---
title: "Everyone wants artificial intelligence; nobody wants the building it lives in"
description: "As data centres multiply across America, 71 percent of residents say: build them somewhere else"
author: "Jan Sokol"
published: 2026-09-28T06:00:40.559Z
modified: 2026-09-28T22:19:05Z
url: https://rews.cc/a/everyone-wants-artificial-intelligence-nobody-wants-the-buil-027973
language: en
tags: ["ai", "infrastructure", "energy", "environment", "politics", "tech"]
publisher: "Rews (https://rews.cc)"
---

# Everyone wants artificial intelligence; nobody wants the building it lives in

*As data centres multiply across America, 71 percent of residents say: build them somewhere else*

By Jan Sokol · September 28, 2026 · https://rews.cc/a/everyone-wants-artificial-intelligence-nobody-wants-the-buil-027973

## In brief

- AI-driven data centre construction has sparked a new wave of NIMBY opposition in the US
- Global data centre capacity is projected to roughly double by 2030, from 103GW to 200GW, per USA Today
- A May Gallup poll found 71% of Americans oppose data centres in their area; 48% strongly oppose
- A large data centre uses some 18.9 million litres of water a day for cooling, equal to a town of 50,000
- Data centre bans and regulation pledges have become midterm campaign issues in the US

The Americans have given an old instinct a new word. NIMBY — Not In My Back Yard — names a familiar species of selfishness: the facility is acknowledged to be necessary for the common good, provided the common good is served in somebody else’s neighbourhood. The waste incinerator, the penitentiary, the psychiatric hospital, the power plant — each has had its turn as the thing everyone needs and no one will house. Citing the environment, the noise, safety, property values, the residents ask the eternal question: why precisely our street?

Now the machine of the age has joined the list. The spread of artificial intelligence has set off a boom in the construction of data centres, and a new NIMBY has been born — not because people doubt the usefulness of AI, but because of what its housing consumes and what it leaves behind: massive amounts of electricity and water, noise and heat, a burden of traffic, and, in the end, very few jobs.

## The appetite of the machine

The scale of the expansion is the first astonishment. According to USA Today, as investment in AI infrastructure grows, global data centre capacity is projected to roughly double by 2030 — from 103 gigawatts to 200. In 2023 data centres accounted for about 4 percent of America’s electricity; by 2030 that share is expected to surge to 11.8 percent. Worldwide, consumption is estimated to rise 26 percent in a single year, from 447 terawatt-hours in 2025 to 565 in 2026, and to approach a doubling by 2030.

Then there is the water. A large data centre uses roughly 18.9 million litres a day for cooling alone — an amount equivalent to the usage of a town of some 50,000 people. The cooling fans and power generation equipment produce noise at the level of a highway, enough to disturb sleep and inflict severe stress. And what the neighbours see is no prettier than what they hear: massive clusters of windowless buildings, raising fears of visual blight and declining property values.

## Jobs that vanish with the scaffolding

The customary defence of an unwelcome neighbour is employment, and here the data centre’s case is weak. The jobs spike temporarily during construction, then fall away sharply once the building is complete. The machine, once housed, largely tends itself.

The residents have noticed, and they have begun to organise their refusal into politics. In a Gallup survey in May, 71 percent of Americans said they would oppose a data centre being built in their area — and of those, 48 percent were strongly opposed. Ahead of this year’s midterm elections in the United States, campaign pledges to ban or regulate data centre construction have multiplied, turning a zoning quarrel into a contentious national issue.

## A question even the machine cannot answer

There is an irony here that needs no exaggeration. The artificial intelligence is said to know everything there is to know about this world — yet before the human psychology of the NIMBY, before the citizen who says “AI is necessary, but not in my neighbourhood,” the AI itself has produced no answer. It can model the weather and compose the letter, but it cannot tell us where to put the building that makes it possible.

Perhaps that is because the question is not technical at all. It is the oldest question of the polis: who bears the cost of what everyone wants. For the incinerator and the prison, the answer was usually the neighbourhood with the least power to refuse. The data centre, vast, windowless and thirsty, now knocks on the same doors.