Nestled between expansive agricultural fields and the stunning Great Lakes, Port Washington, Wisconsin, is home to nearly 13,000 residents. This charming community offers a unique blend of New England allure and Midwestern warmth, making it an inviting small-town retreat just north of Milwaukee.

The town proudly embraces its maritime roots with a variety of annual events. One such highlight is the Polar Bear Plunge held every New Year’s Day, where hundreds brave the icy waters of Lake Michigan. Equally significant is the long-standing Port Fish Day, a major community fundraiser and one of the year’s most anticipated events.

Brick buildings of homes and small businesses line the wide streets that lead to the shores of Lake Michigan and a historic lighthouse. 

The city celebrates the community’s nautical heritage year round with events ranging from the annual Polar Bear Plunge on New Year’s Day, when hundreds of people charge into Lake Michigan amid freezing temps, to the decades-old Port Fish Day, which is one of the largest community fundraisers and events of the year. 

Just outside the city are parcels of dirt plots and lush, green agricultural fields growing soybeans and grains. 

Roughly 2.5 million square feet, or about 43 American football fields, will soon be home to a multibillion-dollar data center “Lighthouse” campus as part of OpenAI’s Stargate project — a facility that runs artificial intelligence ChatGPT. It will join the more than 4,000 other data centers sprawled across the U.S.

The data center, already approved and expected to complete construction in 2028, caused a stir nationwide.

Charlie Berens — a comedian and influencer with a combined 10 million followers across his Instagram, TikTok, YouTube and Facebook accounts — recently began using his platform and popularity to speak out against the data centers in his backyard. 

“I was pissed that people’s voices were being ignored,” he said. “These are my neighbors. People deserve a voice.”

The Oconomowoc, Wisconsin, native made multiple videos on the topic, ranging from encouraging folks to show up to meetings about the data center to updating his audience after council passed it. 

He used his Emmy-winning journalism research skills and comedic bite to deliver the hard news in videos that amassed hundreds of thousands of likes and millions of views. 

“It’s just a slap in the face, with all the AI stuff,” he said. “Anytime you use any kind of social media thing or any app, you’re the product. So, we’re like the lobster in the tank at Red Lobster.”

Other voices echo his frustration. People showed up at city meetings and events in protest. Port Washington city councils, commissions and boards unanimously approved the project despite the outcry. 

In fact, so many people showed up to council meetings, the city had to move to a hotel conference center, something Mayor Ted Neitzke IV said was caused by “social media influencers mobilizing individuals from inside and outside our community.” Multiple people have been arrested during meetings as concerns continued to grow. 

This isn’t something just happening in this small midwestern community. It’s happening nationwide. Take Abilene, Texas; Southaven, Mississippi; and Boardman, Oregon, for example. 

Each rural area is being eyed by tech giants to build their next set of data centers. 

Municipal councils like that of Port Washington argue data centers will boost the economy by increasing the tax base, increasing construction, allowing farmers to willingly sell their land and providing job and education opportunities. 

But as AI and tech ambitions grow, so do concerns about their impact on the world. It leaves people asking how they’re going to make room in their established infrastructures — such as water, power, and land — for new technologies. 

Residents aren’t the only ones asking questions.

In the 1970s, members of Congress established the Environmental and Energy Study Institute. The independent nonprofit was designed with a goal to provide science-based educational resources to policymakers and the public. Its founding was the result of emerging concerns about global and energy problems.  

“(AI) is a topic everyone wants to talk about, but no one seems to know what exactly to talk about,” EESI President Daniel Bresette said. “It involves a little bit of innovation, a little bit of competition with China, a little bit of ‘shiny new object’ syndrome.”

He said there’s a lot of bipartisan interest on Capitol Hill because “it hits on so many different issues that you can’t really find anyone who doesn’t really, really want to talk about it.”

Berens said he speaks up about data centers because they aren’t just taking away community voices. 

