The New Neighbors Nobody Asked For
Communities are paying the price of the data center boom
It started with a sound Scott Montgomery could not identify.
From the back porch of his home in Vineland, NJ, a low mechanical hum drifted constantly across the surrounding farm fields. His son was confused. “He kept asking, ‘What are those helicopter noises in the background?’” Montgomery said. “And it’s just gotten worse over time.”
About a half mile away, construction crews are building what will eventually become one of the largest AI data center campuses in New Jersey — a facility of 2.6 million square feet, tied to a reported $17 billion Microsoft computing contract. The project is nowhere near complete. Yet the cooling systems designed to keep thousands of AI servers from overheating already run loudly enough that neighbors hear them day and night.
“It’s not even fully built yet,” Montgomery said. “That’s the scary part.”
The technology industry likes to think of artificial intelligence as part of “the cloud” — shorthand for the vast networks of remote servers and data centers that store information and run online services. The cloud was once considered frictionless and virtual, but generative AI dramatically changed the scale of demand. It requires enormous clusters of advanced chips to process enormous amounts of data, train AI models on billions of pieces of text and images, and handle millions of user requests simultaneously. This dwarfs earlier generations of cloud computing and requires a massive physical infrastructure: warehouses the size of small towns, industrial cooling systems, diesel generators, substations, and transmission lines.
A single modern “hyperscale” facility can consume as much electricity as 100,000 homes. A sprawling Meta campus in Louisiana recently expanded to 3,650 acres — roughly four times the size of Central Park — and is expected to require roughly triple the electricity consumption of the entire city of New Orleans.
And what makes the situation especially maddening for many residents is that by the time they realize a data center is coming, the approvals are often already complete. Deals are negotiated behind the scenes between developers, utilities, and local governments. By the time the first public hearings happen, many communities discover the real debate is already over.
Increasing resource demands
That dynamic is now unfolding nationwide as tech companies race to build the infrastructure needed to power generative AI systems like OpenAI’s ChatGPT. According to projections from Bloom Energy, US data center electricity demand could nearly double by 2028, reaching 150 gigawatts — the equivalent of adding a country roughly the size of Spain to the American power grid in just a few years.
More than 3,900 data centers already operate in the United States, with nearly 1,500 more planned or under construction. They tend to cluster where land is inexpensive, electricity is available, and local governments are eager for investment. Frequently, that means smaller communities with limited leverage against companies whose market values exceed the GDP of many nations.
In Fayetteville, GA, residents in one subdivision began noticing unusually weak water pressure. County officials eventually discovered that a nearby data center operated by Quality Technology Services (QTS), owned by private equity giant Blackstone, had consumed more than 29 million gallons of water that hadn’t been paid for. An industrial-scale hookup feeding the data center had reportedly been installed without the utility company fully realizing it, and another connection was not properly tied to QTS’s billing account. The county later issued a retroactive bill for roughly $147,000, which the company paid. No fine was imposed.
The controversy became public after residents learned the county had simultaneously urged homeowners to conserve water during drought conditions.
“We get this notification saying you need to stop watering your lawns,” local attorney James Clifton said after posting the utility letter publicly. “Meanwhile, QTS is just absolutely draining us.”
County officials defended their handling of the situation partly on economic grounds.
“They’re our largest customer, and we have to be partners,” said Water System Director Vanessa Tigert. “It’s called customer service.”
The collision between AI infrastructure and water access is now playing out across the country. A Bloomberg investigation found that roughly two-thirds of data centers built or in development since 2022 are located in water-stressed areas.
In Laredo, TX — already under severe drought restrictions and projected to face long-term water shortages — developers announced plans to build a huge server farm that would require five gigawatts of electricity, enough to power several million homes.
A few miles away in the community of La Presa, near the Texas border, residents still haul water to their homes manually. Luz Castillo has spent decades transporting water to her house in a 500-gallon tank in the back of a pickup truck while waiting for permanent hookups that have yet to arrive.
“Money moves mountains,” she said, describing how quickly developers could secure approval for data center projects while local homes still lacked reliable water.
Losing power
Electricity may prove even more contentious.
Residential electricity prices rose nationally by more than 7% in 2025, outpacing inflation. In regions with heavy concentrations of data centers, costs have climbed far faster as utilities have built transmission lines and generating capacity to accommodate AI demand.
For longtime residents of Northern Virginia — already one of the world’s largest data center corridors — the changes have shown up in monthly utility bills.
John Steinbach, who has lived in Manassas for decades, said his electricity bill nearly tripled this winter. “They’re building them like it’s Field of Dreams,” he said. “Build it and the electricity will come.”
In some places, communities fear not just paying higher prices but losing reliable power access altogether.
Residents in California’s Lake Tahoe region recently learned that their longtime electricity provider, NV Energy, plans to stop supplying power there after 2027 as demand surges from new data centers serving companies like Google, Apple, and Microsoft in Northern Nevada.
“We have no leverage,” said Danielle Hughes, a North Lake Tahoe nonprofit leader pressing regulators to act. She described the situation as a form of “resource extraction” — communities are effectively displaced from stable infrastructure by industrial demand tied to the AI economy.
