Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

How Green Can AI Be? A Study of Trends in Machine Learning Environmental Impacts

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read AI training's environmental footprint grew exponentially from 2012 to 2024, and simulated green-energy strategies did not flatten the curve.

desk verdict A useful sector-level LCA study with a known headline: the exponential growth story is likely right, but the trend estimate is shakier than the abstract suggests because of an untested non-random sample. read the letter →

arxiv 2412.17376 v1 pith:J3RV7NSL submitted 2024-12-23 cs.LG cs.CY

Clément Morand (STL) , Anne-Laure Ligozat (ENSIIE , LISN , STL) , Aurélie Névéol (STL , LISN) This is my paper · ORCID
classification cs.LGcs.CY
keywords AIenvironmentalimpactmachinelearningtrainingcarbonfootprintlifecycleassessmentreboundeffectshiftinggraphicscardsproductionmetallicresourcedepletionenergyefficiencytrends
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Machine learning's energy use and environmental impact have been growing exponentially, and this paper argues that current green strategies cannot stop that growth. The authors quantify how graphics-card production and model training evolved from 2013 to 2024, measuring carbon footprint and metallic-resource depletion. They show that even a simulated annual 25% drop in electricity carbon intensity leaves the exponential trend in training emissions intact. They conclude that the AI sector exhibits a rebound effect and that reducing AI's environmental impact requires limiting AI activity, not just improving efficiency.

What carries the argument

The analysis is built on two life-cycle metrics: Global Warming Potential (GWP, in $\mathrm{kgCO_2\,eq}$) and Abiotic Depletion Potential (ADP, in $\mathrm{kgSb\,eq}$), computed with a machine-learning life-cycle assessment model. The hardware side uses a curated set of 167 workstation graphics-card models plus other training accelerators, tracking die area, technological node, memory size, and thermal design power to estimate per-card production impacts. The training side uses a database of notable AI systems; for systems without a reported training duration, the authors convert FLOP counts to GPU-hours using a linear regression calibrated on the systems that report both, then multiply by electricity mixes of the producing countries. The interpretive mechanism is the rebound effect: efficiency gains in compute and hardware do not lower total impacts because they enable larger models and more frequent hardware replacement, which shifts impacts into manufacturing. Scenarios that reduce carbon intensity by up to 25% per year test whether location-shifting and grid decarbonization can outpace this growth.

What would settle it

A complete accounting of all notable AI systems released by 2024 that shows training energy or carbon footprint per model plateauing after 2019 would disprove the exponential claim, as would datacenter-level measurements showing AI training electricity grew linearly while model counts grew exponentially.

Watch

Extended reading notes

Core claim

Using a curated dataset of graphics cards released between 2013 and 2023 and a public database of notable AI systems, the authors estimate the production-phase impacts of the hardware and the full training-phase impacts of each system in terms of Global Warming Potential (GWP, in $\mathrm{kgCO_2\,eq}$) and Abiotic Depletion Potential (ADP, in $\mathrm{kgSb\,eq}$). Their central discovery is that both metrics grew exponentially for model training over 2012–2024, and that this growth persists under a simulated 25%-per-year reduction in the carbon intensity of electricity from 2019 onward. They also find that the production impact of individual graphics cards rose continuously, and that hardware production accounts for essentially all ADP and a meaningful share of GWP, so efficiency gains in energy use do not reduce the full environmental burden. The paper interprets the gap between efficiency gains and total impacts as a rebound effect: cheaper and more efficient compute encourages larger models, and more frequent hardware replacement shifts impacts into manufacturing.

Load-bearing premise

The exponential trend is measured on a subset of the notable-AI-system database that has both hardware and compute information, and the paper does not test whether the missing systems would show the same growth.

