Autors: Stoykova, S. G., Shakev, N. G., Popov, V. L., Ahmed-Shieva, S. A.
Title: Energy Consumption Evaluation for Large Language Models
Keywords: artificial intelligence, energy consumption, environmental impact, large language models

Abstract: This study presents a scalable and reproducible method for estimating per-query at inference energy consumption of popular LLMs using publicly available performance data and hardware specifications, validated against real-time measurements for accuracy across various tasks. Beyond energy use, it highlights the broader environmental footprint of LLMs - including water usage, emissions, and e-waste-and calls for future tools to measure these impacts at the data center level, addressing a critical gap in sustainability research.

References

  1. International Energy Agency, "Energy and AI: World Energy Outlook Special Report", pp. 13-15, pp. 64-65, April 2025, [Online]. Available: https://www.iea.org/reports/energy-and-ai
  2. European Commission, "AI Continent Action Plan", April 2025, pp. 4-8. [Online]. Available: https://digital-strategy.ec.europa.eu/en/library/ai-continent-action-plan
  3. Rotman, D., "AI could keep us dependent on natural gas for decades to come", MIT Technology Review, May 2025, [Online]. Available: https://www.technologyreview.com/2025/05/20/1116272/ai-naturalgas-data-centers-energy-power-plants/
  4. Gholami, A., Yao, Z., Kim, S., Hooper, C., Mahoney, M.W. and Keutzer, K., "AI and memory wall" in IEEE Micro, 44(3), pp.33-39, March 2024, [Online]. Available: https://arxiv.org/pdf/2403.14123
  5. Li, D., Chen, X., Becchi, M. and Zong, Z., "Evaluating the energy efficiency of deep convolutional neural networks on CPUs and GPUs.", IEEE international conferences: SustainCom, BDCloud-SocialCom-SustainCom, pp. 477-484, October 2016.
  6. Anthony, L.F.W., Kanding, B. and Selvan, R., "Carbontracker: Tracking and predicting the carbon footprint of training deep learning models.", July 2020, [Online]. Available: https://arxiv.org/pdf/2007.03051
  7. Thompson, N.C., Greenewald, K., Lee, K. and Manso, G.F., "The computational limits of deep learning", .July 2020, [Online]. Available: https://arxiv.org/pdf/2007.05558
  8. Desislavov, R., Martinez-Plumed, F., Hernandez-Orallo, J., "Trends in AI inference energy consumption: Beyond the performance-vsparameter laws of deep learning", Sustainable Computing: Informatics and Systems, vol 38, April 2023, [Online]. Available: https://doi.org/10.1016/j.suscom.2023.100857.
  9. Latif, I. et al., "Single-Node Power Demand During AI Training: Measurements on an 8-GPU NVIDIA H100 System, " in IEEE Access, vol. 13, pp. 61740-61747, March 2025, [Online]. Available: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10938551
  10. Walsh, N., "How microsoft measures datacenter water and energy use to improve azure cloud sustainability.", April 2022, [Online]. Available: https://azure.microsoft.com/blog/how-microsoft-measuresdatacenter-water-and-energy-use-to-improve-azure-cloudsustainability/
  11. NVIDIA Corporation, "Nvidia DGX A100 Datasheet: The universal system for AI infrastructure.", July 2022, [Online]. Available: https://images.nvidia.com/aem-dam/Solutions/Data-Center/nvidiadgx-a100-datasheet.pdf
  12. Liu, H., Mirando, D, "China ICT Transition BAT Climate Pledges & Action Factsheets" as part of "CWR's Report "China ICT transition-The good, bad & ugly of 5 HKEX ICT listco's net zero pledges & climate action"", June 2023, [Online]. Available: https://cwrrr.org/wpcontent/uploads/2023/09/CWR-2023-Alibaba-Climate-Pledges-Action-Factsheet-FINAL.pdf
  13. Google Cloud, " Datasheet: Architecture and supported configurations of Cloud TPU v4", July 2025, [Online]. Available: https://cloud.google.com/tpu/docs/v4
  14. Google Cloud, "Google Cloud Datacenters 2025 PUE Yearly Report", July 2025, [Online]. Available: https://datacenters.google/efficiency/
  15. NVIDIA Corporation, "Nvidia DGX H800 Datasheet: Highperformance AI configuration: 8x NVIDIA H800 GPUs, 640 GB HBM3 memory, and up to 32 petaFLOPS ", [Online]. Available: https://viperatech.com/shop/nvidia-dgx-h800-systems/
  16. Xie, X., Han, Y. & Tan, H. "Greening China's digital economy: exploring the contribution of the East-West Computing Resources Transmission Project to CO2 reduction.", Humanit Soc Sci Commun 11, 466, March 2024, [Online]. Available: https://doi.org/10.1057/s41599-024-02963-0
  17. Amazon.com Inc., Amazon 2024 Sustainability Report, pp.7, Accessed: July 2025, [Online]. Available: https://sustainability.aboutamazon.com/2024-amazon-sustainabilityreport. pdf
  18. Artificial Analysis Benchmarks, AI Model & API Providers Analysis, Accessed: 15 July 2025, [Online]. Available: https://artificialanalysis.ai/
  19. Chitty-Venkata, K. T. et al, "LLM-Inference-Bench: Inference Benchmarking of Large Language Models on AI Accelerators", In SC24-W: Workshops of the International Conference for High Performance Computing, Networking, Storage and Analysis, pp 1362-1379. IEEE Computer Society, October 2024, [Online]. Available: https://arxiv.org/pdf/2411.00136
  20. P. Patel et al., "Splitwise: Efficient Generative LLM Inference Using Phase Splitting, " ACM/IEEE 51st Annual International Symposium on Computer Architecture (ISCA), Buenos Aires, Argentina, 2024, pp. 118-132, [Online]. Available: https://arxiv.org/abs/2311.18677
  21. Latif, I. et al, "Empirical Measurements of AI Training Power Demand on a GPU-Accelerated Node", December 2024, [Online]. Available: https://arxiv.org/abs/2412.08602

Issue

2025 10th International Conference on Energy Efficiency and Agricultural Engineering, EE and AE 2025 - Conference Proceedings, 2026, Albania, https://doi.org/10.1109/EEAE65901.2025.11273449

Вид: публикация в международен форум, публикация в реферирано издание, индексирана в Scopus