Among the ripples proliferating from rapid AI development and use, few are more complicated than those hitting our natural environment and resources.
As society attempts to mitigate the increasingly devastating impacts of climate change, AI systems—which require vast amounts of energy and water to train and use—are clearly exacerbating the problem. But can they help to solve it as well?
These topics are explored in the IT Professional article, “The Environmental Impact of Artificial Intelligence,” which looks at the resources that AI training and deployment consumes, as well as examining AI’s potential to help mitigate climate change by contributing to energy efficiency and sustainability efforts.
AI Resource Consumption Stages
As the article explains, AI consumes energy and water at three points in the technology lifecycle:
- Producing hardware and infrastructure: Creating the new servers, storage, and networks required to develop, train, and operate AI already contributes substantially to energy consumption, water use, and carbon emissions.
- Model training: Conducting numerous iterations of mathematical operations on vast datasets during AI model training requires considerable computational power and storage.
- Inference phase: After deploying AI systems, the process of predicting outcomes using new input data also consumes significant resources.
AI’s Growing Resource Consumption
Although AI currently accounts for only a smaller fraction of the environmental footprint of global data centers, the article details how this amount is rapidly expanding along with generative AI’s widespread adoption.
In early 2024, for example, AI’s share of the power consumption in global data centers was only 2%, but this consumption was expected to surge to at least 10% by the start of 2025. Further, some tech companies are already well past this percentage. For example, as early as 2021, machine learning accounted for up to 15% of annual energy consumption in Google’s cloud infrastructure.
All this AI energy consumption generates intensive heat, which in turn ripples out into water consumption. Data centers typically use cooling towers and outside air to dissipate heat and prevent server degradation or meltdown—a process that requires increasing amounts of clean, fresh water.
In 2022, for example, the article notes that Microsoft experienced a 34% surge in water use attributed to increased cooling demands in the data centers where tools such as Bing Chat and ChatGPT were being trained.
In addition to cooling down data centers, AI also contributes to substantial offsite water consumption tied to the increased electricity generation that generative AI growth demands. This includes meeting cooling demands at thermal power and nuclear plants, as well as heightened water evaporation from hydropower plants.
Can AI Help Solve These Problems?
Although existing AI resource consumption numbers are expected to rise dramatically over the next few years, AI itself may be able to help solve the resource problems that its increasing ubiquity and demand is creating.
The article details various ways in which AI already is and increasingly could decrease energy and water consumption in various phases, including by reducing a data center’s environmental impact while also increasing its efficiency and equipment lifespan.
The article also mentions other ways that AI can mitigate resource consumption, including the following:
- Enhancing both the cost-effectiveness and precision of energy infrastructure.
- Expediting the process of generating a viable-route short list for pipelines and powerlines from several days to mere seconds, streamlining decision making and project planning.
- Facilitating development of entirely new energy systems with low-carbon footprints that also help suppliers to monitor greenhouse gas sinks.
Dig Deeper
To further investigate AI and its impact (good and bad) on energy and water resources, read “The Environmental Impact of Artificial Intelligence.”
In addition to offering detailed statistics on resource usage, including for specific AI models, the article offers more details on AI’s existing and potential contributions to promoting environmental sustainability.
To dig even deeper into AI and its impact on sustainability efforts, join other AI experts, researchers, government officials, and enthusiasts at the international IEEE Conference on Artificial Intelligence (IEEE CAI) 5–7 May 2025 in Santa Clara, California.
In addition to showcasing the latest AI research and breakthroughs, IEEE CAI emphasizes applications and key subject areas, from sustainability and human-centered AI to issues and industry-specific applications in healthcare, transportation, and engineering and manufacturing.
