from industry leaders.

The rapid rise of data centers running AI workloads is straining power systems worldwide. Eighty percent of utilities expect more extreme and less predictable demand spikes, directly impacting grid resilience. A large majority of industry leaders—70% of electricity executives and 83% of data-center executives—expect high-density AI-led data center sites to significantly increase regional power demand within the next three to five years.

For energy providers, AI is emerging as a force multiplier for grid planning and reliability, with more than 60% expecting it to unlock significant efficiency and operational gains. However, only 45% of utilities today are using AI for grid optimization, revealing significant opportunity to scale digital and AI-driven operations to keep pace with booming demand.

Multiple structural barriers are constraining capacity expansion. Aging infrastructure (cited by 74% of electricity executives), permitting delays (84%), interconnection timelines (76%), insufficient reserve margins (84%), and supply-chain pressures (74%) are slowing deployment and constraining reliable power delivery.

Data-center operators are increasingly turning to on-site or near-site power generation to serve demand that central grids cannot immediately accommodate. Twenty-nine percent of data-center executives globally already deploy on-site generation, 39% plan to add it within one to two years, and more than seven in ten expect it to significantly reduce reliance on the grid within five years.

Both electricity and data-center leaders agree that no single energy source can reliably power the next wave of high-density computing. Diversification is widely seen as critical, with 78% of electricity executives saying renewables alone cannot yet meet 24/7 demand. Sixty-eight percent of electricity and data-center executives view natural gas as key to faster timelines, despite decarbonization tensions.

The research shows a widening gap between data-center demand growth and the pace at which grids can expand capacity. Flexible loads, demand response programs, and dynamic tariffs allow utilities and data-center operators to better manage peak demand and volatility, reducing strain on the grid while enabling faster, more cost-effective capacity planning. AI increasingly enables this optimization through forecasting, grid monitoring, outage prevention, and operational enhancements, unlocking productivity gains while improving reliability and resilience.