The industry’s defenders argue that data centers bring jobs, tax revenue, and investment. In some places, they do. But the benefits often arrive unevenly, and sometimes alongside entirely new pressures local communities did not anticipate.
In Abilene, TX, local officials welcomed Stargate, the massive AI infrastructure initiative backed by OpenAI, Oracle, and SoftBank. President Trump celebrated the project publicly as evidence of American technological leadership.
What followed was a housing crunch severe enough that some workers reportedly slept in cars after thousands of out-of-state construction laborers flooded into a market already struggling with limited housing supply.
Mike Prado, a local community worker who once experienced homelessness himself, recalled distributing blankets one cold morning when a man using a walker approached him outside a shelter.
“The AI plant took all the housing,” the man told him.
The limits of resistance
The resistance to these projects is growing rapidly. According to research firm Data Center Watch, community opposition delayed or blocked roughly $98 billion worth of data center projects between March and June of 2025 alone.
In Indianapolis, Google withdrew a proposed $1 billion project after sustained local backlash.
Across the country, unusual political coalitions are beginning to form as communities confront developments they believe local governments are unequipped to regulate. In Hood County, TX, this spring, the Republican county chair and a Democratic candidate for state agriculture commissioner found themselves on the same side of the fight. “This is one of those issues that can transcend the polar part of party politics,” said Greg Harrell, the Republican.
Similar alliances have emerged in Kansas, Georgia, Virginia, and elsewhere as residents organize through zoning hearings, Facebook groups, and local activist networks. But even organized communities often discover how limited their leverage really is.
In Edgerton, KS, residents packed the city hall earlier this year to oppose a proposal by Dubai-based DAMAC Digital to convert a 400,000-square-foot warehouse into a large-scale data center complete with dozens of diesel generators and its own electrical substation.
“The people of Edgerton, the vast majority, do not want it,” resident Shane Mizer told local officials.
But the city’s planning commission faced a problem that is becoming increasingly common across the country. Under existing zoning rules, the project complied with local code. Denying it without legal justification could expose the city to litigation. Officials delayed the vote to gather more information and buy themselves time, but acknowledged privately that they had little room to maneuver. That sense of powerlessness has become one of the defining features of the AI infrastructure boom.
Meanwhile, nondisclosure agreements between developers and local governments — which apply to roughly 80% of Virginia localities hosting major data center projects, according to one review — often mean residents do not fully understand what is being built until agreements are substantially complete.
Part of the problem is structural. Local governments control zoning but not regional utilities. States regulate pieces of electricity markets but not land use. Federal agencies oversee transmission policy but rarely determine where projects are physically located. Technology companies, meanwhile, operate across all of those jurisdictions simultaneously.
The result is a fragmented system in which responsibility is dispersed widely enough that accountability often disappears altogether — even as the buildout accelerates.
And when opposition does emerge, the imbalance in resources can quickly become stark.
In Utah, opposition to a proposed $100 billion data center campus backed by businessman Kevin O’Leary — known as “Mr. Wonderful” on the television show Shark Tank — grew large enough to become a statewide political issue.
The proposed 40,000-acre project would reportedly require up to nine gigawatts of electricity — more than twice Utah’s current statewide power consumption — while residents and environmental groups warned it could deepen water and energy strains in one of the driest states in the country. Project supporters countered that the campus would use more efficient cooling and water systems than many existing industrial operations.
Governor Spencer Cox, who initially championed the project as essential to Utah’s economic future, later softened his rhetoric after intense local backlash and criticism that a three-member county commission had overridden broad public opposition.
As criticism mounted, O’Leary went on Fox News and accused two local organizers, Gabi Finlayson and Jackie Morgan of Elevate Strategies, of acting as proxies for the Chinese government.
“If we were Chinese operatives,” Finlayson joked in a video response, “we would be the worst operatives in the entire world.”
A trend likely to continue
More than 300 bills related to data centers have already been introduced in states this year, according to the National Conference of State Legislatures. Some proposed legislation aims to attract data centers with tax incentives and infrastructure support, while other bills seek to manage their growing impact through tighter environmental rules, energy and water regulations, and restrictions on future construction. The Trump administration, meanwhile, has pushed to accelerate permitting and environmental reviews in the name of AI competitiveness.
None of that appears likely to slow the broader trajectory. Global spending on data center infrastructure is projected to exceed $7 trillion by 2030.
What connects the hum over Montgomery’s farmland in New Jersey, the water shortages in Georgia, the housing crunch in Texas, and the growing anxiety over electricity in Nevada is not simply technology. It is the realization that the digital economy now behaves much like heavy industry once did — consuming enormous physical resources while dispersing responsibility so widely that no single institution ever fully owns the consequences.
The AI boom is often framed as a race for technological dominance between the United States and China, a contest over chips, algorithms, and economic power. But on the ground, it looks like something else: a nationwide struggle over resources and democratic control, unfolding community by community, often only after the generators have already started running. For a growing number of Americans, artificial intelligence is no longer an abstraction or an online convenience. It has become something they can hear from their back porch at night.
 | A guest post by
|