AI Data Centers Are Getting Blocked Everywhere — Here’s Why $130 Billion in Projects Stalled in Just Three Months

The companies building AI need somewhere to put it.

AI models — the technology behind ChatGPT, image generators, code assistants, and every other AI product that has become part of daily life — do not run on your laptop. They run in massive buildings packed with servers, cooled by enormous amounts of water, powered by enough electricity to supply small cities, and operating around the clock with a low hum that neighbors describe as impossible to tune out.

Those buildings are called data centers. And in 2026, a significant portion of the United States does not want them.

In the first three months of 2026 alone, local opposition blocked or delayed at least 75 data center projects across the United States worth approximately $130 billion — roughly matching the total for all of 2025 in just three months. The number of organized opposition groups has more than doubled, from 396 at the end of 2025 to 833 groups now active across 49 states.

A Gallup poll conducted in March 2026 found that 70% of Americans oppose the construction of a new AI data center in their local area — with 48% strongly opposed.

This is not a niche environmental movement or a small number of vocal residents near specific projects. It is a widespread, organized, well-funded opposition movement that is now one of the most significant practical obstacles to the AI industry’s infrastructure ambitions.

Here is what is driving it, what it means for AI, and what both sides are saying.

What Is a Data Center — And Why Does AI Need So Many?

Before getting into the opposition, it helps to understand what data centers actually are and why the AI industry needs so many of them so quickly.

A data center is a building — or more commonly, a campus of buildings — that houses the computing infrastructure AI companies use to train and run their models. Inside are thousands of specialized processors called GPUs, stacked in server racks, connected by high-speed networking equipment, and cooled by industrial-scale cooling systems.

Training a single large AI model can require months of continuous operation across tens of thousands of GPUs. Running that model at scale — serving millions of users simultaneously — requires constant, reliable computing power available around the clock.

The RTX 5090 and its successors represent the cutting edge of the GPU technology that goes inside these facilities. But even the best individual GPU is only as useful as the infrastructure around it — and that infrastructure requires physical buildings in physical places with physical connections to power grids and water supplies.

The scale of what AI companies are trying to build is genuinely large. Microsoft, Amazon, Meta, and Google have collectively committed to spending hundreds of billions of dollars on AI infrastructure through 2027. Each dollar of that spending represents a physical building somewhere. And each building has neighbors.

Why Are People Opposing Data Centers?

The opposition is not monolithic — different communities are concerned about different things, and the intensity of opposition varies significantly depending on what is being proposed and where. But several themes appear consistently across opposition groups in all 49 states where organized resistance has been documented.

Energy use and electricity prices

This is the most common concern. Roughly half of respondents in a recent Morning Consult survey believe AI data centers negatively affect electricity prices and the power grid.

A single large data center can consume as much electricity as a small city. When multiple data centers are built in the same region — as tends to happen because land, power, and fiber infrastructure cluster together — the cumulative load on local power grids is significant. Residents near proposed data centers frequently report being told by utility companies that new capacity will be needed, which translates into rate increases for existing customers.

The AI industry’s energy demands have also forced uncomfortable conversations about energy transition. Several major tech companies have signed agreements to power data centers with renewable energy — but in practice, the scale of demand has led some to extend the life of coal plants, pursue nuclear power agreements, and seek natural gas connections, all of which conflict with stated climate commitments.

Water consumption

Data centers cool their servers using water — enormous amounts of it. Large facilities can consume millions of gallons per day in regions where water is already a stressed resource. In drought-prone areas of the American West and Southwest, proposed data centers have faced fierce opposition from agricultural communities, environmental groups, and local governments concerned about long-term water availability.

Noise pollution

This is the concern that drives opposition from people who actually live near operating data centers. The cooling systems — fans, chillers, and cooling towers — run continuously and produce a low-frequency hum that residents describe as penetrating walls, disrupting sleep, and degrading quality of life in ways that are difficult to habituate to. Unlike construction noise, which eventually stops, data center noise is permanent.

Environmental impact and land use

Data centers are large buildings that require significant land, often in areas that were previously agricultural or natural. The construction process involves clearing land, pouring enormous concrete foundations, and building the kind of industrial-scale infrastructure that permanently changes the character of an area.

Jobs versus impact

One of the most consistent frustrations in communities that have faced data center proposals is the gap between the economic benefits promised during the approval process and the reality of what data centers deliver. A large data center might employ 50 to 100 people permanently — far fewer than a manufacturing facility of comparable size and cost. Critics argue that the economic case for communities accepting the costs of hosting a data center is weak relative to the disruption it causes.

