Over 70% of Americans Oppose AI Data Centers Near Their Homes — Even Less Popular Than Nuclear Power Plants

Gallup Survey: Data Centers Less Popular Than Nuclear Power Plants

According to the latest survey data from Gallup, over 70% of Americans explicitly oppose the construction of AI data centers near their residential areas. Only 7% of respondents expressed "strong support" for new data center construction. Even more surprisingly, the results show that Americans would rather live near a nuclear power plant than next to a data center — a comparison that profoundly reveals the public's resistance to AI infrastructure expansion.

Gallup is one of the world's most influential polling organizations, founded in 1935 and headquartered in Washington, D.C. Its methodology is renowned for random sampling and rigorous statistical standards, and it has long provided decision-making references for the U.S. government, corporations, and academia. Gallup's findings are typically regarded as authoritative indicators of American public attitudes, and this survey has attracted widespread attention precisely because Gallup's brand credibility gives these data undeniable policy influence.

At a time when the AI wave is sweeping the globe and tech giants are racing to expand computing infrastructure, this survey is like a bucket of cold water — reminding us that the tension between technological progress and community interests is intensifying.

Why Are Americans So Resistant to AI Data Centers?

Energy Consumption and Environmental Concerns Come First

AI data centers are genuine "power black holes." A single large data center can consume as much electricity as a medium-sized city. With the explosive growth of large language models and generative AI, data center energy demands are rising at unprecedented rates. For local residents, this means increased grid pressure, potential electricity price hikes, and higher carbon emissions.

The reason modern AI data centers consume far more energy than traditional ones lies primarily in AI training and inference tasks' heavy reliance on specialized accelerator chips like GPUs (Graphics Processing Units) and TPUs (Tensor Processing Units). Take the NVIDIA H100 GPU as an example — a single card has a Thermal Design Power (TDP) of 700 watts, and a large AI training cluster may deploy tens of thousands of such chips. The International Energy Agency (IEA) reported in 2024 that global data center electricity consumption is expected to grow from approximately 460 terawatt-hours in 2022 to over 1,000 terawatt-hours by 2026, with AI workloads being the primary driver of incremental demand. This growth rate means data center power consumption could double within just a few years — equivalent to adding the entire electricity consumption of Japan.

Noise, Water Usage, and Direct Conflicts with Daily Life

Data center cooling systems consume massive amounts of water resources. In water-scarce regions, this directly competes with residential water supplies. Additionally, the continuous noise from large data centers operating 24/7, intensive construction activities, and impacts on local traffic and land use are all practical reasons for resident opposition.

Cooling systems are the primary source of data centers' water consumption. Traditional cooling methods include chilled water systems and evaporative cooling towers — the latter removes heat through water evaporation, which is highly efficient but extremely water-intensive. By some estimates, a large data center can use millions of liters of water per day, equivalent to the daily water usage of tens of thousands of households. In recent years, liquid cooling technologies (including direct-to-chip liquid cooling and immersion cooling) have been emerging, which can significantly reduce water usage and improve heat dissipation efficiency, but their deployment costs are high and they have yet to become industry mainstream. Companies like Google and Microsoft have publicly committed to achieving "Water Positive" goals — returning more water to the environment than they consume — but the actual implementation progress of these commitments remains under close scrutiny from environmental organizations.

A Textbook Case of NIMBYism

This phenomenon is essentially a classic "NIMBY" (Not In My Backyard) problem. People may acknowledge the value of AI technology and enjoy the conveniences it brings, but when the infrastructure needs to be built at their doorstep, attitudes shift dramatically. The fact that data centers are even less popular than nuclear power plants is a striking comparison — after all, nuclear power plants have long been associated with safety risks in the public mind.

NIMBY is a classic concept in urban planning and public policy, first widely used in the 1980s to describe the contradictory mindset where residents support the construction of certain facilities but oppose having them built in their own communities. Historically, landfills, chemical plants, prisons, and wind farms have all been typical targets of NIMBY opposition. Notably, in recent years academia has also developed the concept of "NIABY" (Not In Anyone's Backyard) to describe a more radical stance of total opposition. Data centers becoming the new NIMBY flashpoint reflects how digital infrastructure is transforming from an "invisible cloud service" into a "tangible physical presence," as the public begins to realize that so-called "cloud computing" is backed by very real steel, concrete, and roaring machines.

