The Cloud Next Door: Investigating the Environmental and Socioeconomic Strain of Datacenters on Local Communities

Elsa Olivetti; Laurent Liote; Michelle Westerlaken; Noman Bashir; Wacuka Ngata; Yasra Chandio

REVIEW 4 major objections 5 minor 42 references

The Cloud Next Door: Investigating the Environmental and Socioeconomic Strain of Datacenters on Local Communities

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Datacenters impose measurable local burdens on adjacent communities—higher noise, degraded power quality, higher bills—and the paper argues these are best documented by mixing quantitative measurements with qualitative stakeholder input…

desk verdict A transparent, well-framed proposal for studying local datacenter impacts, but the quantitative evidence is too thin to support the 'corroborate' claim. read the letter →

arxiv 2506.03367 v1 pith:EVE7QUDY submitted 2025-06-03 cs.DC cs.CYcs.ET

Wacuka Ngata , Noman Bashir , Michelle Westerlaken , Laurent Liote , Yasra Chandio , Elsa Olivetti This is my paper · ORCID

classification cs.DCcs.CYcs.ET

keywords Sustainabilitydatacenterslocalcommunitiesnoisepollutionenvironmentalimpactairblackoutspowerquality

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

This paper argues that the environmental and social costs of datacenters should be measured at the neighborhood level, not only through global energy and carbon metrics, and proposes a mixed-methods approach that treats resident experience as evidence on par with sensor readings. Focusing on Northern Virginia's "Data Center Valley," it identifies four impact dimensions—environmental, social, economic, and infrastructural—and reports early quantitative signs: higher background noise adjacent to a facility and degraded power quality in nearby homes. The core move is to pair these measurements with stakeholder interviews and document analysis to surface who benefits and who bears the burden. The goal is to inform siting and approval decisions so community impacts are considered before, not after, development.

What carries the argument

The load-bearing framework is a mixed-methods matrix: Table 1 maps five impact categories—resource strain, power grid strain, economic effects, noise pollution, and air pollution—to concrete metrics (water use effectiveness, land area, outage frequency, monthly bills, decibels, pollutant concentrations) and data sources (utility records, outage maps, billing data, field measurements, and environmental monitoring). The qualitative arm, built from interviews, document analysis, and community engagement, is treated as equally authoritative. The machinery's output is a triangulated picture that converts diffuse complaints into measurable indicators and then reads those indicators back through stakeholder narratives.

What would settle it

A controlled transect of calibrated noise measurements at multiple datacenter sites across different times of day and seasons, plus a regression of harmonic distortion on distance to datacenters while controlling for local grid age and other demand sources, would settle whether the reported noise and power-quality differences are genuine datacenter effects.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that datacenters place measurable local socio-environmental and economic burdens on adjacent communities, and that these burdens are best surfaced by combining quantitative measurements with qualitative stakeholder analysis on equal footing. As preliminary evidence from Northern Virginia, the paper reports a noise comparison in which a residential area 200 feet from a datacenter measured 28.0 dB(A) versus 22.3 dB(A) in a neighborhood two miles away, and a power-quality analysis finding that more than 6.8% of homes in Loudoun County experienced at least one monthly reading exceeding 8% total harmonic distortion—a level that can damage appliances—with sensors in Prince William County reaching 13%. The paper describes these data points as corroborating the infrastructural strain that has concerned local communities, and its stated contribution is a replicable framework for studying datacenter-community interactions in other regions.

Load-bearing premise

The load-bearing premise is that the early measurements—two afternoon noise readings taken with a smartphone app and a power-quality analysis the authors did not produce—are representative enough to show a real datacenter effect on nearby neighborhoods.

Editorial extensions

If this is right

Datacenter siting and approval decisions should weigh neighborhood-level noise, air, water, and grid-quality data alongside global energy and carbon metrics.
Residents near datacenters can expect higher background noise and a greater chance of appliance-damaging power distortion; the paper's early numbers provide first bounds.
A publicly accessible database combining utility records, field measurements, and interview themes could support both policy evaluation and industry accountability.
The stakeholder attitude map of concerned, incentivized, and indifferent groups offers a way to understand why communities with similar exposure respond differently.