Editorial extensions

If this is right

  • Cutting the carbon intensity of electricity by up to 25% per year from 2019 does not flatten the exponential rise in training emissions; location-shifting alone cannot be the sector's solution.
  • Metallic-resource depletion from training is almost entirely tied to hardware production, so decarbonizing electricity leaves a large share of environmental impact untouched.
  • Hardware production impacts themselves are rising because cards have larger dies, finer nodes, and more memory, so frequent hardware refresh shifts emissions from use to manufacturing.
  • If the rebound effect visible in training extends to inference, then more efficient chips and models will tend to increase, not decrease, total AI energy demand.
  • Because real electricity mixes bottom out near 15–20 gCO2eq/kWh, even a fast global grid decarbonization has a hard physical ceiling against exponential energy growth.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct sensitivity test is suggested by the paper's own data coverage: reconstructing training impacts for the roughly three-quarters of notable systems without complete hardware records would show whether selection bias inflates the exponential trend.
  • Extending the same accounting to inference workloads and fine-tuning runs, which the paper excludes, would likely show a steeper sector-wide curve because inference demand is growing faster than training in many deployments.
  • Extrapolating the fitted trend would yield a crossover date beyond which even a 25%-per-year decarbonization rate no longer offsets the median model's emissions; the paper leaves that date unstated.
  • Applying a year-by-year datacenter-efficiency correction would probably steepen the early trend, since the paper assumes near-optimal infrastructure efficiency for all years.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper studies the environmental impacts of AI hardware and model training over 2013–2023. It curates a dataset of NVIDIA workstation and other graphics cards, estimates production impacts using the MLCA life-cycle assessment tool, and combines this with Epoch AI's notable-systems database to estimate the energy, carbon footprint (GWP), and abiotic depletion potential (ADP) of training individual ML models. The central claim is that training energy and environmental impacts have grown exponentially over time, that hardware production impacts also increase, and that reduction strategies such as shifting compute to less carbon-intensive electricity or improving PUE cannot curb this growth—interpreted as evidence of a rebound effect. The authors share code and data for reproducibility.

Significance. If the exponential-trend result is robust, the paper is a valuable contribution to the Green AI literature: it extends prior compute-trend studies (e.g., Sevilla et al., Thompson et al.) with an environmental multi-criteria assessment that includes embodied hardware impacts, and it explicitly interrogates impact-shifting and rebound effects—issues often ignored in single-phase carbon accounting. The use of MLCA, a previously published tool, applied to new data is appropriate, and the public availability of the curated data and code is a concrete strength. The main significance hinges on whether the fitted exponential trend is an artifact of sample selection and time-correlated modeling assumptions, which the current manuscript does not yet establish convincingly.

major comments (3)
  1. [Section 5.1 and Table 1] The exponential-growth conclusion for training energy and impacts (Figures 6 and 7) is based on only 205 of the 825 systems in the Epoch AI database (112 with GPU-h1, 93 with GPU-h2). This is not a random sample: inclusion requires documented hardware plus either training duration or FLOP, which is systematically more common for large, recent, well-known systems. Section 7 acknowledges that the database is incomplete and that 'excluded models could have an impact,' but the paper provides no test of selection bias. To make the sector-level claim load-bearing, the authors should compare observable characteristics of included versus excluded systems (e.g., release date, reported FLOP) and provide a sensitivity analysis in which missing GPU-hours are imputed under alternative assumptions (e.g., using the fitted GPU-h1~GPU-h2 relation, or assuming excluded systems follow a slower growth path). Without such an analysis, the fitted WLS slope may overstate the sector-wide trend.
  2. [Section 3.2 and Section 7] The modeling assumption of a constant PUE of 1.1 for all years is time-correlated with the trend. The paper itself cites Masanet et al. (2020) and states in Section 7 that 'we supposed that all models were trained in hyper-scale data-centers, leading to using a quasi optimal PUE for all models, masking the increase in PUE over the last decade.' Because older models trained in less efficient datacenters are assigned too low an infrastructure multiplier, their energy is underestimated more than recent models, which can artificially steepen the fitted exponential slope. The authors should report a sensitivity analysis with time-varying PUE (for example, linearly decreasing from 1.8 in 2012 to 1.1 in 2020) and show that the exponential trend and the conclusion that location shifting cannot curb growth remain unchanged.
  3. [Section 3.2, GPU-h2 calibration] The calibration model log(GPU-h1) = 1.31 + 1.00 * log(GPU-h2) is built after excluding 19 anomalies from 106 observations and is then applied to 93 models without direct duration data. The text reports R²=0.98, but the exclusion criteria are not described in enough detail to assess whether the retained 87 observations are representative, and the paper does not report how sensitive the fitted training-energy trend is to the exclusion list or to the constant performance ratio (exp(1.31) ≈ 3.7). The authors should provide the scatterplot (currently only described in words), the list of excluded anomalies, and a sensitivity check in which GPU-h2 is used without correction or with the upper/lower bounds of the regression confidence interval. This would show whether the exponential slope in Figures 6–7 is robust to the calibration choice.
minor comments (4)
  1. [Abstract and Introduction] There are typos: 'reroducibility' should be 'reproducibility', and 'Cummulative' in Section 4.1 should be 'Cumulative'.
  2. [Section 3.2] The sentence 'we choose a close to optimal PUE of 1.1 and a hardware utilization of 50' appears to be missing a unit or percent sign; it should read '50%.'
  3. [Section 3.2] The description of the GPU-h1/GPU-h2 comparison would benefit from separating the model diagnostics from the interpretation: the text moves from the regression equation directly to the statement 'This model correspond to using a constant performance ratio of ≃ 27%,' which is not immediately obvious from the reported coefficients (exp(1.31) ≈ 3.7, i.e., GPU-h1 is about 3.7 times GPU-h2).
  4. [Section 5.2] The scenario analysis multiplies the carbon intensity by (1 - ratio)^n starting from 2019 for all models, including those released before 2019. This means models released before 2019 have their carbon intensity increased (since n is negative), which may visually affect the left side of Figure 8 and the comparison of real versus simulated impacts. The authors should clarify this choice or restrict the counterfactual to models released after 2019.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: the exponential-trend conclusion is not assumed in the inputs; the only self-citation (MLCA tool) is a minor, non-load-bearing dependency.