Tech Giants' Computing Expansion Meets Community Resistance

Hundreds of Billions in Investment Face Grassroots Opposition

Microsoft, Google, Amazon, Meta, and other tech giants are aggressively expanding data centers worldwide to support growing AI computing demands. In 2024 alone, these companies' capital expenditure on data centers reached hundreds of billions of dollars. However, the Gallup survey indicates that this expansion is encountering strong resistance from grassroots communities.

Across multiple U.S. states, cases have emerged of residents self-organizing to oppose data center construction projects. Communities in Virginia, Ohio, Indiana, and elsewhere have protested proposed projects, with some being delayed or even cancelled as a result.

Among these, Loudoun County in northern Virginia is the most representative case. The county is known as the "Data Center Capital of the World," hosting approximately one-third of the world's known data center capacity. This status stems from its geographic proximity to Washington, D.C., abundant fiber optic network infrastructure, and early tax incentive policies for tech companies. However, as data center density has continuously increased, local resident opposition has also grown louder. Between 2023 and 2024, Loudoun County and neighboring Prince William County experienced multiple large-scale community protests, with residents accusing data centers of causing property value declines, destroying rural landscapes, worsening traffic congestion, and placing enormous strain on the local power grid. Prince William County even temporarily suspended approvals for new data center projects, becoming a landmark event of national attention.

Jobs and Tax Revenue Fail to Persuade Residents

Tech companies typically use job creation and increased tax revenue as bargaining chips to win over local communities. But the reality is that highly automated modern data centers create very limited long-term employment. Compared to the land they occupy and resources they consume, the economic returns appear disproportionate in residents' eyes.

Modern hyperscale data centers are extremely automated — from server deployment and fault detection to daily operations and maintenance, much of the work is already handled by software systems and robots. A large data center costing billions of dollars and spanning hundreds of thousands of square meters may require only 50 to 200 long-term staff, primarily including facility engineers, security personnel, and a small number of IT operations staff. By comparison, a manufacturing plant with equivalent investment might create thousands of jobs. This "high investment, low employment" characteristic makes tech companies' strategy of using jobs as leverage to persuade communities largely ineffective. Some local governments have begun reassessing tax incentive policies for data centers, questioning whether these incentives truly deliver economic returns commensurate with the costs borne by communities.

The Social Contract of AI Development Is Being Reexamined

This survey reflects a deeper question: Is the rapid development of AI technology neglecting its "social contract" with the public?

While pursuing computing power growth, tech companies need to seriously consider how to build mutual trust with communities. This is not merely a public relations issue — it involves the fairness of benefit distribution. AI's gains primarily flow to tech companies and their shareholders, while the environmental and quality-of-life costs are borne by local communities — this asymmetry is the root cause of public discontent.

In the future, data center siting and construction may require more transparent decision-making processes, more substantive community compensation mechanisms, and higher standards for energy efficiency and environmental protection. Some companies have already begun exploring bundling data centers with renewable energy projects or locating them in sparsely populated remote areas, but each of these approaches has its limitations.

Regarding renewable energy integration, tech giants are pursuing multiple cutting-edge approaches. Microsoft has signed a power purchase agreement with fusion startup Helion, Google is collaborating with Kairos Power to develop Small Modular Reactors (SMR) to power data centers, and Amazon is investing in large-scale solar and wind projects across multiple locations. Additionally, the concept of "Load Matching" is emerging — aligning data center computing tasks temporally with renewable energy generation peaks to maximize clean energy utilization. However, critics point out that tech companies' massive procurement of renewable energy may crowd out clean power supplies for other industries and residents, creating a "green energy scramble" that could actually drive up overall electricity prices. This means that even seemingly perfect clean energy solutions may trigger new fairness controversies at a broader societal level.

Conclusion: Even the Most Powerful Algorithms Need Physical Servers

This Gallup survey sends an important signal. In the midst of a fierce AI arms race, the pace of technological development is outstripping the formation of social consensus. If the tech industry cannot properly manage its relationship with communities, AI infrastructure expansion may face increasing political and legal resistance. After all, even the most powerful algorithms need servers in the physical world to run — and those servers ultimately need to be built in someone's "backyard."

Key Takeaways

• Over 70% of Americans oppose building AI data centers near their residential areas, with only 7% expressing strong support
• Americans would rather live near a nuclear power plant than next to a data center, reflecting strong public resistance to AI infrastructure
• Data centers' high energy consumption, massive water usage, and noise are the primary reasons for resident opposition
• Tech giants' hundreds-of-billions-dollar data center expansion plans are facing strong resistance from grassroots communities
• The asymmetric distribution of AI's benefits versus environmental costs urgently requires new social contracts and community compensation mechanisms