mlsanityData to 14 Oct 2026

Measuring the Carbon Intensity of AI in Cloud Instances

arXiv 2206.05229 ↗First posted 10 Jun 2022cs.LGcited by 111

Not on the map. Why?

Abstract (arXiv)

By providing unprecedented access to computational resources, cloud computing has enabled rapid growth in technologies such as machine learning, the computational demands of which incur a high energy cost and a commensurate carbon footprint. As a result, recent scholarship has called for better estimates of the greenhouse gas impact of AI: data scientists today do not have easy or reliable access to measurements of this information, precluding development of actionable tactics. Cloud providers presenting information about software carbon intensity to users is a fundamental stepping stone towards minimizing emissions. In this paper, we provide a framework for measuring software carbon intensity, and propose to measure operational carbon emissions by using location-based and time-specific marginal emissions data per energy unit. We provide measurements of operational software carbon intensity for a set of modern models for natural language processing and computer vision, and a wide range of model sizes, including pretraining of a 6.1 billion parameter language model. We then evaluate a suite of approaches for reducing emissions on the Microsoft Azure cloud compute platform: using cloud instances in different geographic regions, using cloud instances at different times of day, and dynamically pausing cloud instances when the marginal carbon intensity is above a certain threshold. We confirm previous results that the geographic region of the data center plays a significant role in the carbon intensity for a given cloud instance, and find that choosing an appropriate region can have the largest operational emissions reduction impact. We also show that the time of day has notable impact on operational software carbon intensity. Finally, we conclude with recommendations for how machine learning practitioners can use software carbon intensity information to reduce environmental impact.

Key facts. Cited by 111 papers in the map (data through 14 Oct 2026). Builds on 22 of its 48 references that are in the map. First posted on arXiv 10 Jun 2022.

Builds on

22 of 48 references.

References · 22

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  3. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  4. 2017A Broad-Coverage Challenge Corpus for Sentence Understanding through Inferencecited by 2,682
  5. 2021Multitask Prompted Training Enables Zero-Shot Task Generalizationcited by 1,470
  6. 2019Energy and Policy Considerations for Deep Learning in NLPcited by 1,254
  7. 2022LaMDA: Language Models for Dialog Applicationscited by 1,220
  8. 2019Green AIcited by 627
  9. 2021Carbon Emissions and Large Neural Network Trainingcited by 567
  10. 2019Quantifying the Carbon Emissions of Machine Learningcited by 521
  11. 2021Sustainable AI: Environmental Implications, Challenges and Opportunitiescited by 242
  12. 2020The Computational Limits of Deep Learningcited by 193
10 more
  1. 2022The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrinkcited by 184
  2. 2021The Values Encoded in Machine Learning Researchcited by 172
  3. 2020Carbontracker: Tracking and Predicting the Carbon Footprint of Training Deep Learning Modelscited by 169
  4. 2020Chasing Carbon: The Elusive Environmental Footprint of Computingcited by 72
  5. 2014DenseNet: Implementing Efficient ConvNet Descriptor Pyramidscited by 66
  6. 2021AutoFL: Enabling Heterogeneity-Aware Energy Efficient Federated Learningcited by 39
  7. 2021Unraveling the Hidden Environmental Impacts of AI Solutions for Environmentcited by 32
  8. 2020SECure: A Social and Environmental Certificate for AI Systemscited by 7
  9. 2021EnergyVis: Interactively Tracking and Exploring Energy Consumption for ML Modelscited by 4
  10. 2020Machine Learning for Gas and Oil Explorationcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 9; 2023: 30; 2024: 29; 2025: 25; 2026: 18.

111 papers.

Sort:

By area:

