S. 4727

[Congressional Bills 119th Congress]
[From the U.S. Government Publishing Office]
[S. 4727 Introduced in Senate (IS)]

<DOC>

119th CONGRESS
2d Session
S. 4727

To require the Administrator of the Environmental Protection Agency to 
carry out a study on the environmental impacts of artificial 
intelligence data centers and associated energy infrastructure, to 
require the Director of the National Institute of Standards and 
Technology to convene a consortium on such environmental impacts, and 
to require the Administrator to develop a reporting system for the 
reporting of the environmental impacts of artificial intelligence, and 
for other purposes.

_______________________________________________________________________

IN THE SENATE OF THE UNITED STATES

June 9, 2026

Mr. Markey (for himself, Mr. Van Hollen, Mr. Wyden, Mr. Booker, and Mr. 
Welch) introduced the following bill; which was read twice and referred 
to the Committee on Environment and Public Works

_______________________________________________________________________

A BILL

To require the Administrator of the Environmental Protection Agency to 
carry out a study on the environmental impacts of artificial 
intelligence data centers and associated energy infrastructure, to 
require the Director of the National Institute of Standards and 
Technology to convene a consortium on such environmental impacts, and 
to require the Administrator to develop a reporting system for the 
reporting of the environmental impacts of artificial intelligence, and 
for other purposes.

Be it enacted by the Senate and House of Representatives of the 
United States of America in Congress assembled,

SECTION 1. SHORT TITLE.

This Act may be cited as the ``Artificial Intelligence 
Environmental Impacts Act of 2026''.

SEC. 2. FINDINGS.

Congress finds the following:
(1) Multiple estimates indicate that the amount of 
computational power being used for artificial intelligence 
applications has increased rapidly over the last decade.
(2) According to the Department of Energy, data center 
energy demand has tripled in the last decade and is expected to 
double or triple again by 2028.
(3) Accelerating use of artificial intelligence greatly 
increases energy consumption due to the power utilization of 
computer hardware required for training and operating 
artificial intelligence models, despite ongoing efficiency 
gains in both artificial intelligence models and hardware.
(4) Rapid growth in data center infrastructure supporting 
artificial intelligence and other computing-intensive 
technologies, including cooling systems and backup power 
equipment, can contribute to air and water pollution, increased 
energy demand, increasing water scarcity, and land-use changes.
(5) According to the Department of Energy, hyperscale 
facilities are projected to consume between 16,000,000,000 and 
33,000,000,000 gallons annually by 2028.
(6) Resource and energy-intensive manufacturing processes 
are required for the hardware that runs artificial intelligence 
and other computing-intensive technologies, leading to 
significant environmental impacts.
(7) Electricity prices have already risen significantly as 
demand from data centers grows, and are projected to continue 
rising rapidly.
(8) According to the Energy Information Agency, between 
January, 2025, and December, 2025, household electricity prices 
increased by as much as 13 percent nationwide.
(9) Prices are projected to rise by another 25 percent in 
certain places over the next five years due to increased demand 
from data centers.
(10) Yearly increases in electronic waste (known as ``e-
waste'') pose environmental and health risks and will likely be 
exacerbated by outdated and discarded hardware used for 
artificial intelligence and other computing-intensive 
technologies.
(11) Certain applications of artificial intelligence may 
have direct and indirect positive environmental impacts, 
including optimizing systems for energy efficiency, developing 
renewable energy, advancing planetary systems research, 
enabling discovery of new materials, and automatically 
monitoring environmental changes. Applications of artificial 
intelligence also have direct and indirect negative 
environmental impacts, including rebound effects, behavioral 
impacts, and accelerating high-pollution activities.
(12) Different communities and regions will experience 
disparate effects from data center infrastructure, with risks 
ranging from higher energy costs to more adverse environmental 
effects, and with certain communities at greater risk from 
cumulative negative impacts, such as low-income communities, 
Black and Brown communities, Indigenous communities, and rural 
communities.
(13) Various options exist to reduce the negative 
environmental impacts of artificial intelligence, including 
using more energy-efficient and water-efficient models, 
hardware, and data centers, using renewable and co-located 
energy, and examining the impacts of artificial intelligence 
applications.
(14) Promoting transparency on energy use and environmental 
impacts and developing and maintaining accurate environmental 
impact metrics may help mitigate negative environmental impacts 
of the rapid growth in artificial intelligence use, while 
promoting artificial intelligence uses with net positive 
environmental impacts.

SEC. 3. DEFINITIONS.

