AI's hidden thirst or how much water does Artificial Intelligence really drink

AI’s hidden thirst or how much water does Artificial Intelligence really drink?

Introduction

Let’s talk about something we don’t often think about when we use AI tools like ChatGPT, Gemini, Claude, DeepSeek, DALL·E, Midjourney, Runway, or even your favorite recommendation algorithms on Netflix, Spotify, TikTok etc: WATER.  

Yes, water.

It turns out that AI has a huge thirst, and it’s not just for data. Behind every AI-generated meme, essay, image, video, playlist suggestion, blog post, social media caption, voice synthesis, chatbot response, deepfake, real-time translation, and even AI-powered code generation lies a massive amount of water used to keep the technology running.

But how much water are we talking about? And can AI actually help solve its own water problem? Continue reading. 

Wait, AI Drinks Water? How?

You might be wondering, “How does AI even use water? It’s just code, right?”

Well, not exactly.

AI runs on powerful computers housed in data centers. These data centers are like the brains of AI, but they generate a lot of heat. To keep them from overheating, they need to be cooled down—and that’s where water comes in.

Cooling systems in data centers use water to absorb and dissipate heat. Think of it like sweating on a hot day, but on a much, much larger scale. For every kilowatt-hour (kWh) of energy used in these cooling systems, up to 9 liters of water can be consumed. That’s enough to fill a large water bottle!

And it’s not just a little water. Big tech companies like Google and Microsoft are using billions of liters of water every year to keep their data centers cool. For example:

  • In 2022, Microsoft’s water consumption surged by 34%, reaching nearly 1.7 billion gallons (6.4 billion liters), largely due to its growing AI operations. This increase was driven by its heavy investment in generative AI and partnership with OpenAI, according to researchers. To put it in perspective, training ChatGPT-3 alone consumed 85,000 gallons (about 700,000 liters) of water. Data centers rely on massive cooling systems to prevent overheating, making AI’s water footprint an increasingly urgent issue.

How much water does AI really use?

Let’s break it down further. When you interact with AI tools like ChatGPT, you’re not just exchanging words—you’re also using water. Researchers estimate that every 20 to 50 prompts you give to ChatGPT can consume about 500 milliliters of water.

Now, imagine every time you asked AI a question, you had to pour out that amount of water. A single prompt might not seem like much—just 500 milliliters as we said.. but now picture millions of people doing this every day, over and over again. Including you.  That water doesn’t just vanish; it gets used to keep massive data centers from overheating.

It’s like filling up a bathtub one drop at a time—except instead of a bathtub, it’s entire lakes disappearing into AI’s growing thirst. By 2027, AI is projected to consume between 4.2 billion and 6.6 billion cubic meters of water annually. That’s enough to reinvent entire water supplies, all for the sake of processing words, images, and data at the speed we’ve come to expect.

But It’s not just about cooling

The water footprint of AI doesn’t stop at cooling data centers. Beyond the massive cooling needs of AI data centers, there’s an even bigger, often overlooked factor: the energy that keeps them running. Most of the world’s electricity still comes from power plants that rely heavily on water. Coal and nuclear plants, for example, use vast amounts of water for steam generation and cooling, while even hydroelectric dams—though renewable—can disrupt ecosystems and consume large water reservoirs. Every time AI processes a request, it’s indirectly pulling water from these power sources to keep the lights (and servers) on.

Then there’s the hardware itself. The AI revolution depends on advanced semiconductor chips, which are produced through an incredibly water-intensive process. Fabricating a single semiconductor wafer can require thousands of liters of ultra-pure water to clean and etch the delicate circuitry. Multiply that by the billions of chips needed for AI servers worldwide, and the water footprint skyrockets long before AI even runs its first prompt.

It’s a chain reaction of consumption: water is used to cool the data centers, generate the power that fuels them, and manufacture the hardware that makes AI possible in the first place. The true water footprint of AI isn’t just about keeping systems cool—it’s embedded in every layer of its existence.

The environmental impact: what’s at stake?

AI’s growing water consumption has real consequences. Freshwater is limited, and many regions already struggle with shortages. As AI expands, its demand for water increases, putting even more pressure on local supplies.

The problem is that data centers are often built in places where electricity is cheap, but those areas don’t always have enough water. This can create competition between tech companies and local communities, especially in regions already dealing with droughts. When AI takes a bigger share of the water, it leaves less for homes, farms, and ecosystems that depend on it.

If AI keeps growing without better water management, the strain could get worse. Finding sustainable solutions is key to making sure innovation doesn’t come at the cost of the planet’s most essential resource.

Can AI help solve its own water problem?

Here’s where things get interesting. While AI is part of the problem, it might also be part of the solution. AI is already being used to optimize water use in various industries, so why not apply it to its own operations?

  • AI can optimize cooling systems in data centers by analyzing data in real time. Google has already used AI to cut cooling energy by 40%, which could save billions of liters of water if applied globally.
  • Water recycling systems powered by AI could repurpose treated wastewater for cooling instead of using fresh water, reducing the strain on local supplies.
  • Predictive maintenance with AI can detect leaks and inefficiencies before they waste water, making operations more sustainable.
  • AI can enhance the use of renewable energy like solar and wind, which require far less water than traditional power plants.

 

What are big tech companies doing?

Tech giants are starting to take notice of their water usage. For example:

  • Google has committed to replenishing 120% of the water it consumes by 2030, meaning it will return more water to the environment than it uses.
  • Microsoft is investing in water-positive technologies and aims to be water-positive by 2030 as well.

These are steps in the right direction, but there’s still a long way to go.

Is AI worth the Water?

So, here’s the big question: Is AI’s water consumption a dealbreaker? The answer isn’t black and white. AI has the potential to revolutionize industries, solve complex problems, and even help us tackle climate change. But its environmental impact can’t be ignored.

The good news is that AI can also help reduce its own water footprint. By using smarter technologies, recycling water, and switching to renewable energy, we can make AI more sustainable.

But it’s not just up to tech companies. As users, we can also play a role. Do we really need to ask ChatGPT for a joke when we could just Google it? Small changes in how we use AI can add up to big savings in water and energy.

So, what do you think? Can AI quench its thirst sustainably, or will it drown in its own water footprint? The answer lies in how we choose to develop and use this powerful technology. 

And if you’re interested to read more about AI, do not miss our previous blogs! 

Ivan Zografski

Most read