Gartner: Data Centres Electricity Consumption Up 26% in 2026

For some time now, the global AI sector has felt like a runaway train fast approaching the end of its tracks.

Data centres have been expanding far more rapidly than the energy infrastructure required to power them, creating an imbalance between supply and demand.

The numbers behind this trend make for stark reading. 

Global data centre electricity consumption is forecast to reach 565TWh in 2026, a 26% increase on the 447TWh recorded in 2025, according to new research from Gartner.

It is a sign, if any were needed, that the infrastructure underpinning the global AI boom is starting to run into a hard physical constraint – the availability of power.

"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.

The power and the pattern

Much of the spike in the demand for electricity can be traced to a single shift in the hardware mix: the rapid uptake of AI-optimised servers.

Gartner estimates that this kind of server will account for 31% of total data centre power consumption through the whole of 2026, up from roughly 20% last year.

The trajectory is striking when mapped against conventional servers, which grew electricity consumption by less than 1% in 2025 and are projected to grow by just 1.2% in 2026.

AI-optimised servers, by contrast, saw consumption climb by more than 83% in 2025 and are forecast to grow by a further 84% this year, reaching 175TWh in total before surpassing conventional servers entirely by 2027.

By that point, Gartner expects AI-optimised server consumption to stand at 258TWh, compared with 200TWh for conventional hardware.

Cooling infrastructure is also a large part of the problem.

The electricity consumption of cooling systems are forecast to jump by 22.6% this year to 195TWh, reflecting both the thermal demands of denser AI hardware and the ongoing expansion of capacity globally.

An energy-constrained future

In the long-term, these trends might see things become difficult.

Gartner estimates total data centre electricity demand will reach 290GW by 2030 and that overall consumption will exceed 1,200TWh by the same year, which is roughly equivalent to the annual power demand of Japan.

The report states plainly that grid supply will be insufficient to meet the demands of future data centre construction as things stand.

“Infrastructure and operations leaders must prioritise efficiency upgrades and secure grid access,” Linglan explains. 

“They also need to invest in high-efficiency cooling systems and edge computing to mitigate power constraints and ensure sustainable, scalable growth.”

That advice carries weight in an environment where the speed of AI deployment has consistently outpaced energy planning.

The pattern is already visible: data centre developers are increasingly treating grid capacity as a site selection criterion rather than an afterthought, and securing long-term power purchase agreements has become as strategically significant as acquiring land or permits.

Whether that level of forward planning can be adopted broadly enough – and quickly enough – to prevent power constraints from becoming a ceiling on AI progress is a question that Gartner's numbers do not answer, but firmly put on the table.