full rationale

The derivation chain is: (i) curate NVIDIA GPU specifications from TechPowerUp/Wikipedia; (ii) compute per-card production impacts with the authors' MLCA tool; (iii) take training hardware, duration, and FLOP from the external Epoch AI notable-systems database; (iv) estimate GPU-hours either directly (GPU-h1) or from FLOP/peak throughput (GPU-h2), with a log-log regression of GPU-h1 on GPU-h2 used as a validation of the ~27% performance ratio; (v) compute training energy as GPU-hours times TDP and multiply by country electricity mixes; (vi) fit WLS trends and simulate carbon-intensity reductions. None of these steps defines the exponential-growth conclusion into its inputs. The GPU-h1 ~ GPU-h2 regression is a calibration, not a fitted parameter renamed as a prediction: the final estimates use GPU-h2 directly for the 93 models without substituting fitted values, so the trend slope is not forced by the calibration. MLCA is a self-citation (Morand, Névéol, and Ligozat 2024), but it is a published tool whose stated limitations (fixed memory density, no end-of-life, no technological-node dependence) do not include the target result; applying it to new Epoch AI data is a genuine application, not a circular derivation. Section 7 explicitly acknowledges database incompleteness and modeling simplifications; these are uncertainty and selection concerns, not circular reasoning. The overall finding is therefore no significant circularity, with only a minor non-load-bearing self-citation of the MLCA tool.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The analysis relies on fitted regression coefficients and several hand-chosen constants (PUE, lifespan, utilization, server configuration) that directly affect the impact estimates. No new physical entities are introduced.

