Data Center Survey Experiment
GoodPower ran a randomized experiment to test what drives perceptions of data centers in local communities. Data for Progress fielded the survey September 8 to 21, 2026 among 2,610 U.S. registered voters. The design oversampled six active buildout states at 200 completes each: Arizona, Georgia, Indiana, Pennsylvania, Texas and Virginia. Each respondent read a short description of a data center proposed for their own community and then answered a series of questions about it.
Block 1 of the experimental design crossed the power source with a ratepayer protection clause. The power source was either new large-scale solar or new natural gas. The protection clause, when present, said the company “has committed, in a binding agreement, to pay for all electric grid upgrades needed for the project and to guarantee the project will not increase electricity rates for local households.” The clause wording was identical across both energy source conditions.
Block 2 of the experimental design came afterward, on the same respondents, using a plainer description that named no fuel source. Half read the rate guarantee and half read a community benefit agreement funding home energy upgrades including efficient appliances and rooftop solar at no cost to the household.
Outcomes measured
- Project favorability (primary outcome)
- Perceived bill increase
- Trust in the company on costs
- Solar favorability
- Town hall attendance intent
- Concern battery (bills, water, health, air, noise, farmland, appearance and local say)
Key Findings
- Rate guarantees drive favorability: A binding promise protecting local electricity rates significantly boosted acceptance (+0.33 points, or a 14-point increase in favorable share). This lever worked almost exclusively on the 78% of voters who distrust data center companies.
- Community benefits tie the guarantee: An agreement funding local home energy upgrades bought the exact same level of project favorability as the rate guarantee, although the rate guarantee was more effective on cost beliefs.
- Respondents broadly prefer data centers to be powered by renewables: 46% of respondents claimed renewable sourcing would make them more favorable. However, pitching a project as solar-powered rather than gas-powered alone yielded no effect (+0.09 points, 95% CI [−0.08, +0.27]). The inclusion of ratepayer protections and/or community benefits is where we see movement.
- No collateral damage to renewables: Connecting data centers to solar energy did not reduce baseline support for solar power.
- What people think about AI predicts support better than party, place or cost strain: Public opinion on data centers is fundamentally driven by how people feel about artificial intelligence, superseding traditional political, environmental, or economic attitudes. A voter’s general outlook on AI is the single strongest predictor of whether they will support a project (+0.31 coefficient), carrying more weight than corporate trust or climate concerns. Furthermore, AI pessimism does not sort by political party; it sits evenly at roughly 31% to 33% across Democrats, Independents, and Republicans alike.
- Renewable energy companies garner high levels of trust on costs: 49% of respondents trusted renewable energy companies to protect them from rising rates, second only to consumer protection agencies and ahead of governors, state regulators and the utilities. Data center developers were last, with just 22% percent trusting them to not raise rates.