A new study has added to concerns over the growing carbon footprint of the increased use of artificial intelligence (AI).

The team of researchers found that continued growth in AI use across the United States could add approximately 900,000 tonnes of CO₂ annually.

They added while not a small amount it equates to a relatively minor increase when viewed in the context of nationwide emissions.

While AI adoption is expected to boost productivity and economic output, researchers note that its environmental footprint can be seen as relatively modest compared to other industrial activities. The study examined potential AI integration across various sectors, estimating the associated rise in energy use and carbon emissions.

Key findings in the report included:

• AI adoption across the U.S. economy may result in an additional 896,000 tonnes of CO₂ emissions per year, which represents just 0.02% of total U.S. emissions.
• Energy use in individual industries could increase by up to 12 petajoules annually, comparable to the electricity consumption of around 300,000 U.S. homes.

Co-author Anthony Harding explained: “While the projected emissions from AI adoption are modest compared to other sectors, they still represent a meaningful increase. This underscores the importance of integrating energy efficiency and sustainability into AI development and deployment, especially as adoption accelerates across industries.”

As AI technologies become more integrated into daily operations, researchers said they are keen to encourage industry leaders to incorporate energy efficiency and sustainability into their AI strategies to ensure responsible growth as adoption scales.

The study comes after Oxford University revealed the rapid rise in the use of generative AI (GenAI) has raised questions about its environmental impact, at the university and beyond.

“AI brings new opportunities to the ways in which we study and work, but it also brings challenges,” it said in a statement. “As AI continues to grow, there is an increasing focus on its environmental implications.”

The University‘s AI Competency Centre has explored these issues and how Oxford is responding to them.

It added: “One of the challenges of considering AI’s energy impact is the difficulty of getting a real understanding of its energy usage and environmental implications from tech companies.

“It’s challenging to get a reliable estimate of the energy consumption of AI because a lot depends on the application and model being used, and available approximations vary considerably. Whilst the sector is expanding rapidly, the technology is also changing over time, bringing improvements in the energy consumption of AI models, hardware and data centres.”

The  university explained AI’s energy usage comes in two parts: the initial energy required for training new models, which is usually a large amount over a short period of time; and the energy cost per individual query, which is relatively low per prompt but can add up over time.

“Understanding how AI will affect not only the University’s carbon footprint, but also everyday teaching and learning, research, and administration, is an ongoing process,” the study added. “The University is actively monitoring and adapting to this fast-changing field, building a clearer picture of how AI is being used and of how this usage aligns with our sustainability targets, in order to inform and shape our approach.”

While AI adoption is expected to boost productivity and economic output, researchers note that its environmental footprint can be seen as relatively modest compared to other industrial activities. The study examined potential AI integration across various sectors, estimating the associated rise in energy use and carbon emissions.