HomeMy WebLinkAbout07 27 2026 - Public Comment - Trella Davis (2)1
Stefan Mercer
To:Stefan Mercer
Subject:RE: EXTERNAL - Submission to Public Record for Jul 27th Gilroy City Council Mtg
From: Trella Davis <trelladavis@gmail.com>
Sent: Friday, July 17, 2026 9:01 PM
To: Kim Mancera <Kim.Mancera@cityofgilroy.org>
Subject: EXTERNAL - Submission to Public Record for Jul 27th Gilroy City Council Mtg
Hi Kim,
I would like to submit the following report to the Gilroy City Clerk's Office on Data Center Impacts for the
July 27th meeting and into public record.
Please confirm receipt of this email and clarify any additional steps to be taken.
Thanks,
Trella Davis
1787 Branham Ln, San Jose, CA 95124
CAUTION: This email originated from an External Source. Please use proper judgment and caution when opening attachments, clicking links, or
responding to this email.
2026 CURRENT KNOWN ENVIRONMENTAL IMPACTS
Trella M Davis, 1TG Research – trelladavis@gmail.com
1
TABLE OF
CONTENTS
1 OVERVIEW
10 COMMON HARMS
44 NEXT STEPS
“Nothing is for free.
Everything has come at a cost.
Protect the home, the sanctuary,
the ship of life that
Nature, Earth and God provide.”
– Trella M Davis
1
Overview
Quick primer of Data Center currently known environmental resources uses and harms (at site
location impacts only). Data Center = DC
What are Data Centers: Data Centers (DCs)
are super-large dense financial value facilities
containing computer servers and GPUs or
TPUs used for cloud computing, data storage,
data analytics, generative AI, and streaming
services. While technically DCs have been
around for 80 years – starting with Mainframes
– conventional modern DCs have been
around since the 1990s for Cloud and
Hyperscale starting in 2006 with Google in
Oregon.
DC Resources Use Quick Overview: Writing a
single 100-word email with ChatGPT
consumes approximately the volume of a
standard bottle of water, uses 0.34 watt of
electricity (the amount to use an LED bulb for
2 minutes) and produces about 4.32 grams of
CO2 and other Green House Gases. This is just
the start.1 Materials mining, electronics
manufacturing, construction of physical site,
resources use and various wastes – there is a
mind-bogglingly array of negative
environmental impacts. It is difficult for most
people to relate to the amount of resources
DCs and especially AI and Hyperscale DCs
use so understanding personal application
and relatable examples is critical for those
looking to evaluate them or understand what
is really behind the tool.
Image Source:
https://www.hecweb.org/2025/06/28/measuri
ng-the-environmental-cost-of-artificial-
intelligence-and-their-data-centers/
1 https://www.hecweb.org/2025/06/28/measuring-the-environmental-cost-of-artificial-intelligence-
and-their-data-centers/
2
Overview
Most DC Developers, Commercial Real Estate
Brokers/Agents, and pro-Data Center
municipalities, may not understand or not
want to know the negative resource impacts
and their various immediate, as well as short
and long-term harms of unsustainable DCs
development. In the US, DCs now consume
nearly 5-6%+ of national electricity use and
that could rise to as much as 12% by 2028.2
Perspective: 2026 20% of US Winter Wheat
crop failed in part due to reduced adequate
water access. Current US corn production is
down by 3% in part due to due to climate
drought and water access availability
problems. Corn, wheat, rice, soybeans and
other major commodities are all shrinking (in
North America, South America, Europe, and
throughout Asia) to due increasing drought,
extremes in weather, rising temperatures,
increasing demand and now more expensive
input costs of fuel, electricity, water, and
fertilizer. DCs are a growing and major
industry to heavily worsen these types of
climate and water scarcity problems.
Basic Operations Overview: All DCs strive to
full operation of “five nines” or 99.999% at
24/7/365. Most DCs can function with 10-50
staff (this includes security and other exterior
personnel – contracted, human employees
and AI agent “employees”). Many DCs use AI
and AI agents to monitor, troubleshoot and
do other DC non-physical jobs like cyber
Image Source:
https://www.visualcapitalist.com/visualizing-
all-of-the-worlds-data-centers-in-2025/
2 https://earthjustice.org/article/what-you-need-to-know-about-ai-data-centers &
https://earthjustice.org/press/2026/naacp-asks-court-for-emergency-action-to-stop-illegal-air-
pollution-from-xais-data-center-power-plant
3
Overview
security. Also, DCs use non-location/fully
remote staff either US or foreign based. Many
share “local physical human” employees or
contractors across multiple sites. Many US DCs
use 2-5 million gallons of water per month for
cooling, generate massive amount of e-waste
monthly (servers are swapped every 1-3
years) and DCs have 30,000 to 200,000+ to
millions depending on site or campus size),
issue highly contaminated wastewater, and
use electricity anywhere from small 5 MW,
medium 20 MW, large 100 MW to campuses
500+ MW. Perspective, a single 50 MW facility
uses the equivalent amount of electricity as
about 40,000 U.S. households.3
DC Harm Health Quick Overview: DCs create
health risks and harm to their builders, workers,
visitors, neighbors, and communities (both
immediate and potentially far away). Air,
water and electricity impacts can cause
harm many miles to hundreds of miles away.
The most commonly measured risks to
neighbors and communities (people, animals,
plants, farms, ranches, waterways, air quality,
etc.) who are not at least 10 miles away, are
many. 7000 feet (1.4 miles) is the minimum
“safe” distance from EMFs, noise, and LF – Low
Frequency vibration. It is 5 to 10+ miles but up
to several hundred, especially if living
downwind, for gas particulate matter / toxic
emissions, and 10-100 miles for water
pollution/direct discharge/”Grey Watering”.
DCs can and routinely have polluted
ground/well/river/stream/aquifer water so a
safe distance could actually be 50-100 miles
or more. Approximately 6 miles is needed for
Image Source:
https://biologicaldiversity.org/w/news/press-
releases/report-ai-data-center-boom-
threatens-us-climate-goals-2025-10-29/
3 https://www.wri.org/insights/us-data-center-growth-impacts
4
Overview
‘Heat Islanding’ or ‘Heat Domes’ as DCs raise
area temperatures laterally by 4-16 degrees
Fahrenheit hotter and as much as 67 degrees
vertically hotter. At least 5 miles for a “safe”
distance is needed for threat of fire,
explosions, attacks, etc. Actual true “safe”
distances for some risks have not been
determined because this is a newish industry
and that would create unsurmountable
problems for getting approval for DC
hyperscale and campus installations next or
imbedded in urban or suburban areas.
Nearby farms, livestock, crops, wildlife,
pollinators, and humans also have various
problems from inability to sleep, still-births,
asthma, reduced growth rates, and more.4
Risks from Extreme Weather Threats &
Earthquakes: Droughts, heat, high heat,
hailstorms, extreme ice/snow, floods, fires, and
tornadoes are all increasing. Expanding and
shifting F5 level tornadoes have become
increasing common in states where never
before common (the entire Tornado Alley in
the US has actually expanded and shifted
significantly towards the East moving into the
Mississippi River valley and the Southeast,
including Alabama, Arkansas, Illinois, Indiana,
Kentucky, Mississippi, Missouri and Tennessee).
Tornado Alley’s eastward shift increases the
risk to human life because the new areas
have a higher population density, more
people living in manufactured homes, limited
road networks, and decreased visibility
caused by more and denser trees and hilly
terrain. These states all have an extremely
Images Source: https://www.apha.org/news-
and-media/multimedia/infographics/how-
climate-change-affects-your-health
4 https://www.nature.com/articles/s41598-018-26185-y & https://www.wsj.com/business/energy-
oil/a-new-foe-has-emerged-for-data-centers-farmers-4a3eda8f
5
Overview
large density of DCs.5 Google, AWS, and other
tech companies have literally built or are
currently building all along the intersection of
3 earthquake fault lines including the San
Andreas Fault line in California’s Bay Area.6
Low/No Buffer Zone: DCs are dangerously
close to or imbedded within towns, cities and
major metropolitan areas. Let’s reference a
similar high value-high risk site: Military Bases.
Bases for various security and safety reasons,
prevent future civilian encroachment like
high-density housing creeping too close to a
base. The military partners with local
governments and land trusts to establish multi-
mile conservation buffers around installations.
