A forecast says global data center power consumption will increase by nearly 30% from a year earlier this year.
On the 12th, market research firm Gartner projected that global data center power consumption will rise 26% from 447 terawatt-hours (TWh) last year to 565 TWh this year. By 2030, it is expected to exceed 1,200 TWh, and Gartner assessed that mounting power constraints at data centers have emerged as the biggest variable for the spread of AI.
Linglan Wang, a senior analyst at Gartner, said, "As demand surges for computation-intensive artificial intelligence (AI) workloads, data center power consumption is expanding to a level different from before," adding, "With AI capabilities now constrained by power availability, securing power for data centers is becoming a new battleground for scale expansion and profitability in the global AI race."
Gartner also forecast that global data center power demand this year will reach 132 gigawatts (GW), up 27% from 104 GW last year. By 2030, it is expected to reach 290 GW. Gartner analyzed, "This shows that the spread of Generative AI is adding a long-term burden across data center infrastructure and power operations."
It identified AI-optimized servers as the key driver of power consumption. Gartner expected AI-optimized servers to account for 31% of data center power consumption this year. It predicted that next year the power consumption of AI-optimized servers will surpass that of existing servers.
Gartner diagnosed that as power grid supply shortages worsen relative to data center construction demand, all data center users will be directly affected.
Wang said, "The top priority for corporations will be improving efficiency and securing reliable access to the power grid," adding, "To ease power constraints and enable sustainable, scalable growth, they should also invest in high-efficiency cooling systems and Edge Computing."