AI Data Centers Face Local Resistance

In the quiet halls of Washington, the drive to build America’s digital future is framed as a matter of national destiny, a modern-day “Manhattan Project” essential to winning the global AI race.

Energy Secretary Chris Wright invokes this powerful analogy, urging a rapid expansion of the nation’s data center infrastructure.

Interior Secretary Doug Burgum goes further, calling the need to power these server farms one of two “existential threats” facing the country, alongside nuclear proliferation.

The message from the top is clear: innovate or be left behind.

Yet, on the ground, far from the polished forums of Silicon Valley and Capitol Hill, a different narrative is unfolding.

Across 28 states, in towns and rural communities, the digital boom is hitting a wall of local resistance.

What Washington sees as a strategic imperative, residents often view as an unwelcome industrial behemoth, consuming staggering amounts of power and water, generating noise, and offering few local jobs in return.

The scale of the proposed build-out is immense.

After years of relatively stable electricity usage, the late-2022 explosion of OpenAI’s ChatGPT sent a shockwave through the energy sector.

Utilities and regulators are now scrambling to meet a projected tripling of data center energy demand by 2028.

This urgency has sparked a construction frenzy, with one commercial real estate firm projecting nearly 4,750 new data center projects breaking ground in 2025 – almost as many as already exist nationwide.

New data centers are, by some measures, the fastest-growing segment of nonresidential construction planning.

But the precise number of existing or proposed data centers remains elusive, a testament to the industry’s rapid, often opaque, growth.

This lack of clear data prompted Robert McKenzie, a former Columbia University adjunct professor, to found Data Center Watch.

His firm, tracking the trend through open-source information, quickly uncovered two significant developments: a far greater number of proposals than anticipated, and a surprisingly robust wave of local opposition.

“There’s more local opposition than certainly we would have imagined,” McKenzie noted, describing how communities are already grappling with projects developers claim are merely “coming.”

The pushback is tangible: a Data Center Watch report from March documented at least 142 ad hoc local groups actively organizing to block or delay data center construction.

Between March 2023 and March 2025, an estimated $18 billion in proposed data centers were outright blocked, and another $46 billion delayed.

This burgeoning local resistance reveals a profound disconnect.

A February survey targeting 16 states slated for AI data center development found 93 percent of respondents agreed that cutting-edge AI data centers are vital for the U.S.

However, a mere 35 percent would vote “yes” to such construction in their own hometown.

As survey author Joe Warnimont put it, “It’s not necessarily about opposing technology in their communities… It’s more that people in these communities want to maintain control over resources and development.”

The objections are varied but coalesce around a few critical concerns.

At the top of the list is the voracious appetite for electricity.

Data centers consume 10 to 50 times more energy per square foot than a typical commercial office building.

The scale is staggering: one Amazon data center campus in Pennsylvania could eventually consume as much electricity as the entire city of Pittsburgh.

This unprecedented demand has utilities and ratepayers bracing for potentially significant cost increases.

While developers express a preference for renewables, the immediate need for baseload power means they will go where energy is available, sometimes even contemplating building their own generators or co-locating near existing power plants.

Then there’s water.

Cooling the legions of servers requires immense quantities, with a single data center potentially gulping up to 5 million gallons a day – enough to supply thousands of households.

While industry proponents like Colleen Gillis, a land-use attorney, argue that newer “data centers 2.0 and 3.0” are more water-efficient, relying on recycling and advanced cooling technologies, the reality on the ground, particularly in water-stressed regions like Arizona, remains a major concern.

A Bloomberg analysis found two-thirds of U.S. data centers are in highly water-stressed areas.

Even in water-rich regions, the impact is felt, especially when coupled with the water demands of fracking for natural gas to fuel these facilities.

Noise is another persistent complaint.

The hum of generators and cooling systems can be comparable to a lawnmower, prompting residents to demand “mufflers” and “dampeners.”

While industry representatives like Jon Hukill of the Data Center Coalition claim most data centers are quieter than many common sounds and rarely violate noise ordinances, the perception of constant industrial drone remains a flashpoint.

Perhaps the most contentious issue for many communities is job creation.

While construction phases generate hundreds of skilled trade jobs, the long-term operational employment of data centers is notoriously low.

Critics often describe them as “one job for every two acres occupied,” leading to skepticism about the economic benefits compared to the significant resource drain.

The industry counters with figures suggesting millions of indirect jobs, but for a community facing the prospect of a massive, power-hungry facility, the direct, visible employment often falls short of expectations.

Adding fuel to the fire is the perceived lack of transparency.

Local governments are often accused of offering generous tax breaks and incentives, frequently shielded by non-disclosure agreements, under the guise of proprietary corporate intelligence.

This secrecy, coupled with instances where state legislatures pre-empt local planning boards from rejecting or modifying proposals, breeds suspicion and anger.

As Kamil Cook of Public Citizen Texas observed, there’s a widespread feeling that the community’s voice isn’t valued, that they lack a genuine choice in these matters.

This backlash is notably bipartisan.

In rural Texas, for example, it’s often “bright red Republicans” leading the charge against data center expansion.

“It’s just, ‘Do I want this in my backyard or not?’,” Warnimont remarked, highlighting that the opposition transcends political affiliation.

Where communities once rallied against factories or retail sprawl, they are now opposing server farms, making data centers the new “not-in-my-back-yard” flashpoint in the evolving landscape of large-scale development.

The industry acknowledges these challenges, particularly as development moves beyond established “Data Center Alleys” like Northern Virginia into “secondary” and “tertiary” markets – places like Mississippi, Alabama, Iowa, and Indiana – where land is cheaper, energy more accessible, and local governments are eager for economic development.

Here, communities often have little historical experience with such industrial operations, making the transition especially jarring.

The national imperative for AI and cloud computing is undeniable, a strategic race with global implications.

Yet, the grassroots resistance to the physical infrastructure required to power this future is equally undeniable.

It’s a complex collision of national ambition and local autonomy, where the abstract world of data finds itself grounded by very real concerns about power, water, noise, and the right of communities to shape their own destiny.

The “Manhattan Project” may be underway, but its foundations are proving surprisingly difficult to lay in America’s backyards.