Technological innovations have an often overlooked downside: the significant energy and water consumption associated with the infrastructure for AI models. The energy requirements of large AI systems such as ChatGPT have multiplied dramatically in recent years, resulting in high CO2 emissions. Microsoft alone increased its emissions in connection with AI infrastructures from around 12 to 15 million tonnes of CO2 between 2020 and 2023. In addition, data centres consume enormous amounts of fresh water, which leads to conflicts, especially in water-scarce regions such as Arizona. Even more critically, global AI demand in 2027 is expected to account for 4.2 to 6.6 billion cubic metresof water abstraction - more than the entire annual water abstraction of Denmark or half of the United Kingdom. The more investment is made in AI, the more public awareness is raised of the environmental impact of powering and cooling huge data centres. Companies such as Microsoft, Google and Amazon are investing heavily in AI technology and therefore also in the expansion of energy-intensive data centres. Although these companies have set themselves sustainability targets, the reality remains sobering: the emissions they cause continue to rise, while at the same time only limited measures to reduce resource consumption can be demonstrated. In fact, the regions where the largest growth in data centres is taking place also tend to be areas where a large proportion of the electricity generated comes from fossil fuels. While some technology companies publicly commit to making their operations carbon neutral, the extent to which these promises are actually realised remains questionable. Even if some of the energy required comes from renewable sources, this does not change the enormous amount of water or materials used to build and maintain the infrastructure. In addition, the high energy consumption of AI systems remains an unresolved problem.