As humanity rushes to integrate Artificial Intelligence (AI) into every facet of daily life, from diagnosing diseases to drafting emails, a new United Nations report casts a stark light on the physical, material substrate of this seemingly "immaterial" technology. The report warns that the rapid expansion of data centers powering Large Language Models (LLMs) could soon consume 3% of the world's electricity, while their thirst for cooling water threatens to eclipse the basic drinking water needs of the global population.
The Energy Abyss of Data Centers
The allure of AI lies in its speed and intelligence, but behind every response from ChatGPT or Gemini lies a massive infrastructure of GPUs (Graphics Processing Units) running non-stop. According to the UN, energy consumption by the Information and Communications Technology (ICT) sector is growing at rates that defy projections from just a decade ago. By the end of the decade, AI alone could require more power than countries like the United Kingdom or Italy consume today.
The problem isn't just in the training of models—which requires thousands of MWh of energy—but in their continuous operation (inference). Every time a user requests an image or a summary, servers are activated that emit heat, requiring further energy for air conditioning and cooling. The report emphasizes that without a radical shift toward more efficient algorithms and green energy, AI could derail the goals of the Paris Agreement on climate change.
The Water Crisis: Cooling at the Cost of Life
Perhaps the most shocking revelation of the report concerns water consumption. Data centers require vast amounts of fresh water to cool processors and prevent overheating. It is estimated that for every 10 to 50 queries we pose to an AI chatbot, about half a liter of water is "consumed" (evaporated or contaminated). On a global scale, this translates into billions of cubic meters annually.
The ethical dilemma is intensified by the fact that many of these data centers are built in regions already plagued by water scarcity. The report cites cases in Arizona, USA, and parts of Latin America, where local communities are coming into conflict with tech giants over access to water resources. AI, often touted as a tool to solve the climate crisis, currently appears to be exacerbating the problem of physical survival for millions.
Digital Colonialism and Ethical Responsibility
The UN report speaks of a form of "digital colonialism." While the benefits of AI are primarily reaped by developed Northern economies and large multinationals, the environmental cost—resource depletion and pollution—is often shifted to the Global South. The mining of rare earth elements for chip manufacturing and the installation of energy-intensive infrastructure in developing nations create an imbalance that the international community can no longer ignore.
The UN's recommendations are clear: a stricter regulatory framework is needed to compel tech companies to disclose their exact environmental footprint. Furthermore, it suggests investing in "circular computing," where the heat generated by data centers is reused for district heating in cities, and the use of non-potable water for cooling systems.
Conclusion: The Need for a "Green" Intelligence
Technological progress cannot be an end in itself if it undermines the foundations of life on the planet. The UN report serves as an urgent wake-up call. The AI industry must prove it can evolve without "thirsting" the planet. The transition from the "largest possible AI" to the "most sustainable AI" is the greatest challenge the industry faces today. Without transparency and accountability, our digital paradise risks turning into an ecological hell.
