Artificial Intelligence Data Center Moratorium Act
- Sponsor
- Sen. Sanders, Bernard [I-VT]
- Committees
- Commerce, Science, and Transportation Committee (primary)
- Last action
- Mar 25, 2026
Bottom line
The bill seeks to halt the physical expansion of AI infrastructure in the U.S. and restrict its global spread until a robust regulatory framework for AI safety, economic equity, and environmental protection is established.
What it actually does
This bill imposes an immediate moratorium on the construction and upgrading of new or existing artificial intelligence data centers. The moratorium would remain in effect until Congress enacts comprehensive legislation addressing AI safety, economic benefits for workers, environmental protection, community approval for data centers, and a prohibition on government subsidies for such centers. It also mandates quarterly reports on AI data centers and establishes export controls on computing infrastructure hardware to countries without comparable AI safety laws.
Proponents argue
Proponents argue that unchecked AI development poses catastrophic risks to humanity, including massive job displacement, privacy erosion, and environmental harm, as warned by leading tech experts. They contend that a moratorium is necessary to create a "pause" for policymakers to enact safeguards, ensuring AI benefits society broadly rather than concentrating wealth and power, and preventing a race to the bottom on safety.
Opponents contend
Opponents would likely argue that the moratorium would severely stifle innovation, cripple the U.S. technology sector, and cede global leadership in AI to other countries. They might contend that such a drastic measure would lead to significant job losses in construction and tech, increase consumer costs, and that existing regulatory frameworks or industry self-regulation are sufficient or preferable to a blanket ban.
The bill is relatively short and clearly written, making it accessible for an informed reader to understand its core provisions and implications within a reasonable timeframe.
Section 3(a)
Definition of Artificial Intelligence Data Center
This section defines what constitutes an "artificial intelligence data center" for the purposes of the moratorium. It includes facilities on single or contiguous sites owned by the same entity that are used for developing or operating AI models at scale, or those with a power capacity over 20 megawatts that deliver 20 kilowatts or more to a single server rack or use liquid cooling systems.
Supporters argue
Supporters would argue that a clear, technically specific definition is crucial to ensure the moratorium effectively targets the infrastructure driving rapid, potentially dangerous AI development, while avoiding unintended impacts on smaller, non-AI-specific data centers.
Critics contend
Opponents might argue that the definition is overly broad or too restrictive, potentially capturing data centers that are not primarily focused on "at scale" AI development or that the technical thresholds are arbitrary and could hinder legitimate technological progress.
Tradeoffs
Balances the need for a precise scope to enforce the moratorium against the risk of inadvertently stifling broader data infrastructure development.
Section 3(b)
Moratorium on Construction and Upgrading of AI Data Centers
This provision establishes an immediate and indefinite halt on the construction or upgrading of any new or existing artificial intelligence data centers. This moratorium will remain in effect until Congress passes one or more laws that explicitly address specific AI safeguards and economic equity, and those laws include a provision that expressly terminates this moratorium.
Section 3(b)(1)
Conditions for Moratorium Termination
The moratorium can only be lifted once Congress enacts laws that establish a federal review and approval process for AI products to ensure safety, mandate that AI's economic gains benefit workers (preventing job displacement and sharing wealth), and impose strict environmental and community-focused standards for any future AI data centers. These future standards include no increased utility bills, no climate harm, community approval rights, no government subsidies, and the creation of union jobs with strong labor standards.
Section 3(c)
Quarterly Reporting and Verification on AI Data Centers
This section requires the Secretary of Energy to submit quarterly public reports to Congress on all artificial intelligence data centers. These reports must include detailed information on financial vehicles, water and energy usage, greenhouse gas emissions (including fenceline air quality), wastewater discharge, cooling chemicals, noise levels, wages, job creation (temporary and permanent), land agreements, and a certification that no government subsidies were used. The Secretary is also granted powers, including subpoenas and inspections, to verify compliance.
Section 4
Export Controls on Computing Infrastructure Hardware
This section mandates that the Secretary of Commerce prohibit the export, reexport, or in-country transfer of computing infrastructure hardware (like semiconductors, integrated circuits, computers, networking equipment) if it's intended for use in an AI data center or for training/deploying AI models at scale. This prohibition applies to any country that does not have laws comparable to the U.S. conditions for lifting the moratorium (i.e., AI safety review, worker benefits, environmental/community standards) or to any person within such a country.
The detailed technical specifications for what constitutes an "artificial intelligence data center," including power capacity thresholds (20 megawatts total, 20 kilowatts per rack) and specific cooling technologies (liquid circulation, submersion).
Section 3(a)
Why it matters:This level of detail is standard legislative practice for defining the scope of a bill's applicability, ensuring clarity and preventing loopholes. It's not necessarily evasive but requires careful reading.
Case for: Supporters would argue that these precise technical definitions are crucial to ensure the moratorium effectively targets the specific, high-power, advanced data centers primarily responsible for large-scale AI development, without inadvertently impacting other types of data infrastructure.
Case against: Opponents might argue that these specific thresholds could become quickly outdated as technology evolves, or that they are arbitrarily set and could be circumvented by slight design modifications, leading to an ineffective moratorium.
Estimated impact: Directly determines which facilities are subject to the moratorium, significantly impacting the investment and operational plans of major tech companies.
The specific, detailed conditions that any artificial intelligence data center built or upgraded after the moratorium is lifted must meet, including not increasing utility bills, not exacerbating climate change, requiring community approval, prohibiting government subsidies, and creating union jobs with prevailing wages and apprenticeship programs.
Section 3(b)(1)(C)
Why it matters:These are specific policy goals embedded within the broader framework for lifting the moratorium. Their detailed nature is typical for legislative requirements, but their placement within the "conditions for termination" rather than a separate "future standards" section might make them less prominent.
Case for: Proponents would argue that these detailed conditions are essential to ensure that any future AI infrastructure development is truly beneficial to society, environmentally sound, and economically equitable, preventing a return to unchecked growth.
Case against: Opponents would likely view these as highly prescriptive and potentially unachievable demands that would effectively make it impossible to ever lift the moratorium, creating a permanent ban on AI data center construction in the U.S.
Estimated impact: If the moratorium were ever lifted, these conditions would fundamentally reshape the economics, labor practices, and community relations for all future AI data center projects in the U.S.