アイコン_インストール_ios_web アイコン_インストール_ios_web アイコン_インストール_android_web

Even if it means facing gunfire and imprisonment, why are small-town Americans still opposing AI data centers?

分析13時間前发布 ワイアット
716 0

The city council signed off without holding a public hearing. Subsequently, the local residents’ group Wake Up JeffCo sued the city government and CRG in St. Louis County Court, and a recall petition against the mayor was also initiated. According to a summary by Tom’s Hardware, around the same time, the home of Indianapolis City Councilor Ron Gibson was shot at with over a dozen bullets at the end of 2025, with a note reading “No Data Centers” left at the door.

Festus is not an isolated case. Not long ago, the home of Indianapolis City Councilor Ron Gibson was targeted by a shooter who fired 13 shots late at night, waking his 8-year-old son. A handwritten note was left at the door: “No data centers.” The FBI has launched an investigation. Jordyn Abrams, a researcher at George Washington University’s Program on Extremism, pointed out that data centers are becoming targets for anti-technology, anti-government extremists.

Even if it means facing gunfire and imprisonment, why are small-town Americans still opposing AI data centers?

Scene of the Ron Gibson shooting case

In its Q2 2025 report, the advocacy group Data Center Watch updated the number of organized opposition groups from 142 (across 24 states) a year ago to 188 (across 40 states). The value of projects halted or delayed rose from $64 billion to $162 billion. On April 1, 2026, Port Washington, Wisconsin, passed the nation’s first referendum explicitly targeting data centers, with 66% of voters approving the addition of a mandatory referendum threshold for Tax Increment Financing (TIF) subsidy projects exceeding $10 million.

Collectively, these events answer the same question: Could the real bottleneck for AI capacity expansion get stuck at the county and city ballot box?

Widespread Backlash

Plotting events from the past 23 months on a U.S. map reveals two levels of backlash. One is at the state level, where eight states have introduced or passed data center moratorium bills, including Maine (passed by the House 82-62, effective until 2027), Vermont (moratorium until July 2030), Virginia (introduced by Democratic Representative Irene Shin, moratorium until 2028), Georgia, Maryland, South Dakota, Wisconsin, and Minnesota. This level involves legislative action, with the broadest impact but the slowest progress.

Even if it means facing gunfire and imprisonment, why are small-town Americans still opposing AI data centers?

The other level is at the county and city level, where the backlash is denser and more intense. The Chandler, Arizona city council unanimously rejected (7-0) Active Infrastructure’s $2.5 billion project in December 2025 (lobbied by former U.S. Senator Kyrsten Sinema). In the same state, the Tucson city council is currently reviewing data center zoning restriction ordinances in April 2026, with public comment open until the end of the month. Hays County / San Marcos, Texas, rejected a $1.5 billion project by a 5-2 vote. Other locations include Cascade Locks, Oregon; Chesterton, Indiana; Catlett Station, Virginia; Peculiar, Missouri; and Lansing, Michigan. According to Data Center Watch’s compilation, at least 10 states have seen direct municipal-level rejections or developer withdrawals.

Over half of the high-conflict incidents are concentrated in the Midwest and Mid-South. This region has had relatively ample surplus grid capacity over the past decade and was a hotspot for the previous wave of data center investment attraction. That the backlash is now concentrated in the same area, from another perspective, means the supply side, expanding from “states with power plant surpluses,” has collided with the most sensitive layer of local politics.

$60 Billion: Not on the Same Scale

Festus’s fiscal scale dictates that the $60 billion figure is simply impossible for the city council to process normally. According to local newspaper myleaderpaper citing municipal budget documents, Festus’s FY2025 general fund plus public safety operating budget is $17.64 million. Total municipal expenditures for FY2024 were $37.41 million, with projected reserve funds at the end of FY2025 being $28.09 million.

The data center project’s $60 billion is approximately 340 times the annual operating budget. Per capita for the city’s 13,200 residents, that’s $450,000. In relative scale, this is not a local development project up for discussion; it’s a small town being plugged into a capital pipeline entirely unrelated to itself.

Even if it means facing gunfire and imprisonment, why are small-town Americans still opposing AI data centers?

Comparing this to the median annual income per capita for Festus residents (around $35,000 for non-metro areas in Missouri) clarifies the issue further. Any decimal point in the data center contract is larger than the entire community’s lifetime per capita disposable income. Local officials lack procedural experience in checks and balances when facing numbers of this magnitude. The reason the Festus city council resolution was criticized as “no public hearing” was held is, technically, because such projects typically operate under commercial confidentiality clauses (neither the developer nor the end client is disclosed), making normal city council deliberation processes unable to review confidential contracts. This is a structural loophole, not the negligence of individual council members.

Precisely because of the scale disparity, breaking down a data center contract into a scale manageable by a local council is inherently unworkable. This is also why, over the past 12 months, the backlash path hasn’t been resolved within the deliberation process but has instead utilized three external weapons: recall, litigation, and referendum. The recall of four Festus council members, the simultaneous acceptance of a lawsuit by a residents’ group in St. Louis County Court, and the initiation of a mayoral recall petition is a rare case where all three paths were triggered simultaneously.

