The debate over data centers in the United States is evolving from local disputes over siting and resources into a broader wave of public opposition. As AI infrastructure expands rapidly, data centers are becoming a new focal point of public concern and are entering the national political arena.

According to a report by US tech media outlet The Information titled America Really, Really Hates Data Centers, public support for data centers is noticeably declining, with the proportion of opponents rising sharply within a year. The latest polling shows that 75% of Americans oppose data center construction, a significant jump from 42% just one year ago.

Furthermore, a joint survey by Embold Research and climate media outlet HeatMap Pro found that this shift in attitude is not confined to a single political party: among Republicans, independents, and Democrats, opponents outnumber supporters by 43, 65, and 75 percentage points, respectively.

Meanwhile, local backlash is beginning to translate into concrete policy actions. Texas Republican Governor Greg Abbott has halted the construction of 1,800 data center projects in the state, while Pennsylvania Democratic Governor Josh Shapiro has signed an executive order tightening restrictions on data center expansion.

For the AI industry, this trend cannot be ignored. Data centers are the physical foundation for AI computing power expansion, and any restrictions on permitting, electricity, or water usage could directly slow down the deployment of new projects. The report notes that negative sentiment surrounding data centers could also spread to consumer-facing AI products and increase tech companies' spending on political lobbying and elections.

The speed of the shift in public opinion has caught the industry off guard

The pace of this emotional shift has even surprised pollsters themselves.

Mike Greenfield, co-founder of Embold Research, said the most surprising aspect of the polling was not the opposition to data centers per se, but how rapidly public attitudes have changed. He noted that people rarely alter their established views in such a short time, but this case is clearly an exception. Greenfield even suggested that such a swift formation of social backlash is uncommon in American history.

Data centers have quickly become a point of contention largely due to their direct and tangible resource consumption: enormous electricity demands, water usage, and impacts on local environments and infrastructure.

Yale University professor Yuan Yao also pointed out that the environmental impact of data centers is real, but the specific effects depend heavily on project location, energy mix, and resource conditions. This means public concerns about data centers are not entirely without a factual basis, yet in media discourse, complex regional differences are often replaced by a simpler narrative of "high energy consumption, high water usage."

The problem is that the AI industry has yet to develop a framework that can effectively persuade average voters.

Tech giants' public relations efforts are struggling

Data centers are inherently complex, and the tech industry's previous communication strategies may have actually amplified public resistance.

The report notes that some companies have attempted to frame data centers as the infrastructure of a new industrial revolution, but this analogy has not resonated. The industrial revolution brought economic growth but also carried historical memories of pollution and labor struggles, which can easily reinforce negative associations with large industrial facilities.

Some companies are already changing their approach.

According to Bloomberg, Oracle has donated to charities near its Project Jupiter data center in rural Texas and sponsored a local children's dinosaur fossil excavation theme park called Dino Dig. Anthropic, OpenAI, and Google are trying to emphasize AI's potential value in fields like cancer research, hoping to shift public attention from energy consumption to the social benefits of the technology.

But this "AI can cure disease" narrative also faces skepticism. Some figures in the traditional scientific community argue that tech companies leveraging major diseases for publicity risks packaging complex scientific achievements as AI marketing stories.

Greenfield stated bluntly that tech companies still have significant shortcomings in explaining AI and its infrastructure to average voters outside Silicon Valley. For the AI industry, the real challenge may no longer be proving to investors how important AI is, but explaining to ordinary voters why they should accept the enormous infrastructure costs behind AI.

From local permitting to midterm elections, political pressure is rising

As opposition sentiment spreads, data centers are beginning to enter the US political agenda.

The National Republican Senatorial Committee recently issued warnings to several major AI companies, stating that data center backlash could affect Republican prospects in key states like Ohio, and has pressured super PACs supporting the AI industry to increase political contributions.

This means the costs facing the AI industry may no longer be limited to land, electricity, and chips. To win support from local governments and voters, tech companies may also need to invest more in policy lobbying and political donations.

Meanwhile, public concerns about AI infrastructure could also spread to the consumer side. The consumer AI hardware being developed by OpenAI in collaboration with Jony Ive, as well as Meta's smart glasses, could be affected by broader AI anxiety. Wearable devices with camera and environmental sensing capabilities, in particular, have already sparked privacy-related controversies.

From data centers to AI hardware, the two may seem far apart but share the same underlying issue: AI is transforming from an abstract cloud-based technology into a physical presence that consumes electricity, land, and resources, and even enters personal living spaces directly.

Therefore, it remains difficult to predict where this backlash will ultimately lead. It could follow the path of hydraulic fracturing in the 2000s, which continued to expand despite intense controversy, or it could mirror the long-term political resistance nuclear power faced in the 1970s and 1980s.

But for now, one thing is certain: data centers are no longer just an infrastructure issue within the AI supply chain, but are becoming a genuine political and social contest.