This insulates the base from complaints about
noise, dust, contaminants, security needs,
attack, and training maneuvers. DCs don’t
seem to care about ‘civilian’ safety buffers
because it vastly increases their infrastructure
connection costs and lowers their
conveniences to pre-built and city-
maintained resources. Infrastructure costs
elimination is one of the most pressing drivers
for DC development as electricity access and
operations profitability will take much longer
to recover or make what is already extremely
risky financially, impossible.
DC Litigation Boom7: Lawsuits due to public
and private nuisance, environmental harm,
property devaluation, disenfranchisement,
medical harm, workers comp, and more may
not be factored into the Developer’s costs,
but are quickly becoming the new Gold Rush
Image Source:
https://www.mercurynews.com/2025/04/09/s
an-jose-data-center-housing-project-
economic-benefits/ Proposed Data Center
with adjacent housing concept art City of San
Jose, CA
5 https://www.usnews.com/insurance/homeowners-insurance/new-tornado-alley-2026-shifting-risks
6 https://www.nbcbayarea.com/news/local/gilroy-amazon-data-center/4091984/
7 https://www.quinnemanuel.com/the-firm/publications/client-alert-emerging-litigation-risks-in-
financing-ai-data-centers-boom/#_ftn53
6
Overview
litigation industry with many firms focusing just
on DCs. Suits (at the municipal level) for
reasonable cause, measurable harm and
agreement violations are quite easy to
document.8
Tax abatements, permit reductions, and
grants have consistently not proven
productive.9 A Tax Revenue per Acre
calculation (Value Per Acre) shows DCs
realized revenues (before subtracting
municipal and community costs and harms)
does not out-perform mixed use low-rise
development (business on ground floor and
next 3 floors are housing). For DC projects
without grants or sales and property tax
abatements, tax revenues might generate
$40,000 to $80,000+ per acre annually vs Low-
rise mixed use $300,000 to $700,000+ per acre
annually (without unique resources costs and
harms).10 A Georgia state audit estimated
that in FY 2025, DCs brought in $41 million in
local tax revenue while receiving $474 million
in state sales tax exemptions. The
environmentally safer path many times is also
the financially sound or best path that is old
but proven.
Economic Development Orgs (EDOs) and
elected municipal officials may unwittingly
subsidize DCs with massive tax abatements
and more. Many times, this is through legally
Image Sources: https://signalohio.org/ohio-
approves-last-data-center-exemption-before-
moratorium/
8 https://www.quinnemanuel.com/the-firm/publications/client-alert-emerging-litigation-risks-in-
financing-ai-data-centers-boom/
9 https://stateline.org/2026/04/15/many-states-dont-report-losses-from-data-center-tax-breaks-
study-says/
10 https://www.dutchessny.gov/Departments/Planning/Docs/June2014-eNews-printerfriendly.pdf &
https://www.nahb.org/blog/2026/01/value-of-land-efficiency & https://goodjobsfirst.org/georgia-
data-center-tax-breaks/
7
Overview
grey to illegal Non-Disclosure Agreements and
with little to no Due Process for review.
Misleading to Fraudulent Information on
Immediate & Long-term Operations: Due to
the never before levels of tech and
investment money being spent, an extensive
amount of real estate brokers and
speculators, electricity/water use agreements
and “Behind-the-Meter Benefits” flippers,
prominent as well as low scruples DC
developers, private equity investors turning
more utilities into for profit organizations, and
doomsday preaching “tech bro” cash-laden
‘leaders’ have made several renown experts
and designers in the DC industry and well as
financial institutions all label this extremely
heavy speculation as “Fake Data Centers” or
“Phantom Data Centers”.11 There are around
5000 DCs across all 50 states.12 Quoting a DC
expert, “All the good locations have long
been gone”. Meaning doing whatever to
secure a location with as much free resources
and financial supports as fast as possible (by
any means necessary sometimes) has been a
key driver and a substantial strategy to
improve DC development margins and to
attract compute clients.
Current AI Policy and Infrastructures are Not
Aligned for Safety, Sustainability or Quality of
Control: AI safety has now become a
mainstream global concern across genders,
ideologies, religions and cultural groups.
Experts and the wider public are united in
their alarm to hatred at emerging risks and
the pressing need to manage AI and the
Image Source:
https://www.youtube.com/shorts/_Ahi5lPmLL
w Johnathan Ross Groq CEO
11 https://www.linkedin.com/pulse/age-fake-datacenters-daniel-golding-dw0je/,
https://www.linkedin.com/pulse/age-fake-data-centers-part-2-faster-faker-daniel-golding-ngxqe/
& https://www.linkedin.com/pulse/age-fake-datacenters-part-3-customers-speak-up-daniel-
golding-7i2re/
12 https://www.datacentermap.com/
8
Overview
infrastructures and resources behind it. AI
regulation has moved the last 10 years from
open-letters of dire warning13, to financial
maneuvers based on doomsday math, to the
beginnings of enforcement.14
Privacy and compliance teams are
increasingly responsible for translating these
requirements or lax grey area laws into
operational controls, documentation, and
ongoing oversight. Controlling AI
development and how it is used would require
effective and punitive consequences driven
oversight by cooperating governments over
AI development (e.g. limiting and slowing
growth of releasing new models/only
operating with ChatGPT 4.0 level power),
charging token rates that include real cost
plus Green House Gas/Carbon Credits
charges, and reigning in risky to
unmanageable financial debts (that have
massive legal/litigation exposure), global
usage standards (e.g. requiring all AI images
to be labeled as such), private sector (e.g.
disclosing AI content controls, labeling and
age restricted access), or rogue actors (e.g.
from developers to nation-states to vigilantes,
having rules around use and access), and
mutual oversight by governments over each
other (e.g. United Nations, etc.).
Image Source:
https://www.byfounders.vc/insights/what-is-
the-existential-risk-of-ai
13 https://futureoflife.org/open-letter/pause-giant-ai-experiments/
14 https://tinyurl.com/bk8f95u7 - The Economist, Business of Evils and Evals, ‘The world wants to
regulate AI, but does not quite know how’, Oct 2023 BLETCHLEY PARK & https://www.cntr-aisle.org/
& https://ainowinstitute.org/publications/data-center-policy-guide
9
Overview
There is no single, binding human-lead global
AI governance agency as of July 2026.15
Instead, an international evolving ecosystem
of transparent to opaque nation-states,
corporations, organizations, councils,
alliances, and grass-roots initiatives influences
AI policy today.16 AI development is
overwhelmingly concentrated and
dominated primarily in the United States
Silicon Valley by for profit companies – both
private and public. These various and
conflicting groups have wildly differing
motivations to guide development and usage
standards, responding to responsible and
criminal AI developments, and working
toward legally binding international
frameworks regarding AI (e.g. halting all AIs
with compute more powerful than ChatGPT
4.0) and the infrastructures behind it
(Developers, Data Centers Placement &
Construction, Electricity Generation &
Delivery, Natural Resource Usage &
Mitigation, Waste & Heat Management).
