When a Data Center Proposal Comes to Your Town

By Peter Montague, SEHN fellow

A new data center has been proposed for your town. What do you think? What do your neighbors think?

A lot of communities doubt the benefits of data centers, and many are opposing them outright. This paper outlines some of the concerns.

What is a data center?

Data centers are huge windowless warehouses stuffed with electronic equipment. They are essentially giant computing facilities shared by millions of people. They run the internet. That’s where you and I get our email and our news (Associated Press, Reuters, USA Today, New York Post, etc.). It’s where Facebook lives, along with WhatsApp, TikTok, Truth Social, Instagram, X, Threads, Bluesky and so on.

Your computer and your hand-held device depend on data centers, as do many or most businesses. But there are already more than 5,000 data centers in the United States, 10 times as many as in China or Germany. So, the internet will not shut down if new ones aren’t built right away. 

Most of the new data center boom is for “hyperscale” (i.e., humongous) data centers to support generative artificial intelligence (genAI, described below). So, the first thing to consider is genAI itself, the main product and purpose of most new data centers.

Without going into detail here, there are important questions about the future of unregulated AI and the global demise of democracy, as detailed recently in a year-long project by the Bulletin of the Atomic Scientists. The Bulletin asserts that “the people and organizations developing artificial intelligence applications are gaining control of the world's governance, information ecosystems, energy resources, military-industrial complex, and more.” Grass-roots pressure to regulate AI is growing. 

There are five kinds of artificial intelligence (AI)

Predictive AI predicts future outcomes, for example weather, or the maintenance requirements of power plants, or how proteins will fold in three dimensions based on two-dimensional pictures of amino acids (for discovery of new drugs).

Computer vision to mimic human vision of the real world—detect and identify objects, recognize humans, interpret x-rays, identify roadway hazards, diagnose the condition of crops, etc.

Physical AI refers to physical objects that interact with the real world, such as autonomous cars and trucks, robots, and drones. 

Agentic AI is a broad term encompassing “agents” designed to do specific tasks, such as virtual voice assistants on mobile devices (Siri, Alexa, Hey Google), systems that optimize the use of energy in buildings or the charging of batteries in electric vehicles, apps that detect spam and malware in emails and text messages, search engines (Google search, Perplexity.ai), fraud detectors in financial services, language translators, facial recognition to unlock devices, and much more. 

Generative AI refers to applications that generate new content, such as text, images, audio and video. This is the AI that has excited the public since late 2022. Examples include OpenAI’s ChatGPT, Anthropic’s Claude, Google’s Gemini, Meta’s Llama, China’s Deepseek, and many more.

Hyperscale hype

Despite mountains of hype (by the corporations that want to build new hyperscale data centers, mostly for genAI), the future financial success of genAI remains very uncertain. For one thing, genAI software inevitably (and spontaneously) produces some wrong answers (formerly known as “hallucinations,” but more recently labeled “fabrications”). This unavoidable defect in genAI will naturally limit its usefulness in many business and governmental applications. 

In addition, genAI suffers from two other unavoidable problems: “scheming” and “untruthfulness.” According to OpenAI, the company that brought you Chatgpt, scheming occurs when a genAI starts “pretending to be aligned [with the user’s wishes and intentions] while secretly pursuing some other agenda.” Finally, genAI has an observed tendency to tell lies, to make stuff up and, when challenged, to provide untrue explanations of how it reached particular conclusions. 

Even without the problems of “fabrications,” “scheming,” and “untruthfulness,” relentless hype about genAI seems likely to cause the construction of some unneeded AI data centers, in which case some or many projects will go bust. See Amory B. Lovins, “Artificial Intelligence Meets Natural Stupidity: Managing the Risks” (May 10, 2025).  See also Michel Gelobter, “AI and Energy Use,” (Sept. 2025). 

For these reasons, data center proposals should always include discussion of what happens if the data center fails to thrive—how would the community be affected?