Reading between the lines

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

Beyond the paper: a systematic measurement campaign with calibrated instruments across many sites, times, and seasons would tell whether the reported noise difference is a general datacenter signature or an artifact of one afternoon.
Beyond the paper: if harmonic distortion near datacenters is as common as the cited analysis suggests, appliance damage and early replacement become a hidden economic transfer from homeowners to cloud operators.
Beyond the paper: the equal-weight mixed-methods template could be adapted to other contested infrastructure—transmission lines, solar farms, natural-gas plants—where local costs are diffuse but real.
Beyond the paper: adding demographic and income data to the stakeholder mapping would let future work test whether datacenter burdens fall disproportionately on lower-income or renter households.

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.

Desk Editor's Note

read the letter

What you should know: this is a late-breaking work that proposes a mixed-methods framework for surfacing local socio-environmental impacts of datacenters, illustrated with a preliminary Northern Virginia case study. The paper is upfront that it is ongoing work.

What is new: it explicitly puts quantitative and qualitative methods on equal footing, and organizes local impacts into four dimensions—environmental, social, economic, infrastructural—with a useful mapping of metrics and data sources. That framing is a legitimate contribution, and the related work is well chosen. The early qualitative findings give a sense of stakeholder attitudes, though the interview pool is self-selected.

Where it is soft: the load-bearing claim in Section 3.3 that "these quantitative data points appear to corroborate" rests almost entirely on Table 2: two smartphone readings, one at 2 miles and one at 200 feet, on one afternoon. No calibration, no replicates, no weather or traffic data. And the values themselves (22.3 and 28.0 dB) are suspiciously low—likely near the noise floor of a phone microphone. So those numbers do not demonstrate a datacenter effect; they are suggestive at best. The THD data from Bloomberg is second-hand, and the link to datacenters is not independently checked. The paper's own limitations note says more measurements are planned, which is honest, but the "corroborate" sentence overstates what is there.

Also minor: the abstract language sometimes reads as if more has been done than is actually reported (e.g., "we highlight," "we examine" versus "we plan" in the body). For a late-breaking paper that is tolerable, but worth tightening.

Overall, the framework has value, the case study is a start, and the authors are transparent about the preliminary nature. My main worry is that the quantitative pillar, as currently reported, is not evidence strong enough to support the claim the section makes. A serious referee would need to push on that.

Recommendation: I would send this to referees—desk reject would be too harsh for a late-breaking venue—but with the clear expectation that the "corroborate" claim needs to be either removed or replaced with a clearly labeled preliminary observation. Worth engaging with, but as a research proposal and taxonomy paper, not as a completed empirical study.

Referee Report

4 major / 5 minor

Summary. This late-breaking work argues that datacenters impose measurable local socio-environmental, infrastructural, and economic burdens on adjacent communities, and that these are best surfaced by a mixed-methods approach treating quantitative measurement and qualitative stakeholder analysis on equal footing. After introducing four impact dimensions (environmental, social, economic, infrastructural), the paper presents a preliminary case study of Northern Virginia's 'Data Center Valley': two noise measurements at different distances from a datacenter, a summary of Bloomberg-reported power-quality (THD) data, an ongoing outage analysis, and early qualitative engagement yielding three stakeholder attitudes (concerned, incentivized, indifferent). Section 3.3 concludes that the quantitative data points 'appear to corroborate the infrastructural strain that has concerned local communities.'

