AI systems rely on data centres that process vast volumes of information. These facilities house thousands of servers that generate constant heat and require continuous cooling to operate safely and efficiently.
This is where water comes in. Many cooling systems use water circuits to dissipate heat, relying on cooling towers or evaporative systems. While effective, these processes involve significant consumption. It is estimated that generating around 100 words with an AI model may consume the equivalent of a small bottle of water (500 ml) – a symbolic figure but revealing when multiplied by millions of daily interactions.
An invisible water footprint
Unlike more tangible uses of water, such as domestic or agricultural consumption, the impact of AI is diffuse and difficult to perceive. End users rarely associate an online search or an automated response with the use of natural resources.
Yet the expansion of AI is directly linked to growing demand for digital infrastructure. More data centres mean greater cooling needs and, consequently, increased pressure on water resources – especially in regions already vulnerable to water scarcity.
The rapid adoption of AI raises a dilemma: how can technological innovation be balanced with environmental responsibility?
Technology companies have been investing in more efficient solutions, such as air-cooling systems and wastewater reuse. Even so, these approaches do not fully eliminate the problem.
There is also a geographical dimension. Many data centres are in areas with easier access to energy and water resources, which may exacerbate regional inequalities and put additional pressure on local ecosystems.
Rethinking the digital world
The discussion around digital sustainability is still at an early stage, but it is becoming increasingly urgent. As with other industries, it is essential to measure, communicate and reduce the environmental impact of AI.
Beyond technological solutions, there is also a role for users and organisations: questioning the necessity of each use, optimising processes and avoiding digital waste. Efficiency depends not only on technology, but also on the choices each of us makes.
Artificial Intelligence promises significant benefits across multiple fields, from science to natural resource management. However, its development is not neutral. Water, a vital and finite resource, lies at the heart of this silent transformation.
Making this impact visible is the first step towards a truly sustainable digital transition. Because in the future of technology, innovation will need to go hand in hand with environmental responsibility.