Typically, whether projects succeed or fail, local people will be forced to pay more for their electricity as utilities and boards of public utilities approve complex, obscure, and subjective rate-making procedures that end up shifting electricity costs from data center owners onto the general public. 

Investor-owned public utilities are regulated. The only way they can earn more profit is by expanding their installed base of equipment. Thus, they have a financial incentive to expand their physical part of the power-distribution system (the grid), such as new power lines and substations and/or the number of power plants within their jurisdiction. For their part, boards of public utilities, which regulate investor-owned utilities, are political bodies that may have political reasons for approving a “gee whiz” data center proposal. No one wants to be seen as “opposing progress.”

If genAI does not meet financial expectations, promised AI data centers will not be built or will shut down after being built, investors may lose their shirts, projected local jobs and tax revenues will not materialize, utilities are likely to increase rates for all their customers to recoup failed investment(s), thus raising electricity costs for everyone unfairly (especially hurting people with low income), and in some cases built-for-purpose fossil-fueled power plants will continue to make global heating worse.

Water, noise, and fossil fuel pollution

Poorly-designed data centers can create unfair and unnecessary burdens on local communities, including:

a. Consuming large volumes of water for evaporative cooling when water-free cooling is already available and being used in well-designed data centers;

b. Dangerous levels of noise. Well-designed data centers would be built to drastically reduce typical data center noise levels. This is a serious problem for data center staff because noise levels inside data centers can exceed the safety standard set by the Occupational Safety and Health Administration (OSHA), which is 85 decibels (dBA). Exceeding the OSHA standard can cause hearing loss. 

Furthermore, data centers typically emit to the outside world a loud low-frequency hum or buzz that is very disturbing to humans and wildlife and is difficult to reduce.  This constant noise can travel one kilometer (0.6 miles) or more and can cause headaches, stress, and sleep disturbance. Lack of sleep and stress, in turn, can cause anxiety, cognitive impairment, and cardiovascular risk. 

c. Reliance on fossil fuels (natural gas, diesel oil) for 24/7 power. Data centers can be, and are being, run entirely on renewable energy with storage, so new well-designed data centers do not need to be powered by dirty fossil fuels. 

Backup Power Requirements

What happens when there’s a power outage?

Data centers need to run 24/7, so backup power must be provided for those times when the main power source fails. 

A tiered power backup strategy could include:

Tier 1 (0 to 4 hours): lithium-ion batteries for immediate response and short-term backup. Other possible battery backup technologies include sodium-ion batteries, vanadium redox flow batteries, iron flow batteries, and organic flow batteries.

Tier 2: (4 to 24 hours): Hydrogen fuel cells or molten salt thermal storage for extended outages.

Tier 3: (24+ hours): Compressed air energy storage or pumped hydro power are possible options for extended outages.  Site selection will determine whether these technologies could be available.

If the chosen site does not allow consideration of compressed air storage or pumped hydro, then perhaps the new data center needs to be sited somewhere else.

When a data center proposal arrives in your town, its owners will most likely propose only fossil-fueled backup power. No modern data center needs to depend on fossil energy as its primary power source, and fossil fuels could be the choice of last resort even for long-term backup power.

Specifically, diesel generators for extended backup should be rejected because they must be run frequently to prove they are still working, they produce lots of fine particle pollution, and they are very noisy.

The owners could be required to show what backup power choices they considered and why they made the choices they made. 

Advantages of renewable energy

Compared to fossil energy, renewable energy with storage is faster and cheaper to build, requires no water, and emits far less carbon dioxide (CO2).  

Renewable energy has an important advantage over the fossil alternatives: If a renewable-powered data center fails to thrive, its renewable power source will still be useful to replace fossil energy on the grid, thus advancing the urgent energy transition from fossil fuels to renewables. 

