Gartner: AI servers hit 258 TWh by 2027, surpassing conventional hardware for first time
Gartner forecasts that AI-optimized server electricity consumption will reach 258 TWh in 2027, a 47.8% increase from 2026's 175 TWh projection, surpassing conventional servers' 200 TWh for the first time. Overall data center consumption is expected to hit 702 TWh in 2027, up from 565 TWh in 2026 (26% growth).
The driver is explosive AI adoption. AI-optimized servers consumed 95 TWh in 2025 and will consume 175 TWh in 2026—an 84% jump in a single year. By contrast, conventional servers grew less than 1% in 2025 and are projected to grow just 1.2% in 2026. Cooling and supporting infrastructure will add another 195 TWh of demand in 2026 alone.
Power availability is now the binding constraint on AI deployment. Gartner estimates 40% of existing AI data centers will be operationally constrained by power by 2027. More than 75 data center projects worth $130 billion were already blocked in H1 2026 due to power and water costs.
Architects must factor power cost escalation, secure long-term grid contracts, and plan for 24/7 reliability (ruling out intermittent renewables). By 2030, total data center electricity demand will exceed 1,200 TWh—roughly equivalent to Japan's entire annual consumption.
Sources
- Primary source
- gartner.com
“AI-optimized server adoption will account for 31% of data center power consumption in 2026, and that by 2027 their power consumption will surpass that of conventional servers”
- tomshardware.com
“More than 75 data center projects worth $130 billion were blocked in the first months of 2026 amid opposition over power and water costs”