The mass-scale buildout of artificial intelligence infrastructure has met something the hyperscalers were not fully prepared for: their own neighbors.
A recent survey conducted by real estate firm Redfin found that 47% of Americans now oppose the construction of an AI data center near their home — a higher rate of opposition than for new apartment buildings (37%) or mixed-use developments (31%). The number marks one of the more striking inversions of US development politics in recent memory. Single-family homeowners, traditionally the most active opposition force against new construction, are now more hostile to data centers than to the housing density that has dominated zoning fights for two decades.
The shift is not happening in isolation. According to public tracking compiled by industry researcher Robert Bryce, the cumulative tally of US data center rejections and restrictions reached 68 facilities as of March 31, 2026. That number has climbed steadily through the year as state legislatures, county boards, and grassroots coalitions have absorbed the same data the hyperscalers themselves are looking at: the US now hosts more than 4,200 data centers, per Data Center Map, with Virginia alone accounting for more than 600.
The Pocketbook Problem
What turned an abstract tech story into a national debate is the electricity bill. Concentrated demand from AI training and inference workloads has begun showing up on residential utility statements across multiple states, and state regulators are now responding.
Florida Governor Ron DeSantis signed SB 484 earlier this month, directing state regulators to ensure that large data center customers pay the full costs of required electrical infrastructure, transmission upgrades, and system expansions rather than shifting those expenses onto residential ratepayers. DeSantis framed the bill as a fairness issue, arguing that households should not pay “one more red cent” to subsidize what he called “the most wealthy companies in the history of the world.”
Maryland’s Office of People’s Counsel — an independent agency representing residential electricity consumers — has filed a complaint with the Federal Energy Regulatory Commission arguing that PJM Interconnection’s transmission cost allocation framework leaves Maryland customers paying for projects largely driven by data center demand, including projects outside the state. The agency estimates that Maryland residents could face a collective $1.6 billion increase in energy bills over the next decade tied to grid costs feeding data centers.
The North American Electric Reliability Corporation (NERC) issued a rare Level 3 Essential Action Alert in May, warning that the surge in large computational loads associated with data centers poses immediate risks to the bulk power system. NERC said it has observed customer-initiated large load reductions and significant oscillations occurring within seconds — events that leave little or no time for real-time intervention and could threaten grid reliability across multiple regions.
Local Flashpoints Multiply
The pushback is no longer confined to spreadsheets and regulatory filings. In Kenilworth, New Jersey, residents rallied against a $1.8 billion, 400,000-square-foot facility under construction near the Garden State Parkway. ABC7 New York reported that residents arrived at a community meeting with cowbells and whistles, saying the project came as a surprise. The developer, CoreWeave, issued a statement describing data centers as “investments in the communities that host them” and saying the facility is designed to be a “responsible neighbor.”
In Andover Township, New Jersey, opposition turned physical. NJ1015 reported that a public meeting about a potential AI data center at a former Newton airport property resulted in a resident being pushed to the ground by police officers during removal, captured on video. The dispute came amid fears that the facility could strain power and water resources.
Water has become its own flashpoint. In Fayetteville, Georgia, residents began raising alarms after noticing unusually low water pressure. A county utility investigation ultimately uncovered two industrial-scale water hookups feeding a data center campus south of Atlanta, according to Politico. Developer Quality Technology Services (QTS) ultimately owed nearly $150,000 for more than 29 million gallons of unaccounted-for water use. QTS disputed the framing of the incident, attributing the metering issues to a procedural error during the county’s smart meter conversion and a separate spike to temporary construction activities, and paid all retroactive charges.
Federal Action Enters the Picture
The AI Data Center Moratorium Act, introduced in late March 2026 by Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez, would temporarily pause new AI data center construction at the federal level to allow lawmakers and communities time to address infrastructure, environmental, and utility impacts. The bill is unlikely to pass in the current Congress, but its introduction signaled that data center politics had crossed the threshold from regional zoning issue to national policy debate.
The Brookings Institution warned in January that backlash could halt US AI infrastructure expansion entirely absent structured community benefit agreements — frameworks under which developers commit to specific local investments, hiring targets, or utility cost protections in exchange for project approval.
How Hyperscalers Are Responding
The industry has begun adapting. Major operators including Amazon, Microsoft, Google, and Meta have shifted toward longer community engagement cycles, more transparent water and power impact disclosures, and pre-negotiated rate structures with utilities. Some developers are moving cautiously in response to community backlash, with one operator publicly delaying construction milestones to accommodate additional regulatory review.
The strategic calculation for the hyperscalers is now clear. Continuing to operate on the assumption that data center siting is a routine industrial real estate transaction has stopped working. The companies that scale AI infrastructure successfully through the second half of 2026 will be the ones that treat community relations and utility cost-sharing as core operational priorities rather than public affairs afterthoughts.
The opposition is not going away. Whether the buildout slows depends on how quickly the industry adjusts.
