Popular sci-fi has long-imagined a future artificial intelligence (AI) gaining consciousness and turning against humankind. However, AI poses huge risks right now, as it threatens our water supplies, our climate goals, and our livable future.

Training and using AI requires huge amounts of computing power in data centers (warehouses filled with huge computers called servers), which in turn need a lot of energy to run and water for cooling. These needs are expected to grow exponentially in just the next few years. 

By one projection, AI server and data center energy demand may triple in the next five years. Given that, we found that by 2028, AI in the US could:

• Consume 300 terawatt-hours (TWh) of energy annually, enough to provide electricity to over 28 million households;
• Require as much as 720 billion gallons of water annually just to cool AI servers, enough water to meet the indoor needs of 18.5 million households.¹Check out our so and methodology in our fact sheet, “A No Brainer: How AI’s Energy and Water Footprints Threaten Climate Progress.”

Worse, the Trump administration plans to expand AI and its environmental impact. On day two in office, Trump announced a $500 billion AI initiative and promised to use his executive powers to hasten AI development. Moreover, Trump’s Environmental Protection Agency has committed to boosting AI expansion. 

Meanwhile, Big Tech leaders, including Elon Musk, are lining up to profit from the AI boom. And oil and gas giants are jumping at the opportunity to supply the power while expanding fossil fuel infrastructure. 

Without change, AI will guzzle more water from water-stressed areas and raise utility bills as companies build more infrastructure to meet demand. The technology will become just another tool for wealthy CEOs to destroy our planet for profit.

AI Guzzles Water in Water-Stressed Areas

At a time of climate-fueled droughts and skyrocketing water bills, more and more people in the U.S. are struggling to access clean water. AI expansion promises to worsen this problem.

AI and its data centers require freshwater to cool servers and produce the electricity that powers them. In 2022, Google, Microsoft, and Meta used an estimated 580 billion gallons of water to provide power and cooling to data centers and AI servers. That’s enough water to meet the annual needs of 15 million households.

If U.S. AI expansion follows projected growth, it could consume as much water as 18.5 million households annually just to cool servers. That water footprint will rise even higher if AI is powered by fossil fuels — which is often the case for data centers — as fossil fuel generation is much more water-intensive than renewables.

Notably, data centers and AI must use clean, treated water and don’t return much of it back to the water source it’s drawn from. For example, Google-owned data centers only discharge 20% of the water withdrawn to wastewater treatment plants. The other 80% is lost to evaporation.

This use will (and has already) stress local water supplies, depleting the water that residents and farmers need. 

In Arizona, for example, data centers withdraw massive amounts of water in areas where farmers fallowed fields and families went without tap water for most of 2023.

Meanwhile, in Oregon, Google’s data centers make up 25% of water use in The Dalles, a dry area that is usually off-limits to new industrial water users. The Dalles facility has increased its water use nearly threefold between 2017 and 2022, and Google has plans to open two more data centers nearby.

AI Will Double Down on Fossil Fuels

Along with its water footprint, the computing power required for generative AI (which creates new images, videos, audio, or text by analyzing vast amounts of data) is incredibly energy-intensive.

Simple ChatGPT searches use nearly 10 times the electricity as a Google search. Creating images is thousands of times more energy-intensive than text searches and can use as much electricity as it takes to charge your phone per image created.

As of 2024, ChatGPT uses over 500,000 kilowatts of electricity every day, as much as used by 180,000 U.S. households. Electricity consumed by AI servers could rise 150-fold in a single decade, to 300 TWh in 2028. That’s enough electricity to power over 28 million households for a year.

While tech companies assure us that this expansion will run on renewables, this has yet to materialize. Big Oil and Gas companies are looking to hop on the AI boom to boost profits. Chevron, for example, recently announced a partnership to build gas power plants for new AI data centers.

By one estimate, from now to 2030, renewables will cover only 40% of new electricity demand from data centers. Expanding fossil fuel infrastructure for AI is unthinkable at this stage in the climate crisis.

Confirming our worst fears, some local leaders are working to keep dirty power plants running to power AI and data centers. In Salt Lake City, UT, lawmakers and utility executives cut investments in renewables while extending the lives of dying coal plants to power new data centers. 

AI Means More Pollution and Higher Bills

Along with impacts to climate and water, AI growth promises greater health impacts from air pollution — especially for Black, Brown, and low-income environmental justice communities. Dirty facilities of all sorts are disproportionately sited near these communities, and power plants are no exception.

To feed data centers, predominantly Black neighborhoods in Randolph, AZ and Memphis, TN are facing more pollution from gas-powered turbines, which is linked to asthma and lung cancer. In the case of Memphis, residents are at the mercy of Elon Musk’s xAI, which is operating at least 18 turbines without permits.

Moreover, the infrastructure buildout for data centers and AI will raise utility bills, even as data centers get discounts on their electricity use. In short, families are subsidizing Big Tech’s AI boom through their electricity bills.

In Arizona, the state utility board passed an 8% rate hike to help fund new electricity demand from data centers. Yet, thousands of residents don’t even have power, and the state rejected a plan to bring electricity to parts of Navajo Nation land.

AI promises higher electricity costs and higher medical bills.

We Can’t Fall For Big Tech’s False Promises on AI and the Environment

Big Tech has claimed that AI will be a boon to climate action, helping us find solutions. But so far, this technology is just derailing these corporations’ own climate goals. 

To cover up their shortcomings, some are pushing misleading claims of their “carbon neutrality” using shady accounting methods. Meanwhile, they’re failing to actually address the environmental harms of AI and data centers.

Microsoft, for instance, has pledged to be carbon-negative and water-positive by 2030. But its AI rollout is driving ever-greater water and electricity use. By Microsoft’s own count, as of 2024, its total planet-warming emissions were about 30% higher than in 2020.

Self-reported emissions from Big Tech are already cause for alarm, but they may only tell part of the story. A Guardian investigation found that from 2020 to 2022, the real emissions from the company-owned data centers of Google, Microsoft, Meta, and Apple were more than 600% higher than officially reported.

Ultimately, AI boosters are promoting this technology as a tool for fighting climate change while sweeping its energy and water impact under the rug. 

Without action, AI will continue to guzzle water unsustainably, keep fossil fuels online, drive more pollution, and raise electricity costs amid today’s cost of living crisis. 

Instead, artificial intelligence should rely on additive renewable sources, meaning its increasing energy demand shouldn’t take renewable power off the grid for other customers. Our leaders must also protect families from rising utility bills driven by AI expansion. As with any technological innovation, our health, environment, and planet must come before corporate profits.