“You take our data and now you want our land, you want our water, you want our air,” he said. “None of it makes sense, and I just want people to be aware that this is happening.”

Data centers holding intangible data need a physical place to live. 

The haste in which data centers are developed concern people like Berens and leave them with more questions like “What’s the rush? Why not wait a little bit?”

“In many ways, it’s an arms race: firms are building aggressively today to avoid being bottlenecked tomorrow as AI, cloud services, and digital workloads continue to scale at unprecedented speed,” wrote ChatGPT, the most popular free AI chatbot available, when asked Berens’ question. 

OpenAI’s ChatGPT launched in 2022 and paved the way for what we see today — tech giants fighting to dominate the new field. 

Hundreds of data centers have been proposed in 2025, bolstered by the Trump administration incentivizing companies to build on and invest in American property. From small communities like Port Washington to the heart of Silicon Valley in San Jose, California. 

The global average size of data centers equals roughly 1.74 American football fields, including endzones. That’s roughly 100,000 square feet, according to the Institutional Real Estate, Inc.

But hyperscale centers are massive.

The proposed location in San Jose plans for 1.7 million square feet of data centers across four two-story buildings. That’s the equivalent of about 30 football fields or six Costco Wholesale stores.  

In Lyon Township, Michigan, citizens experienced something similar to those in Port Washington on the other side of the lake. The township’s planning commission conditionally approved the planned data center consisting of six buildings on more than 1.8 million square feet in September.

Kimberly Killian, who will soon live some 300-feet away from the new building, told WXYZ she was “extremely upset” about the plan. 

“I’m heartbroken,” she said. “Nobody wants that in their backyard,” 

Northern Virginia is home to 668 data centers, according to the Data Center Map. It’s the largest data center hub in the U.S. and where “data center alley” sits. 

Data centers aggregate here because it’s close to policy makers in Washington D.C., has tax incentives, cheap power, and because of the region’s “world class” fiber optic network, Virginia Tech associate professor Landon Marston said in a July podcast.

Dominion Energy spokesperson Aaron Ruby compared Northern Virginia and its data centers to other industry hubs like automotive in Detroit, aerospace in Texas or Silicon Valley in California.

Companies find themselves buying up industrial or agricultural zoned land to build their centers on. 

Raymond Suarez, 60, lives in Florida and fears data centers will be built on land that should be preserved or used for other things like farming, according to a poll from The Associated Press-NORC Center for Public Affairs Research and the Energy Policy Institute at the University of Chicago.

His concerns are already taking place. In central Ohio, 130 data centers are battling farmers for land and water.

“We are literally taking the nation’s breadbasket, where it’s most productive (and) most advantageous to farm, and turning it over for industrial use,” Bryn Bird, a Licking County resident and president of Ohio Farmers Union, which represents more than 2,500 family farms, told the Associated Press. 

So, what do data centers do? 

Data centers are where “the cloud” lives. 

AI and online data is no longer an abstract idea. It is ingrained into communities as a physical object and companies pour billions of dollars to expand. 

Data centers are physical buildings, either a singular structure or a cluster of buildings called a campus. They house servers, networking gear, storage, power and cooling systems, and security infrastructure. 

Buildings store and process data and route and deliver information. They are quite literally the backbone and brain of the internet, which is why there are such detrimental issues when they fail.

Inside data centers are “row upon row upon row of machines all working together” to provide services, an employee in a Google Workspace tour of the company’s South Carolina data center said. 

Data centers hold all information gathered, submitted, and created on the internet. Every “like,” post, password, profile picture, etc., is stored here. While they have been around since the start of the internet, AI data centers are changing the internet and landscape — literally. 

Companies strategically build data centers across the world to make sure everyone can use the internet with decent speeds. Many are located in Northern Virginia, but some of the largest centers in the world are scattered throughout the U.S. in places like Sparks, Nevada; Memphis, Tennessee; and Atlanta, Georgia.