Significance. The topic is timely and important: local impacts of datacenter expansion are understudied relative to global energy and carbon metrics, and the proposed taxonomy plus mixed-methods framework is a useful organizing device for future work. The paper's strengths are the clear IRB clearance, the explicit commitment to treating qualitative and quantitative evidence as complementary, and the compilation of relevant public data sources in Table 1. However, the significance of the concrete empirical claim depends on the validity of the preliminary quantitative evidence, which is currently too thin to support the 'corroborate' language.

major comments (4)

[Section 3.3, Table 2] The noise comparison is based on two measurements taken on a single afternoon with an uncalibrated smartphone app. The reported values of 22.3 and 28.0 dB(A) are at or below typical smartphone microphone noise floors in quiet outdoor settings, so the 5.7 dB difference cannot be attributed to datacenter noise without calibration and without ruling out ambient and device effects. The paper should report calibration data, measurement duration, number of samples, weather and traffic conditions, and replicate measurements across times and sites; alternatively, this evidence should be explicitly downgraded from corroboration to a pilot observation.
[Section 3.3, power-quality paragraph] The THD findings are quoted from a Bloomberg investigation ([25]) and are not independently verified, re-derived, or controlled for confounders such as distance from datacenters, pre-existing grid conditions, or other industrial loads. The claim that distortions are 'strongly linked to proximity to datacenters' is asserted rather than demonstrated in this paper. The authors should either analyze the underlying sensor data with appropriate controls or clearly attribute the statement to Bloomberg and remove it from the paper's own quantitative corroboration.
[Section 3.3, Qualitative Outcomes] The three attitudes (concerned, incentivized, indifferent) are presented as 'dominant' even though the paper acknowledges that the early participant pool is majority environmental-organization and local-community members. No sample sizes, recruitment details, interview protocol, or thematic-saturation evidence are provided. These should be described as emergent themes from an ongoing, non-representative sample rather than as established stakeholder typology.
[Section 3.3, concluding sentence] The sentence 'Thus far, these quantitative data points appear to corroborate the infrastructural strain that has concerned local communities' is load-bearing for the paper's central claim, but it does not follow from the preceding evidence: the noise data are unvalidated and the THD data are second-hand. The sentence should be revised to state clearly that the quantitative results are preliminary and hypothesis-generating, or removed until the measurement program is sufficiently developed.

minor comments (5)

[Table 1] The table lists data sources that are planned or in progress; the paper should clearly distinguish sources already analyzed from those that remain to be collected, for example by adding a status column.
[References] References [20] and [21] are identical; one should be removed or replaced with the intended distinct citation.
[Section 3.2 vs. Section 3.3] Section 3.2 says qualitative research is planned ('we plan to conduct'), while Section 3.3 reports early engagements; the tense should be made consistent to reflect the actual study stage.
[Section 3.3, noise measurements] No map or site coordinates are provided for the two measurement locations; adding a small map with distances, measurement times, and nearby infrastructure would improve reproducibility.
[Section 1, footnote] The footnote thanking reviewers and stating the paper 'will appear at ACM COMPASS 2025' is unusual in a preprint and should be removed or reformatted for the final version.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the quantitative observations are independent of the stakeholder-derived framework, and no fitted parameter or author-imported ansatz is presented as a prediction.

full rationale

The paper does not contain a formal derivation chain; its central claim is that local datacenter impacts can be surfaced through a mixed-methods approach. The impact dimensions in Section 3.1 are described as an early outcome of stakeholder engagement, but the quantitative evidence — the two-point noise comparison in Table 2 and the Bloomberg harmonic distortion data in Section 3.3 — is external empirical observation, not an implication of the stakeholder-derived taxonomy. No parameter is fitted and then relabeled as a prediction; no uniqueness theorem is invoked; and no ansatz is smuggled in through self-citation. Self-citations such as Bashir et al. appear only in related-work or general-context citations about carbon-aware computing and AI sustainability, and they are not load-bearing for the Northern Virginia case study. The fragility of the two uncalibrated smartphone noise readings is a genuine empirical-validity concern, but it is not circularity: those measurements could in principle have disagreed with the qualitative concerns, and the authors explicitly label the work preliminary and plan to expand it. Because the evidence and the organizing framework are not equivalent by construction, the circularity score is 0.

