Artificial intelligence could consume as much water by 2030 as 1.3 billion people currently use [1].
This projection highlights a critical tension between the rapid acceleration of computing power and the sustainability of global freshwater supplies. As data centers expand to support generative AI, the cooling requirements for these facilities place an increasing strain on local ecosystems.
UN scientists authored the report to address the environmental footprint of the technology sector [1]. The findings suggest that current environmental assessments have underestimated the water required to maintain the infrastructure behind AI compute [1].
Data centers rely on vast amounts of water to prevent servers from overheating. This process involves circulating water through cooling systems, which can lead to significant evaporation and local water scarcity. The report indicates that the global scale of this consumption will reach a level equivalent to the water usage of 1.3 billion people by 2030 [1].
Because AI models require massive amounts of energy and processing power, the demand for cooling grows in tandem with the complexity of the software. The UN scientists said the rapid growth of AI compute and data-center cooling demands are being overlooked in current assessments [1]. This gap in reporting may hinder the ability of governments to regulate water usage effectively as the technology scales.
The report emphasizes that these resources are being drawn from worldwide freshwater supplies, potentially impacting regions already facing water stress. The scale of the projected consumption suggests that the digital transformation of the economy carries a physical cost that is not yet fully accounted for in corporate or governmental planning [1].
“AI could consume as much water by 2030 as 1.3 billion people currently use”
The report signals a shift in the AI debate from purely digital concerns, such as algorithmic bias or job displacement, to tangible ecological impacts. By quantifying the water footprint, the UN is framing AI sustainability as a public health and resource management issue, suggesting that future AI growth may be limited not by chip availability, but by the availability of freshwater for cooling.