In 2026, binding AI laws apply across Europe
and with full force by 2027. Germany’s Energy
Efficiency Act (EnEfG), in effect since late
2023, mandates binding energy savings
across public bodies, private companies, and
data centers. It requires firms consuming over
2.5 GWh to implement energy-saving
measures, and enforces strict waste heat
recovery, strict power usage effectiveness
Image Source:
https://www.linkedin.com/pulse/making-
sense-ai-regulations-frameworks-part-1-who-
aiand-steenbeek-tyu7e/
Image Source:
https://hai.stanford.edu/assets/files/ai_index_r
eport_2026_chapter_9_public_opinion.pdf
15 https://www.un.org/global-dialogue-ai-governance/en
16 https://globalaigov.org/ & https://www.akingump.com/en/insights/ai-law-and-regulation-tracker
& https://ai-policy-tracker-tracker.vercel.app/ & https://www.loeb.com/en/general/ai &
https://www.dlapiper.com/en/resources/ai-laws-of-the-world
10
Overview
(PUE) limits, and renewable electricity usage
for data centers.17 As for the United States,
many cities, some states and even a bill in US
Federal government are creating DC
moratoriums. Multiple APAC and Latin
American jurisdictions are working towards
binding laws. China specifically has placed
many controls on development and working
towards sustainable infrastructures (Data
centers offshore ocean cooled, renewables
electricity powered, limits on land
displacement/usage, and China only has
based integrated supply chain from raw
materials to chips to GPUs). India seems to be
following a similar path that let the British East
Indian Company gradually take extractive
control starting in 1608 by simply gifting to
those in charge. State level utilities subsidies,
no public oversite, no environmental hold-
backs, zero-tax until 2047 for foreign data
center companies, construction in drought
and resource strapped areas, land and low-
income communities’ displacements, and
more.18 India does not have AI laws in place
but instead uses existing law and data
controls. Middle East data centers are
primarily in Isreal/Palestine, Saudi Arabia and
UAE (with substantial investments from Saudi
Arabia and UAE) making nearly 200 sites. A
combination of oil-rich Gulf sovereign wealth
funds plus Hyperscale Tech companies
putting in billions have created this
Image Source: https://www.congress.gov/crs-
product/R48555
17 https://www.mayerbrown.com/en/insights/publications/2024/02/sustainable-data-centers-the-
german-energy-efficiency-act-what-data-center-operators-need-to-consider-now-and-in-the-
future &
https://energypartnership.cn/fileadmin/china/media_elements/publications/Energy_Efficiency_Poli
cy_in_Germany_EN.pdf
18 https://www.barrons.com/articles/ai-data-centers-india-75425e19
11
Overview
infrastructure. A mix of many American and
foreign companies have invested in building
Isreal’s data center boom. All Middle East sites
use a variety of water dependent cooling
methods. Many sites were “impacted” by the
Isreal-Iran War – officially only five were
bombed (it is illegal to report any war
damage in these countries). Isreal does not
currently have AI laws and uses existing
privacy laws for governance. Saudi Arabia
also does not have AI laws yet, so
compliance relies heavily on the Personal
Data Protection Law (PDPL), the SDAIA AI
Ethics Principles, and the drafted Global AI
Hub Law.19 The UAE has a combination of AI
Guidance, Cyber Crime, Ethics Framework
and the Regulatory Intelligence Office
among other laws that are being used with AI.
Image Source:
https://www.theguardian.com/technology/20
26/jan/03/just-an-unbelievable-amount-of-
pollution-how-big-a-threat-is-ai-to-the-climate
Composite: Alex Mellon for the Guardian :
Getty Images
19 https://www.twobirds.com/en/capabilities/artificial-intelligence/ai-legal-services/ai-regulatory-
horizon-tracker/saudi-arabia
12
Common Harms
DATA CENTER MEASURABLE HAZARDS
E-Waste
DCs need to be totally transparent, have
a complete verified process and
trustworthy partners for E-Waste
management. An empty standard 42U
data center rack typically weighs
between 150 and 400 lbs (68 to 180 kg).
Uninterruptible Power Supplies (UPS) are
extremely heavy. A single battery unit
can add 50 to 100+ lbs. An individual
enterprise data center server typically
weighs between 30 to 80 lbs (14 to 36 kg),
depending on whether it is a standard
1U/2U unit or a high-density system.20 All
these components which most are
changed every 18-36 months is a massive
amount of e-waste that while toxic also
has value to the right recyclers. All e-
waste management and recycling
expenses should be covered by the DC
Developer/Operator. The following
categories is a partial list only.
E-waste Poor Certification & Lack of
Auditing - Many e-waste providers do not
have full / up-to-date certification, resell
to 3rd parties (who may or may not be
certified), ship overseas or otherwise do
not fully and legally process e-waste
correctly. Auditing of e-waste providers
by DCs and municipalities is necessary.
100MW facility will have 200,000+ servers
resulting in thousands of pounds of
dangerous but potentially valuable e-
waste weekly.
Image source: https://www.human-i-
t.org/how-to-evaluate-data-center-itad-
vendor-2026/
20 https://www.amcoenclosures.com/how-are-server-rack-load-ratings-calculated/
13
Common Harms
Industrial Recycling & Diversion Initiatives
- IT Asset Disposition (ITAD) programs.
These Hardware Circularity programs
resale/donate, repair and recirculate
used equipment, or recycle it to keep
materials out of landfills. A DC Operator
could have these programs in use,
identified, and audited. This should be
part of any DC project proposal.21
Waste
Waste Scale: Data centers are a major
contributor to the millions of tons of e-
waste generated globally every year.22
AI Waste Surge – Estimates show
Generative AI hardware could lead to
1.2 to 5 million metric tons of discarded
server equipment annually by 2030.23 24
Disposal Practices - Many data center
operators still fail to properly track or
recycle retired hardware, resulting in
toxic components being shredded or
dumped.25
Daily Refuse - A typical mid-sized data
center facility (10-50MW) generates
thousands of pounds of operational
waste (which is not including
construction waste) per year (e.g., office
waste and packing materials,
connection cables, cable wipes, etc.).
Estimates on operations waste amounts,
types, and management processes must
be a part of proposal.
Packaging Materials - Wood pallets,
plastic shrink film, and cardboard
consistently rank as the top three physical
Image Source:
https://www.dumpsters.com/blog/data-
center-construction
21 https://www.celestica.com/blog/article/circular-economy-solutions-for-hyperscalers-a-game-
changer
22 Data Center Recycling Alone Can't Solve the E-Waste Crisis | Human-I-T
23 AI will add to the e-waste problem. Here’s what we can do about it. | MIT Technology Review
24 AI-driven data centers risk massive e-waste surge by 2030 - EHN
25 Data Centers Can Achieve +97% Waste Diversion — All About Waste
14
Common Harms
waste streams. Collection and recycling
processes for these materials must be a
part of proposal. During construction,
especially for campuses and hyperscale
data centers 1000s of wood pallets will
be collected in 18-24 months. Special
process recyclers/dedicated process
recyclers for wood pallets must be
identified, audited and included in the
proposal. A waste of 3000 pallets in 18
months is an unmanageable burden on
local municipal waste management, a
strain on local landfills, and
environmentally irresponsible to not
reuse.
Plastic shrink film and Mixed Material Soft
Plastics Recycling - These materials are
not processable at almost all US waste
management or recycling centers.
Specialized dedicated recyclers must be
identified, contracted, and used with
disclosure to provider, and auditing as
part of proposal. 100% of shrink film and
multi-layer soft plastics are to be
recycled.
Landfill Avoidance - On-site waste
segregation of waste is frequently
inconsistent at many businesses including
data centers, causing many of
recyclable and reusable materials to be
sent straight to landfills. Waste
segregation and management will be a
part of project proposal.
Zero-Waste Goals - Some hyperscale
facilities have achieved "Zero-Waste-to-
Landfill" certifications by diverting over
90% of their waste through
comprehensive reuse and recovery
networks. DC Developer will have this as
part of their operations with audits and
include in project proposal.
Image Source:
https://www.ecohome.net/en/guides/3593/is-
it-safe-to-build-beds-or-furniture-with-shipping-
pallets/
15
Common Harms
EIR Environmental Impact Report
Discrepancies, Few Data Points Used &
No Indirect Impacts Studied:
For example, some states have EIR laws
like the California Environmental Quality
Act (CEQA) which requires state and
local agencies to identify, disclose, and
mitigate the significant environmental
impacts of discretionary projects like
data centers. Enacted in 1970, the law
applies to public and private projects
requiring government approval, ensuring
government decision-makers and the
public understand a project's
environmental consequences.
Discrepancies: Facility-specific data—
such as how much water or electricity a
single local data center uses—must be
broken-out to see entire impact. Partial
data provided, as in a focus on Phase 1
data only, when a site may have several
Phases each with additional impacts.
Total full completion site data (with direct
and indirect resource impacts) must be
used in all proposals, communications or
marketing to community and other
relevant information sharing.
Construction, Fully Operational and
Indirect impacts must be presented in
proposal.
Overly “optimistic” EIR assumptions – EIR
Consulting Firms are paid by DC
Developers who may want to hear
certain outcomes that tend to look at
noise, impact on local birds, farmland
displacement, small mammals and so
forth with optimistic biases because they
maybe rewarded with more requests for
EIRs (each Final EIR – FEIR costing as
much as $100,000 or more). Municipalities
should decide which EIR firm to use and
DC developer will pay.
