Rising Data Center Costs Linked to AI Demands

Runaway demand for artificial intelligence is prompting some data center operators to raise commercial leases to cover the added costs of powering stacks of computer servers running increasingly energy-intensive workloads.

Sonal Gupta, a cloud infrastructure manager at Carlsberg AS , a Copenhagen-based brewer, blames the surge in AI for a recent price increase charged by the company’s data center provider. Advanced AI tools, he said, “require more infrastructure, which influences the overall cost,” citing the need for added processing power, advanced chips and other expensive components.

Data centers, the warehouse-size buildings that house multiple racks of servers, routers and other information-technology equipment—providing the underlying infrastructure for cloud computing—already consume massive amounts of energy. Businesses lease space in the facilities to store data and run software applications, with rates typically based on power consumption. 

Data center customers, which range from small businesses to giant cloud providers, are currently eating up power faster than operators can expand capacity. In Northern Virginia, the world’s largest data center market with more than 275 facilities, the amount of power available for lease this year shrank to 38.4 megawatts, from 46.6 megawatts a year ago, according to the latest analysis by commercial real-estate services firm CBRE Group. The declines came even as overall inventories grew 19.5% year-over-year to a total of 2,132 megawatts, CBRE said.

Growing supply constraints—pinched by the increased use of AI—are putting upward pressure on prices, said Pat Lynch, executive managing director of CBRE’s data center business. 

In the first three months of the year, data center customers in Northern Virginia paid up to $140 a month per kilowatt of power, a 7.7% increase from $130 a year ago, according to CBRE. In Silicon Valley, where the vacancy rate currently stands at a near-record low of 2.9%, top monthly rates per kilowatt climbed to a maximum of $250, up 43% from $175 in 2022, the group said.  

Along with higher construction, labor and utility costs, Lynch said, AI is fast becoming a key driver of data center bills. Global spending on AI is expected to exceed $301 billion by 2026, growing at a compound annual rate of 26.5%, according to market research company International Data Corp. 

Artificial-intelligence applications burn more energy than conventional software because they are designed to read through a far larger amount of data. Over several days, a single AI model can consume tens of thousands of kilowatt-hours. Generative AI models, such as the technology underlying OpenAI’s ChatGPT chatbot, can be up to 100 times bigger than standard AI tools.

More than half of some 500 corporate IT decision makers surveyed this year by market research firm Enterprise Technology Research said they planned to evaluate, deploy or commit further resources to ChatGPT-like technology. 

“The GPUs that are used to train generative AI models have large power draws,” said Sean Graham, IDC’s research director for cloud to edge data center trends, referring to graphic processing units helping power AI systems. 

The additional power also requires advanced hardware cooling systems that tend to be more expensive than traditional air coolers—and take up more floor space, Graham said. Sparked by AI, all of these factors are pushing up data center prices, he said.

John Dinsdale, chief analyst and research director at market research firm Synergy Research Group, said data center operators are passing the added costs of running AI applications directly on to their customers. “Any pain that is being felt falls on customers rather than the companies providing their data center facilities,” Dinsdale said.

“AI is influencing both cost dynamics and data center expansion,” said Christina Mongan, global director for innovation ecosystem and emerging technologies at IT services and consulting firm Unisys. Mongan said price hikes should prompt chief information officers and other IT leaders to review their data center strategies, while data center operators need to seek ways to better anticipate future demand and optimize capacity usage—potentially through the use of AI, she said.    

Jon Lin, executive vice president and general manager of data center services at

“The simple economics of supply and demand, coupled with a rising interest-rate environment, has naturally led to increased prices,” Lin said. “Yes, prices are going up.”

Though he believes the shift by businesses to digital technology more broadly is pushing up prices, “AI has come up in almost every conversation I’ve had with CIOs and other enterprise tech leaders over the past couple of quarters,” Lin said. “Generative AI has created huge interest in the ways companies can leverage AI and machine learning,” he said, “but it still feels early in the cycle.”

For corporate technology leaders, rising data center costs are also likely to spill over into higher cloud-computing bills—since many cloud providers also lease space in data centers, said Tracy Woo, senior analyst at IT research firm

“By virtue of generative AI workloads requiring much more compute, and more widely affecting energy efficiency and cooling in the data center, undoubtedly cloud costs will continue to rise as more companies adopt generative AI,” Woo said. Cloud providers operate their own data centers, while also renting space from data center landlords, known as colocators.

“For now, we haven’t seen a huge increase in costs due to generative AI,” Woo said. “But for the future, that will likely change as the technology becomes a standard part of every company’s tool set.”

Write to Angus Loten at [email protected]