WASHINGTON (Realist English). Environmental organizations around the world face a fundamental contradiction: the technology they use to fight the climate crisis is itself a source of growing emissions, electricity and water consumption.
According to a Reuters survey, 15 leading conservation organizations, including Conservation International, Greenpeace International, World Resources Institute, Nature Conservancy and BirdLife International, confirmed that they use AI in their work, and all expressed concern about the environmental footprint of the data centers that power these models.
AI as a conservation tool
Organizations use AI for tasks that were previously difficult or required enormous human resources. Conservation International uses algorithms to assess hotspots detected by satellites and send prompt alerts to community forest patrols about potential fires in remote areas of Cambodia.
Kenya’s Save the Elephants uses machine learning to improve elephant population counts and study their communication: aerial surveys of Kenya’s largest protected area produced about 200,000 images, which would take nine people nine months to analyze manually, while AI models process them in hours.
“The results were astonishing. We found that human counts missed about a third of elephants, about half of zebras and giraffes, and three-quarters of oryx populations,” Reuters quoted the organization’s head, Frank Pope, as saying.
At the institute founded by Jane Goodall, AI is used to digitize decades of field notes on chimpanzee behavior in Gombe National Park in Tanzania. The project, which was expected to take 13–15 years due to handwritten records in the local Kiha language and scientific abbreviations, can be completed in about two years thanks to a tool developed with Amazon Web Services.
Scale of data center impact
The environmental costs that concern organizations have concrete quantitative estimates. According to the International Energy Agency (IEA), by 2030 global data centers will consume about 945 terawatt-hours of electricity per year — twice the current level and roughly the same as Japan consumes. This production is equivalent to more than 300 million tons of carbon dioxide. Water consumption, according to IEA forecasts, will rise from about 560 billion liters in 2023 to 1.2 trillion liters per year by 2030.
Researchers at the UN University Institute for Water, Environment and Health estimate that the global footprint of data centers could more than double by 2030, reaching an area roughly twice the size of Jakarta. A separate study published in Environmental Science & Technology found that 25% of the world’s data centers are already located in regions with at least one month of water scarcity per year, and by the 2040s this share could grow to one-third.
Tech giants and their commitments
Major technology companies, including Microsoft, Amazon and Alphabet (Google), have reported emissions growth partly linked to data center expansion, casting doubt on their ability to meet long-term climate commitments. Amazon told Reuters that its AI products are among the most sustainable due to low-carbon energy purchases and efforts to improve energy and water efficiency. Microsoft and Google did not respond to requests for comment.
Seeking solutions: from “edge computing” to internal limits
Some organizations are already adopting practices to reduce the impact of their AI activity. World Resources Institute (WRI) limits resource-intensive applications such as video and image generation and caches query results so identical searches are not processed repeatedly.
Nature Conservancy uses AI to analyze video from commercial fishing vessels, but instead of uploading huge amounts of data to centralized data centers, it installs AI cameras directly on vessels, reducing energy consumption and data transmission requirements.
Greenpeace International, although it uses some AI tools, mainly because the software platforms it relies on include AI features, would prefer to do without the technology because of its environmental impact.
“While other organizations ask how to use AI responsibly, Greenpeace asks whether all these huge data centers need to be built at all,” Reuters quoted activist Kathrin Sueing Wu as saying.
Political context: COP31 and the call for transparency
The contradiction between AI use and its environmental footprint is expected to be a central topic at the UN climate conference COP31, which will be held this month in Turkey.
The host country will call on governments, businesses and other institutions to sign a commitment on principles for sustainable AI development and use. UN Secretary-General António Guterres has called on major AI companies to publicly disclose the environmental impact of their systems.
Environmental organizations find themselves in a position where moral consistency clashes with practical necessity. On the one hand, they cannot ignore that the data centers powering AI consume energy and water on a scale comparable to entire countries, and that the tech giants behind these facilities are still far from meeting their own climate commitments.
On the other hand, AI tools allow them to do what was unthinkable just a few years ago: process hundreds of thousands of images in hours, predict fires, and make nature “measurable, and therefore investable,” as the head of Save the Elephants put it.
The key question posed by this contradiction is not whether to use AI, but on what terms. If environmental organizations can insist on transparency from tech companies, prioritize edge computing and limit resource-intensive applications, they have a chance to turn AI from an uncomfortable compromise into a tool that does not undermine but strengthens their mission.
If not, they risk finding themselves in a position where their own tools work against the very goal for which they exist. The answer to this question will be shaped not only at COP31 but also in everyday decisions about which models to use, where to locate computing, and what commitments to demand from suppliers.
