Data center electricity consumption accelerates with AI-driven demand
In 2025, data center electricity consumption grew significantly faster than the rest of the global power system. In particular, artificial intelligence demand began to play an increasingly visible role in shaping global energy consumption. This is pushing governments and companies to turn to alternative energy generation and storage solutions.
This is the picture outlined in the new report “Key Questions on Energy and AI” by the International Energy Agency, updating analyses published in April 2025.
According to the study, data center electricity consumption increased by 17% in 2025 alone, compared to global electricity demand growth of just 3%. AI-focused facilities recorded even higher increases.
By 2030, data center energy consumption is expected to double, while electricity use from AI-specific centers could triple. The report also highlights structural constraints, including grid connection delays, shortages of turbines, transformers and advanced chips, pricing pressures and the need for new generation capacity.
+17% in one year for data center electricity consumption
The central finding of the report is the speed of growth. In 2025, data center electricity consumption rose by 17%, a pace nearly six times faster than global electricity demand expansion. The main driver is large-scale investment by technology companies, rapidly increasing computing capacity for new AI models.
The International Energy Agency notes that capital expenditure by five major tech groups exceeded $400 billion in 2025 and could rise by a further 75% in 2026. This investment surge translates into new digital campuses, higher server density and increasingly heavy electricity loads.
The trend is not limited to the United States. Growth is also accelerating across Europe, Asia and emerging markets. However, geographic distribution remains uneven, with new facilities concentrating in regions offering reliable energy supply, land availability, connectivity and faster permitting processes.
Artificial intelligence energy consumption continues to rise
The International Energy Agency observes that each individual AI operation is becoming more efficient. In other words, the energy required per task is decreasing rapidly thanks to improved chips, optimized software and more advanced cooling systems. However, this progress is not enough to offset the explosion in demand.
More users are adopting generative tools, intelligent assistants and automated applications. At the same time, energy-intensive use cases are expanding, including AI agents capable of executing complex sequences of tasks, continuous processing and multimodal functions.
The result is a classic rebound effect: efficiency improves, but usage grows even faster. As a consequence, AI energy consumption continues to rise despite more efficient technologies. According to the report, AI-focused data centers could triple their electricity demand by 2030.
Infrastructure constraints: grids, turbines, transformers and chips
The report dedicates significant attention to industrial bottlenecks. Demand is growing faster than the response capacity of supply chains and infrastructure systems.
Over the past year, supply chains for gas turbines, electrical transformers, advanced semiconductors and IT components have tightened. These technologies are essential to build and power new data centers.
Permitting procedures are also a major constraint. In many countries, project pipelines are putting pressure on planning and regulatory systems. Grid connection waiting times are increasing, as are timelines for environmental approvals.
For the energy sector, this creates a dual challenge: more generation capacity is required, alongside stronger and more resilient grids. Without these upgrades, part of the digital infrastructure pipeline risks delays.
Renewables and new energy sources to power data center growth
The technology sector is not only consuming energy but also driving new electricity supply. According to the International Energy Agency, in 2025 the sector accounted for around 40% of all corporate power purchase agreements for renewable energy.
Many large digital companies are securing long-term contracts for clean electricity, enabling financing for new wind and solar projects. This represents a key lever to expand installed capacity and stabilize revenues.
At the same time, interest is growing in advanced nuclear and next-generation geothermal energy. The pipeline of conditional agreements between data center operators and small modular reactor projects has increased from 25 GW at the end of 2024 to 45 GW today. For the International Energy Agency, this suggests that AI demand could accelerate the commercialization of alternative energy technologies.
The US model: onsite gas and battery storage
In the United States, where grid connection delays are particularly significant in some regions, developers are increasingly considering onsite natural gas generation alongside data centers.
Satellite data collected by the International Energy Agency show that many of these projects remain in early stages, with both technical and financial challenges still unresolved.
One of the main issues relates to the nature of AI workloads, where electricity demand can fluctuate rapidly and significantly. Managing these peaks with traditional generation alone is complex.
As a result, onsite battery storage is becoming a strategic solution. Properly designed storage systems can absorb sudden fluctuations, reduce stress on generation assets and provide flexibility to the grid.
Energy price uncertainty and system impacts
The rapid expansion of data center electricity consumption is becoming increasingly visible and, in some regions, is raising concerns over energy prices and environmental impacts. The report clarifies that rising demand does not automatically lead to higher prices if accompanied by sufficient investment in generation and networks.
However, data centers present unique challenges, including large concentrated loads, rapid growth and the need for dedicated infrastructure. This may require new transmission lines, substations and additional generation capacity, with potential tariff impacts if planning is inefficient.
According to the International Energy Agency, governments still have tools to mitigate these risks, including smart grid integration, dynamic pricing, operational flexibility incentives and location criteria aligned with energy availability.
At the same time, the supply side could also benefit. Mature AI applications could help energy-intensive industries reduce energy costs by 3% to 10% through improved process management, predictive maintenance and consumption optimization.
However, the energy sector is not yet fully leveraging this potential. The main barriers identified are the shortage of digital skills and the limited availability of usable data.