Image Source: City of Gilroy, CA AWS Final
Environmental Impact Report Snapshot
https://tinyurl.com/4y5xjuen
16
Common Harms
Exemptions and Loopholes: Projects can
sometimes “slip through” local review
processes which will result in political and
legal backlash. Project will follow due
process for community input.26
FEIRs – Final Environmental Impact
Reports – often are missing Comment
Letters from local individuals, and local
businesses. They also lack a local citizen
special purpose advisory committee –
which is commonly created for all sorts of
county, city or town priorities or concerns.
Local organizations like Teamsters Unions,
Carpenters Union and other construction
and transportation trades that are
included in “Index of Local
Organizations” that have direct financial
benefits from data center proposals and
do not adequately cover the various
groups that make up a community being
impacted by a data center.
FEIRS should include Required Inputs from
Local and state agencies, local
organizations, local businesses and local
legal individuals who live within 5 to 10-
mile range of data center epicenter.
Based on utilities requirements, local
municipalities (another city within 10-20
miles) that may be impacted must be
included in the Required Inputs Letters of
Comment.
For projects (over 50MW or 95-99MW)
which are nearly 100MW or have the
possibility of meeting or developing into
100MW (due to later advances in
GPUs/TPUs, Racking, etc.) – those
regulations should be included for
reference.
Image Sources:
https://encinoenviron.com/epa-closes-
permitting-loophole-used-by-ai-data-centers/
https://www.epa.gov/stationary-sources-air-
pollution/stationary-gas-and-combustion-
turbines-new-source-performance
26 https://stpp.fordschool.umich.edu/sites/stpp/files/2025-07/stpp-data-centers-2025.pdf
17
Common Harms
Local Stakeholders, Agencies & Utilities:
transportation, utilities (electric and
water), waste, waste water, emergency
services - fire, EMS, police, and sheriff,
local primary healthcare provider,
agricultural agency, local or state fish
and wildlife agency, environmental and
conservation agency, transportation and
traffic agency, parks and recreation
agency, housing authority, business
association, impacted schools and
community areas, and local construction
and transportation trade unions.
Indirect Impacts:
Ranges from not including water and
electricity needed to produce GPUs/TPUs
(and the Green House Gas emissions or
CO2 costs). Actual electricity used or
higher utility bills or contaminated water
for neighboring towns are measurable
indirect impacts. Indirect impacts should
be studied at the cost of the developer
and be submitted with proposal.
The "AI" Disconnect: Environmental
disclosures do not necessarily
differentiate between general cloud
computing and highly resource-intensive
artificial intelligence workloads. DC
Developers and Operators must disclose
type of computing at proposal and any
material change to that will nullify
agreements and be subject to
remediations and fines.
Omitted Byproducts: Proposals and EIR
statements regularly fail to accurately
quantify the total environmental footprint
of a DC by omitting factors like the life-
cycle of waste, pollution and resource
use/reduction. Scope of EIR should be
determined by municipality to include all
direct and indirect environmental
impacts.
Image Source:
https://www.instagram.com/p/DY4ozVTFqED/
18
Common Harms
Over and Undersized Baseline
Assumptions: Energy estimates are based
on historical hardware standards instead
of the version of AI equipment that will
be in operation in 3-7 years from time of
proposal. Estimations on resource usage
projections need to account for local
and state laws (example data center at
100MW+ in state of California requires a
lot more regulations and reviews as
classified as a power station). As GPUs
are getting denser and using more
electricity every day, a DC site rated at
99MW with N+1 redundancies built in can
already operate at 100MW or more and
in a few years after starting operations,
potentially operate substantially over the
100MW maximum electricity threshold.
Twenty-seven states have advanced
state data center legislation that requires
developers to cover DC energy costs
and accurately report usage, with
California, Ohio, and Utah already
enacting laws that go beyond the
federal government's voluntary
Ratepayer Protection Pledge.27
DC’s potential for exceeding MW backup
electricity generation: This must be
provided in proposal (as they already
have the technical and physical
capability to exceed 100MW), as this
serious material and energy change
impacts the municipality along state and
federal laws and regulations (Title 24
Efficiency requirements, Residential Rate
Payer Protections, Stringent Air Quality &
Environmental Impact Reviews, etc.).
Example in California a municipality will
consider any data center project larger
than 90MW to be evaluated with the
Image Source:
https://www.newsweek.com/elon-musks-new-
ai-data-center-raises-energy-water-concerns-
memphis-1937109
Photo-illustration by Newsweek/Getty
27 https://www.whitehouse.gov/releases/2026/03/ratepayer-protection-pledge/
19
Common Harms
same or better safety criteria of a 100MW
site.
General Lack of Transparency/Fraud:
DCs but especially AI DCs have on-going
lack of Transparency, Due Process and
Resource Estimates that are misleading
to fraudulent. A municipality must include
punitive fines to banning operations for
such faulty data or mis-representation as
part of any agreements.
Substations, Transformers, & Lines:28
High & Extreme Heat Days Explosion/Fire
Risks: If degraded insulation or high
ambient heat causes a fault, an
electrical arc-flash can occur. These
events produce localized temperatures
ranging from 5,000°F to 35,000°F, which
can instantaneously ignite the dielectric
cooling fluids, which flash-point at
roughly 300°F to 350°F.
Safety Setbacks: Frequency Noise, EMF
and Electrical Safety, along with flash-
point fire risks cumulatively means 2.5 - 3
miles setback. Warnings must be posted
about sulfur hexafluoride potential leaks
(insulating and arc-quenching gas, SF₆ is
odorless, colorless, and is recognized as
the world's most potent greenhouse
gas—possessing a global warming
potential roughly 23,500 times that of
CO₂). SF₆ is denser than air and can
quickly displace oxygen in low-lying
areas outside or inside in poorly
ventilated substation rooms, creating a
suffocation risk. Risks for Explosion must be
posted as Substation equipment (such as
circuit breakers and transformers) relies
heavily on thousands of gallons of
mineral oil for cooling and insulation.
Image Source:
https://www.pecva.org/region/culpeper/culp
eper-town-poised-to-allow-more-data-center-
noise-county-considers-a-substation-for-yet-
another-data-center/
DC Substation Photo by Hugh Kenny/PEC
28 https://blog.ucs.org/guest-commentary/the-electricity-distribution-system-can-stay-
resilient-in-extreme-weather/
20
Common Harms
Unique Toxic Firefighting Requirements:
Once on fire, substations are connected
to large battery storage facilities
typically. These lithium-ion batteries can
experience "thermal runaway," burning
off electrolytes and venting harmful
gases. Substation fires’ smoke contains
toxic volatile organic compounds (VOCs)
and particulate matter and the cooling
oil may contain toxic polychlorinated
biphenyls (PCBs), creating highly
hazardous smoke and requiring
specialized environmental cleanup.
Firefighter staff must be properly warned
and trained to deal with these extremely
toxic fire environments.
Heat Island, Heat Column & Heat
Feedback Loop:
Heat Island: Data Centers depending on
size, produce more waste heat than
40,000 households; this plus particulate
matter accumulates and forms self-
sustaining "heat islands" rather than
rapidly dissipating. Studies show a
minimum 4-16 degrees temperature raise
in up to a 6-mile range creates “Data
Center Heat Islands”. Thermal Plumes
above data centers further strengthen
the Heat Island impact.29
Concentrated Thermal Plumes: “Heat
Columns” (air above data center) can
rise to 67 degrees hotter than ambient
temperatures impacting localized
weather in immediate area so less rain
and snow or none for those less than 1-2
miles from facilities center. Having a 6-10
mile setback might address this
significant impact somewhat, but as of
Image Sources:
https://www.facebook.com/groups/10628680
21724371/posts/1741577563853410/
https://futurism.com/artificial-
intelligence/data-center-cooks-nearby-
residents-heatwave
29 https://weather.com/news/climate/news/2026-04-14-artificial-intelligence-heat-island-warming
21
Common Harms
2026 it is not a topic with much published
formal studies only incident reports by
sites.30 Impacts of precipitation / rainfall
elimination around and over data
centers may account for lack of
rainwater harvesting – none-the-less,
rainwater harvesting must be
incorporated into site design and on-
going operations.
Heat Energy Feedback Loops: The
elevated temperatures require nearby
residential and commercial properties to
use more air conditioning, which in turn
vents more heat into the atmosphere,
creating an escalating thermal feedback
loop creating a strong Heat Dome.
