AI energy demand in US proves minor climate impact

by Clarence Oxford
Los Angeles CA (SPX) Nov 13, 2025
Growing interest in the environmental impact of artificial intelligence has sparked widespread debate among scientists, energy analysts, and climate advocates. AI systems, while fueling progress in automation and services, are frequently scrutinized for their electricity demand and potential to drive up global greenhouse gas emissions. Recent projections estimate that AI-related data center energy consumption could account for up to 22 percent of annual US household energy by 2028, with the sector's emissions potentially matching those from driving more than 300 billion miles per year. Reports indicate that data centers now consume around 4.4 percent of all US electricity, a figure expected to rise sharply with continued adoption of AI technologies.

Concerns center on the carbon intensity of electricity used in AI-powered data centers. Despite large investments by major technology firms like Google and Microsoft in renewable energy, overall emissions have surged due to rapid AI deployment and increased data center expansion. In recent years, these firms have acknowledged their own greenhouse gas emissions have risen by more than 30 percent as a direct consequence of scaling AI services and infrastructure.

However, the landscape is nuanced. Some researchers note that while surges in local power demand and emissions might be pronounced shortly after new data center deployments, the overall share of global electricity and emissions from the sector remains low - currently around 1 percent of global electricity and 0.5 percent of CO2 emissions. Additionally, AI can contribute solutions by optimizing grid management, improving the energy efficiency of transportation and industrial processes, and aiding the transition to renewable energy systems.

The University of Waterloo and Georgia Institute of Technology investigation challenges the narrative of runaway AI-induced energy demand. Their analysis demonstrates that, in the United States, AI's direct energy consumption, though equivalent to that of Iceland, represents a negligible share at the national level.

The study revealed that roughly 83 percent of the US economy relies on petroleum, coal, and natural gas, primary sources of climate-changing emissions. The incremental energy use attributable to AI does not notably increase overall demand or greenhouse gas output at broad scales.

Dr Juan Moreno-Cruz, Faculty of Environment at Waterloo and Canada Research Chair in Energy Transitions, said: "It is important to note that the increase in energy use is not going to be uniform. It's going to be felt more in the places where electricity is produced to power the data centres. If you look at that energy from the local perspective, that's a big deal because some places could see double the amount of electricity output and emissions. But at a larger scale, AI's use of energy won't be noticeable."

The study did not examine local economic effects linked to data center placement but highlighted potential technological upsides.

Moreno-Cruz stated: "For people who believe that the use of AI will be a major problem for the climate and think we should avoid it, we're offering a different perspective. The effects on climate are not that significant, and we can use AI to develop green technologies or to improve existing ones."

Environmental economists Moreno-Cruz and Dr Anthony Harding intend to extend this research model to other global regions, aiming to clarify AI's long-term role in energy systems and climate solutions.

Research Report:Watts and bots: the energy implications of AI adoption