SACRAMENTO, the United States, June 13 (Xinhua) -- As the United States races to expand data centers needed for artificial intelligence (AI), a growing number of towns, states and voters are moving to slow, restrict or reshape projects over concerns about electricity, water, rising costs and local control.
The latest example came in Arizona, where lawmakers moved to make data center developers pay more in state taxes by suspending a sales tax exemption for three years, Axios reported Wednesday. The measure is expected to save the state about 57 million U.S. dollars through 2029.
The fight over AI in the United States is no longer confined to laboratories, semiconductor controls or corporate investment plans. It is moving into county hearings, utility regulatory cases, water districts and local referendums.
Data centers, once seen mainly as technical facilities, are becoming visible industrial infrastructure. Residents and officials across political lines are asking who benefits from the AI boom and who pays for electricity, water, land and public incentives.
A U.S. Department of Energy-backed report by Lawrence Berkeley National Laboratory estimated that data centers consumed about 4.4 percent of total U.S. electricity in 2023, with the share projected to rise to 6.7-12 percent by 2028.
Goldman Sachs Research said U.S. data center power demand could rise from about 31 gigawatts in 2025 to 66 gigawatts in 2027. It also estimated that only 50-60 percent of the planned capacity for the next one to two years may come online on time due to delays and cancellations.
The U.S. Federal Energy Regulatory Commission, which regulates interstate electricity transmission, has launched a proceeding on how large electricity users, including data centers above about 20 megawatts, connect to the grid and how costs should be allocated.
Public concern is broad. A recent Gallup poll found 71 percent of Americans oppose building an AI data center in their local area, while a Pew Research Center survey published in March found more negative than positive views regarding such facilities' environmental impact, energy costs, and local quality of life.
Resistance is taking different forms across the country. In California, voters in Monterey Park, a prominent city in Los Angeles County, approved a measure banning data centers earlier this month. In Seattle, the city council on Tuesday approved a one-year moratorium on new data centers, citing infrastructure, water use, energy demand, land use, and economic impacts.
In Utah, the proposed Stratos data center project in Box Elder County was scaled down due to concerns over water rights, governance, electricity costs, and local opposition.
In New York state, lawmakers passed a one-year moratorium on large data centers above about 20 megawatts, pending a decision by Governor Kathy Hochul, alongside studies on electricity, water, land use, pollution, and ratepayer protections.
Water is a central issue in western states. Reports note that data centers are expanding into water-stressed areas like California, while disclosure rules make it difficult to determine actual water use.
The California Little Hoover Commission warned that AI data centers could raise electricity bills and urged that technology companies, rather than households, cover grid upgrade costs.
Supporters say data centers are essential for the AI economy, generating construction jobs, tax revenue, and technological capacity. Opponents argue many projects rely on public incentives, strain electricity and water systems, and create relatively few permanent jobs.
The conflict is also about transparency. In several states, residents and lawmakers say projects are often negotiated without full public disclosure of power demand, water use, and infrastructure impacts.
For the United States, the dispute is becoming a practical constraint on AI expansion. Future growth depends not only on chips, capital, and software, but also on power, water, permits, grid access, and local consent.
U.S. federal authorities increasingly treat data centers as strategic infrastructure for AI and national security. However, many communities view them as heavy industrial facilities that may affect housing, farms, utilities, and household costs. ■