Who Wants Data Centers?
Brief #3 from the UVA NSI Data Center Opinion Monitor | Source: UVA NSI survey fielded July 2-3 2026
October 13, 2026
We hear a lot about the people fighting data centers. Dozens of surveys, polls, and headlines announce that people really don’t like data centers. Two in three Americans do not want a data center built near them (that figure is so well established that dozens of surveys could be cited for it, including our own). Opposition is so fierce and widespread that it is reshaping American politics and tearing American towns apart. Localities across the US are voting on data center moratoriums. The landscape of data center opposition is exceedingly well-mapped. So we flipped the question and asked: Who wants data centers?
In early July, we fielded a survey study with a nationally representative sample of 1,991 American adults. We asked all participants how they would feel about a data center being built near them, whether they have a data center in their local area, how often they use AI, and a number of demographic questions. We then randomly assigned each participant to evaluate one of four scenarios for a proposed local data center: three in which the company pledged a $240 million investment in the community, and one with no investment. We analyzed all of the demographic variables and condition assignments, and identified 4 key patterns that sketch a portrait of the data center supporter.
Those that already have a data center nearby are more tolerant of them
People who already have a data center within 5 miles of their home are 19 percentage points more likely to say they would support another one being built in their community. This might sound counterintuitive: a reader might expect that someone who already has a data center in their town would be less willing to have another one taking up space in the same area, compared to someone whose town currently has no data centers at all. However, this pattern is not new.
When Americans began opposing nuclear power, an overwhelming majority of Americans did not want nuclear power near their homes. But the communities that already lived near a reactor were the most willing to accept another one. Familiarity made the risks concrete and manageable rather than abstract and catastrophic, and host communities had seen the jobs and tax revenue firsthand. This pattern shaped the industry’s roll-out for the next few decades: nearly every new reactor built in the U.S. in recent decades has gone up at an existing plant site. Data centers may be following the same path. For people who already live near one, a data center is no longer a looming unknown. It’s a familiar building down the road whose costs and benefits they’ve already experienced.
There are important nuances that are worth mentioning. For example, we asked people to self-report whether they currently have a data center in their town. Many people may not know about data centers near them, and the ones that do may be more likely to have benefit from them, for example by being employed during their construction. Second, only 458 out of 1,991 participants (23%) reported living within near a data center, so our estimates for this group are less precise than for the full sample. Still, the pattern closely mirrors what happened with nuclear power, which gives us some confidence that the effect is real.
Daily AI users are more tolerant of data centers
People that use AI daily are 22 percentage points more likely to support a data center being built in their community. 34% of daily AI users support a local data center, compared with about 12% of everyone else. The gap is just as striking on the opposition side: about half of non-daily users strongly oppose a local data center, while only about a quarter of daily users do.
4 key features significantly predict local support for data centers
Two other features had predictive value: leaning Republican and being promised a cash payout from the data center (in our assigned investment condition). We combined these predictive variables to fit a linear model that predicts likelihood to be a supporter or non-supporter. Each of the four independently and significantly predicted support, and each added a similar boost.
A participant with none of the four had a predicted 15% chance of supporting a local data center. Being Republican raised that by 16 percentage points, living near a data center by another 16, using AI daily by 18, and being offered a personal payout by 13. No single factor gets a community past 50% on its own, but any three of the four do. And because the model is additive, meaning each factor’s effect is the same regardless of the others, their order does not matter. In other words, if you took a sample of participants who live near a data center and are daily AI users - but don’t lean Republican - and added a pledge of a personal payout, you could expect to see 62% support. A sample that has all four of these features would yield a predicted 78% chance of support. This is a massive change from the average levels of support, which hover around 20%.
However, there’s a catch: this group is exceedingly rare or nonexistent. Out of 1,991 people we surveyed, only 84 of them were Republican daily AI users that lived near a data center. That’s about 4% of our nationally representative sample.
A data center can’t change how a community votes or how often its residents use AI. It can choose to build where data centers already exist, and it can pay residents directly. Those two levers alone raise predicted support from 15% to 44%. That’s nearly triple the baseline, but still short of a majority. This means that choices about where data centers go, and policies about how they engage the communities around them, are among the few levers that can move public support.
Data Center Opinion Monitor
A nation’s power is increasingly measured in compute, and compute lives in data centers. This brief is part of our ongoing Data Center Opinion Monitor, which tracks the American public’s support for data centers as they expand their footprint across the country.