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

The paper's claims rest on measurement assumptions, a cited external analysis, and an early interview sample. No fitted numerical parameters or invented physical entities underlie the argument.

assumptions (4)

domain assumption The NIOSH smartphone app noise measurements at two locations are reliable indicators of datacenter-generated noise.
Section 3.3 and Table 2 use the app readings without calibration, error bars, or repeated trials, treating them as evidence of persistent operational sound.
domain assumption The Bloomberg THD analysis correctly attributes power-quality issues to datacenter proximity.
Section 3.3 uses the Bloomberg findings as evidence of grid strain without independent verification or access to the underlying sensor data.
ad hoc to paper The early participant pool, with a majority from environmental organizations and local communities, is sufficiently representative to identify the three attitudes of concerned, incentivized, and indifferent.
Section 3.3 reports these attitudes as dominant while noting the pool is initial, creating a selection-bias risk that could skew the qualitative conclusions.
domain assumption The four impact dimensions (environmental, social, economic, infrastructural) are sufficiently complete to frame the study.
The dimensions emerged from early stakeholder engagement in Section 3.1 and are used as the organizational structure; if additional major impact categories exist outside these four, the framework will miss them.

how reviews work

0 comments

Cite this review

Pith. "Pith review of The Cloud Next Door: Investigating the Environmental and Socioeconomic Strain of Datacenters on Local Communities." pith.science (2026). https://pith.science/paper/EVE7QUDY

@misc{pith2026250603367,
  author       = {Pith},
  title        = {Pith review of: The Cloud Next Door: Investigating the Environmental and Socioeconomic Strain of Datacenters on Local Communities},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EVE7QUDY}},
  note         = {Machine review of arXiv:2506.03367}
}

read the original abstract

Datacenters have become the backbone of modern digital infrastructure, powering the rapid rise of artificial intelligence and promising economic growth and technological progress. However, this expansion has brought growing tensions in the local communities where datacenters are already situated or being proposed. While the mainstream discourse often focuses on energy usage and carbon footprint of the computing sector at a global scale, the local socio-environmental consequences -- such as health impacts, water usage, noise pollution, infrastructural strain, and economic burden -- remain largely underexplored and poorly addressed. In this work, we surface these community-level consequences through a mixed-methods study that combines quantitative data with qualitative insights. Focusing on Northern Virginia's ``Data Center Valley,'' we highlight how datacenter growth reshapes local environments and everyday life, and examine the power dynamics that determine who benefits and who bears the costs. Our goal is to bring visibility to these impacts and prompt more equitable and informed decisions about the future of digital infrastructure.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 32 canonical work pages

[25]

Leonardo Nicoletti, Naureen Malik, and Andre Tartar. 2024. AI Needs So Much Power, It’s Making Yours Worse. https://www.bloomberg.com/graphics/2024- ai-power-home-appliances/

work page 2024
[2]

Noman Bashir, Priya Donti, James Cuff, Sydney Sroka, Marija Ilic, Vivienne Sze, Christina Delimitrou, and Elsa Olivetti. 2024. The Climate and Sustainability Implications of Generative AI. An MIT Exploration of Generative AI (Mar 27 2024). https://mit-genai.pubpub.org/pub/8ulgrckc

work page 2024
[3]

Noman Bashir, Tian Guo, Mohammad Hajiesmaili, David Irwin, Prashant Shenoy, Ramesh Sitaraman, Abel Souza, and Adam Wierman. 2021. Enabling Sustainable Clouds: The Case for Virtualizing the Energy System. In Proc. of the ACM Sym- posium on Cloud Computing (SoCC ’21) . Association for Computing Machinery, 9 pages. https://doi.org/10.1145/3472883.3487009

arXiv 2021
[5]

CDC. 2024. NIOSH Sound Level Meter App. https://www.cdc.gov/niosh/noise/ about/app.html

work page 2024
[6]

Northern Virginia Electric Cooperative. 2025. Storm Center. https://www.novec. com/stormcenter/index.cfm

work page 2025
[7]

Utility Dive. 2023. Data center boom fuels demand for nuclear projects. Util- ity Dive (2023). https://www.utilitydive.com/news/data-center-boom-fuels- nuclear-construction-projects/733603/

work page 2023
[8]

