Is Generative AI the New Major Driver of Climate Change?
18 Feb 2026
Generative AI, such as large language models, is increasingly used for research, drafting, and analysis and is rapidly becoming part of everyday life.
But behind its growing ubiquity in both professional and personal contexts lies a crucial question: what are the climate costs? Is AI a major driver of digital emissions, or do other popular digital applications such as video streaming remain more impactful?
To assess the climate impact of AI, we must distinguish two stages. The first is model training, an “initial” phase of AI use that consumes vast amounts of energy. Some reported “advanced” AI models required about 1,287 megawatt-hours (MWh) of electricity, equivalent to the annual consumption of 120 households in the United States, and generated nearly 552 tons of CO₂ emissions during their training phase, comparable to the yearly emissions of more than 110 cars. Millions of liters of water were also used to cool servers during the process.[1]
The second stage is the daily, “real‑world” use by millions of users. Each AI query consumes on average 0.3 to 3 watt-hours (Wh) of electricity and emits 2.5 to 5 g of CO₂ emissions,[2] which is equivalent to burning a sheet of paper. Though the impact of a single request seems minor in AI, the cumulative effect of billions of queries is not. Estimates suggest that between twenty and fifty queries require about half a liter of water to cool data centers.
Placed in the broader context of digital technologies, these figures are striking. In 2024, the information and communication technology (ICT) sector accounted for approximately 2 to 4% of global greenhouse gas emissions and 6 to 7% of worldwide electricity consumption, a share comparable to that of the aviation industry.[3] According to the International Energy Agency, electricity demand from data centers could double by 2030, driven largely by the expansion of AI.[4]
Streaming video, however, remains the largest source of internet energy consumption today. Between 60 and 70% of global internet traffic is linked to online video and streaming services.[5] Studies show that one hour of high‑definition streaming across platforms can produce between 50 to 100 g of CO₂ emissions. Social media platforms, with their short‑form videos, account for roughly 15% of web traffic and nearly 50 million tons of CO₂ emissions annually.[6]
With generative AI still in its infancy, an important question remains: how soon, if ever, will its emissions overtake those of video streaming? Online video was projected to account for more than 80% of global internet traffic in 2025,[7] yet projections show that generative AI will grow at a much faster pace by 2030. While streaming will continue to expand, its growth is projected to be slower than that of AI. As a result, AI is likely to represent an increasing share of digital emissions over time, even as video remains the dominant source of global web traffic by 2030.
Generative AI represents a major technological breakthrough, but its contribution to climate change cannot be ignored. The substantial energy required for model training, combined with the cumulative impact of everyday use, has already made its footprint comparable to that of entire industries. Innovations in AI must therefore be reoriented to prioritize sustainability and ensure that technological progress does not come at the expense of the planet.
The impact of AI, and of video streaming, on climate change is not only the responsibility of companies that develop and operate the technologies, but also of the millions of users whose collective action creates significant effects. To remain climate-conscious, users must ask whether they are ready to adjust daily habits to reduce their footprint or at least rationalize them. As the saying goes, “small streams make big rivers,” meaning even the smallest actions, when multiplied, can have a meaningful impact.
Ultimately, the growing use of generative AI and digital services is no longer a marginal climate issue. Their energy and water consumption translate directly to greenhouse gas emissions, contributing to global climate change and increasing pressure on electricity systems worldwide. For organizations, understanding and managing these impacts is becoming essential not only to meet regulatory and stakeholder expectations, but also to make informed decisions about digitalization strategies.
We’re here to help!
Intertek Assuris can support organizations’ digitalization strategies by helping them calculate their greenhouse gas output according to international standards, and provide independent verification and assurance of reported emissions. Measuring and validating carbon footprints can move organizations from awareness to action and build climate responsibility into technological innovation.
Discover how Intertek helps businesses move forward with confidence: https://www.intertek.com/assuris/sustainability/ghg-verification/
[1] https://news.mit.edu/2025/explained-generative-ai-environmental-impact-0117
[2] https://www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech/
[3]https://documents1.worldbank.org/curated/en/099121223165540890/pdf/P17859712a98880541a4b71d57876048abb.pdf
[4] https://www.iea.org/reports/energy-and-ai
[5] https://arxiv.org/abs/2401.09854
[6] https://greenly.earth/en-gb/leaf-media/data-stories/the-hidden-environmental-cost-of-social-media
[7] https://www.teleprompter.com/blog/video-marketing-statistics
