Forest’s Climate News
By Forest Olson
Summer 2025
The Impact of AI
AI is everywhere, in the news, on your computer and smartphone, it seems as though we can’t escape it. AI can be a useful tool that can do many things to help us in our day-to-day lives and has the potential to help us solve big scale problems too. However, AI is also having a devastating impact on the environment. Generative AI chatbots are asked to process hundreds of millions of queries each day. This takes a lot of energy. In fact, it takes 5 times as much energy for AI to answer a question or summarize an email as it does to do a simple web search. Huge data centers are being built to process these millions of messages. These temperature-controlled buildings and the super-computers that they hold are problematic for a couple of reasons.
First, these centers use massive amounts of energy. AI, particularly AI training, requires much more power density, and uses up to 8 times more energy than normal computing. For example, it is estimated that training Chat-GPT 3 (2021) used 1,287 megawatt hours of electricity (enough to power roughly 120 homes for a year), and generated 552 tons of carbon dioxide. Data centers run 24/7, so they cannot run on renewable energy. By 2026 data centers will use enough energy to power the entirety of Japan, and by 2027, half the United Kingdom. At the current rate it is estimated that by the end of the decade the global energy demand of AI will be equivalent to 2 to 6 Californias.
Second, cold water is used to cool and absorb heat from these super computers. It is estimated that for every kilowatt hour of energy data centers use, a liter of water is used for cooling. Freshwater must be used for cooling, as any other kind of water could damage the equipment. Around two thirds of new data centers built since 2022 are in locations with high water stress. While the southwest United States has become a hub for these data centers, the region, as well as large swaths of Latin America are experiencing mega drought periods, making freshwater ever more vital. This shows how even the computing hardware used for generative AI has its own, less direct impact on climate change.
While many people tend to believe that AI is all in the cloud, it is also having a devastating impact on our physical world, through massive water consumption and carbon emissions. If we are not careful, AI could counter everything we have done to slow the climate crisis. So, all things considered, AI can be a very useful tool, but we must be conscientious about our consumption to lessen its environmental impact. We must ask ourselves what are ethical uses of AI given the climate impacts? On the other hand, might AI be a tool to help us solve the climate crisis?
