The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink
Open accessPublished version of a preprint
The DOI matches its Crossref record; 18 of the 23 references it cites were found in a registry or an index.
Abstract
Machine learning (ML) workloads have rapidly grown, raising concerns about their carbon footprint. We show four best practices to reduce ML training energy and carbon dioxide emissions. If the whole ML field adopts best practices, we predict that by 2030, total carbon emissions from training will decline.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“The energy consumption and the hardware requirements are substantial and simply train -ing a single advanced LLM can be equivalent to the carbon footprint of multiple vehicles throughout their lifetime (Patterson et al. 2022 ).”
From A review of LLMs and their applications in the architecture, engineering and construction industry (2025) · cites this paper as Patterson et al. 2022
In the citing paper
“Recent advances in large language models (LLMs) have shown great potential in solving complex tasks with reasoning capabilities (Huang and Chang, 2022; Patterson et al., 2022; Achiam et al., 2023; Mao et al., 2023; Dutta et al., 2025)”
From Thought Space Explorer: Navigating and Expanding Thought Space for Large Language Model Reasoning (2024) · cites this paper as Patterson et al., 2022
In the citing paper
“Although emissions are amortized over the model lifetime, the recent ChatGPT model was estimated to consume 1287 MWh that translates into a cost of 522 tCO2e [98].”
From Connecting the dots in trustworthy Artificial Intelligence: From AI principles, ethics, and key requirements to responsible AI systems and regulation (2023) · cites this paper as [98]
In the citing paper
“training GPT-3 (the system on which the popular ChatGPT chatbot is based17) produces approximately 552 metric tons CO2e11.”
From The carbon emissions of writing and illustrating are lower for AI than for humans (2024) · cites this paper as [11]
Citations
Cites
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Kaugeranna, R., Kaugeranna, E., & Aion, (. S. 4. 6. (2023) · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · doi:10.4230/lipics.giscience.2023.43 · Cited by 13,690
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445922 · Cited by 6,988
Energy and Policy Considerations for Deep Learning in NLP
Strubell, E., Ganesh, A., & McCallum, A. (2019) · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/p19-1355 · Cited by 3,684
Schwartz, R., Dodge, J., Smith, N. A., & Etzioni, O. (2020) · Communications of the ACM · doi:10.1145/3381831 · Cited by 1,647
Recalibrating global data center energy-use estimates
Masanet, E., Shehabi, A., Lei, N., Smith, S., & Koomey, J. (2020) · Science · doi:10.1126/science.aba3758 · Cited by 1,545
The real climate and transformative impact of ICT: A critique of estimates, trends, and regulations
Freitag, C., Berners-Lee, M., Widdicks, K., Knowles, B., Blair, G. S., & Friday, A. (2021) · Patterns · doi:10.1016/j.patter.2021.100340 · Cited by 570
The Computational Limits of Deep Learning
Thompson, N. C., Greenewald, K., Lee, K., & Manso, G. F. (2020) · arXiv · doi:10.48550/arxiv.2007.05558 · Cited by 538
The Computational Limits of Deep Learning
Thompson, N., Greenewald, K., Lee, K., & Manso, G. F. (2023) · Ninth Computing within Limits 2023 · doi:10.21428/bf6fb269.1f033948 · Cited by 495
Carbon Emissions and Large Neural Network Training
Patterson, D., Gonzalez, J., Le, Q., Liang, C., Munguia, L.-M., Rothchild, D., So, D., Texier, M., & Dean, J. (2021) · arXiv · doi:10.48550/arxiv.2104.10350 · Cited by 486
Self-citationQuantifying the Carbon Emissions of Machine Learning
Lacoste, A., Luccioni, A., Schmidt, V., & Dandres, T. (2019) · arXiv · doi:10.48550/arxiv.1910.09700 · Cited by 408
Towards the Systematic Reporting of the Energy and Carbon Footprints of Machine Learning
