Global electricity consumption by data centres is expected to increase by 26 per cent in 2026, driven largely by the rapid expansion of artificial intelligence workloads, according to new research from Gartner.
The research firm forecasts worldwide data centre electricity consumption will reach 565 terawatt hours (TWh) in 2026, up from an estimated 447 TWh in 2025. The increase reflects growing demand for compute-intensive AI applications and the accelerating deployment of AI infrastructure across industries.
According to Gartner, global data centre power demand is projected to rise from 104 gigawatts (GW) in 2025 to 132 GW in 2026, representing annual growth of approximately 27 per cent. Looking further ahead, demand is expected to reach 290 GW by 2030 as organisations continue investing in generative AI and advanced computing capabilities.
AI infrastructure driving power consumption growth
The report identifies AI-optimised servers as the primary contributor to rising electricity consumption.
Gartner estimates that AI-optimised servers will account for 31 per cent of total data centre power consumption in 2026. By 2027, electricity usage from AI-focused systems is expected to exceed that of conventional servers for the first time.
Power consumption from AI-optimised servers is forecast to increase from 95 TWh in 2025 to 175 TWh in 2026, before reaching 258 TWh in 2027. In comparison, conventional server consumption is expected to remain relatively stable, rising from 193 TWh in 2025 to 200 TWh in 2027.
The increasing density of AI workloads is also expected to drive higher energy requirements for cooling and supporting infrastructure. Gartner forecasts electricity consumption from cooling and related infrastructure systems will increase from 159 TWh in 2025 to 243 TWh by 2027.
Linglan Wang, Director Analyst at Gartner, noted that power availability is emerging as a key constraint on AI infrastructure expansion, with access to reliable electricity becoming a critical factor in determining how quickly organisations can scale AI deployments.
Growing pressure on energy infrastructure
The report highlights mounting concerns around the long-term sustainability of data centre growth as AI adoption accelerates.
Gartner estimates that total data centre electricity consumption could exceed 1,200 TWh by 2030, creating significant pressure on existing power grids and energy infrastructure.
The firm warns that electricity supply constraints could increasingly affect future data centre construction projects and limit the pace at which organisations can expand digital infrastructure.
As a result, infrastructure and operations leaders are being encouraged to focus on improving operational efficiency, securing long-term power access and investing in technologies that reduce energy consumption.
Recommended strategies include deploying high-efficiency cooling systems, improving data centre utilisation rates and expanding the use of edge computing architectures to distribute workloads more effectively.
AI boom reshaping infrastructure planning
The findings underscore a growing challenge facing technology providers, hyperscalers and enterprise IT organisations as they seek to balance AI-driven growth with energy availability and sustainability goals.
While demand for AI infrastructure continues to rise, power constraints are increasingly becoming a strategic consideration for data centre operators and cloud providers. Industry observers note that future competitiveness in the AI market may depend as much on access to energy and infrastructure as on advances in computing technology itself.
The report reflects a broader shift in enterprise infrastructure planning, where power efficiency, cooling innovation and energy resilience are becoming critical components of long-term AI and digital transformation strategies.