One Data Center, Consuming a Small Town

The electricity consumption of an AI data center is most palpable when compared to a small American city. A fully loaded 200 MW AI data center, operating at an 86% load factor, consumes about 1,500 GWh annually. A small U.S. city of 100,000 residents consumes about 420 GWh per year (based on EIA’s average annual residential electricity use of 10.5 MWh per household, with 2.5 persons per household). The data center uses 3.6 times the electricity of the city’s residents. And this is just electricity, not counting cooling and associated water use.

The water comparison is reversed but even more intuitive. Based on typical residential water use values from the USGS (100 gallons per person per day), a small city of 100,000 uses about 3.65 billion gallons annually. A hyperscale AI data center, using the Google Council Bluffs (the largest Google data center in the U.S.) as a benchmark, consumes 500 million gallons annually. In absolute terms, the data center uses 13.7% of the small city’s water. But on another scale, it’s equivalent to the annual drinking water for 14,000 people. For a small town with a permanent population of 10,000 to 50,000, this means dedicating a significant portion of the city’s water supply system to a single user. According to Lawrence Berkeley Lab’s 2024 data center energy report, U.S. data centers used 17 billion gallons of water for direct cooling in 2023 and 211 billion gallons of indirect water (for power generation). Direct water use is projected to double or quadruple by 2028.

Even if it means facing gunfire and imprisonment, why are small-town Americans still opposing AI data centers?

The most common protest chant is “Our wells will run dry.” Looking at the numbers, this isn’t just emotional rhetoric. In Loudoun County, Virginia (the county with the highest density of data centers in the U.S.), the potential water use by data centers within the county in 2023 was 899 million gallons, accounting for about 10% of the county’s total water use, according to local water utility data cited by the Sierra Club and Grist. If this is the situation at the county level, the numbers for towns will only be more extreme.

Planned Capacity is Entering the Backlash Window

The actual operational capacity of U.S. data centers, according to FERC and Wood Mackenzie Q4 2025 data, is about 50 GW. The total pipeline under planning is 241 GW, with 33% (about 80 GW) in active development and the remaining 67% (about 161 GW) not yet started. BloombergNEF predicts the U.S. will add 97 GW of capacity from 2025 to 2030, with data center peak power demand reaching 106 GW by 2035. All these numbers point to one fact: the vast majority of capacity is still on the drawing board, not yet built.

Even if it means facing gunfire and imprisonment, why are small-town Americans still opposing AI data centers?

According to data disclosed by Sightline Climate via TechRadar, 30% to 50% of the 16 GW originally scheduled for commissioning in 2026 is expected to be canceled or delayed. Meanwhile, Data Center Watch data shows that from May 2024 to March 2025, organized opposition blocked or delayed $64 billion worth of data center projects. In Q2 2025 alone, that figure was $98 billion, corresponding to 20 projects. The amount blocked in a single quarter already exceeded the cumulative total of the previous 10 months.

This creates a timing mismatch. Capital has committed to multiplying U.S. data center capacity several times over the next five years, but the new capacity must pass through county and city-level approvals one hurdle at a time. The more planned capacity, the larger the surface area vulnerable to backlash. The reason a case like Festus could escalate from a city council vote to recalls and lawsuits within a month is not because it’s special, but because the number of opposition groups increased by 46 in one year (according to Data Center Watch’s Q2 2025 report), and they are sharing templatized legal tools across state lines, including TIF subsidy referendums, zoning lawsuits, and councilor recall petitions. Whether the long-term power contracts signed by frontier labs will be fulfilled depends on which counties these contracts land in and what kind of residents are watching those counties’ councils.

The bottleneck for AI capacity expansion has, for the first time, moved beyond the power contract negotiation table and appeared on the recall ballots cast by 13,200 people.

この記事はインターネットから得たものです。 Even if it means facing gunfire and imprisonment, why are small-town Americans still opposing AI data centers?

Related: U.S. and Iran Brew New Round of Talks, Bitcoin Recovers Losses Alongside U.S. Stocks

On April 14th, BTC surged from around $70,500, reclaiming the $74,000 level. ETH rose 7.63% to approximately $2,367. The total crypto market cap recovered to about $2.52 trillion, with Bitcoin’s dominance holding steady between 57% and 58%. According to data from Alternative.me, today’s Crypto Fear & Greed Index rose to 21 (yesterday’s reading was 12, indicating “Extreme Fear”), suggesting the market remains in an “Extreme Fear” state. It hit a low of 8 in early April, a rare reading since 2022. In traditional financial markets, the S&P 500 rose 1.02% to 6,886 points, recovering all losses incurred since the US and Israel provoked the Iran conflict and surpassing its February 27th closing level of 6,878 points. The Nasdaq gained 1.23%, marking its ninth consecutive day of gains and its longest…

© 版权声明

相关文章