Water and AI: two major 21st-century challenges that are in conflict
When the "new intelligence" makes the planet thirsty
There is something profoundly ironic, almost tragic, about our time. We are this civilization that looks at the sky anxiously, aware that the meteorite is approaching, but continues to debate the color of the living room curtains.
Do you remember Don't Look Up , the film where Leonardo DiCarpacchio plays a scientist desperate in the face of collective indifference?
Replace the meteorite with the water crisis , and you get our current reality.
Except this time, we're not looking at the sky: we're looking at our screens, our smartphones, our chatbots .
And meanwhile, the water evaporates .
Water: this vital element that we forget
Water is the matrix of all life on Earth.
Without it, there would be no photosynthesis, no ecosystems, no civilization.
It makes up the majority of the molecules in our body through its hydration potential, it nourishes our soils, it regulates our climate.
It is present in every living organism as a source of life.
I talk about it enough in this blog to better understand its properties, states and virtues.
Based on my work on 432 Hz, we refined with Professor Marc Henry in 2020 the bioresonance frequency of the water molecule H2O to better understand the impact of sound and frequencies on biology.
I was able to delve even deeper into the frequencies of water with the work of Professor Gerald Pollack in early 2024 with the [H3O2]- experiment
We know that.
We teach our children that water is an important and essential resource.
And yet, we are witnessing a staggering contradiction: while we claim to want to preserve this vital resource, we are developing technologies that consume it at a dizzying rate.
The figures speak for themselves.
According to the International Energy Agency, data centers consumed approximately 560 billion liters of water in 2023. This figure could more than double by 2030, reaching nearly 1.2 trillion liters per year. To put this into perspective: a large data center can consume between 3.8 and 19 million liters of water per day. That's equivalent to the consumption of tens of thousands of households.
https://www.connaissancedesenergies.org/afp/un-rapport-appelle-les-acteurs-de-lia-mieux-controler-leurs-consommations-delectricite-et-deau-250207
And here's the absurd paradox: 20 questions asked of ChatGPT require the equivalent of a 0.5-liter bottle of water . We're conversing with machines while literally draining our groundwater.
And how many questions do you think are asked of this software every day?
2.5 billion !!!
Calculation :
If 20 questions = 0.5 liters of water
So 1 question = 0.5 ÷ 20 = 0.025 litre (or 25 ml)
For 2.5 billion questions: 2,500,000,000 × 0.025 = 62,500,000 liters
That's 62,500 cubic meters or 62.5 million liters per day
To put this into perspective:
- That's the equivalent of 25 Olympic swimming pools per day (an Olympic swimming pool = 2,500 m³)
- Approximately 22,800 cubic meters per year (if this daily rate is maintained)
- The water consumption of approximately 140,000 people per day (assuming 450 liters/person/day including all uses)
These figures illustrate the significant environmental impact of the infrastructure required to cool the data centers that host these AI models. This is, in fact, a growing concern in the technology sector.
https://www.numerama.com/tech/2036695-le-nombre-de-questions-posees-chaque-jour-a-chatgpt-est-hallucinant.html
Artificial intelligence: an insatiable thirst
Data centers dedicated to artificial intelligence are particularly energy-intensive: their electricity consumption is four to five times higher than that of a traditional data center . In 2022, data centers, cryptocurrencies, and AI already accounted for nearly 2% of global electricity consumption, or 460 TWh. By 2030, these needs could double to reach 945 TWh, with a growth rate of approximately 15% per year.
But that's not all.
The carbon footprint is equally alarming.
Training the BLOOM AI model emits 10 times more greenhouse gases than a French person in a year .
CO₂ emissions related to the manufacture of AI GPUs could jump 16 times between 2024 and 2030, reaching 19.2 million tonnes of CO₂ equivalent.
The disturbing philosophical question
Let's ask ourselves a fundamental, almost embarrassing question: how can an "artificial intelligence" be more energy-intensive than a real intelligence?
Our brain, this marvel of evolution, operates on approximately 20 watts. Twenty little watts to think, create, dream, compose symphonies, or solve complex equations. A query to an AI consumes about 0.3 watt-hours, potentially 10 times more than a typical Google search.
Multiply that by billions of daily queries and you understand the absurdity we are facing.