Land Cover Conversion Required: The
replacement of natural, vegetated
landscapes (which absorb less heat) with
concrete, metal buildings, and parking
lots (which absorb and trap heat) plus
Thermal Exhaust (The active release of
warm air from mechanical chillers and
server fans) creates 3.4 degrees raise but
in some cases up to 16 degrees
Fahrenheit increases about 6 miles in
radius.31 Land and plant designs to
mitigate heat retention is required.
Setbacks of 6 miles from center of facility
may not be adequate.
Image Sources:
https://www.instagram.com/p/DP2NRoLjCbI/
https://www.theguardian.com/world/2026/ju
n/09/worlds-first-wind-powered-underwater-
datacentre-starts-operating-in-china
30 https://asmedigitalcollection.asme.org/sustainablebuildings/article/7/2/024501/1233035/Data-
Center-Waste-Heat-as-an-Emerging-Urban
31 https://www.cnn.com/2026/03/30/climate/data-centers-are-having-an-
underrported
22
Common Harms
Light Pollution:
DCs generate large high intensity
unnatural light pollution. Regular
and Hyperscale DC facilities require
extensive all-night full coverage high
Intensity “security prison lighting” across
the entire compound, adjoining
complexes and buffer areas due to
extreme risk security requirements.
This harms the natural (circadian)
rhythms of the body, including melatonin
production (the hormone that regulates
sleep), insomnia, fatigue, anxiety,
cardiovascular stress, low testosterone,
chronic diseases complication or
instigation, and sleep-wake cycles
disorders. Excessive Light pollution also
harms migration patterns, habitat
development, fertility, predator-prey
relationships, population size/health, and
mating32 among various species and
animals like birds, insects, bats, cats, farm
animals and reptiles as well as many
plants to name only a few. It creates
systemic ecosystem imbalance. This type
of unnatural light obviously also
negatively impacts site workers. Glare
from unnaturally bright lighting creates
potential visibility and safety problems for
those even driving by DCs.
Setbacks for commercial flood lights
should be 1000+ feet.
Image Source:
https://x.com/Newsforce/status/205102147435
8587779
32 https://www.sciencedirect.com/science/article/abs/pii/S0169534722003329
23
Common Harms
Water Use & Wastewater - Scarcity,
Contamination & Pollution:
40% of a Unites States based DC
electricity usage can be for cooling.
Newer DCs use liquid cooling (Closed
or Open Loop). Liquid cooling is more
expensive up front but cheaper in
operations and requires water to cool
computer servers so they do not
become too hot to function. A large
DC will use up to 5+ million gallons of
water daily. Drawing from local water
supplies, these facilities are slurping
up resources that should be available
to residents, agriculture and general
water table health in water-scarce
regions. No DC needs water. Water in
some form of cooling use is typically
the cheapest and easiest means to
cool a data center due to the fact
that municipalities and water utilities
are allowing water access and at
rates that are at or below residential
rate tiers. The drilling of wells for water
use is problematic as even substantial
historic ground water levels or
aquifers do not replenish at data
center or hyperscale usage rates.33
Water & Closed-loop systems - These
still require ‘Blow-down’ or changing
of liquid cooling mixture on a regular
basis. This is wastewater which
includes various bacterium, metals,
and toxic chemicals that many
wastewater treatment facilities are
not capable to treat. “Hauling
Image Sources:
https://www.yahoo.com/news/science/articl
es/epa-writing-permits-could-let-
182418731.html
33 https://www.eesi.org/articles/view/data-centers-and-water-consumption &
https://www.wri.org/insights/us-data-center-growth-impacts
24
Common Harms
cooling waste liquid offsite” is not a
clear option. Who, what and how this
waste will be fully treated or burned
must be identified with the name of
treatment provider, audit cycle and
clarity what blow-down schedule is
required to be in compliance and
following Insurance Coverage
agreements. Direct and indirect
water use by DCs is adding to various
water resource problems. Even
Nvidia’s newest AI servers which can
operate at 45°C34 are connected to
a Closed-Loop system would still
require regular maintenance with a
parallel example being getting an oil
change for a conventional Internal
Combustion Engine car; it is closed-
loop but still gets “dirty”.
Water scarcity is increasing while
drought is also increasing. Nearly 98%
of the water consumed by major DC
operators is drawn from municipal
drinking water systems—many of
which are already operating near or
at capacity.35 A hyperscale facility
can use 1-5+ million gallons of direct
water daily—equaling the daily
consumption of a city with up to
50,000 residents. Additionally
landscaping at large hyperscale
facilities can use 20,000 gallons per
day; and due to the DC
environmental conditions requires
substantially more water. Globally,
DCs rank among the top 10 water-
consuming commercial industries,
with collective usage straining local
Image Sources:
https://www.facebook.com/100069223329155
/posts/understanding-closed-loop-cooling-at-
the-qts-data-center-in-york-
countyintroduct/1231593165824783/
https://www.change.org/p/stop-mega-data-
centers-in-the-rgv-protect-our-water-grid-and-
public-health/u/34247184
34 https://blogs.nvidia.com/blog/liquid-cooling-ai-factories/
35 https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2025AV002140
25
Common Harms
supplies and environments. Example:
Tech giants like Google report using
6.4 billion gallons across their data
centers in the US and offices in a
single year. Basic industry metrics,
such as Water Usage Effectiveness
(WUE), a measure of water use
efficiency for data centers, are
disclosed by fewer than one-third of
operators.36
Indirect water footprint can be as
large or even larger, as power
generation for DCs requires
additional water. Electricity
generation in the US requires on
average 0.075 gallons per watt. GPUs
are extremely water intensive with
chip makers needing to distill
municipal water before using. It takes
approximately 8 to 10 gallons of
water to manufacture a single
computer chip, including a GPU. This
volume goes toward the ultra-pure
water required to wash silicon wafers
and prevent microscopic
contamination during fabrication. A
“99” megawatts MW (approximately
99 million watts data center is using
7.5M gallons of electricity generation
indirect water per day, with
approximately 60K GPUs (600,000
gallons of water replaced every 24
months). Additionally, more water is
needed to treat and manage
wastewater. As all hyperscale DCs
are built on N+1 or 2N standards for
redundancy in operations it is easy to
assume operators could and would
Image Sources:
https://www.sciencedirect.com/science/artic
le/abs/pii/S0921344925001892?via%3Dihub
https://eta.lbl.gov/publications/water-use-
data-center-workloads
https://uwm.edu/freshwater/data-centers-
consuming-water-the-conversati/
36 https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2025AV002140
26
Common Harms
increase total capacity of compute
by simultaneously drawing electricity
from grid connections while running
at full original or greater capacity on-
site electricity generation (on-site fuel
generators mobile or stationary).37
Exceeding permitted backup MW
usage has legal, regulatory,
permitting and other oversights in
many countries, regions and cities.38
Exceeding backup or grid drawn
electricity usage creates a material
change in data center agreements
with municipalities.
Known Wastewater Contaminant
Types: There may be other
contaminants currently unknown
which must be disclosed and will be
included once detected. Closed
loop evaporative cooling loses 70-
80% of second loop water for a
typical AI DC using 5M gallons a day
in a Closed Loop Cooling system
produces at least 1M gallons a day of
grey sludge “blowdown” must be
filtered, treated and recycled at the
DC and not pushed off to local
wastewater treatment facility or
dumping (river), sent to a dump site
(local or not), or pumped into ground
water/aquafers. EPA has testified to
congressional oversite subcommittee
that it as well as municipal
wastewater treatment facilities in the
US and otherwise are not capable to
handle the various types of
contaminates the volume of
Image Sources:
https://www.instagram.com/reels/DYSMGNti
G4n/
37 https://www.nytimes.com/2026/05/14/climate/elon-musk-xai-data-center-doj.html
38 https://www.wsmv.com/2026/07/08/nashville-data-center-restrictions-advance-after-record-
setting-council-meeting-heres-what-you-should-know/ & https://www.energy.ca.gov/programs-
and-topics/topics/data-centers
27
Common Harms
wastewater contamination or
continuous rate of flow of
contamination. Closed loop cooling:
A two “loop” process system. Loop
one stays within the system, floods the
servers to remove heat through a
heat exchange chiller to then
continue on and back through the
loop to be recycled through an
ongoing heat, cool heat cool
process. Servers get up to 200
degrees F but must be kept below
170 to avoid damage. Loop two is
part of the heat exchange process as
well but water and contamination
travels to the roof where it goes
through evaporative coolers with
water (instead of air conditioning
cooling which is more expensive).
