Nobody talks about water when they talk about artificial intelligence. They talk about energy, chips, the speed of the models and how much they cost. But there is one resource that is silently being depleted every time someone asks Gemini, Google’s search engine, a question or trains a model from scratch.
Fresh water is at the center of one of the most urgent environmental debates in the technology sectorand Google has just taken a step that no one expected so soon: publicly recognizing the problem and presenting a plan with figures, deadlines and real commitments to reverse it before the end of this decade.
The ecological problem that nobody wanted to see
Every time you use Gemini, ask a search engine a question, or train an AI model, the servers behind that action generate heat. And that heat needs to cool down. The most efficient method is still water, which absorbs heat and evaporates. The result is brutal and increasing water consumption.
The numbers are hard to ignore. Google consumed around 22.7 billion liters of water in 2024 only in their data centers. A study from the University of California, Riverside, estimated that asking up to 50 questions on an AI like ChatGPT can consume half a liter of water. And if we extrapolate that to a global scale, with billions of queries a day, the impact stops seeming anecdotal and becomes a real environmental crisis.
By 2030, according to recent projections, The water consumption associated with artificial intelligence could be equivalent to that of 1.3 billion people. It is a figure that deserves pause, reflection and, above all, action.
The conception of Google in five commitments
On June 3, Google published a weblog with five specific commitments on water matters. These are not vague promises or pretty statements of intent for sustainability reporting. They are measurable commitments, with figures, dates and projects underway.
The most ambitious is the first: Replenish more water than your data centers consume by 2030. That does not mean equaling consumption, but exceeding it. Google has set the goal of replenishing the 120% of the water you usewhich means returning more to the environment than it extracts. To achieve this, the company already has in place 165 replacement projects distributed in 97 hydrographic basinswhich are expected to return more than 19 billion gallons of water per year by 2030an amount sufficient to supply the entire city of Los Angeles for more than 40 days.
The second commitment aims to modernize the water infrastructure of the communities near its facilities. Google has allocated more than $500 million dollars to improve supply systems, wastewater treatment and reuse projects. Because the problem is not only the water that Google exhausts: it is also the critical state of many pipes and distribution networks that have been without maintenance for decades.
The third point is perhaps the most technical but equally relevant. The company promised Protect the most vulnerable watersheds by choosing air cooling systems instead of water when local resources are under pressure. If an area has high water risk, liquid cooling is not installed.
The fourth commitment speaks of Complete transparency in water consumption reporting. Google was the first major cloud provider to publish annual data on the water use of its data centers, and pledged to continue to do so in greater detail.
The fifth and final axis is to explore alternative sources, such as regenerated waste waterto reduce dependence on fresh water. In some counties in Georgia, for example, they are already testing this model with promising results.
Additionally, the company announced an additional investment of 17 million dollars for new water management projects in seven US states, including Georgia, Iowa, Michigan, Minnesota, Missouri, Nebraska and Texas.
Is this effort enough?
The honest answer is that we don’t know yet. By 2025, Google has already replaced more than 7 billion gallons of water thanks to its active projects. That is real progress. But consumption also continues to grow as AI scales, and the gap between what is exhausted and what is replenished remains wide.
What is undeniable is that Google is responding to growing public pressure. In many communities in the United States there are active protests against the construction of new data centers, precisely because of their impact on local water resources. Ben Townsend, Google’s head of infrastructure and sustainability, said the company wants to provide a “reference model” that other communities can use when evaluating proposals for new data centers.
It’s an important step forward, and in the world of technology, a five-point concept with real investments and specific dates is not something you see every day. It remains to be seen if Google—and the rest of the industry—can make the promise a reality before AI moderately spills the beans on those who need it most.
Keep reading:
• University of California revealed how much water ChatGPT exhausts when running
• Synthetic Intelligence is leaving Iowa rivers without water: what is happening?
• Chinese scientists create a floating device that transforms each drop of rain into electric current
