Power consumption in global data centers is set to see a significant increase, with projections indicating a rise to 565 terawatt hours (TWh) by 2026, which is a 26% jump from 447 TWh in the previous year. This growth in power consumption is largely driven by the rise of AI-oriented data centers, which are expected to take up a substantial proportion of the overall power usage.
According to Gartner, global data center power demand will increase by 27% in 2026, reaching 133 gigawatts (GW), up from 105 GW in 2025, with an estimate of 291 GW by 2030. These figures reflect the incredible demand for compute-intensive AI workloads, which is outpacing available power resources. Linglan Wang, a director analyst at Gartner, noted that rising demand for efficient AI operations has created a competitive environment for securing sufficient power for data centers.
Among the highlights of this report, AI-optimized servers are expected to represent 31% of total data center energy consumption by 2026—an increase from the current figures, where conventional servers are only expected to see minimal growth. Specifically, conventional servers are projected to consume 195 TWh in 2026, a mere 3% growth from 2025.
In contrast, AI-optimized servers are anticipated to surge from 95 TWh in 2025 to 175 TWh in 2026, marking an 84.2% increase. By 2030, it’s forecasted that these servers will account for nearly half of all data center power consumption globally.
Considering the predicted increase in overall data center electricity consumption, which could exceed 1,200 TWh by 2030, the current grid infrastructure will likely be unable to meet future demands. Consequently, leaders in infrastructure and operations are urged to prioritize energy efficiency, secure better access to the grid, and invest in advanced cooling systems and edge computing solutions to navigate upcoming energy constraints.
