Utilities Redesign Grid Planning as AI Data Centers Demand Gigawatts

Electric utilities across the United States are overhauling how they plan, price, and connect large customers as AI data centers request power at scales that exceed two decades of grid planning assumptions.

At a Thompson Hine energy forum in Cleveland, executives from utilities, grid operators, and engineering firms described an industry scrambling to accommodate facilities that can demand hundreds of megawatts—or even gigawatts—at a single location. The challenge extends far beyond generation capacity to encompass transmission lines, substations, transformers, and the interconnection studies required to integrate massive loads safely.

"The data centers are ubiquitous. They are coming," Brian Thiry, director of strategic engagement and reliability at ReliabilityFirst, said at the June event. "We're going to need all the megawatts from all the resources to power the future."

Why it matters

The shift from flat electricity demand to rapid load growth is colliding with infrastructure constraints that take years to resolve. Transformer lead times now stretch four to five years, transmission projects require extensive permitting, and existing interconnection processes were designed for incremental additions—not campuses that rival small cities in power consumption. For business leaders evaluating AI infrastructure investments, power availability has become the primary site selection criterion, often trumping traditional factors like labor markets or tax incentives.

Demand forecasts rewritten

Utilities spent years planning for minimal growth in electricity consumption. That assumption collapsed as hyperscale AI projects, electrification, and advanced manufacturing entered development pipelines simultaneously.

Thiry pointed to PJM Interconnection forecasts that have climbed sharply as utilities incorporate AI campuses into long-range models. ReliabilityFirst had already identified resource adequacy concerns tied to power plant retirements before AI demand accelerated the timeline.

"Directionally, even if we get a portion of that load forecast, it's still much more than we've seen for the last couple of decades," he said.

The bottleneck is no longer just generation. Transmission capacity, substations, and switching equipment—assets that require years to permit and install—are now the binding constraints.

Utilities adapt processes

FirstEnergy has introduced a two-stage load study process that provides developers with preliminary cost and schedule estimates earlier while filtering speculative projects before they consume engineering resources, according to Rachel Lindesmith, the utility's director of national accounts.

"The earlier the better," Lindesmith said, urging developers to begin utility conversations before finalizing site plans.

Equipment shortages are extending project timelines significantly. "Some of these transformers and breakers and things that are required take four or five years," Lindesmith noted.

FirstEnergy is also investing in larger transmission infrastructure, including a joint venture with American Electric Power to build more than 300 miles of new 765-kilovolt transmission across Ohio.

Power becomes the first question

Commercial real estate broker Terry Coyne described a fundamental shift in industrial development conversations. Developers once prioritized highway access, land availability, and workforce. Today, they lead with power capacity.

"It's all about gas and it's all about power," Coyne said.

Manufacturers are now competing with hyperscale operators for electrical capacity. Robotics, defense manufacturing, and advanced industrial projects increasingly request 20 megawatts or more—levels once reserved for only the largest facilities.

"The gold rush is kind of over," Coyne said of Ohio's data center land market. "The power is tapped out."

Behind-the-meter generation not a complete solution

While panelists discussed behind-the-meter generation and small modular reactors as necessary developments, none viewed them as complete solutions to grid constraints.

"Bringing your own generation doesn't necessarily solve all of our reliability questions," Thiry said. Even privately developed power plants must be integrated into the broader grid, requiring utilities to conduct modeling, operating studies, and transmission upgrades to manage power flows during outages or system disturbances.

ReliabilityFirst is working with the North American Electric Reliability Corporation on new standards for computational load entities—recognizing that AI campuses have become operating elements of the bulk power system rather than simply large customers.

"You are now part of the grid," Thiry said.

The discussion came one week after the Federal Energy Regulatory Commission ordered regional transmission organizations to justify or revise their large-load interconnection rules, reflecting concern that existing processes cannot keep pace with AI development.

Panelists welcomed the regulatory attention but cautioned that policy alone cannot shorten transformer lead times, accelerate transmission construction, or expand manufacturing capacity for critical equipment. Those constraints remain governed by engineering, permitting, and supply chains.

Details were first reported by Shane Snider at Data Center Knowledge.