The U.S. overwhelmingly has the most data centers, with more than 4,000 on domestic soil. The United Kingdom falls in second place with 499, according to Statista. 

The closer a user is to a data center, the more quickly they’re able to access the internet because distance increases latency, or delay.

The main difference between a data center and an AI data center are the use of Graphics Processing Units, or GPUs. 

Typical data centers need a computer’s “brain,” the Central Processing Unit, or CPU, to operate while AI data centers need a computer’s “soul,” the GPU. 

CPUs can operate without GPUs, but GPUs must have CPUs to function. And they take a lot energy to run. 

Alex S. uses multiple AI tools on a daily basis for work. She is among nearly 800 million others who use ChatGPT weekly, according to a National Bureau of Economic Research study. 

She uses half a dozen AI tools in her work to build custom AI solutions for companies using the new technology, and ships and delivers generative AI to them. ChatGPT is the main tool she uses as a “pretty central part” of her workflow.

She estimates sending 15 to 20 or more prompts or questions a day, ranging in tasks from transcribing her out loud brainstorming sessions to summarizing her Slack threads or taking notes during meetings. 

OpenAI estimated one average query, defined by ChatGPT as a user sending between five and eight messages when asked for clarification, uses roughly 0.0034 kilowatt-hours of electricity. A typical 700-watt household blender uses about the same amount of electricity to run for 15 seconds.

It may seem like a small amount, but when combined with the estimated 18 billion queries sent to ChatGPT on a weekly basis, power use grows exponentially. 

Dominion Energy, an electric services company serving 20 states along the east coast and south, supplies power to roughly 450 data centers, according to Ruby. This makes Dominion the world’s biggest data center-serving electric utility.

“Virginia, particularly Northern Virginia, is the largest data center market in the world and it has been for decades now,” he said. “It’s the birthplace of the internet.”

Ruby said all the data centers within Dominion’s service area combined use the same amount of energy as about a million homes. 

“One single data center, give or take, might use the same amount of power as 25,000 homes,” he said. 

Data centers are plugged into the grid or getting their power from an energy supplier like Dominion. 

It causes concern about energy consumption, but EESI worries about the already complicated transmission of energy and dated infrastructure that is rapidly being forced to accommodate data centers.

Ruby said this isn’t a concern if energy suppliers are doing their job right. 

“It’s our job to make sure there’s always enough reliable power to serve our customers when they want to use it,” he said. 

Ruby highlighted how they also have many large customers like military bases, schools and hospitals among millions of residents and hundreds of data centers in their service territory. 

Energy demand was relatively stagnant for a few decades, but is forecast to more than double by 2045.  

This is due in large part to “the digitization of the economy served by data centers and electrification of energy needs,” according to Dominion Energy’s updated Integrated Resource Plan. 

These are historic levels of growth and have spread across the country into states like Oregon, Iowa and Wyoming, which are seeing energy use by data centers in the double digits. 

Energy costs are increasing along with this new demand. Millions of people in 41 states across the country are seeing energy costs increase, according to a Center for American Progress study. 

The energy-hungry data centers aren’t solely to blame. Other causes include increased use of electric vehicles and appliances and massive rollbacks of green energy tax breaks from the Biden-era Inflation Reduction Act. Also to blame: price increases for natural gas, which generates more than 40% of U.S. electricity, as President Donald Trump’s tariffs make exports to international customers grow more expensive.

The changes are already impacting people across the country, like 58-year-old Kevin Halsey of Normal, Illinois.

He told the Associated Press his monthly electricity bill used to be $90 in the summer because he had solar panels. That bill has risen to roughly $300. 

“I’ve got to be pessimistic,” he told AP. “I don’t see this as getting better.”

The U.S. Energy Information Administration found that the average price of electricity in America has increased 13% since 2022. The EIA said retail electricity prices have increased faster than the rate of inflation since 2022 and are expected to continue on this trajectory until 2026. 