free parameters (7)
  • log-log regression intercept for GPU-h1 vs GPU-h2 = 1.31
    Fitted to 87 observations; used to calibrate GPU-hour estimates for models lacking direct training duration data.
  • log-log regression slope = 1.00
    Fitted to the same 87 observations; effectively a constant multiplicative factor of about 3.7 (exp(1.31)).
  • Hardware lifespan = 3 years
    Chosen from Ostrouchov et al. 2020; affects the amortized production impact per training hour.
  • Power Usage Effectiveness (PUE) = 1.1
    Chosen as a close-to-optimal hyper-scale value; understates infrastructure energy for older datacenters.
  • Hardware utilization over life-cycle = 50%
    Assumed to allocate production impacts across training hours.
  • GPU power draw during training = 100% of TDP
    Assumed for energy consumption, acknowledged as an overestimate by up to 2x.
  • GPUs per server = 4 for workstation, 2 for non-workstation
    Assumed to compute server-level production impacts; memory is excluded.
assumptions (4)
  • domain assumption MLCA production impact factors for integrated circuits scale with die area and memory surface, independent of technological node.
    Used to compute GPU production impacts; the paper notes MLCA does not account for node and assumes fixed memory density, which limits accuracy.
  • domain assumption The Epoch AI dataset and its subset with complete information represent the trends of the AI sector.
    The paper uses 205 of 825 systems; representativeness is untested and the database is acknowledged to be incomplete.
  • ad hoc to paper Carbon intensity reduction can be modeled as a fixed annual percentage decrease from 2019, independent of geography.
    The scenario is an idealized representation of location shifting and grid decarbonation; the paper chooses up to 25% per year.
  • standard math Standard statistical assumptions of linear regression are met.
    The paper reports Shapiro-Wilk, Breusch-Pagan, and Durbin-Watson tests for the GPU-hours regression.

how reviews work

0 comments
Cite this review

Pith. "Pith review of How Green Can AI Be? A Study of Trends in Machine Learning Environmental Impacts." pith.science (2026). https://pith.science/paper/J3RV7NSL

@misc{pith2026241217376,
  author       = {Pith},
  title        = {Pith review of: How Green Can AI Be? A Study of Trends in Machine Learning Environmental Impacts},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J3RV7NSL}},
  note         = {Machine review of arXiv:2412.17376}
}
read the original abstract

The compute requirements associated with training Artificial Intelligence (AI) models have increased exponentially over time. Optimisation strategies aim to reduce the energy consumption and environmental impacts associated with AI, possibly shifting impacts from the use phase to the manufacturing phase in the life-cycle of hardware. This paper investigates the evolution of individual graphics cards production impacts and of the environmental impacts associated with training Machine Learning (ML) models over time. We collect information on graphics cards used to train ML models and released between 2013 and 2023. We assess the environmental impacts associated with the production of each card to visualize the trends on the same period. Then, using information on notable AI systems from the Epoch AI dataset we assess the environmental impacts associated with training each system. The environmental impacts of graphics cards production have increased continuously. The energy consumption and environmental impacts associated with training models have increased exponentially, even when considering reduction strategies such as location shifting to places with less carbon intensive electricity mixes. These results suggest that current impact reduction strategies cannot curb the growth in the environmental impacts of AI. This is consistent with rebound effect, where the efficiency increases fuel the creation of even larger models thereby cancelling the potential impact reduction. Furthermore, these results highlight the importance of considering the impacts of hardware over the entire life-cycle rather than the sole usage phase in order to avoid impact shifting. The environmental impact of AI cannot be reduced without reducing AI activities as well as increasing efficiency.

Figures

Figures reproduced from arXiv: 2412.17376 by the authors.

Figure 1
Figure 1. Evolution of the characteristics of NVIDIA work [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Evolution of the energy consumption of NVIDIA [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Evolution of the production impacts (left in GWP, right in ADP) of NVIDIA workstation graphics cards. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Evolution of the production impacts (left in GWP, right in ADP) of the graphics cards used for training ML systems [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Evolution of the number of graphics cards used [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 7
Figure 7. Figure 7: Environmental impacts in GWP (a) and ADP (b) of training ML models over time. Value intervals account for ambigu [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Estimated carbon footprint of training ML models [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Exploring the sustainable scaling of AI dilemma: A projective study of corporations' AI environmental impacts

    cs.AI 2025-01 conditional novelty 5.0 of 10

    A corporate AI portfolio LCA model projects that high generative-AI adoption could increase AI electricity use about 24-fold by 2030.