Dustin Edwards and Zane Griffin Talley Cooper. 2024. The Making of Critical Data Center Studies. Convergence: The International Journal of Research into New Media Technologies 30, 1 (2024), 3–17. https://doi.org/10.1177/13548565231224157

work page doi:10.1177/13548565231224157 2024
[9]

Melissa Farney. 2025. The Power Play: How Data Centers and Utilities Are Reinventing Energy Strategies. https://www.datacenterfrontier.com/ energy/article/55277929/the-power-play-how-data-centers-and-utilities-are- reinventing-energy-strategies

arXiv 2025

Show all 42 references

[10]

Brody Ford, Riley Griffin, and Josh Saul. 2025. Meta Plans Nearly $1 Billion Data Center Project in Wisconsin. https://www.bloomberg.com/news/articles/2025- 04-04/meta-wisconsin-data-center-to-cost-nearly-1-billion

2025
[11]

Anshul Gandhi, Dongyoon Lee, Zhenhua Liu, Shuai Mu, Erez Zadok, Kanad Ghose, Kartik Gopalan, Yu David Liu, Syed Rafiul Hussain, and Patrick Mcdaniel
[12]

Viktor Urban Gsteiger, Pin Hong (Daniel) Long, Yiran (Jerry) Sun, Parshan Javanrood, and Mohammad Shahrad. 2024. Caribou: Fine-Grained Geospatial Shifting of Serverless Applications for Sustainability. In Proceedings of the ACM SIGOPS 30th Symposium on Operating Systems Princi...

2024
[13]

Lee, David Brooks, and Carole-Jean Wu

Udit Gupta, Mariam Elgamal, Gage Hills, Gu-Yeon Wei, Hsien-Hsin S. Lee, David Brooks, and Carole-Jean Wu. 2022. ACT: designing sustainable computer systems with an architectural carbon modeling tool. In Proceedings of the 49th Annual International Symposium on Computer Archite...

2022
[14]

Yuelin Han, Zhifeng Wu, Pengfei Li, Adam Wierman, and Shaolei Ren
[15]

Hanafy, Qianlin Liang, Noman Bashir, David Irwin, and Prashant Shenoy

Walid A. Hanafy, Qianlin Liang, Noman Bashir, David Irwin, and Prashant Shenoy. 2023. CarbonScaler: Leveraging Cloud Workload Elasticity for Opti- mizing Carbon-Efficiency. Proc. of the ACM on Measurement and Analysis of Computing Systems 7, 3 (Dec 2023). arXiv:2302.08681 [cs.DC]

2023 arXiv
[16]

PJM Interconnection. 2025. Data Miner 2. https://dataminer2.pjm.com/list

2025
[17]

Data Centers in Virginia

JLARC 2024. Data Centers in Virginia . Technical Report. Joint Legislative Audit and Review Commission. https://jlarc.virginia.gov/pdfs/reports/Rpt598.pdf

2024
[18]

Danielle Keading. 2025. Power-hungry data centers are prompting new gas plant proposals. Critics say that would lock in pollution and higher costs for decades. https://www.wpr.org/news/power-wisconsin-data-centers-gas-plants- pollution-higher-costs

2025
[19]

Islam, and Shaolei Ren

Pengfei Li, Jianyi Yang, Mohammad A. Islam, and Shaolei Ren. 2024. Making AI Less "Thirsty": Uncovering and Addressing the Secret Water Footprint of AI Models. Commun. ACM (2024). https://arxiv.org/abs/2304.03271

2024 arXiv
[21]

Diptyaroop Maji, Noman Bashir, David Irwin, Prashant Shenoy, and Ramesh K Sitaraman. 2024. The Green Mirage: Impact of Location- and Market-based Carbon Intensity Estimation on Carbon Optimization Efficacy. In Proceedings of the 15th ACM International Conference on Future and ...

