Geography of U.S. Data Center Deployment: An OnLocation Energy Horizons Report

The latest in OnLocation’s Energy Horizons report series sheds new light on the transformation occurring in the U.S. energy system. The transformation is driven in part by unprecedented growth of data centers for artificial intelligence (AI) training and user applications. OnLocation, a division of KeyLogic, produced Geography of Data Center Deployment in the U.S. to provide insights into the evolving power grid response to AI data center growth and Administration policy changes. It also explores several critical decision factors used by developers when siting data centers. The report features updated and timely scenario analysis using OnLocation’s revised version of the National Energy Modeling System (OL25-NEMS) and focuses on two scenarios illustrating the range of uncertainty in future demands.

Prior reports in the Energy Horizons series include the September 2024 Data Center Demands & Impacts on the Energy System and the March 2025 Energy Horizons Data Center Report Update: Spotlight on Policy Impacts. The September 2024 report provided a first look at how data center demands may impact the future of the U.S. energy system, and the March 2025 report focused on an alternative policy scenario for a snapshot of potential power grid responses under the new Administration’s priorities.

What is the Motivation for Releasing the New Data Center Report?

Deployment, especially for hyperscale AI data centers, continues to be a driving force in electricity demand growth. Our latest report highlights the significance of three factors that may alter the future electricity generation mix:

• The expected pace of growth. Significant investment in multiple energy sources will be essential for meeting the growing energy demand for data centers, including conventional fossil and nuclear generation, and advanced nuclear, renewables, and battery storage, and may require investments in microgrids.

• Evolving Policy Landscape. Congressional legislation such as the One Big Beautiful Bill Act (OBBBA), along with the expected repeal or substantial modification of EPA greenhouse gas regulations in the power and transportation sectors, will influence capacity investments and utilization of various energy sources. In addition, state & local policies that offer business-friendly incentives to attract new AI data centers, along with other factors, will influence the location of new centers.

• Regional decision factors will drive the locations that developers select for deployment of hyperscale data centers. Priority factors include Power Availability and Affordability, Favorable Policy and Business Environment, Infrastructure Readiness, Workforce Accessibility, and Environmental Resilience. Different regions of the country have access to different energy resources, so the location of growth matters.

What Does the Report Reveal?

OnLocation’s research and analysis identified the most favorable regions for new data center development by creating a ranking based on the key decision factors.

Figure 1: U.S. Map of Most Favorable Electricity Regions for Data Centers

As shown in the map, our analysis determined that the top five electric reliability regions that stand to gain the most load growth from data centers, shown in dark green on the map, are: Texas Reliability Entity, PJM/West, SERC/Southeast, Midcontinent ISO/West, and Northwest Power Pool. While these regions appear to be most favorable for development, all states and regions are likely to experience new development in the next several years. The rankings were used to inform our scenario analysis, with the highest-ranking regions assumed to have the highest future growth rates. Based on recent news reports, the top regions have seen increasing interest from developers in recent months, which provides some validation for our findings. However, this ranking is a snapshot of current circumstances and the importance of each critical decision factor will continue to evolve.

Two OnLocation scenarios, OL High Data Center Growth and OL Low Data Center Growth, were created using the OnLocation’s version of NEMS (OL25-NEMS). Both scenarios include policy changes reflecting the new Administration priorities, updated regional representation of existing centers, and applied regional growth rates for new data center deployment.

Figure 2: National Electricity Generation by Source and Scenario

The High Growth scenario assumes nearly 1800 TWh of growth in data center demand over the next 25 years, while the Low Growth scenario includes half of that, about 900 TWh.

As shown in the National Electricity Generation chart, demand in both scenarios leads to increasing generation from low-cost renewables and natural gas. Many hyperscalers continue to prefer low carbon power where available due to their sustainability goals even though speed to power tends to be the more dominant driver.

However, there are important differences between the scenarios:

• The Low Growth scenario shows a continued preference for new renewables, especially wind and solar, to meet increasing demands.

• The High Growth scenario relies more on gas and coal generation to meet the growing demand, including increased use of both new and existing natural gas generation over time, in addition to adding new renewable generation.

The report also highlights select regional results (ERCOT, PJM/West) that show differences in the regional response to growth including the choice of technologies for capacity expansion and changes in regional electricity trade.

Conclusions & Further Research

This new report underscores the expected significant growth in energy demand driven by the growing usage of artificial intelligence, presenting a critical challenge for the national energy infrastructure. Meeting this demand will require investment in a variety of technologies including conventional fossil fuels, conventional & advanced nuclear power, renewables, and battery storage to maintain grid reliability. The evolving policy landscape and the geography of new development will impact the future generation mix. The growth of cryptocurrency centers and other large loads, which have different determining factors for location and operating characteristics, requires further study to develop a comprehensive view of future power sector requirements.

This analysis is a first step in assessing the future deployment of data centers in the U.S. and the interrelationship between the energy and IT sectors. Additional priority analyses include:

• Are we overbuilding or underbuilding for AI demand?
• How will resource availability (labor, land, water) change regional data center deployment? 
• How will data center growth affect grid reliability?
• What is the impact on electricity prices from data center energy demand?
• Will microgrids become an attractive solution for powering data centers, improving reliability and security?
• How will cryptocurrency data center demands evolve?
• How will new policies & regulations change as more data centers are developed?