Reference graph

Works this paper leans on

45 extracted references · 42 canonical work pages · cited by 1 Pith paper

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Ambrose, J. 2024. Ireland’s datacentres overtake electricity use of all urban homes combined. The Guardian, https://www.theguardian.com/world/article/2024/jul/23/ireland-datacentres-overtake-electricity-use-of-all-homes-combined-figures-show

    work page 2024

  4. [4]

    Anthony, L. F. W.; Kanding, B.; and Selvan, R. 2020. Carbontracker: Tracking and Predicting the Carbon Footprint of Training Deep Learning Models. arXiv:2007.03051

    arXiv 2020

  5. [5]

    Bannour, N.; Ghannay, S.; N \'e v \'e ol, A.; and Ligozat, A.-L. 2021. Evaluating the carbon footprint of NLP methods: a survey and analysis of existing tools. In Moosavi, N. S.; Gurevych, I.; Fan, A.; Wolf, T.; Hou, Y.; Marasovi \'c , A.; and Ravi, S., eds., Proceedings of the Second Workshop on Simple and Efficient Natural Language Processing, 11--21. V...

    work page 2021

  6. [6]

    M.; Gebru, T.; McMillan-Major, A.; and Shmitchell, S

    Bender, E. M.; Gebru, T.; McMillan-Major, A.; and Shmitchell, S. 2021. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, FAccT '21, 610–623. New York, NY, USA: Association for Computing Machinery. ISBN 9781450383097

    work page 2021

  7. [7]

    Bol, D.; Pirson, T.; and Dekimpe, R. 2021. Moore's Law and ICT Innovation in the Anthropocene. In 2021 Design, Automation & Test in Europe Conference & Exhibition (DATE), 19--24

    work page 2021

  8. [8]

    Bouza, L.; Bugeau, A.; and Lannelongue, L. 2023. How to estimate carbon footprint when training deep learning models? A guide and review. Environmental Research Communications, 5(11): 115014

    work page 2023

Show all 45 references
  1. [9]

    Chu, A. 2024. US slows plans to retire coal-fired plants as power demand from AI surges. Financial Times, https://www.ft.com/content/ddaac44b-e245-4c8a-bf68-c773cc8f4e63

    2024

  2. [10]

    P.-V.; Jacquemotte, Q

    Cl \'e ment, L.-P. P.-V.; Jacquemotte, Q. E.; and Hilty, L. M. 2020. Sources of variation in life cycle assessments of smartphones and tablet computers. Environmental Impact Assessment Review, 84: 106416

    2020

  3. [11]

    Comber, R.; and Eriksson, E. 2023. Computing as Ecocide . In Ninth Computing within Limits 2023 . LIMITS. Https://limits.pubpub.org/pub/a8h46wqy

    2023

  4. [12]

    C.; and Mattern, F

    Coroama, V. C.; and Mattern, F. 2019. Digital Rebound - Why Digitalization Will not Redeem us our Environmental Sins. In Wolff, A., ed., Proceedings of the 6th International Conference on ICT for Sustainability, ICT4S 2019, Lappeenranta, Finland, June 10-14, 2019 , volume 2382...

    2019

  5. [13]

    Dauvergne, P. 2021. The globalization of artificial intelligence: consequences for the politics of environmentalism. Globalizations, 18(2): 285--299

    2021

  6. [14]

    de Vries , A. 2023. The growing energy footprint of artificial intelligence. Joule, 7(10): 2191--2194

    2023

  7. [15]

    Desislavov, R.; Martínez-Plumed, F.; and Hernández-Orallo, J. 2023. Trends in AI inference energy consumption: Beyond the performance-vs-parameter laws of deep learning. Sustainable Computing: Informatics and Systems, 38: 100857

    2023

  8. [16]

    Epoch AI . 2022. Parameter, Compute and Data Trends in Machine Learning. Accessed: 2024-07-19

    2022

  9. [17]

    Forster, P.; Storelvmo, T.; Armour, K.; Collins, W.; Dufresne, J.-L.; Frame, D.; Lunt, D.; Mauritsen, T.; Palmer, M.; Watanabe, M.; Wild, M.; and Zhang, H. 2023. The Earth’s Energy Budget, Climate Feedbacks and Climate Sensitivity. In Masson-Delmotte, V.; Zhai, P.; Pirani, A.;...

