AI's Hidden Thirst: A Tech Founder Warns About Water Consumption

As the world rapidly adopts Artificial Intelligence (AI) technologies across various aspects of life, from writing and analysis to education and digital services, growing questions are emerging about the hidden environmental cost of this technological revolution. This is especially true concerning the water consumption required to operate the giant data centers that power these systems.

Tariq Al Hosani, founder and chairman of ZeroGravity Group, points out that the increasing use of AI fundamentally relies on a massive infrastructure of servers and data centers operating around the clock. This demands enormous amounts of energy and cooling. With the high temperatures generated by complex computational processes, cooling systems become essential to maintain device efficiency. This often involves using large quantities of water in industrial cooling towers.

According to academic estimates and studies, a significant portion of the water used in these operations evaporates during the cooling process and does not return to the water cycle that humans benefit from. Figures indicate that about 78% of the water drawn by major data centers is potable (drinkable) water. This raises serious questions about the priorities of water resource use, especially with increasing global challenges related to water scarcity.

Some estimates reflect the scale of consumption linked to daily interaction with AI technologies. A study by Professor Shaolei Ren from the University of California, Riverside, suggests that an average conversation with an AI model, involving 20 to 50 questions, could consume around 500 milliliters of fresh water due to the cooling processes needed for data processing. Furthermore, training a single AI model from scratch might require approximately 700,000 liters of water in one training session.

This consumption volume only grows with the rapid expansion of these technologies. According to circulating estimates, the ChatGPT model alone processes at least one billion queries daily. Meanwhile, Google reported drawing about 37 billion liters of water in 2024, with approximately 29 billion liters evaporating during cooling operations in its data centers.

Water consumption isn't limited to just operating servers; it extends to the entire lifecycle of the technology itself. Industry estimates suggest that manufacturing a single smartphone can consume around 12,670 liters of water. A single cryptocurrency transaction, like Bitcoin, might deplete nearly 16,000 liters of water – an amount sufficient to fill a small swimming pool.

In this context, Al Hosani believes the challenge is no longer just about developing faster or more powerful AI models. Instead, it's about making these technologies more efficient in consuming natural resources, especially non-renewable ones like water. He proposes the concept of "Green AI" as a framework that integrates environmental sustainability principles from the earliest stages of designing and developing digital systems.

He emphasizes that companies successfully achieving this balance between innovation and environmental efficiency could gain a future competitive advantage. Sustainability has become a pivotal factor in evaluating the performance of technology companies globally, both by investors and users.

Regarding technical solutions, some technology companies have already started looking for alternatives to reduce reliance on water for cooling. These solutions include closed-loop cooling systems, immersion cooling technologies that minimize water needs, and the development of more energy-efficient electronic chips.

In 2024, Microsoft also announced a new data center design aiming for zero water consumption in cooling operations. This step reflects a growing awareness within the technology sector of the scale of environmental challenges associated with digital infrastructure.

Al Hosani believes that a crucial path to enhancing sustainability involves developing clear metrics to measure the "water footprint" of AI models. A global standard for measuring water consumption could boost transparency and motivate companies to develop more efficient solutions. It would also give users a better understanding of the environmental impact of the technologies they rely on daily.

Amidst the accelerating global digital transformation, calls are growing for a balance between technological advancement and the preservation of natural resources. While AI represents one of the most significant drivers of innovation in the modern era, managing its environmental impacts could become one of the most important challenges that will shape the future of this industry in the coming years.