1This section focuses on two crucial dimensions of AI’s environmental footprint: energy use and water consumption. There has been much public attention over the last few years on AI’s energy use. More recently, concerns have been raised about data center water use. Understanding these impacts is essential for designing AI systems that align with sustainability goals and for mitigating the environmental consequences of their rapid growth.

Zoom Original (jpeg, 933k)

2Jonathan Koomey and Eric Masanet examine how artificial intelligence (AI) affects energy use, emphasizing that its impacts extend beyond the electricity consumed during model training and inference. While the direct energy use of AI operations in data centers is the most visible and measurable effect, the authors stress that AI’s broader impacts arise from how it is applied across energy systems and how it interacts with the wider economy. AI deployments could either increase or reduce overall energy demand, depending on whether they support efficiency improvements and renewable energy integration or drive rebound effects, expanded fossil fuel extraction, and increased consumption. These complex systemic interactions between AI systems, energy, and the economy contribute to the uncertainty about the net effect of AI on energy use and emissions. Because of the unpredictable evolution of AI technology, Koomey and Masanet argue for systematic measurement, improved data and analytical frameworks, and scenario-based approaches rather than single-point projections to understand AI’s long-term implications for energy use and emissions.

3Ana Pinheiro Privette shifts the focus to a less well understood dimension of AI operations: the water footprint of data centers. Privette explains that training and deploying large-scale AI models require substantial computational power, generating heat that is often managed through water-intensive cooling systems. In addition, the electricity generation needed to power these AI systems can also involve significant indirect water use. Although data centers’ global water use remains modest compared to other sectors, their rapid expansion can intensify localized water stress, particularly in water-scarce regions. To address these challenges, Privette emphasizes the need to align AI growth and data center development with local and regional water resource planning, alongside adopting circular water management practices such as wastewater reuse and closed-loop cooling. The article also underscores the importance of standardized water-use metrics and transparent reporting to support effective governance and more water-conscious AI development. Additionally, AI itself can be leveraged for predictive monitoring and adaptive water management, enabling smarter water resource management. Privette concludes by emphasizing the urgency of a multi-faceted, cross-sector approach to ensure AI’s sustainable evolution, positioning AI as a potential enabler of broader environmental stewardship.

4Together, these two articles provide an overview of AI’s environmental impacts in terms of both energy and water use. Many of these impacts appear in local communities where data centers operate. Articles in Sections 1, 2, 4, and 5 examine community-level impacts and opportunities, including access to data, inclusive governance, and workforce capacity.

5While AI holds tremendous potential for driving sustainability ‒ particularly through optimizing management of energy and water systems ‒ it also poses significant challenges, primarily local challenges, as it continues to expand. Address these challenges requires a combination of continued technology innovation, regulations, transparency, and a community first approach to AI infrastructure development.