forbes.com via Reddit

Americans prefer nuclear plants over AI data centers nearby

ai infrastructure compute ai-infrastructure public-opinion

Key insights

  • Americans now view AI data centers as less desirable neighbors than nuclear power plants, a direct head-to-head preference documented by survey data.
  • Noise, water consumption, and grid strain are the primary local grievances driving opposition to data center siting.
  • Virginia, Texas, and Georgia, the top three US data center markets, already have active state-level siting legislation in motion.

Why this matters

Hyperscalers are racing to deploy gigawatts of AI compute capacity under the assumption that capital and federal permitting reform are the binding constraints, but this survey signals that local public opposition is becoming a structural bottleneck that capital cannot simply outspend. State legislators in high-density data center markets now have explicit polling cover to impose water-use caps, grid-impact fees, and siting moratoriums without political cost. For AI infrastructure teams and the real estate arms of major cloud providers, site selection models that ignore community sentiment are now materially mispriced.

Summary

Americans rank AI data centers below nuclear power plants as preferred neighbors, according to a new survey that inverts one of the most durable assumptions in public energy opinion. Nuclear has carried a toxic public image for decades, shaped by Three Mile Island, Chernobyl, and Fukushima. The fact that hyperscaler infrastructure now ranks worse speaks to how fast the backlash against data center siting has accelerated. The finding adds a precise comparative frame to what had previously been a diffuse NIMBY sentiment. Earlier Gallup data showed opposition percentages, but this dataset draws a direct head-to-head: given the choice, communities would absorb the historically feared technology over the new one. Essentially: (Amazon, Microsoft, Google, Meta) are building gigawatt-class campuses across the US into a permitting and public-opinion environment that is actively hardening against them. - Survey respondents cited noise, water consumption, and grid strain as primary concerns about data centers, complaints that map directly onto hyperscaler operational footprints. - Nuclear plants, by contrast, have become associated with stable employment and low visible disruption once operational, shifting local calculus. - Zoning fights and state-level siting legislation are already pending in Virginia, Texas, and Georgia, the three largest US data center markets. The public opinion gap gives municipal and state governments political cover to impose stricter siting conditions, water-use caps, and grid-impact fees on hyperscaler builds through 2027.

Potential risks and opportunities

Risks

  • Amazon Web Services and Microsoft Azure face project delays in Virginia and Texas if county commissions cite public opposition data to justify permitting holds on campuses already in planning.
  • Hyperscalers that have publicly committed to 2026 and 2027 capacity targets could miss those targets if siting litigation, backed by this kind of survey evidence, adds 12-to-18-month delays.
  • Water utilities in data-center-dense regions (Northern Virginia, Central Texas) face political pressure to reject or reprice large industrial water contracts, raising operational costs for facilities already under construction.

Opportunities

  • Nuclear-adjacent land developers and co-location operators near existing nuclear sites (Constellation Energy, Talen Energy) gain direct negotiating leverage as hyperscalers seek pre-permitted, community-acceptable alternatives.
  • Modular and underground data center vendors (Nautilus Data Technologies, Green Mountain) can use this survey data as primary sales collateral to differentiate low-footprint designs to site-selection teams.
  • Public affairs and community-relations consultancies specializing in infrastructure siting gain a clear opening to sell hyperscalers on proactive local engagement programs before permitting applications are filed.

What we don't know yet

  • Survey methodology undisclosed in coverage: whether respondents were shown operational data centers or proposed greenfield campuses, which would significantly shift perceived risk.
  • Whether the nuclear preference holds in states with no active nuclear plants nearby, where abstract preference may differ from communities with direct operational experience.
  • Which specific hyperscalers (Amazon, Microsoft, Google, Meta) have active siting applications pending in jurisdictions where local opposition is now documented above 50 percent.