Last Wednesday, President Trump unveiled a broad “AI Action Plan” that supplies the administration’s regulatory intent, or lack thereof, around AI development, deployment, and its supporting infrastructure. While some of the plan’s ambitions and recommendations hold merit, the plan raises critical concerns regarding climate and environmental impacts that cannot be overlooked.
A key section outlines the administration’s recommendations around meeting the energy demand for AI infrastructure. It includes categorical exclusions of “data center-related actions” from National Environmental Policy Act (NEPA) review, circumventions of the Clean Air and Clean Water Acts, and the use of federal land for data center construction. The administration’s short-sighted policies on climate change are well documented, yet the implications of these policies in tandem with the growing AI revolution will have far-reaching consequences.
The exact scale of AI’s projected impact on the electrical grid, water, and other climate impacts remains speculative relative to the technology’s possible rate of growth. However, the energy and water demand necessary to facilitate the development and deployment of AI will lead to unavoidable impacts regardless of the growth case.
A comprehensive study conducted by the UC Berkley in 2024 found their models projected AI data centers will make up 6-12% of the nation’s energy consumption in 2028, compared to 4.3% in 2024. According to a 2025 International Energy Agency report, a 100WH hyperscale data center in the United States consumes 2 million liters of water per day, equivalent to about 6,500 households. The same report shows global data center water usage at 560 billion liters per year, with an expected 1,200 billion liters per year in 2030, roughly double the annual water usage of the entire city of Los Angeles.
Though AI models and data centers continue to advance in energy efficiency, without corresponding action, increased emissions and environmental harms remain a stalwart outcome. Under current climate policies, the International Monetary Fund projects that AI-driven electricity demand could add 1.7 gigatons to global greenhouse gas emissions between 2025 and 2030. A Bloomberg News report found that two-thirds of all U.S. data centers built or planned will be built in regions already facing high water stress. Additionally, the necessary minerals and other materials needed to produce AI computing hardware will likely exacerbate unsustainable and dangerous mining practices.
The immanence of these potential futures is daunting, but they need not be fatalistic. Where AI presents immense climate risk, it also inspires great opportunity. Not only to address its energy demands and climate externalities, but also as a tool to support broader climate just transition goals and initiatives.
Innovations in AI’s usage to bolster energy efficiency in its infrastructure and models’ design illustrate the technology’s double-edged potential. In 2024, Microsoft unveiled a new data center design that integrates AI and smart cooling technology, requiring no water for cooling, potentially eliminating 125 million liters of water per year per data center. MIT and Northeastern University Researchers in 2023 designed a software tool that uses carbon-friendly machine learning inferences to build sustainable AI models with drastically reduced emissions
The avenues for a sustainable AI ecosystem are evident and continue to grow exponentially with each passing year. In tandem with AI’s acceleration of scientific innovation, there is immense potential for AI to reshape our energy future.
“Smart Grid” projects have been funded throughout the country and globe that seek to integrate AI solutions to produce stable and optimized grid management. For example, researchers at the Argonne National Laboratory have developed AI software that predicts expected failures of grid components, limiting outages and energy waste from inefficiencies and maintenance. AI is already being leveraged to predict inflows and outflows of energy systems, but could potentially revolutionize grid integration and management of complex energy ecosystems, providing energy savings from both fossil fuels and renewable energy sources.
While the U.S regulatory environment is signaling a step back in public investments towards combating climate change, the global appetite for climate solutions continues to surge. “The 2024 UN Adaptation Gap Report, estimated financing needs of climate adaptation between $215 billion and $387 billion a year until 2030 for just the countries of the Global South, with the global projected demand for spending rising to between $0.5 trillion and $1.3 trillion a year by 2030. This investment need is coupled with an equally attractive return potential for investors.
A 2025 working paper by the World Resources Institute found that from 320 surveyed climate investments that every $1 invested yielded a return of $10.50 over a 10-year period, with an average 27% ROI.
Ultimately, the relationship between AI and a climate just transition does not have to be a zero-sum game. Advancements in the former need not detract from the latter. Investors are in a crucial position to drive the development of an AI ecosystem that integrates the alignment of climate goals rather than sacrifices them. The tremendous investment opportunities for novel climate adaptation technologies and strategies are attractive not just from an innovation standpoint, but from a return perspective as well.
If you are interested in learning more about you and your firm can engage on AI and just climate transition initiatives, reach out to our team at compass@rfkhumanrights.org