“People want a more affordable Virginia, and they have every right to demand it,” Virginia Lt. Gov. Ghazala Hashmi said in an energy regulation meeting. “Over the past few years, Virginians have felt the sting of rising costs everywhere.”

Studies credit the higher costs not only to increased use but also to transmission and distribution, infrastructure upgrades, and new lines to meet new demand.

A Lawrence Berkeley study found that more customers equals more people to pay for these costs, yet the need for new infrastructure and power sources to run data centers will “likely increase system costs for all customers,” according to the Joint Legislative Audit and Review Commission. 

There’s another aspect to data centers that’s been less of the focus: emissions and air-quality impacts. 

There are many data centers across the U.S. that are low emission and run on clean energy, but a 2024 study found roughly 56% of the electricity used to power data centers comes from fossil fuels.

Fossil fuels, oil, natural gas and coal are derived from fossilized plant and animal remnants from millions of years ago, according to the Smithsonian National Museum of Natural History. 

An EESI fact sheet says fossil fuels result in “significant climate, environmental, and health costs that are not reflected in market prices” due to their CO2e emissions. Greenhouse gases can trap heat in the Earth’s atmosphere, which in turn contributes to climate change — defined by the United Nations as long-term shifts in temperatures and weather patterns. 

They also impact local air quality and health. 

University of California at Riverside and Caltech found the public health burden is a hidden toll of AI data centers. 

“If you look at those sustainability reports by tech companies, they only focus on carbon emissions, and some of them include water as well, but there’s absolutely no mention of unhealthful air pollutants, and these pollutants are already creating a public health burden,” Shaolei Ren, a corresponding author of the study, said in a statement. 

Take xAI’s Memphis, Tennessee data center “Colossus,” for example. It’s the supercomputer that powers Elon Musk’s X, formerly Twitter, AI chatbot Grok. 

The data center, which was built in just 122 days, partially uses gas-powered turbines to keep its systems operational. Gas generators are cleaner than their diesel counterparts, but because centers operate around the clock, they create mass emissions. 

The Southern Environmental Law Center accuses xAI of violating the federal Clean Air Act by using unpermitted methane gas turbines without public notice, permits or pollution control. 

When asked for comment about the Colossus data center and future plans, xAI responded with an automated email reading “Legacy Media Lies.” 

The UC Riverside study found generating the electricity needed to train OpenAI’s ChatGPT-4 model. ChatGPT is now in model GPT-5, OpenAI’s self described “smartest, fastest, most useful model yet.” It produced an air pollution equivalent of more than 10,000 round trips by car from Los Angeles to New York City.

The Memphis Community Against Pollution, an environmental justice organization based in Memphis, works to combat and prevent environmental racism.

“What’s happening in Memphis is a human rights violation,” KeShaun Pearson, executive director of the organization, told Democracy Now in April. “Elon Musk and xAI are violating our human right to clean air and a clean, healthy environment.” 

Colossus isn’t the only data center using gas powered generators for their electricity. OpenAI’s Stargate project in the old railroad community of Abilene, Texas, also uses them to partially power the campus. 

“We’re burning gas to run this data center,” Altman told the Associated Press in September but added he hopes it will rely on other power sources in the future.

Other campuses, like Microsoft’s $3.3 billion data center Mount Pleasant, Wisconsin, get permission for backup gas-generators meant to keep the facilities, which are online 24/7, running in the case of a power outage. 

Amanda Garcia, a senior attorney at SELC, said air pollution issues often go underreported because data center impacts on water and energy take more public attention. No one thinks they will run out of clean air. 

“The air impacts have been less of a focus because, a lot of times, the air quality isn’t happening in the same area where the data center is,” she said. “It’s being built in a different community, but it is happening and is impacting the people living around those centers.”

The Environmental Protection Agency officially created new rules for data centers’ power generators Jan. 16, 2026, specifying that they must comply with Clean Air Act regulations.