Potable water is cheap and usually
has been negotiated to residential or
even below residential rates. When
that water evaporates as part of the
cooling process, there is no recapture
process. Most municipalities cannot
provide Recycled Water for DC use
because it is not clean enough. DCs
do not build water recycling into their
processes due to added incremental
cost.
Wastewater Pollutants from “Closed
Loop” cooling systems “Blowdown”:
Total dissolved solids (TDS): The
most common effluent from data
center cooling systems is a high
concentration of dissolved
minerals and salts, like chlorides.
This can put stress on local
wastewater treatment plants,
which are often not designed to
handle a large volume of these
concentrated pollutants.
Image Source: https://ketos.co/closed-loop-
cooling-water-saver-or-chemical-time-bomb
28
Common Harms
Corrosion inhibitors: These
chemical additives protect
cooling equipment but can
pollute waterways. Examples
include phosphates and
molybdates, which contribute to
the nutrient overgrowth of algae
(eutrophication) and negatively
affect aquatic life.
Untreatable bio-hazards like new,
rare (e.g. Cupriavidus gilardii),
common bacteria, or other single-
cell organisms that municipal
wastewater treatment facilities
are not currently capable to
treat39
Biocides: These chemicals, such as
isothiazolinones and
glutaraldehyde, are used to
prevent the growth of bacteria
and other microorganisms in
cooling towers. Toxic to any life
but especially aquatic
ecosystems.
Glycols: Some liquid-cooled
systems use ethylene or propylene
glycol as a coolant. Ethylene
glycol is toxic and can cause
serious environmental damage if it
leaks or is disposed of incorrectly.
Per- and polyfluoroalkyl
substances (PFAS): Some
advanced two-phase immersion
cooling systems use refrigerants
that fall under the category of
PFAS, or "forever chemicals,"
which are persistent and
bioaccumulative in the
environment.
Image Source:
https://www.bizjournals.com/atlanta/news/20
25/03/06/data-center-water-use-georgia.html
Atlanta Business Chronicle
39 https://www.forbes.com/sites/victoriaforster/2026/07/09/meta-ai-data-center-linked-to-rare-
bacteria-in-citys-water-system/
29
Common Harms
Heavy metals: These can leach
from cooling equipment as it
corrodes. Examples include lead,
zinc, copper, and chromium,
which are harmful at these levels
to wildlife, marine life and
humans.
Sulfuric acid, lime, and anti-
scaling agents like phosphonate
How often blow-down (dumping
the coolant water and replacing
it) happens must be monitored,
recorded and reported, because
a "closed loop" system could be
blow-down multiple times per day
essentially turning a "closed loop"
promise into an effective open-
loop well-to-sewer system.
Blow-down dumping. For those
DC sites where wastewater
treatment at local municipalities
or on-site operations cannot or will
not properly and safely treat
blow-down water, many DC
operators truck this waste for off-
site “removal”, “relocation”, or
other potential scenarios for
possible dumping.40
Wastewater Pollutants from backup
power (generators or BESS):
DCs typically use methane or
diesel-powered backup
generators to ensure continuous
operation during grid failures.
While their primary output is air
pollution, water contamination is
normal.
Image Source:
https://www.usdanalytics.com/industry-
reports/water-treatment-chemicals-market-in-
data-centers-and-electronics-manufacturing
40 https://www.techrepublic.com/article/news-meta-ai-data-center-water-scare/
30
Common Harms
Diesel fuel spills: Accidental leaks
or spills of diesel fuel from on-site
storage tanks can contaminate
soil and groundwater.
Heavy metals from batteries:
Uninterruptible power supply (UPS)
systems use batteries to bridge the
gap between a power outage
and the startup of backup
generators.
Lead-acid batteries contain lead
and sulfuric acid, both of which
are hazardous materials that
require strict management to
prevent water contamination at
end-of-life.
Lithium-ion batteries release toxic
chemicals during manufacturing
and at the end of their lifecycle.
Wastewater Pollutants from e-waste,
maintenance and emergency/fire
response:
Electronic waste (e-waste):
Servers, batteries, and other
hardware are regularly upgraded
in data centers, producing
significant amounts of e-waste.
Toxic leachates: If not disposed of
properly, e-waste can leach
heavy metals like lead, mercury,
copper, and cobalt into
groundwater.
Cleaning agents & degreasers:
Maintenance produce
wastewater contaminated with
industrial-grade chemicals used
for cleaning & degreasing.
Clean Agents specialized for
firefighting data centers create
substantial PFAS and other
pollutants.
Image Source:
https://www.tycorun.com/blogs/news/data-
center-
batteries?srsltid=AfmBOooqZzxzgMDX969lOli0v
IQ--icap7ExwEVduJi16TQ7Y0idIJQ6
31
Common Harms
Air Pollution:
10+ Miles Setback for DC Air
Pollution:41 DCs, especially gas-
powered and backup DCs, emit
significant pollutants, such as nitrogen
oxides, methane, volatile organic
compounds, and course to fine
particulate matter. According to a
September 2025 study, these
pollutants increase rates of respiratory
diseases and cardiovascular
conditions, and they elevate cancer
risks among nearby communities. A
2025 model indicates that U.S. data
centers in 2030 could cause
approximately 600,000 asthma
symptom cases and 1,300 premature
deaths, exceeding 1/3 of asthma
deaths in the U.S. each year, resulting
in a public health burden of more
than $20 billion. Air quality measures
that do not meet agreed upon levels
are subject to penalties, moratoriums
and legal liabilities.
Impacts to adults and children
include but not limited to42:
Abnormal brain development,
reduced math test scores, and
increased ADHD or behavioral
risks
Asthma exacerbations,
hospitalizations, and even
structural DNA changes in the
lung’s airways
Obesity, childhood obesity,
diabetes, and metabolic changes
Image Source:
https://www.nytimes.com/2025/10/20/technol
ogy/ai-data-center-backlash-mexico-
ireland.html
Marcos Zegers for The New York Times
41 https://hsph.harvard.edu/news/analyzing-air-pollution-health-economic-risks-from-ai-data-
centers/
42 https://www.epa.gov/research-grants/niehsepa-childrens-environmental-health-and-disease-
prevention-research-centers
32
Common Harms
Waste Particles can be carried over
long distances by wind and then
settle on ground or
water. Depending on their chemical
composition, the environmental
effects include but not limited to43:
making ground water, lakes &
streams acidic
changing the nutrient balance in
coastal waters, large river basins &
surrounding lands
depleting the nutrients in soil & soil
viability
damaging sensitive forests &
wildlife, farm crops & ranch
animals
affecting the diversity of
ecosystems
contributing to acid rain & its
effects
EPA alongside the NIEHS, has
determined this is no safe level of
diesel gas or nitrous oxide
emissions
Transboundary Air Pollution - DC Air
Pollution Traveling Distance:
Communities living within one mile of
EPA-regulated DCs face levels of
particulate matter, nitrogen dioxide,
diesel and other types of particulate
matter above the national median.
However, the impact is confined to
just those living within such close
proximity. Under certain conditions,
PM2.5 particles can travel hundreds
or even thousands of miles away
from their source, meaning that data
center emissions can affect non-
fenceline communities because
Image Source:
https://loudounclimate.org/data-centers/
43 https://www.epa.gov/pm-pollution/health-and-environmental-effects-particulate-matter-pm
33
Common Harms
particulates can stay in the
atmosphere for days to weeks.
Nitrogen oxides have a relatively
short atmospheric lifetime, but they
can still travel far with strong winds.
PM10 particulate can stay in the air
for minutes to hours and can travel
between 100 yards and about 30
miles.44
Carbon Credits & Offsets:
Carbon offset allows an individual or
organization to compensate for their
GHGs greenhouse gas emissions by
funding environmental projects that
reduce or remove an equivalent amount
of carbon dioxide elsewhere. One
carbon offset represents the reduction or
removal of 1 MT (metric ton) of carbon
dioxide equivalent (CO₂e). DC should
have their construction, operations and
emissions off set by Nature-Based local or
in state Certified Carbon Capture
Projects annually. A continuously running
1 MW IT load draws significantly more
power to operate (accounting for
cooling and power distribution overhead,
known as Power Usage Effectiveness or
PUE). A 1 MW facility generally emits
between 3,000 and 6,500 MT metric tons
of CO₂ equivalent (CO₂e) per year. A
Data Center of 100MW would require 3-
7M+ MT metric tons of CO₂ equivalent
(CO₂e) per year. The DC Developer must
provide names of Carbon Offset Projects
chosen or work with municipality to select
local certified programs. Additionally, the
land that was cleared is extensive.
