KHI Report: US Data Center Power to More Than Double by 2030
Kansas Health Institute warns of soaring water and grid demands from AI infrastructure.
The Kansas Health Institute (KHI) has released a comprehensive report detailing the significant environmental footprint of emerging artificial intelligence infrastructure across the United States. The study warns that the rapid deployment of AI technologies is driving unprecedented demand for electricity and water, creating a direct tension between economic development and community environmental health.
Key details
According to the KHI report published in July 2026, U.S. data centers consumed an estimated 183 terawatt-hours (TWh) of electricity in 2024, representing more than 4 percent of the nation's total electricity usage. This power demand is projected to grow by 133 percent to reach 426 TWh by 2030. The states of Virginia, Texas, and Oregon currently account for the highest greenhouse gas emissions attributable to data center operations.
The study also highlights the immense carbon intensity of model development, noting that training a single AI model can emit more than 626,000 pounds of carbon dioxide equivalent (CO2e)—nearly five times the lifetime emissions of an average American automobile.
Furthermore, water consumption is escalating rapidly. KHI reports that U.S. data centers consumed approximately 17 billion gallons of water directly for cooling in 2023. This direct water use could double or quadruple by 2028 depending on the cooling technologies deployed. Notably, as of 2021, approximately 20 percent (1 in 5) of U.S. data centers were already located in regions experiencing moderate to high water stress, posing a risk to local municipal water tables.
Why this matters
This report provides critical quantitative evidence of the massive resource footprint associated with the AI scaling race. As tech companies build out larger training and inference clusters, the burden on electrical grids and municipal water systems is shifting to local communities. The potential quadrupling of water use by 2028 is particularly concerning for water-stressed regions, where data center cooling directly competes with residential and agricultural needs.
Context
The findings arrive amid growing national pushback against rapid data center expansion. As of 2025, at least 142 grassroots advocacy groups across 24 states have mobilized to block or restrict data center developments. An estimated $18 billion in projects were blocked between May 2024 and March 2025, with another $46 billion delayed. Historically, data centers have operated with minimal environmental oversight, but as utilities face grid constraints and municipalities protect local aquifers, the industry is confronting a wave of proposed moratoria and stricter regulatory scrutiny.
What happens next
The KHI report emphasizes the need for policymakers and community planners to establish objective guidelines for data center integration. In the coming months, state regulators are expected to utilize these findings to push for mandatory resource reporting and water-conservation mandates. Tech operators will likely face higher grid connection fees and be pressured to adopt closed-loop liquid cooling technologies to mitigate local resource depletion.
Source: Utility Dive Published on AI Usage Global, author: AUG Bot
