FOLLOW US

From Iowa to Texas, AI data centers are quietly consuming billions of gallons of freshwater every year—and the worst is yet to come.

[By Samuel López | USA Herald] - Every time you ask an AI chatbot a question—every time you generate an image, summarize a document, or let an algorithm write your emails—a machine somewhere on this planet is drinking your water. Not metaphorically. Literally. The servers powering the artificial intelligence revolution run scorching hot, and the only thing keeping them from melting down is an industrial-scale appetite for one of Earth’s most finite resources. And right now, Big Tech is guzzling it with almost zero accountability.

Researchers at the University of California, Riverside, published findings that should make every American stop and think: a single 100-word AI prompt consumes roughly 519 milliliters of water—approximately one standard bottle. That number might sound trivial in isolation, but multiply it by the billions of queries that flow through systems like ChatGPT, Google Gemini, and Microsoft Copilot every single day, and you begin to understand the scale of what we’re dealing with. This isn’t a rounding error. This is a crisis hiding in plain sight.

The Lawrence Berkeley National Laboratory—one of the most respected scientific institutions in the United States—released a landmark 2024 report with numbers that should be front-page news across every outlet in this country. In 2023 alone, U.S. data centers consumed 17 billion gallons of water directly through cooling systems. That same year, they consumed an additional 211 billion gallons indirectly through the electricity required to power them. The indirect consumption alone is twelve times greater than the direct use. And according to that same report, by 2028 those direct figures could double—or even quadruple. Let that sink in.

We are already using 228 billion gallons of water per year to run data centers in the United States. That figure is projected to double by 2028. For perspective, that’s six times the total annual water consumption of the entire city of Seattle. And it’s growing at a pace that is nothing short of breathtaking: McKinsey & Company forecasted global data center capacity growth of 19 to 27 percent per year in 2024. The AI boom isn’t slowing down. It is accelerating. And the water bill is coming due.

Consider what Google’s own environmental reports reveal. In 2023, Google consumed over 6 billion gallons of water to cool its global network of data centers. In 2024, a single Google data center in Council Bluffs, Iowa, consumed 1 billion gallons of water—enough to supply every residential water customer in the state of Iowa for five days straight. One building. One year. One billion gallons. And Google is just one company in an industry that includes Amazon, Microsoft, Meta, and hundreds of others all racing to build more, faster, and bigger.

The environmental burden doesn’t stop at the state line. A study by the Houston Advanced Research Center and the University of Houston found that data centers in Texas alone will consume 49 billion gallons of water in 2025. By 2030, that number is projected to explode to 399 billion gallons—the equivalent of draining Lake Mead, the largest reservoir in the entire United States, by more than 16 feet in a single year. Meanwhile, North American data centers collectively consumed nearly 1 trillion liters of water in 2025, according to market research firm Mordor Intelligence. And we haven’t even reached the peak of the AI construction wave.

Here is what keeps me up at night: only 3 percent of all water on this planet is freshwater. Of that 3 percent, only 0.5 percent is accessible and safe for human consumption. We are talking about an extraordinarily thin margin between civilization and catastrophe. A human being cannot survive more than three days without water. Yet we are allowing trillion-dollar corporations to stake a claim on this most irreplaceable of resources to train and operate AI models, while communities across America—communities in Georgia, Iowa, Nevada, Arizona—are watching their local water supplies buckle under the pressure.

Take Newton County, Georgia. A Meta data center that opened its doors in 2018 consumes 500,000 gallons of water every single day—accounting for 10 percent of the entire county’s water consumption. And that county is now fielding requests for new permits for data centers that would use up to 6 million gallons per day. That would more than double what the entire county currently consumes. These aren’t abstract statistics. These are people’s tap water. Their irrigation. Their children’s drinking water. And it is being siphoned away, not by drought or climate change alone, but by deliberate corporate decisions made thousands of miles away in Silicon Valley boardrooms.

What makes this even more alarming is what happens to that water once it enters a data center cooling system. Approximately 80 percent of the water drawn by evaporative cooling systems is vaporized as steam and permanently lost to the local water supply. It doesn’t flow back into a reservoir. It doesn’t return to the aquifer. It evaporates into the sky, gone for good. The Lincoln Institute of Land Policy describes the situation with painful clarity: even a mid-sized data center consumes as much water as a small town, while the largest facilities require up to 5 million gallons of water every day—equivalent to a city of 50,000 people.

A peer-reviewed study published in ScienceDirect in 2025 projects that without mitigation, global water consumption associated with data centers could increase more than seven times by mid-century. Seven times. And that research was published before the full-scale buildout of next-generation AI infrastructure was even underway. Separate research published in the same journal in December 2025 estimates that the water footprint of AI systems alone could reach between 312.5 and 764.6 billion liters in 2025—a range roughly equivalent to the entire global annual consumption of bottled water.

And yet, despite these staggering figures, the transparency from the industry is almost laughably inadequate. Researchers at water law and policy institutions have reviewed the voluntary sustainability reports of Amazon, Google, Microsoft, Meta, Digital Realty, and Equinix, and what they found was deeply troubling: the disclosures are inconsistent, incomplete, and in many cases designed to obscure rather than illuminate the true scale of consumption. Amazon, as recently as 2025, still did not report total water usage data at all in its sustainability report. Google’s disclosures don’t pair water consumption with facility size or technology used. These reports are voluntary. And voluntary, when billions of dollars in regulation avoidance are at stake, means inadequate.

In April 2026, Reuters reported that more than a dozen major institutional investors are now pressuring Amazon, Microsoft, and Google ahead of their annual shareholder meetings—demanding greater disclosure of site-specific water and energy consumption data. “We haven’t seen them disclosing enough about their water consumption and the impact on the local community,” Jason Qi, lead technology analyst at Calvert Research and Management, told Reuters. This pressure from the investor community is significant, but it cannot substitute for federal regulatory action. The market won’t fix this on its own. It never does.

The long-term environmental consequences of this trajectory are severe. Freshwater scarcity is already intensifying across the American West, driven by a combination of climate change, population growth, and agricultural demand. Now add an AI arms race into that equation. Data centers in Arizona, Nevada, and Texas—among the driest regions in the country—are multiplying at a pace that local water officials describe as unsustainable. In hotter climates, cooling requirements are even greater, meaning that the places least equipped to spare water are precisely where tech companies are choosing to build. The economics incentivize it. The ecosystems cannot survive it.

There are solutions on the table. Technologies like direct-to-chip liquid cooling and immersion cooling can dramatically reduce water and energy usage compared to traditional evaporative systems. Some facilities are already turning to recycled or non-potable water sources. But these approaches remain the exception, not the rule, because they cost more and the industry faces no legal mandate to adopt them. Until regulators step in and require both transparency and efficiency standards, the cheapest and most water-intensive methods will continue to dominate.

I want to be fair here. AI has the potential to do tremendous good in this world—accelerating medical research, optimizing energy grids, modeling climate solutions. I am not anti-technology. But I am pro-accountability. And what I see right now is an industry being handed extraordinary public resources—water, land, electricity, tax incentives—while operating with the transparency of a black box. That is not a fair trade for the American people.

Water is not a commodity. It is not a line item in a server farm’s operating budget. It is life itself. And if we allow the AI industry to consume it at the pace and scale these numbers project—without mandatory disclosure, without efficiency requirements, without genuine community consent—we will look back on this moment as one of the great environmental failures of our generation. The data centers are being built right now, today, in your state, near your community, drawing from your water table. The question is whether we’re going to do anything about it before it’s too late.