Carbon Offset Projects for that impact
will need to be included as well.
Image Source:
https://energytracker.asia/what-are-carbon-
credits-and-how-do-they-work/
44 https://www.clarity.io/blog/how-far-can-air-pollution-travel &
https://energy.sustainability-directory.com/term/transboundary-air-pollution/
34
Common Harms
Wildlife, Plants, Farms & Ranches:
DCs impact trees, land and animals
primarily through massive land clearing,
heat islanding/heat domes, water
reduction/contamination, and pollution,
which destroys native habitats and the
ability for the land to act as a Carbon
sink as well as ground water support. DCs
threaten wildlife through habitat
destruction, intensive water use, water
pollution, ecosystem drought
exacerbation, noise pollution, heat
islanding, and grid-related air pollution
for several miles up to 10 or more. These
facilities stress local ecosystems, displace
native species, and endanger vulnerable
or migratory wildlife. Wildlife is also
impacted by distressed trees and other
plant life.
Plants whether native or landscaped
suffer. Thermal Pollution and Heat
Islands/Heat Domes from the continuous
emission of hot air from data center
cooling systems creates localized
"microclimates" of 4 to 16+ degrees
Fahrenheit hotter laterally. Vertically
temperature increases of up to
additional 65+ degrees Fahrenheit hotter
(reducing or eliminating snow and other
precipitation) have been measured. This
extreme elevated temperature dries out
the surrounding soil, forcing plant life to
adapt to increased drought stress and
altered seasonal growth cycles. Normal
watering for landscaping needs to be
increased substantially to sustain plants.
Most plants in perimeter show slow to
stunted growth. High water-consumption
models and cooling towers can alter
local humidity. Some local agriculture,
like nearby Christmas tree farms, have
reported stunted growth and lost viability.
Routine testing and emergency use of
Image Source:
https://www.instagram.com/reels/DXuSh3aEv
VL/ Clayton Tucker Texas Agriculture
35
Common Harms
diesel backup generators emit nitrogen
and sulfur oxides and other particulate
matter that damage tree foliage,
degrade air safety and produce acid
rain.
Farms of any type45 can suffer – Those
farms located within 6-10+ miles and
ranches where livestock experience on-
going negative impacts similar to people.
Farms and ranches next to or near DCs
across Texas have been experiencing
some consistent impacts: no pregnancies
or only still births with cattle and goats,
reduced growth rates and weights of
around 30% less consistently with various
types of livestock, reduced dairy yields,
and egg production rates reduced by
50%. Agricultural farms typically are
negatively impacted by contaminated
water, low water pressure, ‘Habitat
Parching’ (decreased water table dries
out entire areas), increased utilities rates,
competition for land, and increased
water scarcity.46 Combined negative
impacts alters soil composition, promotes
desertification, acid rain (from nitrous
and sulfuric oxide emissions) and disrupts
delicate underground mycorrhizal
(fungal) networks that plants rely on for
nutrient sharing.47
Sound & Vibration Pollution:
There are several types of sound and
vibration pollution all with their own
harmful impacts.
Image Source:
https://www.wsj.com/business/energy-oil/a-
new-foe-has-emerged-for-data-centers-
farmers-4a3eda8f
45 https://news.vcu.edu/article/northern-virginia-data-center-air-pollution-rivals-power-plant-
emissions
46 https://www.foodandwaterwatch.org/2026/06/24/ai-data-center-boom-farmland/
47 https://www.clarity.io/blog/what-is-the-impact-of-ai-data-centers-on-air-quality-and-the-
environment
36
Common Harms
Low-Frequency Noise or LFN - They
frequently propagate through the air
and ground, penetrating buildings, with
significant impacts reported over 2 miles
from sources like industrial sites, wind
turbines, and transit hubs. LFN is not
common naturally as usually only from
pre- earthquakes and earthquakes.
Minimum safe buffer zone of 2-3 miles
needed for LFN, and vibration.
Electromagnetic Fields (EMF) -
Substations and high-voltage power lines
emit EMFs. EMF Electromagnetic
Frequencies - DCs generate non-ionizing
electromagnetic fields (EMFs), primarily in
the form of Extremely Low Frequency
(ELF) fields from power distribution
systems and radiofrequency (RF) signals
from wireless infrastructure. Surrounding
electrical infrastructure—including new
transmission lines and substations
required to power the sites—can
increase exposure to electromagnetic
fields (EMFs), which some community and
health groups monitor for potential links
to increased long-term health risks.48
While the science on EMF and human
health remains debated, many studies
over the years starting from 1979,49 link
chronic high exposure to increased
cancer risks, such as childhood leukemia.
Power quality issues (harmonic distortions)
from large data centers can also cause
electrical fluctuations that increase
indoor EMFs. Most people and property
buyers tend to devalue places with
substations nearby. An electrical
substation typically devalues nearby
Image Source:
https://www.sensear.com/markets/data-
centers/noise-levels-infographic
48 https://ehsciences.org/ai-data-center-health-impacts/
49 https://pmc.ncbi.nlm.nih.gov/articles/PMC12441653/
37
Common Harms
property by 2% to 10%, and up to 20% to
30% for homes directly bordering the
perimeter fence due to buzzing noise,
aesthetic blight, fear of health risks and
security lighting.
Audible Spectrum Noise - the heavy
equipment used to construct the facilities
are loud. Then, once they are up and
running, diesel generators plus heating,
ventilation, and air conditioning (HVAC)
systems create a constant hum that can
be audible to neighboring residents and
wildlife. DCs generate noise levels that
may exceed 90-95 decibels. Noise levels
above 85 decibels are harmful to
hearing. Depending on wind-direction
and topography, .5 to 1.5 miles safe
buffer zone distance is required for
Audible spectrum noise.
Zoning & Ordinances Noise Levels – Most
US municipalities only allow Construction
Noise Monday–Friday from 7:00 AM to
7:00 PM and Saturdays from 9:00 AM to
7:00 PM. Construction is prohibited on
Sundays and city-observed holidays.
Construction & Operation Property &
Equipment: Fixed outdoor mechanical
equipment (like AC units, pool pumps,
and spas) for most municipalities must
adhere to city decibel limits and are
typically expected to quiet down by
10:00 PM depending on the property
zone.
Human & Animal Harm:
Mental Health – Due to ongoing health
harms, noise and other documented
negative impacts, mood disorders, low-
no sleep, depression, anxiety, and other
metal health problems are common for
those near and in DC cities and towns.
Continuous exposure contributes to
Image Sources: Food and Water Watch
https://x.com/foodandwater/status/20271114
85147681007
Business Insider, Nov 2023
38
Common Harms
insomnia, hypervigilance, and clinical
depression. Many DC neighbors report
that the noise triggers severe anxiety
attacks and forces them to avoid their
own outdoor spaces.50 The rapid
replacement cycles of hardware create
millions of tons of e-waste, causing severe
neurodevelopmental and psychological
damages in vulnerable recycling
communities. Solastalgia (Loss of
Environment) - Rapid construction often
converts green or rural spaces into
industrial facilities and the destruction of
natural environments and loss of
community character creates a deep
sense of emotional loss and distress,
known as solastalgia.51 Removal of green
space, heat islanding/heat dome and
other impacts add to both mental and
physical health negative short- and long-
term impacts.
Chronic exposure to DC environmental
noise – It is linked to sleep disruption, high
blood pressure, elevated stress
hormones, and cognitive issues in
children.52 Inside the facilities, workers
face daily risks such as arc flash incidents,
exposure to harsh water treatment
chemicals or battery acid, extremes in
temperature, noise at damage levels, air
pollution, general high-stress, and
musculoskeletal strains from repetitive
lifting and maintaining heavy server
hardware.