The EPA’s new policies state the temporary or portable large power generators, specifically gas turbines, cannot produce air pollutant emissions exceeding the levels mandated by the federal Clean Air Act. Permitting for the turbines would also fall under federal law, rather than local or state guidelines.

Previous guidance listed these power generators exempt from Clean Air Act standards because they were generally used in short-term or emergent situations before the proliferation of AI data centers. Those large power generators will now be required to meet Clean Air Act standards, while medium and small turbines will have new exemptions.

“We expect local health leaders to take swift action to ensure they are following federal law and to better protect neighbors from harmful air pollution,” Garcia said in an SPLC press release. 

Poor air quality can also directly impact water quality. But while the quality of water isn’t the focus around data centers, the sheer volume of water they use is notable. 

There is no publicly available data measuring how much water ChatGPT uses on a daily basis because OpenAI does not release the information. 

Some research estimates the AI “chugs a bottle of water” for each search. OpenAI’s CEO Sam Altman said a query uses roughly one-fifteenth of a teaspoon of water. 

Electronic devices can overheat, which can cause problems, so they need to stay cool to work properly.

Think of when your phone heats up and slows down when it’s charging and you’re watching a video, your gaming console getting slower when you run graphics too high, or your laptop not working as well when you use it on your bed instead of a desk.

Many devices use air as a coolant — think internal fans that make a laptop sound like a plane is taking off — but electronics running more intense programs typically use water to combat overheating. 

Large centers can consume up to 5 million gallons of water a day, according to EESI. That equals roughly the daily demand for a community of 50,000 people. 

That doesn’t account for the impacts on agriculture, which also needs water and electricity. 

“Don’t forget that it is also required for us to grow food,” Gerard Lim, CEO of vertical farming startup Agroz, said at a November Fortune Innovation Forum in Kuala Lumpur, Malaysia.

AI data centers are never off. They’re always running, always generating, always learning and always using water to cool down. Some data centers use recycled wastewater. Others use fresh water, which is typically used for human consumption and use. 

How does one measure this use in a way that makes sense? 

There are too many variables to get an accurate count of how much water data centers use. 

Should the indirect use of water in making computer chips be included in counting water use? What about the water used to lay concrete the data centers sit atop? Or perhaps the water used in hydropower plants to generate nearly 6% of the U.S.’s electricity supply that could be powering data centers, too? 

Back to Port Washington and the Lighthouse campus, the city agreed to supply as much as 1.2 million gallons of Lake Michigan freshwater a day. If the average American uses 100 gallons of water a day like the EPA estimates, this data center is allowed to use the same amount as nearly the entire city population per day. 

Cooling systems are kept trade secret for the most part, making it difficult if not impossible to gauge how much water is being used up completely. The Lighthouse campus will feature a “closed-loop chilled water system” and be “near zero,” thanks to the “latest cooling designs,” Vantage said on its website. But no one outside those companies really knows what this means. 

Humans use a lot of water, from raising cows for hamburgers to watering crops to feed said cows or even showering and having internal plumbing, and there isn’t always going to be enough to go around.

Even in the face of these worries and future unknowns, not all that comes out of AI is bad. 

Data centers impact water, air and energy. They also impact how first responders do their jobs, as well as public accessibility and climate change — but perhaps not in the way you may be thinking. 

CAL FIRE uses AI elements with the help of cameras managed by Alert California at University of California San Diego. The systems involve around 1,144 cameras in a public-private partnership that scan the horizon every two minutes for “anomalies,” like fire or smoke.

David Acuña, CAL FIRE Battalion chief and spokesperson, said this AI has been helpful for firefighters responding to California’s infamous fires, but it took some time to get there. 

“It has improved over time,” he explained. “When it started, it would ping a fire multiple times a day, and we’d be like, ‘You’re right, that is a fire. It’s called the sun.’ So it had to learn to ignore that.” 

