With cloud technology advancing at a brisk clip, artificial intelligence has become something of a double-edged sword for data centers.

Advanced algorithms are placing new demands on servers, requiring heightened power, space and cooling in data centers. But they’re also part of the solution, with software boosting efficiency and raising the bar for data center operators.

“With businesses of every size looking to leverage AI, particularly medium-sized businesses that can’t afford extensive in-house data center networks, the demand for modernized data centers will only increase,” Nucleus Research analyst Daniel Elman said. “Of course, a large part of that increase will be serviced by the largest cloud vendors ... but it leaves opportunity for other players to enter the market as well to meet the growing demand."

The rise of AI creates new challenges for data centers, including increased demand for storage, more heat to manage and distribution of resources. Keeping up can get costly, both for the data center operators themselves and for their customers.

“Companies have an expectation that data centers can handle these workloads ... the ones that can’t will lose business,” Elman said.

Historically, data center operators used standardized guidelines on how much power and cooling resources to allocate to certain workloads. Those guidelines also built in “massive safety margins,” Future Facilities Managing Director Sherman Ikemoto said.

Future Facilities develops so-called “digital twin” software for data centers, which are virtual simulations that help engineers, designers and managers to optimize space and resources.

“Data centers are a physical platform, and they have constraints,” Ikemoto said. “They’re constrained physically in their ability to support IT equipment. We’re seeing a lot of these high-powered computer systems coming in, and it’s sent a shock wave through the data center management world ... and really highlights the sustainability problem.”

As with many technologies, major companies are driving the most influential advances in AI. Google’s DeepMind division, which is focused on artificial intelligence, in 2016 delivered a breakthrough in data center technology. By applying a machine learning model to Google data centers, DeepMind reduced the energy expended on cooling technology by up to 40%.

“AI is being deployed within Google and some of the other major hyperscale data centers as a tool to help improve their energy efficiency,” said Marc Cram, director of sales at Server Technology, which supplies power solutions to data centers. “That’s one potential application, and another is as a security mechanism, allowing you to watch the traffic on the network and be able to identify what is friend or foe.”

Hyperscalers like Google, Microsoft and Amazon may have the most to gain by deploying AI in their data centers. They generate, consume and process vastly more data than the average enterprise, and their business models depend in part on how quickly, reliably and efficiently they can manage data at a global scale. That’s especially pronounced in the case of Google, which operates the world’s leading search engine, among other ubiquitous products, and processes a mind-boggling number of requests each day.

It’s likely that colocation providers will follow suit adopting similar AI tools over the next few years as demand grows for better, more efficient data centers. In the case of colocation providers, which control the physical environment but not their customers’ computing systems, AI adoption will look a bit different compared to hyperscalers. But with efficiency considered a major competitive advantage in the industry, colocation firms have a clear incentive to ensure they can pack in as much computing muscle as possible.

Despite the potential of AI, widespread adoption of new technology may also create risk for those operating and funding data centers. The rapid pace of development means there’s a chance of obsolescence over time, even for dominant colocation providers, that their customers will need less space.

For now, however, data center firms say the advent of AI is much more of a boon than a burden, and are spinning up products and partnerships responding to increased AI adoption among their customers. Opportunity abounds for suppliers of data center technology, as well.

“As AI is leveraged for more and more critical workloads, protecting the data where it is stored and analyzed will become even more essential, so look for a continued boom in automated data center, network security, and monitoring tools in the future," Elman said.