Utilities face challenges, opportunities from AI-driven data center power demand growth: report

Global power consumption from data centers was roughly 460 TWh in 2022 and could double by 2026 to more than 1,000 TWh, approximately equal to Japan's total electricity use, offering utilities risks and opportunities for efficiency and credit improvement, Morningstar analysts said April 1.

Specifically, growing power demand will "stress utility company balance sheets as capital spending escalates to upgrade the infrastructure and integrate renewable energy resources," the analysts said in the commentary titled "Watts Up With AI: Strategies for Utilities in an Era of Surging Demand."

"However, this also presents an opportunity for improved credit profiles if managed with forward-looking investments and cost-efficient strategies," the analysts said.

The 2023 total power capacity of the top ten largest data center markets globally was led by Northern Virginia with over 2,500 MW of capacity, according to the report. Beijing was the second largest market with a little below 2,000 MW of capacity.

Power demand from operational and currently planned data centers in US power markets is expected to total about 30,694 MW once all the planned data centers are operational, according to analysis of data from 451 Research, which is part of S&P Global Market Intelligence. Investor-owned utilities are set to supply 20,619 MW of that capacity.

Dominion Energy serves the largest data center market in the world in Loudoun County, Virginia, about 30 miles west of Washington. The Richmond, Virginia-headquartered investor-owned utility has pointed out that electricity demand from data centers in Virginia increased by about 500% from 2013 to 2022.

Since 2019, 81 data centers with a combined capacity of 3.5 GW have connected to Dominion's power system, the utility said in a presentation to Mid-Atlantic grid operator PJM Interconnection.

In the summer of 2023, Dominion told data center operators about delays in powering new facilities because of the constraints in the utility's transmission infrastructure. The delay was "directly caused" by a rapid surge in power demand from data centers, "overwhelming the grid's capacity to accommodate new connections," Morningstar said.

Challenges, opportunities for utilities

That example highlights the need for utilities to expand their transmission and distribution system networks to accommodate higher power demand from data centers increasingly being used for AI applications, but getting regulators to approve those power system upgrades could be a challenge.

"Ratepayers will have to incur higher electricity rates solely to accommodate the growth of AI without receiving any direct benefits from the higher rates," the Morningstar analysts said.

Additionally, the power intensive nature of training machine learning models used for AI has been increasing greenhouse gas emissions levels. "We expect that, without improvements in energy efficiency, there will be a substantial escalation in emissions attributable to the use of AI," Morningstar said, adding that in response to heightened power demand driven by AI, utilities may be compelled to re-engage the use of fossil fuels for power generation, "which would be antithetical to achieving net zero emissions."

Power grid pressure and environmental impacts can be alleviated by developing on-site renewable energy generation capabilities.

For example, Apple's data center in Maiden, North Carolina, is powered by the Maiden Solar Park, which is an on-site solar project with 58 MW of capacity, according to the report.

Additionally, power purchase agreements, or PPAs, will play an important role in mitigating the challenges of high power demand, by allowing data centers to lock in their electricity consumption or power pricing, ensuring more predictable and stable demand on the grid, the analysts said.

The proliferation of AI and data centers could be transformational for the utility industry, bringing opportunities and challenges, Morningstar predicted.

For utilities with data centers in their service territories, proactively investing in renewable energy power developments and grid modernization could provide "operational or financial improvements," driven by increased efficiency and grid stability along with higher revenue from greater power demand. However, failing to adapt may bring greater operational risks, potential losses from grid congestion, and transition risk over time due to the increased GHG emissions from powering data centers which could negatively affect credit profiles, the report said.