TRG发布美国24州AI数据中心就绪度排名弗吉尼亚州居首

Aerial view of data centers in Northern Virginia

Aerial view of a dense cluster of data centers, including facilities under construction, in Loudoun County’s Data Center Alley in Northern Virginia.Getty Images

Power supply and water availability could help states outside established data center hubs compete for AI infrastructure investment, according to a new survey of US states from TRG DataCenters. The report, published in August 2026, compares 24 states using measures of infrastructure scale and environmental exposure and points to a more geographically diverse market as AI demand expands.

Virginia topped the August ranking of states positioned for AI data center growth, but the findings also suggest smaller markets can compete by offering a favorable combination of electricity supply, lower-carbon generation, and water availability.

“When planning infrastructure for the long term, that balance between scale and resource efficiency matters as much as raw megawatts,” Chris Hinkle, CEO of TRG DataCenters, told Data Center Knowledge.

TRG’s analysis considered mapped large-scale AI data center sites, power capacity, electricity use, estimated annual carbon emissions, and the proportion of mapped facilities in areas of high water stress. Capacity figures reflect the AI infrastructure included in TRG’s dataset and do not represent each state’s total data center market.

Virginia’s Lead Comes with Water Constraints

Virginia received a readiness score of 100, followed by Ohio at 88.2, Tennessee at 83.7, Georgia at 81.8, and Texas at 80.8. TRG’s dataset mapped seven large AI sites in Virginia totaling 1,085 MW of capacity, while Ohio and Tennessee showed higher mapped capacities – 1,410 MW in Ohio and 1,373 MW in Tennessee. Virginia’s advantage reflects performance across several measures, including lower estimated grid carbon intensity serving the mapped facilities.

Hinkle said Ohio’s mapped AI capacity was around 30% greater, but the grid serving Virginia’s facilities had roughly one-third lower carbon intensity than Ohio’s. “As a result, the AI facilities in our dataset are associated with around 2.3 million tons of CO2 emissions annually, compared with roughly 4.5 million tons in Ohio and 4.3 million in Tennessee,” he said.

Water – or lack thereof – is a complicating factor. About one-third of Virginia’s mapped data centers are located in areas of high water stress, compared with none in Ohio and just 7% in Tennessee. “One factor Virginia will need to continue managing as the market expands is site selection around water availability,” Hinkle said. Those differences make the ranking a starting point for evaluating markets rather than evidence that development conditions are uniform within a state.

Smaller Markets Offer Competing Advantages

South Carolina placed sixth with a score of 80.5, just behind Texas, despite having only one large AI site and 142 MW of capacity in the study. “What surprised us most was South Carolina,” Hinkle said. “It has just one large AI data center site and around 142 MW of capacity, yet it finished almost level with Texas in our overall readiness ranking, despite Texas having 13 large sites.” He attributed that performance partly to the state’s nuclear-heavy electricity supply and the absence of mapped facilities in areas of high water stress.

Hinkle also pointed to New York, Iowa, and Mississippi. New York combined relatively limited large-scale AI capacity with one of the lowest-carbon grids in the dataset, while Mississippi already supported more than 500 MW of AI capacity. “States that can combine reliable power with a relatively low-carbon electricity mix and favorable water conditions could become increasingly attractive as demand expands,” he said.

Texas Faces a Siting Challenge

Texas had the largest number of AI sites in the study – 13, with a combined capacity of 1,301 MW. However, 58.8% of those facilities were located in areas of high water stress. Hinkle said he expects those conditions to influence cooling choices and project locations. “That is likely to encourage more sophisticated siting and infrastructure strategies rather than simply slowing development,” he said. “We could see greater use of approaches such as closed-loop or air-based cooling, recycled water, and locations where local water conditions are more favorable.”

TRG notes that the water-stress measure describes facility locations and does not establish actual water consumption or the impact of individual cooling systems. Developers still need to assess those conditions at the project level.

From Readiness to Deliverable Infrastructure

Statewide advantages must also translate into infrastructure that developers can use. A relatively small share of statewide electricity use does not, by itself, establish that a particular site has power available for expansion.

“Grid connections, transmission capacity, fiber connectivity, build-ready sites, and long-term power availability can all influence where major projects ultimately land,” Hinkle said. For emerging markets, those requirements determine whether attractive resource conditions can support projects at scale. For established hubs, the requirements shape where additional development can proceed.

“The result could be a more geographically diverse AI infrastructure market as developers match projects with the locations best equipped to support them.”

About the Author

Contributor

Nathan Eddy covers data center trends and technologies across multiple industries. A graduate of Northwestern University’s Medill School of Journalism, he is also a documentary filmmaker specializing in architecture and urban planning. He currently lives in Berlin, Germany.