In this Act:
(1) Administrator.--The term ``Administrator'' means the 
Administrator of the Environmental Protection Agency.
(2) Artificial intelligence.--The term ``artificial 
intelligence'' has the meaning given the term in section 5002 
of the National Artificial Intelligence Initiative Act of 2020 
(15 U.S.C. 9401).
(3) Artificial intelligence data center.--The term 
``artificial intelligence data center'' means any facility, or 
group of facilities that--
(A) is located on a single site or on contiguous, 
adjacent, or otherwise connected sites;
(B) has a maximum rated power capacity or total 
peak power load in excess of 50 megawatts; and
(C) houses and operates infrastructure for the 
purpose of artificial intelligence inference, training, 
pre-training, fine-tuning, or deployment at scale.
(4) Artificial intelligence model.--The term ``artificial 
intelligence model'' means a component of an information system 
that implements artificial intelligence technology and uses 
computational, statistical, or machine-learning techniques to 
produce outputs from a given set of inputs.
(5) Artificial intelligence system.--The term ``artificial 
intelligence system'' means any data system, software, 
hardware, application, tool, or utility that operates in whole 
or in part using artificial intelligence.

SEC. 4. STUDY ON ENVIRONMENTAL AND ENERGY-RELATED IMPACTS OF ARTIFICIAL 
INTELLIGENCE.

(a) In General.--Not later than 1 year after the date of enactment 
of this Act, the Administrator, in collaboration with the Secretary of 
Energy, the Director of the National Institute of Standards and 
Technology, the Director of the Office of Science and Technology 
Policy, the Secretary of the Interior, and the Federal Energy 
Regulatory Commission, shall carry out, submit to Congress, and make 
publicly available a report describing the results of, a comprehensive 
study on the environmental and energy-related impacts of artificial 
intelligence and related infrastructure.
(b) Requirements.--The study required under subsection (a) shall 
include an examination of the following issues and a forecast for how 
those issues are expected to change during the 2-year, 5-year, and 10-
year periods beginning on the date on which the study is completed:
(1) Full lifecycle impacts of artificial intelligence.--The 
energy and water use, effects on air and water quality, effects 
on local ecosystems, local noise and light impacts, energy cost 
impacts, and the direct, indirect, and value chain greenhouse 
gas emissions associated with the full lifecycle of artificial 
intelligence, including effects and impacts associated with--
(A) hardware needed for artificial intelligence, 
including the extraction of raw materials, 
manufacturing, electronic waste, and transportation 
between all stages associated with that hardware;
(B) the design, development, deployment, and use of 
artificial intelligence models; and
(C) the operation of artificial intelligence data 
centers to power artificial intelligence, including 
effects relating to--
(i) energy infrastructure, including new or 
existing grid infrastructure, behind-the-meter 
primary onsite power sources, and backup onsite 
power sources;
(ii) water use, high-temperature water 
discharge, and the use of chemical biocides;
(iii) land use for artificial intelligence 
data centers and related energy infrastructure, 
particularly with respect to scenic, 
historical, and cultural landscapes and units 
of the National Park System or locations 
included on the National Register of Historic 
Places; and
(iv) cooling systems and other 
infrastructure.
(2) Design and location.--The effect that design and 
deployment decisions for artificial intelligence models and 
associated artificial intelligence data centers have on the 
impacts described in paragraph (1), including design and 
deployment decisions with respect to--
(A) the artificial intelligence model to be used;
(B) hardware;
(C) location;
(D) energy mix;
(E) behind-the-meter generation;
(F) the use of power purchase agreements;
(G) demand flexibility;
(H) cooling system methodology; and
(I) artificial intelligence data center user 
ownership and leasing arrangements.
(3) Disaster resilience.--The impacts of artificial 
intelligence data centers on disaster resilience, including 
with respect to local water and energy stress during extreme 
weather events.
(4) Positive impacts.--The potential positive environmental 
and energy-related impacts associated with applications of 
artificial intelligence, which may include optimizing systems 
for energy efficiency, developing renewable energy, advancing 
planetary systems research, enabling discovery of new 
materials, and automatically monitoring environmental changes.
(5) Negative impacts.--The negative environmental and 
energy-related impacts associated with applications of 
artificial intelligence and related infrastructure, which may 
include direct impacts from data center construction and 
operation, higher local energy costs, rebound effects, 
behavioral impacts, and accelerating high-pollution activities, 
such as fossil fuel extraction.
(6) Updated study results.--The results of the updated data 
center study carried out under section 453(e)(2) of the Energy 
Independence and Security Act of 2007 (42 U.S.C. 17112(e)(2)).
(7) Other environmental and energy-related impacts.--Other 
environmental and energy-related impacts, as determined by the 
Administrator, in consultation with the heads of Federal 
agencies described in subsection (a).
(c) Public Input.--
(1) Public comment required.--In conducting the study 
required under subsection (a), the Administrator shall solicit 
and consider public comments.
(2) Public hearings.--Not later than 1 year after the date 
of enactment of this Act, the Administrator shall convene 
public hearings with respect to the environmental and energy-
related impacts of artificial intelligence and related 
infrastructure in each region of the Environmental Protection 
Agency to capture unique regional considerations.

SEC. 5. ARTIFICIAL INTELLIGENCE ENVIRONMENTAL AND ENERGY-RELATED 
IMPACTS CONSORTIUM.