2024
[22]

Eliza Martin and Ari Peskoe. 2025. Extracting Profits from the Public: How Utility Ratepayers Are Paying for Big Tech’s Power. Environmental & Energy Law Program (2025)

2025
[23]

Steven Gonzalez Monserrate. 2022. The Staggering Ecological Impacts of Com- putation and the Cloud. The MIT Press Reader (2022). https://thereader.mitpress. mit.edu/the-staggering-ecological-impacts-of-computation-and-the-cloud/

2022
[24]

National Institute for Occupational Safety and Health. 2024. NIOSH Sound Level Meter App. https://www.cdc.gov/niosh/noise/about/app.html

2024
[26]

Darrell Proctor. 2025. Natural Gas, Energy Storage at Heart of New Group’s Data Center Power Concept. https://www.powermag.com/natural-gas-energy- storage-at-heart-of-new-groups-data-center-power-concept/

2025
[27]

Arman Shehabi, Alex Hubbard, Alex Newkirk, Nuoa Lei, Md Abu Bakkar Siddik, Billie Holecek, Jonathan Koomey, Eric Masanet, Dale Sartor, et al. 2024. 2024 United States Data Center Energy Usage Report . Technical Report. Lawrence Berkeley National Laboratory. https://eta-publica...

2024
[28]

Brad Smith. 2025. The Golden Opportunity for American AI. https: //blogs.microsoft.com/on-the-issues/2025/01/03/the-golden-opportunity-for- american-ai/

2025
[29]

Abel Souza, Noman Bashir, Jorge Murillo, Walid Hanafy, Qianlin Liang, David Irwin, and Prashant Shenoy. 2023. Ecovisor: A Virtual Energy System for Carbon-Efficient Applications. In Proc. of the 28th ACM International Confer- ence on Architectural Support for Programming Langu...

2023
[30]

Bhargs Srivathsan, Marc Sorel, Arjita Bhan, Pankaj Sachdeva, Haripreet Batra, Raman Sharma, Rishi Gupta, and Surbhi Choudhary. 2024. AI Power: Expanding Data Center Capacity to Meet Growing Demand . Technical Report. McKinsey & Company. https://www.mckinsey.com/industries/tech...

2024
[31]

Emma Strubell, Ananya Ganesh, and Andrew McCallum. 2020. Energy and Policy Considerations for Modern Deep Learning Research. In Proceedings of the AAAI conference on artificial intelligence , Vol. 34. 3 pages

2020
[32]

Tessum, Jason D

Christopher W. Tessum, Jason D. Hill, and Julian D. Marshall. 2017. InMAP: A model for air pollution interventions. PLOS ONE 12, 4 (04 2017), 1–26. https: //doi.org/10.1371/journal.pone.0176131

2017 doi
[33]

Alexandra Tremayne-Pengelly. 2024. Amid the A.I. Boom, These States Have Become Data Center Hubs. Observer (2024). https://observer.com/2024/12/ai- demand-where-data-centers-using-most-energy/

2024
[34]

United States Environmental Protection Agency. 2023. Air Quality System (AQS). https://www.epa.gov/aqs

2023
[35]

Upwind. 2024. U.S. Data Center Powerhouses: The 5 Fastest-Growing Hubs. https://www.upwind.io/industry-research/data-center-powerhouses

2024
[36]

Bureau of Labor Statistics

U.S. Bureau of Labor Statistics. 2024. Northern Virginia, VA: Nonfarm employ- ment and labor force data. https://www.bls.gov/regions/mid-atlantic/data/xg- tables/ro3fx9538.htm

2024
[37]

Census Bureau

U.S. Census Bureau. 2024. QuickFacts: Loudoun County, Virginia. https: //www.census.gov/quickfacts/fact/table/loudouncountyvirginia/POP645223

2024
[38]

Energy Information Administration

U.S. Energy Information Administration. 2025. Electricity Data Browser. https: //www.eia.gov/electricity/data/browser/

2025
[39]

Jaylen Wang, Daniel S. Berger, Fiodar Kazhamiaka, Celine Irvene, Chaojie Zhang, Esha Choukse, Kali Frost, Rodrigo Fonseca, Brijesh Warrier, Chetan Bansal, Jonathan Stern, Ricardo Bianchini, and Akshitha Sriraman. 2025. Enabling Sustainable Cloud Computing with Low-Carbon Serve...