Direct Physical Health – This report details
physical harmful impacts but this is
Image Source:
https://www.wired.com/story/the-data-
center-resistance-has-arrived/
Photo-Illustration: WIRED Staff; Getty Images
50 https://www.usnews.com/news/national-news/articles/2026-04-28/living-in-hell-data-center-
neighbors-grapple-with-noise-air-pollution
51 Data Centers: Examples of Adverse Health and Other Effects
52 https://phys.org/news/2026-06-ways-centers-endanger-local-communities.html
39
Common Harms
extensively growing in case and types to
the point that it is impractical to list all
possible medical problems as the list is
not inclusive and up to date as possible.
Particulate Matter, off-gassing and
hazardous exhaust during routine testing
and power outages adds to or can
create increases in respiratory diseases or
lengthen common viral or bacterial
respiratory infections. Particulate Matter
(PM) and air pollution are primary drivers
of respiratory morbidity, triggering or
worsening conditions like asthma,
Chronic Obstructive Pulmonary Disease
(COPD), and lung cancer.53 Microscopic
particles bypass natural defenses,
lodging deep within the lungs to cause
chronic inflammation and oxidative
stress. Poor air quality has substantial
impact on school attendance, gross
local productivity (workers taking more
sick days) and general decreases in
health requiring more or additional
medical treatments. problems, including
but not limited to:
o premature death in people with
heart or lung disease
o nonfatal heart attacks
o irregular heartbeat
o aggravated asthma
o decreased lung function
o increased respiratory symptoms, such
as irritation of the airways, coughing
or difficulty breathing.
Public Health Costs to Cities:
The negative impacts to local GDP,
health insurance, and social services due
to poor air quality and other related
environmental impacts is significant.
Image Sources: Legionella Bacteria
https://t-safe.com/t-safe-in-focus/what-is-
legionella-transmission-symptoms-and-
prevention-explained/
https://www.forbes.com/sites/victoriaforster/2
026/07/09/meta-ai-data-center-linked-to-rare-
bacteria-in-citys-water-system/
https://legionellacontrol.com/legionella/legio
nella-control-for-data-centres/
53 https://www.epa.gov/pm-pollution/health-and-environmental-effects-particulate-matter-pm
40
Common Harms
Frequency of illness, length of illness and
types of illnesses all can increase.
“Stay Inside” unsafe air quality days and
other DC pollution derived problems can
impact populations at risk (elderly,
special needs, young, pre-existing
conditions, etc.) driving up medical care
costs, social services costs, and
decreasing local business productivity
due to sick children not in school, sick
employees not at work, sick customers
not shopping and loss of valued citizens
due to moving away.54
Low Livability Leading to Blight and
Environmental Degradation:
Substantial to very severe increases in
costs (utilities bills similar to rent or
mortgage payments) and increases in
costs to growing lack of access to basic
utilities (water, waste, sewage, electricity,
fire and police services) are extensively
documented and are increasing at such
a substantial rate that it is impractical to
list average figures in this document.
Some communities are experiencing
polluted utility water, polluted ground
water, rolling-black outs to no electricity
provided as more extreme cases. This
creates a negative compounding
environmental spiral effect.
Long-term blight—characterized by
vacant lots, abandoned buildings, and
neglected infrastructure—causes severe
environmental harm by fostering toxic soil
Image Sources: Illustration Rose Wong
https://www.commonwealthfund.org/publica
tions/2023/jan/state-strategies-controlling-
health-care-costs-implementation-guides
https://www.caltech.edu/about/news/air-
pollution-and-the-public-health-costs-of-ai
https://arxiv.org/abs/2412.06288
54 https://www.next10.org/sites/default/files/2025-11/ai-environmental-public-health-costs.pdf
41
Common Harms
and water contamination, escalating
localized air pollution, and creating
havens for disease-carrying pests. These
decaying spaces act as localized
ecological disasters, disproportionately
threatening human and wildlife health.55
Soil and Groundwater
Contamination: Abandoned
residential and industrial sites
frequently leak hazardous
materials into the earth.56
Vector and Pest Infestations
Increase: Dilapidated structures
and unmaintained vegetation
create ideal breeding grounds for
disease vectors.57
Localized Air Quality Degradation
Urban Heat Dome Growth
Unforeseen Environmental Harms:
New environmental harms not already
documented could very well exist. All DC
project agreements should include a
caveat to account for this.
Data Centers Are Draining Public Funds &
Risking Global Financial Markets:
Data centers introduce a host of
environmental and health harms, and yet
states are forgoing many millions of
dollars in tax revenue to entice Big Tech
to set up shop. At the same time, the
federal government may shell out billions
Image Source:
https://www.foodandwaterwatch.org/2026/0
3/04/the-top-10-reasons-data-centers-must-
be-stopped/
Image Source:
https://www.barrons.com/articles/ai-data-
centers-backlash-stocks-8d564b5f
Photograph by Nell Pittman
55 https://pmc.ncbi.nlm.nih.gov/articles/PMC11349791/
56
https://www.urban.org/sites/default/files/publication/89491/2017.04.03_urban_blight_and_public_h
ealth_vprn_report_finalized.pdf
57 https://www.detroitgreentaskforce.org/post/the-state-of-blight-in-our-built-environment
42
Common Harms
to prop up the AI data center boom. In
August 2025, the Trump administration
pledged an $8.9 billion investment in Intel
stock to expand U.S. manufacturing of
semiconductors (a key component in
computers). Meanwhile, OpenAI
is calling on Trump for support to finance
its growth.58
Public Money spent on AI and DCs
means less for managing environmental
impacts and natural consequences.
Image Source:
https://roddubitsky.substack.com/p/the-
everything-bubble-ai-data-centers
58 https://www.foodandwaterwatch.org/2026/03/04/the-top-10-reasons-data-centers-must-be-
stopped/
43
Common Harms
“THE ‘SUPER INTELLIGENCE’ TECH
SEEKS IS ALREADY IMBEDDED IN
THE EARTH, NATURE AND GOD.
THE PROBLEM IS THEY ONLY
WANT A ‘SUPER INTELLIGENCE’
THEY CREATE AND CONTROL.”
- T R E L L A M D A V I S
44
Next Steps
CLARITY & SUSTAINABLITY
Full Clarity: Vague Data Center Developer proposals, and overly limited EIRs are
potentially misleading legally fraudulent on environmental impacts for many reasons.
Data Center impact communities’ resources including soil, plants, water, air, animals,
humans, electricity, and society and none of these represent a “Trade Secret”.
Transparency: NDAs with local or state governments are in fact a document that is
legally part of the public record. It can be requested by citizens or courts. Elected
Officials or proxy Non-governmental organizations like Economic Development
Organizations (EDOs) have little to no legal legs with NDAs and Data Centers. Courts
overwhelmingly rule them un-enforceable to outright illegal violation of Due Process,
Public Notice Requirements, Public Records Acts, and Freedom of Information Act.
Citizen advisory group for on-going active electorate participation in addition to
following Public Notice and Due Process requirements.
Proactive: Data Centers require specific Zoning. No “Industrial”, “By-right” or
mislabeling should be allowed. Data Centers must have their own unique “Data
Center Zone” criteria. Data Center Moratoriums – whether indefinite to 5 years, are
critical for understanding the real impacts on the environment and communities as
well as the financial viability of the companies that are building/owning them (whose
leaders have built dooms-day bunkers in other nations and remote island locations).
Require Revenue Per Acre metric studies to determine best use of land allotments
and competitive bidding (almost all DC proposals have no competitive bids).
Requirements & Sustainability: 100% Clean electricity self-generation (partnerships
with local farms agrivoltaics, curved low elevation wind turbines, rainwater
harvesting), use of salt-ion BESS batteries, removal of generator use, aggressive use of
passive cooling and ground cooling techniques, isolated locations for sound,
vibration and radiation emissions, zero public tax abatements or grants, committed
measurable local employment numbers for construction and operations, complete
GHGs and CO2 footprints with full coverage nature-based Carbon Credit offsets,
100% e-waste reuse programs, sorting of other waste streams with verified recycling
partners, impacts to city services (health care, fire, police, roads, etc.) fully funded,
audited quarterly reporting on shared resource inputs as well as waste outputs, on-
going monitoring based on extensive EIRs that address all areas of negative
environmental hazards, and punitive, “claw back”, or operations stoppage
reparations for not meeting agreements, abandoned sites or default on debts.
45