The Numbers Behind the Opposition

The scale of what is happening in 2026 represents something genuinely new — not just an increase in opposition activity but what researchers are calling a structural shift in how communities engage with data center proposals.

Data Center Watch, which tracks local data center activity, found that in Q1 2026 the total number and value of data centers blocked or delayed was roughly equal to the total for all of 2025 — meaning the opposition movement accomplished in three months what it previously took a year to achieve.

300 bills were introduced in statehouses across 14 states in just the first six weeks of 2026. Senators Bernie Sanders and Representative Alexandria Ocasio-Cortez introduced a federal moratorium bill. While none of these proposals has been signed into law, they signal that data center opposition has moved from local zoning fights to state and federal legislative strategy.

142 protests were staged in 42 states — with organizers describing the AI infrastructure buildout as an “unacceptable infringement on our liberty” and calling the companies behind it “unaccountable.”

Morgan Stanley flagged community opposition as a market risk comparable to energy and chip supply constraints — the first time a major financial institution had formally identified community consent as a bottleneck in AI infrastructure planning.

The Surprising Finding — Most Opponents Do Not Live Near a Data Center

One of the most revealing findings from recent polling is also one of the most counterintuitive.

Only a small fraction of data center opponents actually live near one, according to polling by Milltown Partners, a consulting firm that works with leading AI labs and tech startups. The findings highlighted how data centers have become a stand-in for broader anger at an AI future many Americans do not want but fear they will have to pay for.

In other words, the fight over data centers is not primarily a fight about data centers. It is a fight about AI itself — its energy costs, its economic disruption, its concentration of power in a small number of large corporations, and the absence of meaningful public input into how AI is developed and deployed.

Data centers are the most visible physical manifestation of AI. Unlike the software and models themselves — which are invisible — data centers are buildings that show up in planning applications, require local approvals, and change the physical landscape of the communities where they are built. They are where public opposition to AI’s broader trajectory has found a practical leverage point.

This is why the opposition has grown even in communities where no data center is currently proposed. People who have concerns about AI in general are joining opposition movements not because they are directly affected but because opposing a local data center is the most accessible form of civic action available to express those concerns.

What AI Companies Are Doing in Response

The industry has not been passive in the face of rising opposition. Several strategies have emerged.

Community benefit agreements

Some developers have moved toward negotiating formal community benefit agreements before proposing projects — committing to local hiring targets, contributions to community funds, energy subsidies for residents, or environmental mitigation measures in exchange for community support.

Smaller footprints and better siting

The era of proposing 500-acre mega-campuses in suburban or rural areas is running into practical limits. Some developers are shifting toward smaller facilities in industrial zones or areas already zoned for heavy commercial use — where opposition tends to be lower and permitting more straightforward.

Renewable energy commitments

Attaching genuine renewable energy commitments to data center proposals — rather than carbon offset arrangements — has become more common as a response to energy-related opposition. Some developers have announced plans to build dedicated renewable generation capacity alongside proposed data centers.

Working with utilities earlier

The electricity price concern has driven some developers to work with utilities and regulators before filing planning applications — establishing rate structures and grid upgrade plans that insulate local residents from cost increases before the project becomes public.

None of these approaches has resolved the fundamental tension between the scale of AI infrastructure demands and the preferences of the communities where that infrastructure needs to be built. But they represent a shift from treating community opposition as an obstacle to be overcome toward treating it as a stakeholder relationship to be managed.

What This Means for AI’s Future

The data center opposition movement has real consequences for the AI industry’s timeline and ambitions.

Every delayed project represents months or years of lost compute capacity. The $130 billion in projects stalled in Q1 2026 is not money lost — it is money waiting for approvals that may or may not come. For an industry that has built its competitive narrative around the speed of capability improvement, delays measured in years are significant.

The geographic concentration of viable data center locations is also narrowing. As opposition spreads and legislative moratoriums are proposed in more states, the number of communities willing to approve large data center projects decreases. This could drive more development to a smaller number of locations — increasing the concentration risks associated with critical AI infrastructure.

The best free AI tools that millions of people use every day depend on this infrastructure. ChatGPT, image generators, code assistants — all of them run on servers in buildings that need to be somewhere. The community opposition movement is, in effect, a public negotiation over where those buildings go and on what terms.

That negotiation was largely invisible for the first several years of the AI boom. In 2026, it has become one of the defining bottlenecks of the industry — not chips, not talent, not capital, but consent.

Frequently Asked Questions