Water footprint assessment of Artificial Intelligence related applications in Europe

This master’s thesis investigates the water footprint associated with artificial intelligence (AI) applications in Europe, with a specific focus on data centers, which represent the core infrastructure supporting AI systems. While the energy consumption of digital technologies has been widely studied, their water-related impacts remain less explored, despite their growing relevance in the context of environmental sustainability and resource management. The study adopts a Life Cycle Assessment (LCA) approach, in accordance with ISO 14046 guidelines, and uses the SimaPro software combined with the Ecoinvent database and the AWARE method to quantify water consumption. Three main components are analyzed: i. Water footprint associated with electricity generation, ii. Water footprint in hardware production, particularly GPUs, iii. Water footprint for cooling systems in data centers. The results highlight significant geographical differences across Europe. Countries such as Spain and Italy emerge as the main contributors, due to a combination of factors including energy mix, climatic conditions, and demand for digital infrastructure. In contrast, Nordic countries show a consistently lower impacts, benefiting high shares of renewable energy and favorable climate. The findings demonstrate that the water footprint of AI is a multi-dimensional issue, influenced by the interaction of energy systems, global supply chains, and local operational factors. However, the analysis is subjected to limitations related to data availability, modeling assumptions, and the use of average values. Overall, the study emphasizes the need for integrated strategies to reduce water consumption in digital infrastructures, highlights the importance of renewable energy adoption, efficient cooling technologies, and improved transparency in resource use.