AI-Driven Data Centers Push Power Grids Into Era of Unpredictable Electricity Demand: Capgemini Report

The rapid expansion of AI-driven data centers is not only increasing electricity demand, but making it significantly harder to predict, challenging how power systems are planned and delivered. A large majority of electricity executives expect more extreme and less predictable demand spikes, while more than three quarters say they struggle to forecast future needs accurately, according to the Capgemini Research Institute’s latest report, AI meets the grid: shaping the data center power play. The research, which surveyed over 600 senior electricity executives from organisations with annual revenue exceeding USD 500 million, highlights how power systems are entering a new phase as AI workloads become increasingly unpredictable. According to the report, forecasting has become significantly harder, but AI is also part of the solution with a majority of executives expecting AI to unlock significant efficiency and operational gains.

Beyond growth, the bigger challenge is uncertainty. Utilities are increasingly planning for demand that may never materialise. The report highlights a growing disconnect between projected and actual demand: a majority (67 percent) of electricity executives refer to ‘phantom’ data-center load requests, with around two in ten (19 percent) of them never materialising, distorting forecasts and increasing the risk of both over- and under-investment.

This forecasting uncertainty creates a significant capital allocation dilemma. Utilities must decide not only how much capacity to invest in, but where and when to prioritise grid modernisation investments to support future demand while avoiding stranded assets. For hyperscalers, the challenge is equally acute, requiring major infrastructure decisions to be made against a backdrop of uncertain demand forecasts, grid availability and connection timelines.

Furthermore, over three-quarters (77 percent) of utilities are forecasting difficulties predicting future demand accurately, as consumption patterns from AI become less stable and more difficult to model. As a result, they expect demand variability to emerge as a major system challenge, requiring new approaches to planning and operations.

In addition, 68 percent of electricity executives also anticipate shortages due to data-center demand growing faster than they can expand supply.

The challenge is compounded by the geographic concentration of data centers, which places significant strain on local grids: more than half of electricity executives identify load concentration as a major obstacle to reliable service, while large clusters of high-density facilities are creating localised bottlenecks that affect system stability and investment planning.

Claire Gauthier, Global Head of Energy and Utilities at Capgemini, said, “AI is transforming electricity systems far beyond demand growth. It is exposing structural constraints in grid capacity, planning and power availability, while making demand more dynamic and harder to predict. The challenge is no longer only how much power is needed, but whether it can be delivered reliably, where and when it is required. Utilities have a defining role to play as system orchestrators, leveraging AI-enabled insights to balance grid and customer-owned resources, accelerate deliverable capacity, and enable the next phase of data-center growth.”

According to the report, electricity consumption from AI training and inferencing is expected to rise significantly from 25 percent to 60 percent of total data center electricity demand in the next three to five years, largely displacing other IT workloads.

At the same time, electricity executives see AI as a force multiplier for grid planning and reliability: around six in ten expect advanced AI analytics to deliver over 10 percent improvements in failure reduction, operational productivity and preventing and restoring outages.

As per the report, less than half (45 percent) say they are currently using AI for grid optimisation, and only 16 percent of electricity organisations have implemented more advanced AI-driven approaches to optimise power flows, enhance resilience and improve real-time system performance to keep pace with booming demand.

According to the report, grid infrastructure construction timelines are also a critical constraint in accommodating rapid demand growth from AI data centers. This underscores the urgent need to accelerate grid modernisation through AI itself and climate tech to deliver reliable, affordable and sustainable power.

On-site power as a structural shift toward hybrid and decentralised energy systems

Facing grid constraints and delays, data centers are increasingly shifting from backup-only approaches toward primary Behind-the-Meter (BTM) and near-site solutions. Nearly three in ten say they already deploy on-site power solutions and 39 percent plan to add on-site/BTM within the next one to two years; more than seven in ten expect these solutions to significantly reduce reliance on the grid within five years.

The majority (86 percent) see the ability to operate independently from the grid as a competitive advantage. This evolution is reshaping the traditional relationship between utilities and large energy consumers, introducing both opportunities and coordination challenges.

A balanced, diversified energy mix at the core of reliable and sustainable data-center growth

A diversified energy mix is emerging as essential to ensure reliability and long-term resilience while renewable energy alone cannot yet provide continuous power at scale for large data centers and AI workloads - according to 78 percent of electricity executives and 73 percent of data-center executives. Both report on active investment in Battery Energy Storage Systems (BESS) to help bridge the gap.

They also agree that long-term solutions such as nuclear (Small Modular Reactors) will take time to deploy. As a result, more than two-thirds (68 percent) of electricity and data-center executives globally see natural gas as a near‑term, transitional solution until renewable energy and storage technologies can scale, creating tensions with decarbonization goals.

“For both energy providers and data-center operators, the key challenge is no longer only scaling capacity, but doing so under uncertainty, speed constraints, and rising system complexity,” concludes Claire Gauthier. “Success will depend on the ability to align infrastructure investment, energy sourcing, and AI-enabled operations to manage both the scale and volatility of demand, while balancing reliability, cost, and sustainability.”