Power Grids infrastructure buildout is another major constraint. Construction timelines are not keeping pace with the speed of AI-driven demand growth, which makes it even more important to accelerate grid modernization through AI and climate technology.
The rapid rise of AI-driven data centers is pushing electricity demand higher, but it is also making that demand much harder to predict. That is creating a serious challenge for how power systems are planned, built, and delivered. According to Capgemini Research Institute’s latest report, AI meets the grid: shaping the data center power play, a large majority of electricity executives expect more extreme and less predictable demand spikes, while more than three-quarters say they are struggling to forecast future needs accurately.
The report is based on a survey of more than 600 senior electricity executives at organizations with annual revenue above $500 million.
It shows that the power sector is entering a new phase as AI workloads become increasingly volatile. Forecasting has become much more difficult, but AI is also seen as part of the answer, with most executives expecting it to unlock major efficiency and operational gains.
A new era of volatile demand
The biggest issue is no longer just growth — it is uncertainty. Utilities are now planning for demand that may never actually show up.
The report points to a widening gap between projected and real demand, with 67% of electricity executives referring to “phantom” data-center load requests. Around 19% of those requests never materialize, which distorts forecasts and raises the risk of both overinvestment and underinvestment.
This uncertainty creates a serious capital allocation problem. Utilities have to decide how much capacity to build, as well as where and when to prioritize grid modernization investments so they can meet future demand without creating stranded assets.
Hyperscalers face a similar challenge, making major infrastructure decisions while dealing with uncertain demand forecasts, limited grid availability, and unclear connection timelines.
Forecasting is getting harder too, with 77% of electricity executives saying they struggle to predict future demand accurately.
As AI consumption patterns become less stable and harder to model, demand variability is emerging as one of the biggest system-level challenges. On top of that, 68% expect shortages as data-center demand grows faster than supply can be expanded.
The problem becomes even more difficult because data centers are often concentrated in specific locations. More than half of the executives surveyed say load concentration is a major obstacle to reliable service.
Large clusters of high-density facilities are creating local bottlenecks that affect grid stability and investment planning.
“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,” said Claire Gauthier, Global Head of Energy & Utilities at Capgemini. “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.”
AI as both problem and solution
The report says electricity use from AI training and inference is expected to rise sharply, growing from 25% to 60% of total data-center electricity demand over the next three to five years. That growth will largely replace other IT workloads.
At the same time, electricity executives see AI as a powerful tool for improving grid planning and reliability.
Around six in ten expect advanced AI analytics to deliver more than 10% improvements in failure reduction, operational productivity, and outage prevention and restoration.
Adoption is still limited
Even with those benefits, AI adoption is still relatively modest. Fewer than half of respondents, or 45%, say they are already using AI for grid optimization.
Only 16% say their organizations have implemented more advanced AI-driven systems to optimize power flows, improve resilience, and boost real-time performance in line with demand growth.
Power Grids infrastructure buildout is another major constraint. Construction timelines are not keeping pace with the speed of AI-driven demand growth, which makes it even more important to accelerate grid modernization through AI and climate technology.
The goal is to deliver reliable, affordable, and sustainable power without falling behind demand.
The shift to on-site power
Because of grid constraints and delays, data centers are moving beyond backup-only setups and increasingly adopting primary behind-the-meter and near-site power solutions.
Nearly 30% already use on-site power, while 39% plan to add on-site or BTM capacity in the next one to two years. More than 70% expect these solutions to significantly reduce their dependence on the grid within five years.
This shift is changing the traditional relationship between utilities and large energy users. The report says 86% view grid independence as a competitive advantage. That creates both opportunities and coordination challenges as the energy ecosystem becomes more decentralized.
The need for a mixed energy strategy
A diversified energy mix is becoming essential for reliability and long-term resilience. According to 78% of electricity executives and 73% of data-center executives, renewable energy alone still cannot provide continuous power at the scale required by large data centers and AI workloads.
Both groups say they are actively investing in battery energy storage systems to help fill the gap.
They also agree that longer-term solutions such as nuclear power, including Small Modular Reactors, will take time to deploy. In the meantime, more than two-thirds of both electricity and data-center executives globally see natural gas as a transitional option until renewable energy and storage technologies can scale. That creates a clear tension 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,” Claire Gauthier said. “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.”
Survey methodology
Capgemini Research Institute surveyed 612 senior electricity executives at organizations with annual revenue above $500 million that are actively working with data centers.
It also surveyed 175 senior executives from data-center-owning and operating organizations with revenue above $250 million.
Respondents came from 21 countries across North America, Europe, APAC, and Latin America. The global survey was conducted in January 2026.
To Read The Full AI meets the grid: shaping the data center power play Report: CLICK HERE
