Data Center Planet
Back to News
Datacentre Growth Faces Potential Power Constraints by 2030

Datacentre Growth Faces Potential Power Constraints by 2030

By Data Center PlanetWednesday, 10 June 2026

Energy consumed by datacentres is projected to grow 26 percent this year driven by AI advancements, with concerns that grid supply may not meet demand by 2030, according to Gartner.

The research firm anticipates global datacentre electricity consumption reaching 565 terawatt-hours (TWh) by 2026, with power demand increasing from 104 GW in 2025 to 132 GW this year.

This projection surpasses the 500 TWh per annum Gartner had previously estimated for AI-optimised servers by 2027.

The increasing requirement for computational power to support AI workloads is largely responsible, as companies invest heavily in AI projects amidst competitive pressures, often with limited returns.

Gartner highlights that AI-optimised servers are a key driver of rising datacentre power consumption, with significant investment from hyperscalers and other buyers directed towards systems configured for AI processing.

The firm estimates that AI-optimised servers will constitute 31 percent of overall datacentre power consumption this year, and by next year, their energy use is expected to exceed that of all conventional servers.

This aligns with earlier predictions that AI workloads would surpass other server tasks, such as databases and analytics, becoming the leading workload by server deployment by 2027.

Nevertheless, this growth trajectory signals a potential challenge. Gartner predicts total datacentre electricity consumption could exceed 1,200 TWh by 2030, indicating that grid supply may be inadequate for supporting future datacentre capacity expansions.

Previous warnings have highlighted the risk of datacentre energy demands outstripping grid capacity. Goldman Sachs predicted energy use would more than double by the decade's end, yet Gartner's current figures suggest demand is already surpassing those earlier estimates for 2027.

Energy infrastructure company Schneider Electric proposed four future scenarios for AI datacentre electricity consumption, but Gartner's latest 2030 forecasts exceed even Schneider's most aggressive outlook.

Power grid operators and datacentre developers, particularly in the US, face significant challenges, with industry analysts noting a lack of straightforward solutions.