Key Takeaways
- 71 % of U.S. adults oppose a local AI data centre (48 % strongly), surpassing the 53 % opposition to nearby nuclear plants.
- Top concerns include strain on water and energy systems, loss of farmland, increased traffic, noise, and higher utility bills.
- Industry leaders attribute opposition to limited public understanding and insufficient outreach about data‑centre benefits.
- Data‑centre electricity use has risen from 1.9 % of national consumption in 2018 to 4.4 % today, with projections of 12 % by 2028.
- Expected U.S. data‑centre power demand will grow from 80 GW in 2025 to 150 GW in 2028.
- Water consumption jumped to roughly 17 billion gallons in 2023—triple the 2014 level.
- AI‑related CO₂ emissions could reach 24–44 million metric tons annually by 2030.
- Local moratoria may merely shift projects elsewhere, as demand is already pre‑booked years in advance.
Overview of Gallup Survey Findings
A Gallup poll revealed that 71 % of U.S. adults oppose the construction of an AI‑powered data centre in their neighbourhood, with 48 % expressing strong opposition. This level of resistance exceeds the 53 % who oppose having a nuclear power plant nearby, highlighting a growing apprehension toward AI infrastructure. The survey underscores that, while nuclear opposition has historically hovered below 63 % since 2001, AI data‑centre scepticism is already surpassing that benchmark, signalling a shift in public perception of emerging technology risks.
Historical Context of Nuclear Opposition vs. AI Concerns
Gallup has tracked attitudes toward nuclear plants since 2001, noting that opposition has never crossed the 63 % threshold. In contrast, the recent AI data‑centre opposition figure of 71 % marks a new high for infrastructure‑related worries. This comparison suggests that fears about AI’s immediate environmental and community impacts are resonating more strongly than longstanding concerns about nuclear safety, reflecting anxieties over resource consumption, visual intrusion, and perceived lack of control over new tech installations.
Community‑Specific Concerns Driving Opposition
Follow‑up questioning showed that roughly half of those opposed cited potential strain on local water and energy systems, as well as the possible loss of farmland. About 22 % mentioned quality‑of‑life issues such as increased traffic, while one in five worried about higher utility bills. Noise pollution also emerged as a frequent complaint. These concrete, everyday impacts—rather than abstract technological fears—appear to be the primary drivers of local resistance to AI data‑centre siting.
Industry View: Need for Better Education and Communication
Wannie Park, CEO of PADO AI and an energy‑industry veteran, argued that much of the pushback stems from a lack of understanding about how data centres operate and their economic upside. He called the opposition “insane” and attributed it to uninformed stakeholders who have not been shown the opportunities these facilities bring. Park emphasized that better marketing, transparent communication, and education campaigns are essential to bridge the knowledge gap and alleviate unfounded fears.
Rising Electricity Demand of AI Data Centres
The report notes that AI workloads have made data centres increasingly resource‑intensive. In 2023, U.S. data centres accounted for about 4.4 % of national electricity consumption, up from just 1.9 % in 2018. Analysts project this share could climb to 12 % by 2028 if current trends continue. This steep rise underscores the growing energy appetite of AI‑driven computing and highlights the urgency of improving efficiency and sourcing renewable power for these facilities.
Projected Power Demand Growth 2025‑2028
Total electricity demand from U.S. data centres is expected to increase from approximately 80 gigawatts (GW) in 2025 to 150 GW by 2028—a nearly 90 % rise in just three years. This growth trajectory reflects the relentless expansion of AI applications, cloud services, and high‑performance computing needs. Such projections place considerable pressure on grid planners and policymakers to ensure sufficient, sustainable generation capacity to meet the looming demand.
Water Consumption Trends and Impacts
Water use has also surged alongside electricity demand. Data centres consumed an estimated 17 billion gallons of water in 2023, which is triple the volume used in 2014. Much of this water is employed for cooling systems that prevent overheating of densely packed servers. Local communities, particularly in arid regions, worry that such withdrawals could exacerbate water scarcity, affect agriculture, and strain municipal supplies, reinforcing opposition rooted in resource stewardship concerns.
Carbon Emissions Projections for AI Workloads
Studies referenced in the report estimate that AI‑related data‑centre operations could generate between 24 million and 44 million metric tons of CO₂ annually by 2030. This range reflects variations in energy mix, efficiency gains, and the scale of AI deployment. If realized, these emissions would represent a notable fraction of U.S. greenhouse‑gas output, prompting calls for cleaner energy adoption, advanced cooling technologies, and stricter environmental standards for new facilities.
Limits of Local Bans and the Persistence of Demand
Park cautioned that imposing local moratoria or outright bans may not halt development; instead, projects could simply relocate to jurisdictions with fewer restrictions. He noted that developers often secure compute capacity years in advance, meaning that even if a site is blocked today, the contracted workload is already pre‑booked for future delivery. Consequently, sustained demand for AI infrastructure is likely to persist, pushing the industry to seek compromises—such as improved community benefits, transparent impact assessments, and greener designs—rather than relying solely on local opposition to shape siting outcomes.