You may have caught a clip in your Instagram feed of U.S. House Representative Alexandra Ocasio-Cortez (AOC) holding up a mason jar full of brown water during a congressional hearing last month. In the hot seat was Jessica Kramer, the Environmental Protection Agency’s assistant administrator for water, who was being questioned about impacts to local residents after the construction of a new Meta data centre in rural Georgia. The water in the jar was taken directly from a residential tap in that neighborhood.
Since the work began, families who live in the area have seen their water pressure significantly decrease (in some cases rendering their appliances unusable) and the quality of water has become so contaminated that even boiling it for use isn’t an option. They rely on bottled water to cook meals, bathe, brush their teeth, etc., which is not sustainable nor fair to the locals that didn’t ask for such an intrusion. Economically, this can also cause property values in the area to plummet.
A director from Newton County Water and Sewage Authority said that the Georgia data centre in question already uses 10% of the county’s entire water supply to cool the facility. If the usage continues at this rate, the area will be in a water deficit as soon as 2030, rendering the area completely uninhabitable.
This is just one example of countless complaints from rural communities who have had their lives upturned by new data centres. But the U.S. isn’t the only place it’s happening. Citizens in Chile just lost a battle against Amazon to build in a green space that is beloved to the locals. In Germany, the power grid is already in jeopardy. The UK is fighting back against the rapid data centre development in their towns and villages.
And it’s not just water or green spaces that are at risk—many of the data centres also produce significant sound pollution, disrupting both humans and wildlife.
On this World Environment Day, we assess the impact of the data centre boom; explore the AI behind the increase in construction and share resources on how to reduce the need for these centres… at least until they can produce them to be more environmentally sound.
Fast Facts
Some history and details about how far the data centre development has come…
- The very first data centre was built in Pennsylvania, USA in 1945 for military purposes and was cooled using huge fans and vents.
- The data centre landscape today is massive. See the locations on this comprehensive database, which drops a map pin on all 115,000+ facilities already spread across 179 countries.
- The rate of development has rapidly accelerated. As of late 2025, there were 3,000 more centres under construction in the U.S. alone. In Texas, 161 billion gallons of water could be consumed by data centres in the next four years.
- Large data centers have the capacity to consume up to 5 million gallons per day, which is equal to the water use of a town populated by 10,000 – 50,000 people.
- Over 100 of the current proposed data centre projects would violate sacred Indigenous lands. The Seminole Nation of Oklahoma were the first to pass a Native land moratorium on new data centres; other tribes are following suit and opposing these developments.
Enter Erin Brockovich
The activist, famously portrayed by Julia Roberts in the movie about her life, which focused on her case against Pacific Gas and Electric for groundwater pollution, has a new website set up to track data centres across the U.S.
On the site, she tracks current news reports, outlines key concerns, and also invites citizens to share their issues.
Devil in the Details
Why is there such a demand for new data centres?
The rapid adoption of Artificial Intelligence (AI) in nearly every commercial industry across the world is the primary driver of this widespread construction.
The Positive Side of AI
- In healthcare, AI is progressing the accuracy and speed of body scans. In some documented cases, it’s also detecting cancers and disease faster than their human counterparts.
- Many of us have AI to thank for the speed with which the COVID-19 vaccines were formulated. Typically, vaccines can take 5 – 15 years to develop. Machine learning algorithms accelerated this process during the global pandemic, greatly reducing hospitalizations and deaths from the virus.
- AI may also soon solve the mysteries behind the treatments (or cures), as the Harvard model TxGNN has already identified drug candidates for over 17,000 rare diseases that do not currently have therapies.
- With an even more ambitious mission, Google Deepmind’s creators aim to solve every disease on earth and just raised major funding to work toward that goal.
- For those with disabilities, AI is providing an easier way to navigate work, school and daily life.
- In the wellness sector, AI massage robots offer a safe space for treatment for victims of assault and others who may not be comfortable with human touch from strangers.
The Problematic Side of AI
- In education, a study last year found that AI may diminish cognitive functions, decrease critical thinking skills and create unnatural bias due to the lack of independent contemplation.
- In transportation, some of the most prominent vehicles that use AI are being called out for safety concerns—by the very people who train them.
- In tech, parents are is suing OpenAI, claiming their chatbot assisted in the suicide of their teenage son. This family is not the only family to experience this horrific reality.
- In social media, romance scams, which take millions from victims each year, have become harder to detect because of AI.
- In machine learning, respected journalist Ronan Farrow reported on the shockingly dystopian result of a recent chatbot test.
How can we help?
There are several ways, both directly and indirectly that we can slow the expansion of AI and help prevent new data centres from harming our natural world:
Advocacy. In the U.S., you can download a free toolkit, call your leaders and be sure to vote against any legislation that favours the development. In the UK, take action with the Global Action Plan.
AI Adoption. Before engaging ChatGPT or Grok to ask a question (which uses 10x more electricity than a simple Internet search), consider the damage you are causing (not to mention how inaccurate the answer is likely to be). Ensure that the end justifies the means if you really feel it necessary to engage a chatbot.
Search Engines. Though Google has been the gold standard for internet searches since its inception, for the first time in 25 years, it’s altered its search box to be ‘powered by AI’. You can switch your default browser to Ecosia (which plants trees for every search). Or, if privacy is your main concern, try DuckDuckGo instead.
Energy Efficiency. You can lessen data centre consumption by reducing your online carbon footprint. A few ways to do so: Empty junk email often; regularly delete old emails and cloud backups; consolidate digital photos into one online album with a secure backup or transfer to a physical hard drive.
As stated above, we can’t deny AI has worked wonders to preserve the longevity of life, and continues to do so in leaps and bounds, but there must be a better way forward that doesn’t sacrifice the health of the land and our communities to do so.
These challenges to save the earth are not insurmountable and can co-exist peacefully with a better solution than the current data centres provide.
Perhaps we should use AI itself to create a less damaging solution?
It may just be able to self-diagnose a better way forward…
One can only hope.
‘Joshua Tree #2, 2017’ by Julian Lennon is available framed in a Limited Edition of 3 (Dye Sublimation on Aluminum with DiBond Backing). For purchase inquiries, contact Fremin Gallery.
Our fund dedicated solely to Environmental projects is called The Desert Bloom. To become a monthly member or make a one-time donation, start here.