Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Review
. 2025 Feb 28;6(5):100866.
doi: 10.1016/j.xinn.2025.100866. eCollection 2025 May 5.

Tracking the carbon footprint of global generative artificial intelligence

Zhaohao Ding  1 , Jianxiao Wang  2 , Yiyang Song  3 , Xiaokang Zheng  1 , Guannan He  4 , Xiupeng Chen  5 , Tiance Zhang  1 , Wei-Jen Lee  6 , Jie Song  2   4
Affiliations
Review

Tracking the carbon footprint of global generative artificial intelligence

Zhaohao Ding et al. Innovation (Camb). .
No abstract available

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1
Figure 1
Quantitative results of the energy consumption and carbon footprint of global GAI (A) Regional distributions of the energy consumption and carbon emissions of GAI models. The color depth of different regions indicates their carbon emission levels, whereas the size of the bubbles represents their energy consumption levels. (B) Proportion of the computational load of 64 typical GAI models in different countries. The top chart shows the computational load proportion for GAI models with more than 10 billion parameters, whereas the bottom chart shows that for GAI models with fewer than 10 billion parameters. Notably, owing to space constraints, typical large models with a proportion of less than 0.5% are not shown and are, instead, detailed in the supplemental information. (C) Energy consumption and carbon emissions of large domain-specific models. The right side illustrates the distinct domain-specific models. The left side shows the energy consumption, carbon emissions, and economic scale of these models. (D) Prediction of GAI footprint in four scenarios. The top and bottom whiskers of the boxplot represent the quartiles, whereas the yellow line indicates the median. The square represents the mean value. The scatter points depict GAI carbon emissions for low-, medium-, and high-level parameters; daily active users; and tokens across 27 scenarios. The curves illustrate the normal distribution of carbon emissions. (E) Global GAI average carbon emissions in 2035 under four scenarios.
See this image and copyright information in PMC

References

    1. Zhao T., Wang S., Ouyang C., et al. Artificial intelligence for geoscience: Progress, challenges, and perspectives. Innovation. 2024;5 doi: 10.1016/j.xinn.2024.100691. - DOI - PMC - PubMed
    1. Xu Y., Wang F., Zhang T. Artificial intelligence is restructuring the new world. Innovation. 2024;6 doi: 10.1016/j.xinn.2024.100725. - DOI - PMC - PubMed
    1. Kaack L.H., Donti P.L., Strubell E., et al. Aligning artificial intelligence with climate change mitigation. Nat. Clim. Change. 2022;12:518–527.
    1. Brown T., Mann B., Ryder N., et al. Language Models are Few-Shot Learners. Adv. Neural Inf. Process. Syst. 2020;33:1877–1901.
    1. Ding Z., Wang J., Song Y., et al. Tracking the carbon footprint of global generative artificial intelligence: Supplementary information. ResearchGate. 2024 doi: 10.13140/RG.2.2.24123.12326. Preprint at. - DOI - PMC - PubMed

LinkOut - more resources