The AI scans the horizon for a glow, fire or smoke, then pings a captain in the emergency command center to confirm what it’s “seeing.” The captain will then make the decision whether to mobilize crews. The decision to use resources is made by humans, not the AI.  

Acuña said this AI can detect fires, especially at night, around the size of a quarter-acre, or two NBA basketball courts side by side. The AI can do it before a 911 caller spots and reports the fire. This is particularly helpful in the foothills, where fires can burn undetected for hours before growing large enough for someone to spot them and call it in. 

CAL FIRE launched a new AI chatbot that searches the information within official websites and comes up with an answer to questions. The chatbot, “Ask CAL FIRE,” also gives officials a look at what people are searching for and provides the most up-to-date answers for those looking. 

The chatbot is available in 70 languages, and CAL FIRE is improving it constantly, Acuña said.

“Our goal is to make sure people have the best information to make decisions for themselves and their families,” he said. 

Cities across the country are also turning to AI in cameras to improve road safety. Mounted cameras are tracking potholes in California, inspecting guardrails, road signs and pavement markings in Hawaii and building annual road congestion reports.

EESI said the U.S. Department of Energy’s National Laboratories AI supports innovations in crucial disaster forecasting, preparedness and risk mitigation. The National Cancer Institute says AI helps cancer research and detection.

The United Nations uses AI to track pollution and minimize environmental impact while maximizing efficiency.

Yes, artificial intelligence is being used to combat climate change. 

“AI is a tool, right? If you take a hammer, someone can misuse a hammer and harm another person, but someone can also take a hammer and build a house,” Acuña said. “It just depends on how it’s used.”

There’s an emerging trend referred to as “AI shame,” in which people are embarrassed to admit they use AI, even if they do so frequently.  

A study by WalkMe, surveyed 1,000 U.S. workers and 48.8% admit to hiding their use of AI at work to avoid judgment. 

Alex doesn’t hide her use, per se, but she said she uses the programs more than she wants to.

“It’s like using a packaged salad kit instead of making it from scratch,” she said. “It feels a little less good and kinda like cheating.”

She feels her industry moves too quickly for her to produce things from scratch, even if that’s something she would much prefer to do. 

Despite this, the Trump administration and tech giants are pushing AI into social media apps, basic communication, and every aspect of the internet. Trump’s administration said the U.S. should be a “global leader” in AI and views the new technology as the means to “promote human flourishing, economic competitiveness and national security.”

Meanwhile, big tech companies are dumping billions of dollars into this infrastructure while promising to find ways to minimize concerns on the environment. In January, Microsoft CEO Satya Nadella took to the World Economic Forum to urge users embrace AI instead of shun it away. 

“We will quickly lose even the social permission to take something like energy, which is a scarce resource, and use it to generate these tokens, if these tokens are not improving health outcomes, education outcomes, public sector efficiency, private sector competitiveness, across all sectors, small and large, right?” he said. “And that, to me, is ultimately the goal.”

These companies are still developing new ways to use AI and integrate it into daily life outside work. 

Altman admitted to Jimmy Fallon on “The Tonight Show” that he uses his company’s AI for a very personal reason: helping raise he and his husband’s 8-month-old child. 

“The idea is like a general-purpose, sort of, life adviser,” he said. “I cannot imagine having gone through figuring out how to raise a newborn without ChatGPT.”

Policymakers and the public are stuck navigating the complex environment that changes on a daily basis, balancing the promise of productivity gains and life-saving technologies against real, immediate impacts on air quality, water supplies and general affordability.

“Data centers are both a challenge and an opportunity,” ChatGPT wrote when asked about weighing the impacts of AI. “They undeniably consume large amounts of energy and water, but they’re also catalysts for cleaner infrastructure, renewable energy investment, and more efficient global computing. The key is how responsibly — and transparently — they’re built and operated.”