    2023

  10. [18]

    Gossart, C. 2015. Rebound Effects and ICT: A Review of the Literature. In Hilty, L. M.; and Aebischer, B., eds., ICT Innovations for Sustainability, 435--448. Cham: Springer International Publishing. ISBN 978-3-319-09228-7

    2015

  11. [19]

    Greenfield, P. 2023. Revealed: more than 90\ certifier are worthless, analysis shows. The Guardian, https://www.theguardian.com/environment/2023/jan/18/revealed-forest-carbon-offsets-biggest-provider-worthless-verra-aoe

    2023

  12. [20]

    G.; Lee, S.; Tse, J.; Lee, H.-H

    Gupta, U.; Kim, Y. G.; Lee, S.; Tse, J.; Lee, H.-H. S.; Wei, G.-Y.; Brooks, D.; and Wu, C.-J. 2022. Chasing Carbon: The Elusive Environmental Footprint of Computing

    2022

  13. [21]

    Jay, M.; Ostapenco, V.; Lefevre, L.; Trystram, D.; Orgerie, A.-C.; and Fichel, B. 2023. An experimental comparison of software-based power meters: focus on CPU and GPU. In 2023 IEEE/ACM 23rd International Symposium on Cluster, Cloud and Internet Computing (CCGrid), 106--118

    2023

  14. [22]

    H.; Brown, L.; Cheng, J.; Khan, M.; Gupta, A.; Workman, D.; Hanna, A.; Flowers, J.; and Gebru, T

    Jiang, H. H.; Brown, L.; Cheng, J.; Khan, M.; Gupta, A.; Workman, D.; Hanna, A.; Flowers, J.; and Gebru, T. 2023. AI Art and Its Impact on Artists. In Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society, AIES '23, 363–374. New York, NY, USA: Association for ...

    2023

  15. [23]

    Keyes, O. 2018. The Misgendering Machines: Trans/HCI Implications of Automatic Gender Recognition. Proc. ACM Hum. Comput. Interact. , 2( CSCW ): 88:1--88:22

    2018

  16. [24]

    Lannelongue, L.; Grealey, J.; and Inouye, M. 2021. Green Algorithms: Quantifying the Carbon Footprint of Computation. Advanced Science, 8(12): 2100707

    2021

  17. [25]

    Lohmann, L. 2009. Toward a different debate in environmental accounting: The cases of carbon and cost–benefit. Accounting, Organizations and Society, 34(3): 499--534

    2009

  18. [26]

    S.; Viguier, S.; and Ligozat, A.-L

    Luccioni, A. S.; Viguier, S.; and Ligozat, A.-L. 2023. Estimating the Carbon Footprint of BLOOM , a 176B Parameter Language Model. Journal of Machine Learning Research, 24(253): 1--15

    2023

  19. [27]

    Masanet, E.; Shehabi, A.; Lei, N.; Smith, S.; and Koomey, J. 2020. Recalibrating global data center energy-use estimates. Science, 367(6481): 984--986

    2020

  20. [28]

    Morand, C.; N \'e v \'e ol, A.; and Ligozat, A.-L. 2024. MLCA: a tool for Machine Learning Life Cycle Assessment . In 2024 International Conference on ICT for Sustainability (ICT4S) . Stockholm, Sweden

    2024

  21. [29]

    Mytton, D. 2021. Data center water consumption. npj Clean Water, 4

    2021

  22. [30]

    Ortar, N.; Taylor, A. R. E.; Velkova, J.; Brodie, P.; Johnson, A.; Marquet, C.; Pollio, A.; and Cirolia, L. 2023. 4 Powering ‘smart’ futures: data centres and the energy politics of digitalisation, 125--168. Berlin, Boston: De Gruyter. ISBN 9783110745641

    2023

  23. [31]

    A.; Engelmann, C.; Shankar, M.; and Rogers, J

    Ostrouchov, G.; Maxwell, D.; Ashraf, R. A.; Engelmann, C.; Shankar, M.; and Rogers, J. H. 2020. GPU Lifetimes on Titan Supercomputer: Survival Analysis and Reliability. In SC20: International Conference for High Performance Computing, Networking, Storage and Analysis, 1--14

    2020

  24. [32]

