Artificial Intelligence Environmental Impacts Act of 2026
- Sponsor
- Sen. Markey, Edward J. [D-MA]
- Committees
- Environment and Public Works Committee (primary)
- Last action
- Jun 9, 2026
Bottom line
This bill establishes a federal framework for studying, standardizing measurement of, and reporting on the environmental and energy impacts of artificial intelligence and its data centers.
What it actually does
This bill mandates that the Environmental Protection Agency (EPA) conduct a comprehensive study on the environmental and energy-related impacts of artificial intelligence (AI) and its associated infrastructure. It also requires the National Institute of Standards and Technology (NIST) to convene a multi-stakeholder consortium to develop measurement standards for AI's environmental impacts, and directs the EPA to establish a mandatory annual reporting system for large AI data centers. The bill aims to provide data and foster collaboration to address the environmental footprint of the rapidly growing AI sector.
Proponents argue
Supporters argue that the rapid growth of AI and its supporting data centers presents significant, yet poorly understood, environmental risks, including increased energy and water consumption, air and water pollution, and e-waste. This legislation is crucial for gathering the necessary data, developing consistent metrics, and fostering collaboration to mitigate these negative impacts while identifying and promoting AI applications that offer positive environmental benefits. It represents a proactive step towards responsible technological development.
Opponents contend
Critics might contend that the bill's requirements could impose substantial administrative and financial burdens on the AI industry, potentially hindering innovation or increasing operational costs. They may argue that existing environmental regulations are sufficient or that the broad scope of the mandated studies and reporting system is premature, given the evolving nature of AI technologies, and could lead to overregulation.
The bill is relatively short and clearly structured, making it accessible for an informed reader to understand its core provisions within a reasonable timeframe.
Section 4
EPA Study on AI Environmental and Energy Impacts
This provision requires the Administrator of the Environmental Protection Agency (EPA), in collaboration with several other federal agencies, to conduct a comprehensive study on the environmental and energy-related impacts of artificial intelligence (AI) and its related infrastructure. The study must be submitted to Congress and made publicly available within one year of the bill's enactment. It will examine the full lifecycle impacts of AI, including hardware manufacturing, model design and deployment, and data center operations, as well as the effects of design and location decisions, disaster resilience, and both positive and negative applications of AI. Public input, including regional hearings, is mandated as part of the study process.
Supporters argue
Supporters argue that this comprehensive study is an indispensable first step to establish a foundational understanding of AI's rapidly evolving environmental footprint. They contend that such data is critical for developing informed, evidence-based policies to mitigate negative impacts and strategically leverage AI's potential for environmental good. The multi-agency collaboration ensures a holistic and authoritative assessment.
Critics contend
Critics might argue that conducting such a broad and detailed study within a one-year timeframe could be overly ambitious and resource-intensive for federal agencies, potentially diverting resources from other critical environmental work. They may also suggest that the scope is too wide, making it challenging to produce actionable insights promptly.
Tradeoffs
This provision balances the urgent need for comprehensive data on AI's environmental impacts with the practical challenges of conducting a wide-ranging, inter-agency study within a tight deadline. It also navigates the tension between potential regulatory burdens on industry and the public's right to understand the environmental consequences of emerging technologies.
Section 5
NIST-Convened AI Environmental Impacts Consortium
This section directs the Director of the National Institute of Standards and Technology (NIST) to convene a consortium of diverse stakeholders. This group will include representatives from Indian Tribes, local governments, academia, civil society, and industry. The consortium's primary goal is to identify the necessary measurements, methodologies, and standards for accurately assessing and reporting the full range of environmental and energy-related impacts of artificial intelligence. It also aims to facilitate consistent reporting, develop open-source tools, and provide recommendations for mitigating negative impacts and promoting positive AI use cases. The consortium mandates equal representation among stakeholder groups, with specific representation from geographically distinct Tribal and local communities where AI data centers are located.
Section 6
EPA Reporting System for AI Data Centers
This provision establishes a mandatory annual reporting system for entities operating "artificial intelligence data centers," defined as facilities with a maximum rated power capacity or total peak power load exceeding 50 megawatts that house AI operations at scale. These "covered entities" must submit reports to the EPA detailing the full range of their environmental impacts. The EPA, in consultation with NIST, DOE, and the consortium, is tasked with developing specific reporting requirements, including guidelines for calculating energy and water consumption, local air and water pollution, electronic waste, and noise/light pollution. The reports must be prepared for public disclosure, and the EPA is authorized to seek corrective action and issue administrative penalties for non-compliance, with penalty amounts based on facility size and duration of non-compliance.
Section 7
Joint Agency Report to Congress
This section mandates a joint report to Congress from the Administrator of the EPA, the Secretary of Energy, and the Director of NIST. This report is due two years after the EPA finalizes the reporting requirements under Section 6(c) and after the NIST consortium has been convened. It must detail the main findings of the consortium, summarize the information reported by AI data centers under Section 6, provide recommendations for legislative or administrative actions to mitigate the negative environmental impacts of AI, and suggest beneficial use-cases for AI in environmental applications based on net-benefit assessments.
The bill defines an "artificial intelligence data center" as any facility or group of facilities on a single or connected site with a maximum rated power capacity or total peak power load exceeding 50 megawatts, which houses and operates infrastructure for AI inference, training, pre-training, fine-tuning, or deployment at scale.
Section 3(3)
Why it matters:This is standard legislative practice to define key terms upfront to ensure consistent interpretation throughout the bill. However, its specific thresholds and criteria are highly consequential for the bill's reach.
Case for: Proponents argue that this specific definition provides clarity and appropriately limits the scope of the reporting requirement to large-scale AI operations, focusing regulatory efforts where environmental impacts are likely most significant. The 50MW threshold targets major contributors to energy and water consumption.
Case against: Critics might contend that the 50MW threshold is arbitrary and could exclude a substantial number of smaller data centers which, in aggregate, could still have significant environmental impacts. They might argue that focusing only on 'at scale' deployment could miss distributed AI computing impacts.
Estimated impact: This definition directly determines the universe of entities subject to the EPA's mandatory reporting requirements and potential administrative penalties under Section 6, significantly shaping the practical reach of the bill.
This subsection grants the Administrator of the Environmental Protection Agency (EPA) the authority to seek corrective action and, if reports are not submitted within six months of their due date, to issue administrative penalties to covered entities that fail to comply with the annual reporting requirements. The penalty amount is to be determined by the Administrator based on the total size of the applicable facilities and the length of non-compliance.
Section 6(e)
Why it matters:It is standard legislative practice to place enforcement details after the main obligations, but the specific authority to levy administrative penalties is a significant power granted to the EPA, which could be politically sensitive.
Case for: Proponents argue that this enforcement authority is crucial for ensuring compliance with the reporting requirements. Without the ability to impose penalties, the reporting system could be ineffective, undermining the bill's goals of transparency and data collection. It provides necessary teeth to the regulatory framework.
Case against: Industry critics might argue that granting the EPA broad authority to determine penalty amounts, based on criteria like 'total size' and 'length of noncompliance,' could lead to arbitrary or overly punitive enforcement. They may contend that such penalties could disproportionately impact businesses and create an unpredictable regulatory environment.
Estimated impact: This provision creates a strong financial incentive for covered entities to comply with the reporting requirements, potentially leading to significant financial impacts for non-compliant operators through administrative penalties.