Looking for the Wizard of Oz

If my town
says no Who gains when a community builds or blocks an AI data center? John Lee Ohio tariff Hypothetical 15 GW campus · AI-generated illustration Concept rendering of a hypothetical 15 GW data center campus beside a town, with rows of data halls, substations and transmission lines.

Railroad construction averaged about 2.4% of U.S. GDP per year from 1865 to 1890. Stijn Van Nieuwerburgh estimates that if U.S. AI infrastructure reaches 188 GW by 2032, investment would average 3.2% of GDP per year over 2025–2032, about one-third more relative to the size of the economy. The paper estimates nearly $9 trillion for that nationwide scenario. The illustration depicts one hypothetical campus. The comparison concerns the size of the investment and does not predict its financial outcome.

A utility may need to build transmission lines or substations for a proposed campus. If the customer uses less power than it reserved, the tariff helps determine who pays for the unused capacity. AEP Ohio’s data center tariff requires covered customers to pay for a minimum share of their contracted capacity. The image illustrates a hypothetical 15 GW campus.

Sources and credits
https://www.aepohio.com/company/about/rates/data-center-tariff/
AI investment scenario: https://www.nber.org/papers/w35865
Historical railroad benchmark: https://business.columbia.edu/sites/default/files-efs/imce-uploads/svannieuwerburgh/papers/FinancingAIBuildout_03192026.pdf

Virginia’s 41 GW includes projects on paper Reported power by stage 9 GW Operational 8 GW Under construction 23 GW Planned October 2026 · Rounded figures; reported power varies in scope and coverage. Virginia data U.S. totals DCMap methodology

By mid-2022, Virginia had more hyperscale data center capacity than all of Europe. Synergy compared the power available to IT equipment across 19 large cloud and internet service providers. Amazon, Microsoft and Google already operated infrastructure in Northern Virginia.

DCMap’s October 2026 Virginia inventory reports 9.0 GW at operating sites, 8.3 GW under construction and 23.3 GW planned. The rounded 41 GW total includes proposals whose completion is uncertain. The figures describe reported capacity rather than metered electricity consumption. Published figures vary in scope, and some sites lack power data.

Synergy and DCMap cover different facilities and use different power measures. Their figures establish Virginia’s importance before the generative AI boom and the scale of its current infrastructure. They do not establish how much of the 9 GW existed in 2022 or how much was subsequently built for AI.

Sources and credits
https://dcmap.us/states/virginia/
https://dcmap.us/stats/
https://dcmap.us/methodology/
https://www.srgresearch.com/articles/virginia-still-has-more-hyperscale-data-center-capacity-than-either-europe-or-china

Why would you want one nearby? Among supporters Among opponents Gallup · April 2026 · Open-ended responses Percent within each group mentioning a reason. Responses can overlap.

More people expected harm to neighbors’ quality of life Share saying data centers were mostly bad for each area 0% 10% 20% 30% 40% 50% 60% Nearby quality of life +19 Environment +15 Home energy costs +12 Local jobs +9 Local tax revenue +9 Bars run from January to August. Whiskers show approximate 95% intervals for the change. Assumes independent waves. Calculation and sources in speaker notes. Pew Research Center · January and July–August 2026

Pew asked U.S. adults how data centers affect five aspects of local life. From January to August, the share answering “mostly bad” rose from 30% to 49% for nearby residents’ quality of life, 39% to 54% for the environment, 38% to 50% for home energy costs, 15% to 24% for local jobs and 12% to 21% for local tax revenue. Among rural adults, the corresponding shares rose from 28% to 45% for quality of life, 32% to 50% for the environment and 32% to 46% for energy costs. The share of all adults who had never heard of data centers fell from 25% to 12%. These group-level changes do not identify which individuals changed their views.

Pew reports full-sample 95% margins of ±1.4 percentage points in January (n = 8,512) and ±1.3 in July–August (n = 10,548). Assuming independent waves, SE(change) = √[(1.4/1.96)² + (1.3/1.96)²] ≈ 0.975 points. The approximate 95% half-width is 1.96 × SE ≈ 1.91 points. The intervals for the increases are 17.1–20.9 points for quality of life, 13.1–16.9 for the environment, 10.1–13.9 for energy costs and 7.1–10.9 for both jobs and tax revenue. Each whisker shows uncertainty in the change, positioned at its August endpoint on the absolute percentage axis. These calculations use rounded survey-wide margins. Panel overlap and item-specific design effects could alter the intervals.

Sources and credits
https://www.pewresearch.org/short-reads/2026/09/22/americans-views-of-data-centers-have-turned-more-negative/
https://www.pewresearch.org/wp-content/uploads/sites/20/2026/03/SR_26.03.12_data-centers_topline.pdf
https://www.pewresearch.org/wp-content/uploads/sites/20/2026/09/SR_26.09.22_data-centers_methodology.pdf

Which would you want nearby? AI data center 71% Nuclear power plant 53% Share of U.S. adults opposed to local construction · March 2026 Gallup survey November AP Voter Poll: about ¾ of voters opposed a new data center nearby. AP Voter Poll

How data centers figured in the vote United States Oregon, Ohio ≈25% 61% said AI data centers were crucial to their congressional vote. voted for a 25 MW limit on data centers. AP Voter Poll Campaign funding Vote & Vantage response

Nationally, about a quarter of voters told the AP Voter Poll that AI data centers were crucial to their congressional vote. That response measures the importance voters assigned to the issue. It does not identify which position or candidate they favored.

In Oregon, Ohio, Jim and Laura Davis helped gather signatures to put a 25 MW data center cap on the ballot. About 61% voted for it. Vantage contributed $250,000 to an effort to defeat the cap. Four union groups contributed $85,000 to a separate committee on the same side. Two days after the vote, Vantage said the project would not proceed as proposed.

Sources and credits
https://www.cleveland.com/open/2026/11/ohio-voters-reject-data-centers-despite-hundreds-of-thousands-in-opposition-spending.html
https://www.13abc.com/2026/11/05/vantage-data-centers-says-it-will-not-move-forward-after-oregon-voters-approve-issue-25-ban/
https://apnews.com/article/63b33be8884d867d3ac831040d683a2c

Ohio tariff Lease economics · SEC Photo: jbthemilker CC BY-SA 4.0 Who gains when fewer data centers get built? In Oregon, Ohio, residents organized and stopped a proposed data center. If fewer centers get built and demand stays high, companies that already own them could charge more. Who stands to gain most from this shift in public opinion? What reason might they have to shape it? Electrical substation in Ashtabula, Ohio, illustrative photograph

Oregon’s vote changed the outcome for one proposed campus. A project that relocates takes potential jobs and tax revenue to another community. If many projects stall and demand stays high, the owners of existing centers may gain bargaining power as leases come up for renewal. The shift in public opinion raises a broader question about who gains most and what they stand to gain by influencing the debate.

The photograph shows the Ashtabula substation in Ohio. Photo by jbthemilker, licensed under CC BY-SA 4.0.

Sources and credits
https://www.aepohio.com/company/about/rates/data-center-tariff/
https://www.sec.gov/Archives/edgar/data/1297996/000110465926015365/dlr-20251231x10k.htm
https://commons.wikimedia.org/wiki/File:Ashtabula_substation.png
https://creativecommons.org/licenses/by-sa/4.0/
Oregon vote and developer response: https://www.13abc.com/2026/11/05/vantage-data-centers-says-it-will-not-move-forward-after-oregon-voters-approve-issue-25-ban/