Seven in ten Americans say they would not support an AI data center being built in their local area, according to a Gallup survey, with environmental concerns driving the strongest opposition. This public resistance is emerging as a powerful organizing force that is reshaping how communities worldwide approach AI infrastructure development, from New York to the Netherlands.

What's Driving Public Opposition to AI Data Centers?

When Gallup asked a national sample of Americans whether they would support an AI data center in their local area, 71% opposed it, with 48% expressing strong opposition. The reasons people cited most often centered on environmental impacts. Respondents pointed to concerns about water consumption, energy demand, land use, noise pollution, and air quality, alongside frustration over the lack of community consent in planning decisions.

Interestingly, opposition to AI data centers actually exceeds opposition to nuclear power plants. When asked about living near a nuclear facility, 53% of Americans opposed it, a markedly lower figure than the 71% who reject data centers. This suggests that public concern about AI infrastructure is distinct from general anxiety about energy production.

A randomized controlled trial conducted by Social Change Lab with 3,467 participants tested which arguments about AI's harms actually motivate people to take action. Researchers evaluated 11 different arguments, ranging from environmental effects and job losses to algorithmic bias, surveillance risks, and existential risks from superintelligent AI. The findings revealed that environmental impacts were the strongest predictor of whether someone would move from concern to action. Job displacement, the issue people mention most often when discussing AI unprompted, ranked much lower in motivating behavior. Extinction risk came last.

The emotion most strongly associated with willingness to act was not fear or anxiety, but anger. This emotional dimension helps explain why data centers have become focal points for public mobilization.

How Are Communities Organizing Against Data Centers?

Data centers have become tangible targets for climate organizing in ways that abstract environmental metrics cannot match. Unlike discussing "CO2 parts per million," every data center requires a planning application, a zoning decision, a water permit, and approval from elected officials. These local decision points create concrete opportunities for community organizing and political leverage.

The movement to halt or restrict data centers has grown with remarkable speed in the United States:

• New York State: The state assembly recently passed what could become the country's first statewide moratorium on data center construction.
• Seattle: The city became the largest American city to adopt a data center moratorium after residents generated more than 96,000 emails to the city council.
• Coalition Building: More than 500 organizations from 47 states have joined the Stop Data Centers Coalition.

Deeda Seed of the Center for Biological Diversity observed that concerns about data centers are mobilizing people who have never taken a public stand on any issue before. "They're willing to take a stand on this," Seed noted.

What Happened in the Netherlands Data Center Fight?

One of the clearest victories for the anti-data-center movement occurred in Zeewolde, a Dutch town of just over 22,000 people. In February 2021, Meta proposed building a 200-megawatt hyperscale data center on 410 acres of reclaimed farmland outside the town. The facility would have consumed at least twice as much electricity as the entire town itself.

Residents, farmers, and Extinction Rebellion campaigners objected on multiple grounds. Beyond the sheer scale of the project, they were angered by the secrecy surrounding it. Officials had spent two years negotiating with Meta while keeping the company's identity out of the public record. Many residents felt that scarce renewable energy was being handed to a foreign corporation with little local benefit.

The campaign operated on two fronts simultaneously. Locally, organizers plastered posters on the town hall, organized a residents' petition, and staged nonviolent direct action. Days before the December vote, 20 Extinction Rebellion activists occupied the council building's public hall with banners reading "We want Greta, not Meta," until police cleared them out. Nationally, allies in The Hague pressed a different lever. Part of the land Meta needed belonged to the state, and motions in both chambers of parliament urged ministers to use that leverage.

Although campaigners lost the initial vote when the council approved the project in December 2021, they won almost everything that followed. Within months, the Dutch government imposed a moratorium on new hyperscale developments, and local elections delivered results favoring opponents of the project.

How Is Power Infrastructure Adapting to AI Data Center Demands?

While communities resist new data centers, the infrastructure supporting existing facilities is undergoing rapid innovation. AI data centers face unprecedented power challenges that standard backup systems cannot handle. Sudden power spikes threaten hardware, extreme heat risks thermal runaway, and grid connections are strained to their limits.

To address these challenges, China-based energy storage firm Highstar unveiled a comprehensive battery cell portfolio designed specifically for AI data centers at the 2026 GGII Energy Storage Industry Summit. Rather than forcing a single battery type to handle all demands, Highstar deploys custom chemistries tailored to each operational layer of the data center.

The system covers three distinct power layers, each with specialized technology:

• Server-Rack Layer: High-density GPU racks require instantaneous, high-rate discharge to protect servers against power spikes and dropouts. Highstar introduced tabless cylindrical cells in 18650 and 21700 formats, which reduce internal electrical resistance and heat buildup in tightly packed rack spaces.
• Facility-Wide Layer: Centralized Uninterruptible Power Supply (UPS) and high-voltage direct current systems use an 85-ampere-hour lithium iron phosphate cell for frequent, rapid pulse loads, alongside a 50-ampere-hour sodium-ion cell that operates reliably across extreme temperature ranges.
• Grid-Side Layer: Massive 314-ampere-hour lithium-ion and 160-ampere-hour sodium-ion utility-scale cells store bulk energy and smooth out regional power demands to shield the wider electrical grid from sudden data center load surges.

The combination of lithium-ion and sodium-ion chemistry is strategically important. Lithium provides high energy density, while sodium offers thermal stability, extreme-cold endurance, and supply-chain resilience against lithium price spikes. Power grids worldwide, including those in the United States, Europe, and Asia, are delaying data center connections due to transformer shortages and capacity limits. On-site battery storage allows operators to ramp power safely without tripping local substations.

What Does the AI Infrastructure Investment Cycle Reveal About Market Signals?

The scale of AI infrastructure commitments in 2026 has expanded into extraordinary physical quantities: gigawatts of capacity, billions of dollars of hardware, and data-center campuses built around the future requirements of frontier models. This week, AMD agreed to provide the systems behind up to two gigawatts of Anthropic's future infrastructure, alongside an investment that could reach $5 billion.

However, the structure behind these commitments invites critical reflection. When a supplier provides capital to finance a buyer whose orders then drive the supplier's growth, some of the industry's most impressive demand signals may be considerably less independent than they appear. Investors may be treating financially dependent AI commitments as independent evidence of demand, which could distort how the market measures the true scale and sustainability of the AI buildout.

As AI training loads double every few months, the battle for dominance in AI infrastructure will eventually come down to power. The convergence of public resistance to new data centers, innovation in power backup systems, and complex financial interdependencies between suppliers and customers suggests that the next phase of AI infrastructure development will be shaped as much by community politics and power engineering as by raw computing ambition.