As Amory Lovins and his RMI consulting firm point out, large-scale renewable energy sources can be built within a couple of years but connecting them to the grid requires a wait of 5 to 7 years (depending on location). RMI recommends a solution: build renewable energy with backup storage near a shuttered fossil fuel plant that was previously connected to the grid.  They call these situations “power couples.”

Jobs

Like any project, data centers will hype the jobs being created. But it’s worth looking closely at such claims. Most of the jobs are probably in construction. Of course, construction jobs are usually good jobs, but they probably only last 1 to 2 years. And how many of those jobs will be filled by local people, versus workers brought in from somewhere else?

More fundamentally, how many jobs is AI displacing? Workers, even those getting short-term jobs, must consider the long-term impacts of AI on employment and on union membership.

Public subsidies

Thirty-six states currently use tax dollars to subsidize data centers being built by and for some of the wealthiest corporations on the planet—Apple, Alphabet [owner of Google], Amazon, Meta [owner of Facebook], Microsoft, Nvidia, and Tesla—all of which can easily afford to pay for their own infrastructure expansion. 

Local communities should not be asked to offer tax incentives to data center owners on the promise of local jobs and local tax revenues. Data center owners should pay for all the costs their proposals entail. 

Subsidizing data centers reflects the political power of AI corporations, not public policy that serves the public interest.

There is abundant evidence that federal and state regulatory bodies routinely approve legal agreements that shift electricity costs created by new data centers onto residential customers. This is highway robbery endorsed by governments and should be called out and forbidden. Data centers could easily pay all of their own costs with not one dime of public subsidy.

The big picture

The 2024 United States Data Center Energy Usage Report from the Lawrence Berkeley National Laboratory (LBNL) concluded that U.S. data centers consumed 176 terawatt-hours (TWh) of electricity in 2023, representing 4.4 percent of total U.S. electricity consumption. A terawatt-hour is a unit of power equal to a trillion (1012) watts.

The LBNL study projects U.S. data center electricity use (excluding crypto) in 2028 as somewhere between 325 and 580 TWh, representing somewhere between 6.7 percent and 12.0 percent of total U.S. electricity use in 2028. 

These numbers imply the need for 30 to 80 new power plants (each rated at 900 megawatts, average). In our present (2025) political climate, these would likely be mostly natural-gas power plants, contributing to global heating.  

To put into perspective the projected growth in electricity demand by data centers, according to International Energy Agency researchers, between 2023 and 2030, compared to growth in U.S. data center demand, space heating demand will grow twice as much; space cooling (air conditioning) demand will grow three times as much, demand by electric vehicles 3.25 times as much.  

On the other hand, data centers tend to cluster in locations near population centers with fast internet access and low-cost electricity, thus putting unusual cumulative stress on local power systems. Given that data centers can be built within 2 or 3 years but power plants and grid expansion may take far longer, the risk of grid failures is real.

Conclusion

To try to make data centers acceptable to local communities, data center owners can propose centers that offer substantial financial benefits to the host community, use almost no water, create no noise pollution, emit no dangerous fine-particle air pollution, emit no carbon dioxide during normal operations, destroy no farmland, and are sited in the least obtrusive way possible, far from schools, homes, places of worship, historic or sacred sites, cultural sites, and community amenities. 

In some locations, data center owners may have “by right” permission to build a center without asking for special zoning, which means they can build without any public hearings. However, if they want to earn trust from the community, owners will voluntarily hold public meetings to explain their project and hear the community’s concerns. 

Owners of proposed data centers should be prepared to show—in terms that community members can understand—how their proposal incorporates best available technology to minimize environmental and social harm. They could analyze the “cumulative impacts” of their project in combination with existing conditions and other pending proposals and publish their findings for the community. Consideration of cumulative impacts will then suggest reliance on the precautionary principle as the basis for decisions.     

Finally, data center owners could voluntarily complete and publish an “environmental justice impact statement” to clarify who will bear most of environmental and public health stressors from their proposed industrial warehouse, on top of the stressors local people are already enduring. 

Then let the community decide.