This bill would pause construction or upgrades of new "artificial intelligence data centers" until Congress enacts one or more laws that set safety, economic, labor, community, and environmental rules for large AI computing facilities. The bill defines covered data centers by site control and by either use for AI model development at scale or by size and design (over 20 megawatts total peak power, delivering 20 kilowatts or more to a single rack, or using liquid cooling).
The moratorium stays in place until laws are passed that require: federal review and approval of AI products before release; policies to prevent AI-driven job loss and share economic gains with workers; limits so new centers do not raise consumer utility bills, worsen climate change, or harm the environment; community approval rights; no government subsidies for new centers; and creation of union jobs with prevailing wages and apprenticeship requirements. The Department of Energy (DOE) must submit quarterly public reports on AI data centers with specified information (energy, water, emissions, jobs, wages, subsidies, contracts, cooling chemicals, noise, and more) and is authorized to issue subpoenas, interrogatories, and inspections for verification. The bill also directs the Commerce Secretary to prohibit exports, reexports, and in-country transfers of computing infrastructure hardware for AI end-uses to countries that lack comparable statutes and to persons in those countries.
No publicly available information on the bill's costs or budgetary estimates. The bill requires the Department of Energy to produce quarterly public reports and authorizes verification tools (subpoenas, interrogatories, inspections), but it does not include federal cost or budget projections.
The bill text cites statements from AI researchers, industry leaders, and a large open letter calling for pauses or stronger regulation of advanced AI development. It frames the moratorium as a way to ensure AI product safety, protect workers, share economic gains, protect communities and the environment, and require democratic oversight before large-scale AI infrastructure expands.
No publicly available information.