Data center operators are increasingly concerned about future access to electricity amid the increasing historic demand driven by the AI race.
Concerns about data center capacity or power needs planning have increased 9% since 2023. The Uptime Institute was found in an annual survey of data center owners and operators. Industry advice said it considers Jump “important.”
The survey found that rising costs for data center operators this year, like in 2024, is a major concern, but concerns about power sources are quickly catching up.
Thirty-eight percent of survey respondents said they were “very concerned” about rising costs, 36% said they were “very concerned” about electricity constraints, and the same number of respondents said they were “very concerned” about demand forecasts.
According to the International Energy Agency, electricity demand from data centers around the world has doubled by 2030.
In the US, large tech companies spend hundreds of billions on the infrastructure needed to deploy and scale AI. Amazon, Microsoft, Meta, Google and Apple are projected to collectively spend more than $350 billion on data centers in 2025.
Last week, four of the five reported quarterly revenue, each told investors they would spend more than the initial forecast explained.
The rising costs caused by inflation, labor shortages and rising energy prices can be resolved, but data center operators do not control the electrical grid.
Related Stories
Business Insider tells the innovative stories you want to know
Business Insider tells the innovative stories you want to know
According to CEO Andy Jassy, it is the biggest constraint that Amazon faces as AI grows its AI business.
Energy efficiency attempts do not help ease data center power constraints, the Uptime study found.
As AI demands more power, industry power usage (the metric used to track facilities' energy efficiency levels) has been stagnating for the past six years.
While new cooling technologies for data centers continue to emerge, studies have found that older facilities and “local-specific barriers” are hampering significant advances in energy efficiency.
Data center operators don't fully employ AI when they help them use technology to manage their facilities.
“The trust in AI for data center operations depends on use cases. Most operators can use it to analyze sensor data and predictive maintenance tasks, but not configuration changes, equipment control, or administrator money,” the study states.
Do you have any tips for this reporter? Contact Ellen Thomas Ethomas@insider.com.