    R.; Texier, M.; and Dean, J

    Patterson, D.; Gonzalez, J.; Hölzle, U.; Le, Q.; Liang, C.; Munguia, L.-M.; Rothchild, D.; So, D. R.; Texier, M.; and Dean, J. 2022. The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink. Computer, 55(7): 18--28

    2022

  25. [33]

    P.; Pip, A

    Pirson, T.; Delhaye, T. P.; Pip, A. G.; Le Brun, G.; Raskin, J.-P.; and Bol, D. 2023. The Environmental Footprint of IC Production: Review, Analysis, and Lessons From Historical Trends. IEEE Transactions on Semiconductor Manufacturing, 36(1): 56--67

    2023

  26. [34]

    Roussilhe, G.; Pirson, T.; Xhonneux, M.; and Bol, D. 2024. From silicon shield to carbon lock-in? The environmental footprint of electronic components manufacturing in Taiwan (2015–2020). Journal of Industrial Ecology, n/a(n/a): 15

    2024

  27. [35]

    A.; and Etzioni, O

    Schwartz, R.; Dodge, J.; Smith, N. A.; and Etzioni, O. 2020. Green AI . Commun. ACM, 63(12): 54–63

    2020

  28. [36]

    Sevilla, J.; Heim, L.; Ho, A.; Besiroglu, T.; Hobbhahn, M.; and Villalobos, P. 2022. Compute Trends Across Three Eras of Machine Learning. In 2022 International Joint Conference on Neural Networks (IJCNN), 1--8

    2022

  29. [37]

    Shi, M.; Currier, K.; Liu, Z.; Janowicz, K.; Wiedemann, N.; Verstegen, J.; McKenzie, G.; Graser, A.; Zhu, R.; and Mai, G. 2023. Thinking Geographically about AI Sustainability. AGILE: GIScience Series, 4: 42

    2023

  30. [38]

    Strubell, E.; Ganesh, A.; and McCallum, A. 2019. Energy and Policy Considerations for Deep Learning in NLP . In Korhonen, A.; Traum, D. R.; and M \` a rquez, L., eds., Proceedings of the 57th Conference of the Association for Computational Linguistics, ACL 2019, Florence, Ital...

    2019

  31. [39]

    Thompson, N.; Greenewald, K.; Lee, K.; and Manso, G. F. 2023. The Computational Limits of Deep Learning . In Ninth Computing within Limits 2023 . LIMITS. Https://limits.pubpub.org/pub/wm1lwjce

    2023

  32. [40]

    B.; and Heijungs, R

    van Oers, L.; Guin \'e e, J. B.; and Heijungs, R. 2020. Abiotic resource depletion potentials (ADPs) for elements revisited—updating ultimate reserve estimates and introducing time series for production data. The International Journal of Life Cycle Assessment, 25: 294--308

    2020

  33. [41]

    Velkova, J. 2019. Data Centres as Impermanent Infrastructures. Culture Machine, 18

    2019

  34. [42]

    Verdecchia, R.; Sallou, J.; and Cruz, L. 2023. A systematic review of Green AI. WIREs Data Mining and Knowledge Discovery, 13(4): e1507

    2023

  35. [43]

    World Health Organization . 2021. Children and digital dumpsites: e-waste exposure and child health. World Health Organization

    2021

  36. [44]

    A.; Huang, J.; Bai, C.; Gschwind, M.; Gupta, A.; Ott, M.; Melnikov, A.; Candido, S.; Brooks, D.; Chauhan, G.; Lee, B.; Lee, H

    Wu, C.; Raghavendra, R.; Gupta, U.; Acun, B.; Ardalani, N.; Maeng, K.; Chang, G.; Behram, F. A.; Huang, J.; Bai, C.; Gschwind, M.; Gupta, A.; Ott, M.; Melnikov, A.; Candido, S.; Brooks, D.; Chauhan, G.; Lee, B.; Lee, H. S.; Akyildiz, B.; Balandat, M.; Spisak, J.; Jain, R.; Rab...

    2022

  37. [45]

    Wu, C.-J.; Acun, B.; Raghavendra, R.; and Hazelwood, K. 2024. Beyond Efficiency: Scaling AI Sustainably. arXiv:2406.05303

    2024 arXiv

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.