Gartner: AI Drives 26% Data Centre Power Surge to 565TWh

Data centre developers are navigating physical infrastructure constraints as the electricity required to support global AI deployments outpaces available utility supply. 

The technology sector is currently experiencing a structural imbalance between the rapid construction of new server facilities and the underlying energy infrastructure necessary to keep them continuously operational. 

Recent industry analysis from Gartner indicates that global data centre electricity consumption is projected to reach 565TWh by the end of 2026.

This forecast represents a 26% increase from the 447TWh recorded throughout 2025. The physical availability of power now strictly dictates the pace of facility expansion across key global markets – effectively acting as a hard constraint on development.

"Surging demand for compute-intensive AI workloads is driving unprecedented data centre power growth, while AI capacity is now constrained by power availability, making data centre power security the new battle ground for scaling and protecting margins in the global AI race," says Linglan Wang, Lead Economist at Gartner.

Changing server hardware mix

This upward trajectory in electricity demand is heavily linked to an alteration within the facility hardware environment. Operators are rapidly installing AI-optimised servers to support machine learning and heavy computational workloads. 

Gartner estimates that these advanced servers will account for 31% of total data centre power consumption during 2026, marking a significant increase from approximately 20% in the previous year.

The consumption profile of this new hardware contrasts sharply with conventional enterprise deployments. Conventional server electricity consumption grew by less than 1% in 2025 and is projected to increase by a marginal 1.2% in 2026.

AI-optimised servers recorded an 83% consumption increase in 2025. This category is forecast to expand by a further 84% this year to reach a total of 175TWh globally.

Based on the current installation rate, AI hardware will entirely surpass conventional servers in terms of energy use by 2027. Gartner projections indicate AI-optimised server consumption will hit 258TWh at that stage, with conventional hardware reaching a lower total of 200TWh.

Thermal demands increase cooling load

Alongside rising rack power densities, the mechanical infrastructure required to maintain safe operating temperatures is drawing significantly more energy from the grid. Cooling systems are operating under heavier loads to manage the thermal output generated by denser chipsets.

Data centre cooling infrastructure electricity consumption is forecast to jump by 22.6% this year, reaching a total of 195TWh. This increase reflects both the immediate thermal demands of running compute-intensive hardware and the sheer volume of new whitespace capacity coming online internationally.

Securing long-term grid access

The long-term outlook outlines a challenging environment for operators reliant on existing utility networks. By 2030, Gartner estimates that total data centre electricity demand will reach 290GW. The associated consumption is expected to exceed 1,200TWh by the end of the decade, a figure roughly equivalent to the entire annual power demand of Japan.

“Infrastructure and operations leaders must prioritise efficiency upgrades and secure grid access,” explains Linglan. “They also need to invest in high-efficiency cooling systems and edge computing to mitigate power constraints and ensure sustainable, scalable growth.”

Current utility grid supply plans remain insufficient to support this projected volume of future data centre construction. Developers are increasingly treating grid capacity as a primary site selection criterion.

Securing long-term power purchase agreements has become equally as strategically significant as acquiring physical land and facility building permits.