The artificial intelligence boom has a people problem, and it is getting worse faster than most investors have noticed.
While Wall Street has spent the better part of three years fixating on chip stocks, hyperscaler spending, and the relentless march of AI valuations, a less glamorous drama has been unfolding in suburban town halls, county commission meetings, and online petitions from New Jersey to Michigan.
Ordinary Americans, armed with electricity bills, noise complaints, and a generalised anxiety about what artificial intelligence is doing to their lives, are pushing back against the physical infrastructure of the AI boom — and they are beginning to win.
In the first quarter of 2026, 75 data-centre projects worth a combined $130 billion were blocked or delayed by local opposition, according to Data Center Watch, a research firm backed by AI security company 10a Labs.
That is as many projects as faced that fate across the entirety of 2025.
The pace of resistance is accelerating precisely as the pace of construction is accelerating, creating a collision that the industry has been slow to take seriously.
Why communities are saying no
The grievances are varied, but they cluster around a handful of recurring concerns.
Power consumption sits at the top. Between 2018 and 2023, the share of total US electricity consumption represented by data centres rose from 1.9% to 4.4%, according to a study published in the journal Environmental Research Letters.
Projections for what comes next are stark: by the end of the decade, national average wholesale electricity costs could rise between 6% and 29%, with the increase driven primarily by data-centre expansion.
In Virginia, one of the epicentres of the country’s data-centre boom, electricity generation costs could spike by as much as 57%.
Water usage is a second flashpoint.
Data centres use enormous volumes of water for cooling, and in communities already managing drought risk or ageing infrastructure, the addition of a facility consuming millions of gallons annually is not an abstraction.
Residents have also cited the constant low-frequency hum emitted by large facilities, which critics argue could fundamentally alter the character of surrounding neighbourhoods and pose health concerns with prolonged exposure.
Then there is something harder to quantify but no less real.
A general psychological resistance to artificial intelligence has fused with the more concrete grievances, giving the movement an ideological dimension that purely economic arguments cannot easily address.
About 44% of Americans now oppose data-centre construction in the United States, against just 21% who support it, according to a Reuters/Ipsos poll conducted in June.
The gap widens sharply when the question becomes personal: asked whether they would support a data centre in their own community, 57% said no, while only 14% said yes.
“Something that has changed right now is that now we have people that are against data centres even though they don’t have a data centre in their backyard, because they see data centres as the embodiment of AI,” Miquel Vila, lead analyst at Data Center Watch, told Fortune.
“What they oppose is AI. They consider that stopping data centres is the way to stop AI development.”
Why Wall Street is beginning to pay attention to the protests
To appreciate why the financial stakes are significant, it helps to understand how thoroughly the data-centre buildout has underpinned the broader economy and equity markets.
Morgan Stanley estimates that hyperscalers like Microsoft, Amazon, Alphabet, and others will spend $800 billion on capital expenditures in 2026 — roughly the same amount that all non-technology S&P 500 companies combined spent on capex in 2025.
The Semiconductor Industry Association projects that government and industry will spend a further $4 trillion on data-centre infrastructure through 2028.
Data-centre construction spending has already topped $50 billion in a single month, surpassing total US public spending on transportation infrastructure including airports and subways, as Bloomberg reported.
AI enthusiasm has been almost entirely responsible for the S&P 500’s 84% rise since ChatGPT’s public launch in November 2022.
Goldman Sachs expects the AI investment theme to account for roughly half of all earnings growth over the next two years.
The lofty valuations of companies across the AI supply chain rest, to a significant degree, on the assumption that planned capacity will materialise.
A large portion of it may not.
“A lot of the commitments and the build-out of data centers where it’s easy has kind of been done, so you’re getting marginally more difficult,” said Todd Castagno, a managing director at Morgan Stanley in a New York Times report.
“From a markets perspective, expectations might be, maybe not reset, but realigned with the fact that it’s hard to put a couple trillion dollars in the ground in a short time.”