(a) In General.--The Director of the National Institute of 
Standards and Technology shall, in consultation with the Administrator, 
the Secretary of Energy, and such others as the Director considers 
appropriate, convene a consortium of stakeholders, including members 
from Indian Tribes, units of local government, academia, civil society, 
and industry, to identify the future measurements, methodologies, 
standards, and other appropriate needs, in order to measure and report 
the full range of environmental and energy-related impacts of 
artificial intelligence.
(b) Location.--The Director may determine the location of the 
consortium within the National Institute of Standards and Technology.
(c) Goals.--The goals of the consortium shall include the 
following:
(1) Facilitating consistent, comparable reporting on the 
environmental impacts of the full lifecycle of artificial 
intelligence models and artificial intelligence systems.
(2) According to technical feasibility, the development or 
cataloging of open-source software and hardware tools and other 
resources designed to facilitate the measurement of 
environmental impacts of artificial intelligence models and 
artificial intelligence systems.
(3) Providing recommendations on how to mitigate the 
negative environmental impacts and deploy the positive use 
cases, of artificial intelligence.
(d) Composition.--The consortium shall--
(1) be composed of representatives with equal 
representation between each stakeholder group; and
(2) include at least 3 representatives each from--
(A) geographically distinct Tribal communities 
within which artificial intelligence data centers 
reside; and
(B) geographically distinct cities or counties 
within which artificial intelligence data centers 
reside.

SEC. 6. SYSTEM FOR REPORTING OF ENVIRONMENTAL AND ENERGY-RELATED 
IMPACTS OF ARTIFICIAL INTELLIGENCE.

(a) Definition of Covered Entity.--In this section, the term 
``covered entity'' means an entity that operates an artificial 
intelligence data center.
(b) Reports to EPA.--A covered entity shall annually submit to the 
Administrator a report that describes the full range of environmental 
impacts of the artificial intelligence data centers of the relevant 
entity in accordance with this section.
(c) Report Requirements.--
(1) In general.--The Administrator, in consultation with 
the Director of the National Institute of Standards and 
Technology, the Secretary of Energy, the consortium convened 
under section 5, and such other entities as the Administrator 
considers appropriate, shall develop requirements for covered 
entities to report under subsection (b) the full range of 
environmental impacts of artificial intelligence data centers, 
which shall--
(A) include guidelines on how to calculate and 
report--
(i) the resource consumption and intensity 
with respect to energy use, including 
electricity use and capacity, resource mix, 
power sourced from the grid, power sourced from 
on-site generation, power sourced from backup 
systems, power usage effectiveness, performance 
per watt, energy reuse, and carbon intensity 
per task;
(ii) the resource consumption and intensity 
with respect to water use;
(iii) local air and water pollution;
(iv) electronic waste associated with the 
full lifecycle of artificial intelligence 
models and hardware;
(v) local noise and light pollution 
impacts; and
(vi) other positive and negative impacts of 
artificial intelligence use, as determined by 
the Administrator;
(B) require sufficient detail to allow the 
Administrator to verify the accuracy and completeness 
of the submission; and
(C) require reports to be prepared in a manner that 
enables public disclosure under subsection (d).
(2) Public comment.--Before finalizing the requirements 
under paragraph (1), the Administrator shall solicit comments 
from the public on a draft version of the requirements.
(3) Publication.--The Administrator shall make available to 
the public all requirements developed under paragraph (1).
(d) Availability.--The Administrator shall, to the maximum extent 
practicable, make the information reported pursuant to subsection (b) 
available on a publicly available website, except for any information 
that is exempt from disclosure pursuant to section 552(b)(4) of title 
5, United States Code.
(e) Enforcement.--If a covered entity fails to comply with the 
reporting requirements under this section, the Administrator shall seek 
corrective action and, if no reports under subsection (b) are submitted 
within 6 months of the date on which a report was due, may issue an 
administrative penalty in an amount determined by the Administrator 
that is based on--
(1) the total size of the applicable facilities of the 
covered entity for which reports have not been submitted; and
(2) the length of noncompliance.
(f) Savings Provision.--Nothing in this section abridges, alters, 
or modifies the obligations of the Administrator to disclose 
information pursuant to any other applicable law, including section 552 
of title 5, United States Code (commonly known as the ``Freedom of 
Information Act'').

SEC. 7. REPORT TO CONGRESS.

After the Director of the National Institute of Standards and 
Technology convenes the consortium under section 5 and not later than 2 
years after the date on which the Administrator completes development 
of the requirements for reporting under section 6(c), the 
Administrator, the Secretary of Energy, and the Director of the 
National Institute of Standards and Technology shall jointly submit to 
Congress a report detailing the following:
(1) The main findings of the consortium convened under 
section 5.
(2) A description of the information reported under section 
6.
(3) Recommendations for legislative or administrative 
action to mitigate the negative impacts of artificial 
intelligence.
(4) Recommended use-cases for artificial intelligence in 
environmental applications based on net-benefit assessments.
<all>