Data center operators recognize that AI is the future, but also recognize that the future won’t arrive until the present is past.
This is the fourth of a six-part series focused on Uptime Institute’s Global Data Center Survey 2024.
Perhaps the biggest takeaway from looking at the numbers in the Innovation and Impact section of this year’s Global Data Center Survey from Uptime Institute is that data center operators seem to have embraced AI to the point that they’re using it to help them run their operations.
In response to a choose-all-that-apply question — “Which of the following — if any — do you consider to be benefits for using AI in your data center? — every benefit listed, except “none,” received a substantial number of votes — from 37%-58%.
“I found it interesting that 94% at least see some benefits to using AI in operations, that it’s a very small 6% that say that none of these would really be improved or be seen as benefits.” Doug Donnellan, research analyst pointed out in an online briefing Uptime sponsored to discuss the survey.
Another survey question asked for a simple yes or no response: “Would you trust artificial intelligence to make operational decisions in a data center, assuming the AI had been adequately trained with historic data?”
Fifty-eight percent gave that one a thumbs up, although the numbers for trusting AI are on a three-year decline. In 2023 64% answered yes, and in 2022 the number was 76%.
Uptime’s CTO, Chris Brown, says he thinks the declining numbers are because data center operators have learned that despite AI’s promise, it’s a technology that’s not yet ready for prime time.
“They [generative AI applications] do most things well, but there’s some places where they get it wrong, and sometimes spectacularly wrong,” he said. “People are starting to look at this in their own data centers and think, for a high availability data center that has to get everything right, one or two mistakes from a system that has control over my critical data center infrastructure could be a big problem.”
“The fact is that they’re saying it’s not necessarily as mature as they would want it to be to trust the cost associated with it, the risk associated with it, and all those things,” he added. “It’s not just the machine learning, it’s not just the AI, it’s not just the brain behind it, but it’s the actual automation part of it, the actual independent control that we’ve gotten away from in the industry that most people just aren’t comfortable with. All those things are playing together. The closer they get to having to say go and use this, the more concerned people are getting about it.”
Tomorrow: Data Center Resiliency and Outages.
Christine Hall has been a journalist since 1971. In 2001, she began writing a weekly consumer computer column and started covering Linux and FOSS in 2002 after making the switch to GNU/Linux. Follow her on Twitter: @BrideOfLinux
