Cost, AI, and staffing are biggest concerns for data centers
Top ways AI could benefit data center operations:
- Increased facility efficiency: 58%
- Lower risk of human error: 55%
- Increased staff productivity: 45%
- Improved IT performance: 43%
- Lower risk of equipment failure or outages: 42%
- Reduced maintenance/service costs: 37%
Yet the results also show that trust in AI has fallen for the third consecutive year; 42% of respondents in this year’s survey said they would not trust AI to make operational decisions in a data center, even assuming the AI has been adequately trained with historic data. This represents an increase from 2022, when 24% of respondents said they would not trust AI, and 2023, when 37% said they do not trust AI to make decisions.
Despite the hype around AI, the densest IT workloads supported today across data centers are primarily business applications and high-powered computing (HPC).
Workloads driving the highest density deployments in data centers include:
- Densified infrastructure for business applications: 49%
- High performance computing with accelerators: 33%
- High-performance computing without accelerators: 19%
- Generative AI training: 15%
- Generative AI inference: 8%
- AI training (other than generative): 5%
With the rise of AI workloads, data center operators are challenged to accurately predict and plan for future capacity requirements, according to Uptime.
“The million-dollar question the industry is struggling with right now is how much of this AI is there really going to be, how much bigger will the installation be, how much denser is the installation going to be. It’s something we’re working and trying to bring some clarity to, but it’s a difficult challenge,” said Chris Brown, chief technical officer at the Uptime Institute. “It’s something data center operators are going to be struggling with for the next few years.”