2025
[40]

WattTime. 2025. SIGNAL: Health Damage. https://watttime.org/data-science/ data-signals/health-damage/

2025
[41]

Carole-Jean Wu, Ramya Raghavendra, Udit Gupta, Bilge Acun, Newsha Ardalani, Kiwan Maeng, Gloria Chang, Fiona Aga, Jinshi Huang, Charles Bai, Michael Gschwind, Anurag Gupta, Myle Ott, Anastasia Melnikov, Salvatore Candido, David Brooks, Geeta Chauhan, Benjamin Lee, Hsien-Hsin L...
[42]

Madelyn Zander. 2024. Futuring the Cloud: An Ethnographic Study of Critical Data Center Infrastructure Growth, Environmental Threat, and Community Risk . Ph. D. Dissertation. University of Colorado at Boulder. https://www.proquest. com/docview/3100387207

2024
[2022]

In Proceedings of Machine Learning and Systems, D

Sustainable AI: Environmental Implications, Challenges and Opportunities. In Proceedings of Machine Learning and Systems, D. Marculescu, Y. Chi, and C. Wu (Eds.), Vol. 4. 795–813. https://proceedings.mlsys.org/paper_files/paper/2022/ file/462211f67c7d858f663355eff93b745e-Paper.pdf

2022
[2023]

SIGENERGY Energy Inform

Metrics for Sustainability in Data Centers. SIGENERGY Energy Inform. Rev. 3, 3 (Oct. 2023), 40–46. https://doi.org/10.1145/3630614.3630622

2023
[2024]

arXiv:2412.06288 [cs.CY] https://arxiv.org/abs/2412.06288

The Unpaid Toll: Quantifying the Public Health Impact of AI. arXiv:2412.06288 [cs.CY] https://arxiv.org/abs/2412.06288

arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [25]

Leonardo Nicoletti, Naureen Malik, and Andre Tartar. 2024. AI Needs So Much Power, It’s Making Yours Worse. https://www.bloomberg.com/graphics/2024- ai-power-home-appliances/

work page 2024

[2] [2]

Noman Bashir, Priya Donti, James Cuff, Sydney Sroka, Marija Ilic, Vivienne Sze, Christina Delimitrou, and Elsa Olivetti. 2024. The Climate and Sustainability Implications of Generative AI. An MIT Exploration of Generative AI (Mar 27 2024). https://mit-genai.pubpub.org/pub/8ulgrckc

work page 2024

[3] [3]

Noman Bashir, Tian Guo, Mohammad Hajiesmaili, David Irwin, Prashant Shenoy, Ramesh Sitaraman, Abel Souza, and Adam Wierman. 2021. Enabling Sustainable Clouds: The Case for Virtualizing the Energy System. In Proc. of the ACM Sym- posium on Cloud Computing (SoCC ’21) . Association for Computing Machinery, 9 pages. https://doi.org/10.1145/3472883.3487009

arXiv 2021

[4] [5]

CDC. 2024. NIOSH Sound Level Meter App. https://www.cdc.gov/niosh/noise/ about/app.html

work page 2024

[5] [6]

Northern Virginia Electric Cooperative. 2025. Storm Center. https://www.novec. com/stormcenter/index.cfm

work page 2025

[6] [7]

Utility Dive. 2023. Data center boom fuels demand for nuclear projects. Util- ity Dive (2023). https://www.utilitydive.com/news/data-center-boom-fuels- nuclear-construction-projects/733603/

work page 2023

[7] [8]

Dustin Edwards and Zane Griffin Talley Cooper. 2024. The Making of Critical Data Center Studies. Convergence: The International Journal of Research into New Media Technologies 30, 1 (2024), 3–17. https://doi.org/10.1177/13548565231224157

work page doi:10.1177/13548565231224157 2024

[8] [9]

Melissa Farney. 2025. The Power Play: How Data Centers and Utilities Are Reinventing Energy Strategies. https://www.datacenterfrontier.com/ energy/article/55277929/the-power-play-how-data-centers-and-utilities-are- reinventing-energy-strategies

arXiv 2025