Pressure rises to bolster data center energy efficiency as IT leaders worry demand is pushing the grid to breaking point
IT leaders are increasingly concerned about both data center energy efficiency and supply, with the rapid expansion of AI putting strain on national infrastructure.
Research from Cadence found that three-quarters of data centers face increased pressure from AI-driven demands. As a result, seven-in-ten IT decision makers warn that the national grid is being pushed to breaking point.
“AI is driving increased capacity demands,” said Mark Fenton, Cadence’s product engineering director.
“However, the reality is that its current footprint is relatively modest compared to its future size. This will change as adoption grows, and, in tandem, a rising percentage of storage is brought on-premises for regulatory reasons and due to privacy concerns.”
With three-quarters of the decision makers saying data centers are under increased pressure from AI, 88% are actively working to enhance energy efficiency. However, just three-in-ten believe that they’re doing enough.
Hyperscale and colocation data centers are rather more confident about their energy efficiency, with 42% and 38% of providers respectively believing they’re doing enough compared with just three-in-ten enterprise data centers.
“This reflects the fact that most hyperscalers have already announced carbon neutral targets for 2025 or 2030 — goals they intend to reach through offsetting emissions, improving energy efficiency, and embracing sustainable energy sources,” said Fenton.
“It’s irrefutable that they have greater funds to pursue these measures, but their efforts do illustrate what’s needed across all facility types to make the industry more sustainable.”
Data center transformation faces challenges
Eight-in-ten decision makers said their organization has the capabilities to capitalize on the technologies that could transform their data centers, the study found.
However, this may have been a bit of an exaggeration. While three-quarters are using AI and machine learning, only 63% are using renewable energy sources, with 45% using liquid cooling and 42% digital twins.
Cadence also found there are significant hurdles to innovation, with 42% of decision makers citing the cost of implementing new technologies as the biggest barrier.
Other obstacles include a lack of skilled staff, incompatibility with legacy systems, and uncertainty about future technologies, all of which are affecting more than three-in-ten.
AI is increasingly being used within data centers, with six-in-ten using it for fault detection and natural language assistants, for example. More than half are using it for demand forecasting and automating capacity management.
But it’s not all positive, with 10% of decision-makers say that they had been using AI but aren’t anymore.
“AI is expensive, complex, and used improperly may not deliver actual value, especially in small-scale applications. The key is understanding what to ask of it to attain meaningful outcomes,” said Fenton.
“When it comes to complex problems, such as training AI to assist with capacity fulfilment, it can take years of research to identify the right questions.”
Digital twins are attracting increasing interest, with three-quarters of decision-makers saying they’re a game changer for driving technological innovation in data centers.
Those that are using them already are even more enthusiastic, with 81% saying the same. And 21% of decision-makers who aren’t currently using digital twins said they plan to introduce them in the next 12 months.
MORE FROM ITPRO
Source link