Henderson, P., Hu, J., Romoff, J., Brunskill, E., Jurafsky, D., & Pineau, J. (2020) · arXiv · doi:10.48550/arxiv.2002.05651 · Cited by 393
So, D. R., Liang, C., & Le, Q. V. (2019) · arXiv · doi:10.48550/arxiv.1901.11117 · Cited by 277
Self-citationGLaM: Efficient Scaling of Language Models with Mixture-of-Experts
Du, N., Huang, Y., Dai, A. M., Tong, S., Lepikhin, D., Xu, Y., Krikun, M., Zhou, Y., Yu, A. W., Fırat, O., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2112.06905 · Cited by 157
Self-citationPrimer: Searching for Efficient Transformers for Language ModelingPreprint
So, D. R., Mańke, W., Liu, H., Dai, Z., Shazeer, N., & Le, Q. V. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2109.08668 · Cited by 145
Self-citationDeep Learning's Diminishing Returns: The Cost of Improvement is Becoming Unsustainable
Thompson, N. C., Greenewald, K., Lee, K., & Manso, G. F. (2021) · IEEE Spectrum · doi:10.1109/mspec.2021.9563954 · Cited by 139
Does not compute: Avoiding pitfalls assessing the Internet's energy and carbon impacts
Koomey, J., & Masanet, E. (2021) · Joule · doi:10.1016/j.joule.2021.05.007 · Cited by 53
Searching for Fast Model Families on Datacenter Accelerators
Li, S., Tan, M., Pang, R., Li, A., Cheng, L., Le, Q. V., & Jouppi, N. P. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.00799 · Cited by 29
Self-citation
Research works printed but not linked: 3
- jouppi, 0, Ten lessons from three generations shaped Google’s TPUv4i, Proc Int Symp Comput Arch, 1
- evans, 2021, Amazon shocker CEO Jassy says cloud less than 5% of all IT spending
- schechner, 2021, Amazon and other tech giants race to buy up renewable energy
Cited by
Recent Advances in Natural Language Processing via Large Pre-trained Language Models: A Survey
Min, B., Ross, H., Sulem, E., Veyseh, A. P. B., Nguyen, T. H., Sainz, O., Agirre, E., Heintz, I., & Roth, D. (2024) · ACM Computing Surveys · doi:10.1145/3605943 · Cited by 1,387
Díaz-Rodríguez, N., Del Ser, J., Coeckelbergh, M., López de Prado, M., Herrera-Viedma, E., & Herrera, F. (2023) · Information Fusion · doi:10.1016/j.inffus.2023.101896 · Cited by 800
The growing energy footprint of artificial intelligence
de Vries, A. (2023) · Joule · doi:10.1016/j.joule.2023.09.004 · Cited by 597
Jouppi, N., Kurian, G., Li, S., Ma, P., Nagarajan, R., Nai, L., Patil, N., Subramanian, S., Swing, A., Towles, B., et al. (2023) · Proceedings of the 50th Annual International Symposium on Computer Architecture · doi:10.1145/3579371.3589350 · Cited by 508
Self-citationEstimating the Carbon Footprint of BLOOM, a 176B Parameter Language Model
Luccioni, A. S., Viguier, S., & Ligozat, A.-L. (2022) · arXiv · doi:10.48550/arxiv.2211.02001 · Cited by 378
Metadata
- Authors
- Patterson, D., Gonzalez, J., Holzle, U., Le, Q., Liang, C., Munguia, L.-M., Rothchild, D., So, D. R., Texier, M., & Dean, J.
- Affiliations
- University of California, Berkeley (US)
- Google (United States) (US)
- Venue
- Computer
- Published
- 2022-07
- Volume
- 55
- Issue
- 7
- Pages
- 18-28
- ISSN
- 0018-9162 (ISSN-L)
- 1558-0814
- OpenAlex
- W4283704460
- Cited by
- 400
- References
- 23
- Citations by year
- 2026: 110 · 2025: 104 · 2024: 77 · 2023: 93 · 2022: 16
- FWCI
- 19.84
- Type
- journal article
- Language
- en
- Open access
- green
- doi.org
- Licence
- Locations
- Computer (publishedVersion)
- Figshare (submittedVersion, cc-by)
- Topics
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Green IT and Sustainability
- Physical Sciences › Computer Science › Hardware and Architecture › Energy Efficiency in Computing
- Physical Sciences › Computer Science › Information Systems › Cloud Computing and Resource Management
- Keywords
- carbon footprint, machine learning training, energy consumption, CO2 emissions, emission reduction
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.