This disproportion reveals a disturbing truth: we call "intelligence" something that is fundamentally inefficient, wasteful, excessive.
It's like we've created a wheel that needs a fuel truck to turn on its own axis. Where's the logic? Where's the wisdom?
A revealing geographical paradox
The absurdity reaches its peak when we examine where these data centers are located. A study reveals that 45% of the world's data centers are located in river basins with a high risk of water availability, and 47% in drought-prone areas .
Furthermore, 65% are located less than 5 km from key areas for biodiversity.
It's like building Olympic-sized swimming pools in the middle of the Sahara Desert.
The image is ridiculous? Yet that's exactly what we're doing.
When these facilities discharge heated cooling water into rivers or lakes, they disrupt aquatic ecosystems through thermal pollution. Fish, aquatic plants—an entire millennia-old balance is upset to allow servers to function.
The deafening silence of transparency
And as in Don't Look Up , there's the same code of silence, the same organized denial. Currently, no mandatory rules exist for the AI industry regarding the disclosure of environmental information. Tech giants can consume, pollute, and deplete resources without having to publicly account for it in a standardized way.
Microsoft quietly revealed that its water consumption increased by 34% between 2021 and 2022. Google, for its part, recorded a 20% increase over the same period. These figures are buried in hundreds of pages of sustainability reports, invisible to the general public.
An expert points out that " water hoarding is aggressive in the sense that it is unplanned and sometimes undiscussed, with a real problem of democratic debate regarding water use. " We are being deprived of our right to collectively decide on the use of our vital resources.
The impossible duality
We are thus faced with a dizzying duality. On the one hand, there are the discourses on ecological transition, the preservation of biodiversity, and the fight against global warming. International conferences are multiplying, sustainable development goals are being proclaimed, and companies are communicating about their environmental responsibility.
On the other hand, there's a frantic race toward digital "progress." Each new generation of AI must be more powerful, faster, more impressive. Investments are in the billions. Promises of a bright future thanks to AI are flying from all sides.
Between the two: water. This finite, irreplaceable resource, which is being depleted while we debate.
This reveals a deep-seated economic flaw. We have created a system where technological growth takes precedence over preserving what sustains life itself. It's like sawing off the branch we're sitting on, but with a smile on our faces and talk of innovation.
Can the trend still be reversed?
Indirect emissions related to electricity consumption by data centers will increase from 180 million tonnes of CO₂ today to 300 million tonnes in 2035, an increase of 66%.
In an uncontrolled takeoff scenario, this figure could peak at up to 500 million tonnes.
Faced with this trajectory, a few initiatives are tentatively emerging. In France, a general framework for frugal AI has been developed to provide organizations with a methodology for assessing environmental impact. In 2025, 91 partners created a Coalition for Environmentally Sustainable Artificial Intelligence.
Drops of water in the ocean?
Perhaps. But it is these drops that create the waves.
Some arguments suggest that AI could, paradoxically, help solve environmental problems. Google has reportedly increased the efficiency of its solar and wind farms by 20 to 30% using AI. In smart buildings, AI-based systems could reduce energy consumption by up to 40%. According to the IEA, the emissions generated by AI could be offset by emission savings three to five times greater.
But here's the thing: we're betting on the future. We're mortgaging the present for an uncertain future. This is precisely the argument of the characters in the film who, faced with the meteorite, propose to "manage" it rather than avoid it.
The intelligence we are really looking for
In this opposition between water and AI, between nature and technology, between life and algorithms, perhaps lies the most fundamental question of our century: what kind of intelligence are we really looking for?
The intelligence of nature, which has taken millions of years to create remarkably efficient ecosystems where nothing is lost and everything is recycled? Or artificial intelligence, voracious and inefficient, devouring vital resources to clumsily simulate what a human brain does with 20 watts?
Water gives us a clear answer, if we are willing to listen to it.
It flows, it adapts, it nourishes, it transforms.
She is the very intelligence of life.
And we are sacrificing it for servers that are overheating.
The meteorite is there, above our heads. It has the shape of a data center.
And we continue not to look up.
At Naturasounds, we believe that true intelligence lies in harmony with life, not in its programmed destruction. Water is our most precious common good. It's time to choose: preserve the source of all life, or look the other way while it dries up.