  1. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  2. 2024OLMo: Accelerating the Science of Language Modelscited by 613
  3. 2024MOMENT: A Family of Open Time-series Foundation Modelscited by 482
  4. 2023Harnessing the Power of LLMs in Practice: A Survey on ChatGPT and Beyondcited by 411
  5. 2023Power Hungry Processing: Watts Driving the Cost of AI Deployment?cited by 191
  6. 2023Large language models can accurately predict searcher preferencescited by 131
  7. 2023AdapterSoup: Weight Averaging to Improve Generalization of Pretrained Language Modelscited by 84
  8. 2022Zeus: Understanding and Optimizing GPU Energy Consumption of DNN Trainingcited by 79
  9. 2022Efficient Methods for Natural Language Processing: A Surveycited by 78
  10. 2023A Systematic Review of Green AIcited by 77
  11. 2023LLMCarbon: Modeling the end-to-end Carbon Footprint of Large Language Modelscited by 72
  12. 2024ALLaM: Large Language Models for Arabic and Englishcited by 55
38 more
  1. 2022Efficient Diffusion Models for Vision: A Surveycited by 51
  2. 2023Counting Carbon: A Survey of Factors Influencing the Emissions of Machine Learningcited by 42
  3. 2024When Large Language Models Meet Vector Databases: A Surveycited by 39
  4. 2022Compute-Efficient Deep Learning: Algorithmic Trends and Opportunitiescited by 38
  5. 2023COBRA Frames: Contextual Reasoning about Effects and Harms of Offensive Statementscited by 34
  6. 2023How to estimate carbon footprint when training deep learning models? A guide and reviewcited by 34
  7. 2023On the application of Large Language Models for language teaching and assessment technologycited by 27
  8. 2025Holistically Evaluating the Environmental Impact of Creating Language Modelscited by 26
  9. 2024Cross-Modal Safety Alignment: Is textual unlearning all you need?cited by 25
  10. 2023Reducing Energy Bloat in Large Model Trainingcited by 22
  11. 2024Comparing Retrieval-Augmentation and Parameter-Efficient Fine-Tuning for Privacy-Preserving Personalization of Large Language Modelscited by 21
  12. 2023AI-native Interconnect Framework for Integration of Large Language Model Technologies in 6G Systemscited by 19
  13. 2024A Survey of Sustainability in Large Language Models: Applications, Economics, and Challengescited by 18
  14. 2023Leftover Lunch: Advantage-based Offline Reinforcement Learning for Language Modelscited by 17
  15. 2023Transformers in Healthcare: A Surveycited by 17
  16. 2024MBIAS: Mitigating Bias in Large Language Models While Retaining Contextcited by 15
  17. 2023DVFO: Learning-Based DVFS for Energy-Efficient Edge-Cloud Collaborative Inferencecited by 14
  18. 2024Toward Sustainable GenAI using Generation Directives for Carbon-Friendly Large Language Model Inferencecited by 13
  19. 2023Efficiency is Not Enough: A Critical Perspective of Environmentally Sustainable AIcited by 12
  20. 2022EC-NAS: Energy Consumption Aware Tabular Benchmarks for Neural Architecture Searchcited by 11
  21. 2023FedZero: Leveraging Renewable Excess Energy in Federated Learningcited by 11
  22. 2023Energy and Carbon Considerations of Fine-Tuning BERTcited by 11
  23. 2025Towards Sustainable NLP: Insights from Benchmarking Inference Energy in Large Language Modelscited by 11
  24. 2024Gradient descent with generalized Newton's methodcited by 9
  25. 2024Model-GLUE: Democratized LLM Scaling for A Large Model Zoo in the Wildcited by 9
  26. 2023Green Federated Learningcited by 8
  27. 2024Developing Safe and Responsible Large Language Model : Can We Balance Bias Reduction and Language Understanding in Large Language Models?cited by 8
  28. 2023Efficiency Pentathlon: A Standardized Arena for Efficiency Evaluationcited by 7
  29. 2024Learn to Code Sustainably: An Empirical Study on LLM-based Green Code Generationcited by 7
  30. 2026Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarkscited by 7
  31. 2024Sociotechnical Implications of Generative Artificial Intelligence for Information Accesscited by 6
  32. 2022Embedding Recycling for Language Modelscited by 5
  33. 2022Bridging Fairness and Environmental Sustainability in Natural Language Processingcited by 5
  34. 2023Multi-Objective Optimization for Sparse Deep Multi-Task Learningcited by 5
  35. 2023Towards Message Brokers for Generative AI: Survey, Challenges, and Opportunitiescited by 5
  36. 2024How green is continual learning, really? Analyzing the energy consumption in continual training of vision foundation modelscited by 5
  37. 2023Is TinyML Sustainable? Assessing the Environmental Impacts of Machine Learning on Microcontrollerscited by 4
  38. 2023Unlocking Bias Detection: Leveraging Transformer-Based Models for Content Analysiscited by 4

50 most cited of 111 All 111 (JSON)