Cities including Tulsa, New Orleans, Birmingham and Ypsilanti Township in Michigan have implemented temporary bans on permitting or construction, as have dozens of other counties and towns, according to a database maintained by hedge fund Interconnected Capital.
Democrats and Republicans in 14 states have proposed construction pauses.
Maine’s legislature passed a temporary statewide moratorium in April, though it was subsequently vetoed by Governor Janet Mills.
Why the tech industry’s charm offensive may not be enough
The technology industry has responded with a concerted public relations effort.
Late last year, Meta spent more than $6 million on an advertising campaign across eight states and Washington DC, promoting the economic benefits of data centres to local communities.
OpenAI and Microsoft have publicly pledged to absorb the energy costs their facilities generate, a gesture aimed at defusing consumer anxiety about rising electricity bills.
Nvidia, Amazon, and Google have each announced technological advances they claim will significantly reduce data-centre water consumption.
Whether any of this is sufficient is genuinely unclear.
“The AI boom is fast approaching a moment of truth, as rapid growth and soaring valuations collide with ballooning capital expenditure, a public backlash and the challenges of real-life adoption,” Deutsche Bank analyst Cox wrote in a recent report.
The resistance, as Vila and others have noted, is no longer purely local.
It has taken on the character of a broader social movement, and social movements are not easily neutralised by folksy advertising.
Analysts debate magnitude of risk to AI-related stocks
For investors, the distribution of risk matters as much as its existence.
“Data-centre opposition is more of an emerging risk than an immediate pressure on AI-related stocks,” Gil Luria, head of technology research at DA Davidson, said in a Barron’s report.
The largest hyperscalers — Microsoft, Google, Amazon — have global footprints and enough redundancy to route investment around hostile localities. They are inconvenienced, not threatened.
The same cannot be said for smaller operators dependent on a handful of large projects.
“The smaller AI clouds are small enough, and have projects that are big enough, that losing a few projects is material,” Luria says.
CoreWeave, for instance, is facing organised resistance to a proposed facility in Kenilworth, New Jersey, that would draw 250 megawatts of electrical capacity — roughly a quarter of the company’s active capacity today.
An online petition calling for the project’s cancellation has gathered more than 11,000 signatures.
Logan Purk, a technology industry analyst at Edward Jones, believes that already extended construction timelines will lengthen further, ultimately reducing the total amount of capacity built.
The ripple effects would travel up the supply chain. “I do think the difficulty is not fully baked in,” Purk said in a New York Times report.
“If we assume tomorrow that data-centre construction stops because there’s no access to new power, the ripple effects across the semiconductor industry would be pretty substantial.”
The picks-and-shovels companies — the equipment and infrastructure suppliers whose fortunes are pegged to the volume of construction — are the most directly exposed.
The resistance might also create some winners
The backlash, however, is not without its beneficiaries.
Mark Guberti of The Motley Fool argues that operators who already have data centres built and generating revenue are quietly positioned to benefit.
“The presence of fewer data centers helps these companies charge higher prices for their AI infrastructure,” he says.
Among the names he points to are Iren and Terawulf, both of which have operational sites and a revenue base that a construction freeze would only make more valuable.
Edge data centres represent a separate category of potential winner.
“These types of data centers are much smaller than large-scale AI data centers that eat up multiple gigawatts of energy,” Guberti says.
“Protesters are less likely to rally against these types of data centers, and zoning requirements for them are less complex.”
These facilities consume far less power and water, present a significantly smaller target for organised opposition, and are considerably less likely to trigger the kind of community mobilisation that is stalling larger projects.
One Stop Solutions, which designs the hardware that forms the backbone of edge data-centre sites, is among the companies analysts have identified as a direct beneficiary of that shift.
Honeywell offers exposure to the same theme through its building automation division.
The business grew 8% year over year in the fourth quarter and accounted for roughly a fifth of the company’s total sales.
However, Honeywell is diversified across multiple industrial businesses, making it a less concentrated play on the edge data-centre theme than One Stop Solutions, which carries more risk but offers purer exposure for investors seeking growth.
The post Americans' revolt against data centers is growing: how it could disrupt the AI trade appeared first on Invezz