- AI data center power consumption will soon surpass that of conventional data centers
- Global data center energy consumption has increased by 26% from 2025
- In the US, AI data centers account for 36% of all data center energy consumption
Data centers built to add capacity for AI will soon consume more energy than conventional data center hardware. With consumption of 175 terawatt-hours (TWh) in 2026, predictions raise that figure to 258 TWh in 2027, at which point AI-optimized data centers will surpass conventional data centers in terms of consumption.
Compared to 2025, the energy consumption of AI data centers has increased by about 84%. These are the findings of a Gartner forecast that also predicts that AI-optimized servers will account for 31% of data center energy consumption by 2026.
This level of consumption, combined with an excess of electricity demand compared to production, will be the main limitation to the future expansion of AI, Gartner predicts. But total data center consumption is expected to be 565 TWh in 2026, an increase of 26% year over year.
Consumption increases, but capacity expands slowly
When it comes to global consumption, the United States accounts for 36% or around 204 TWh of the total 565 TWh of global demand. Within that portion of US demand, AI data centers account for a third, and expected consumption by 2026 will be around 68 TWh.
“Rising demand for compute-intensive AI workloads is driving unprecedented growth in data center power, while AI capacity is now limited by power availability, making data center power security the new battleground for scaling and protecting margins in the global AI race,” said Linglan Wang, direct analyst at Gartner.
By 2030, Garter predicts that supply will not be able to meet demand once consumption exceeds the 1,200 TWh mark.
To address this limitation, Wang suggested that business leaders and infrastructure providers should focus on improving the efficiency of power grids and hardware that consumes the most energy, such as cooling systems.
A recent Google Cloud report further suggested that to address the rising cost of energy consumption, enterprises should move from running AI models in a centralized cloud to deployments at the edge, where the efficiency of these systems is increased with the added benefit of avoiding a global disruption of services if the centralized cloud system fails.
But these levels of increased consumption will hardly help cool growing anti-data center sentiment in the US, which, combined with hardware and power production shortages, has seen nearly half of all data centers delayed or canceled by 2026.
Through Tom Hardware
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds.
