How the Proposed Federal Moratorium on AI Data Centers Would Work

The physical infrastructure required to power the artificial intelligence boom has become the focal point of a major legislative push in Washington. Senator Bernie Sanders and Representative Alexandria Ocasio-Cortez have introduced the Artificial Intelligence Data Center Moratorium Act, a sweeping proposal that would temporarily halt the construction and expansion of large-scale AI data centers across the United States. The legislation aims to pause the rapid buildout of computing facilities until the federal government can establish comprehensive safeguards regarding energy consumption, consumer utility costs, and environmental impact. While the bill faces a steep climb in a divided Congress, it represents the most aggressive federal attempt to date to regulate the physical footprint of the generative AI industry.[1][2]

At the core of the proposed legislation is a strict threshold targeting the most energy-intensive facilities. The moratorium would apply specifically to data centers with a peak power demand of 20 megawatts or more that utilize high-performance server racks or liquid cooling systems designed for training and operating AI models at scale. Traditional cloud storage facilities and smaller enterprise data centers—such as those used by hospitals, schools, and municipal governments—would be exempt from the pause. By drawing this technical distinction, the bill's authors are attempting to isolate the unique infrastructural demands of advanced machine learning from everyday internet operations.[2]

The urgency behind the proposal stems from unprecedented strain on the U.S. electrical grid. Domestic electricity consumption hit a record high in 2024, driven in large part by the explosive growth of hyperscale computing facilities. A single AI-focused data center can consume as much electricity as 100,000 households, forcing utility providers to rapidly revise their long-term capacity forecasts. Lawmakers argue that without federal intervention, this surging demand will inevitably lead to higher utility bills for everyday consumers, as power companies pass on the costs of grid upgrades and new power generation.[1]

Under the terms of the Sanders-Ocasio-Cortez bill, the nationwide pause would remain in effect until Congress passes a comprehensive regulatory framework. This framework must ensure that AI data centers do not increase electricity prices for local communities, that developers utilize union labor and pay prevailing wages, and that the facilities do not compromise environmental standards. The legislation also includes a transparency mandate, requiring the Department of Energy to collect and publish detailed operational data from data center operators, including their water usage, wastewater discharge, and overall energy consumption.[1][2]

Beyond domestic construction, the legislation seeks to prevent tech companies from simply moving their operations offshore to avoid regulation. The bill includes a provision that would ban the export of advanced AI computing infrastructure—such as specialized graphics processing units (GPUs)—to any country that lacks equivalent safety and environmental laws. This export control mechanism is designed to stop a "race to the bottom," ensuring that the environmental and labor standards demanded by the U.S. government cannot be bypassed by relocating server farms to jurisdictions with looser regulations.[1][2]

While the federal moratorium has been characterized by some critics as a fringe idea, it mirrors a rapidly growing grassroots movement across the country. Since August 2025, more than 100 local communities and municipalities have enacted temporary bans or zoning restrictions on data center buildouts. Furthermore, at least 15 states—including New York, Georgia, and South Carolina—have debated or advanced statewide moratorium proposals as local officials grapple with the noise, water usage, and visual blight associated with hyperscale facilities. In New York, for instance, lawmakers have pushed a bill that would prohibit state agencies from issuing permits for any data center capable of using 20 megawatts or more until new environmental regulations are drafted.[1][3]

The tech industry and its political allies have mounted a fierce defense against the concept of a moratorium, framing the rapid deployment of AI infrastructure as an economic and national security imperative. Opponents argue that halting data center construction would effectively cede global leadership in artificial intelligence to geopolitical rivals, particularly China. Democratic Senator John Fetterman and Interior Secretary Doug Burgum have both publicly rejected the idea of a pause, with Fetterman characterizing a moratorium as waving a "surrender flag" in the global technology race.[3]

The White House has also signaled resistance to blunt moratoria, instead favoring a strategy that encourages innovation while mitigating grid impacts. The administration recently issued guidance suggesting that Congress should preempt state-level AI laws that it views as overly burdensome. Rather than halting construction, federal energy officials and bipartisan policy groups have advocated for a "Bring Your Own Generation" approach. Under this model, hyperscalers like Google, Microsoft, and Amazon are encouraged—and in some proposals, required—to build or purchase new, dedicated sources of clean power to offset their massive energy draws, rather than siphoning capacity from the existing public grid.[3]

Tech giants have already begun making voluntary commitments to address these concerns. Following meetings at the White House, several major AI developers agreed to cover the expenses of infrastructure upgrades and invest in advanced nuclear, geothermal, and solar projects to power their operations. However, proponents of the Moratorium Act argue that voluntary pledges are insufficient to protect ratepayers and the environment. They maintain that the sheer scale and speed of the AI revolution require binding, democratic oversight rather than relying on the goodwill of technology executives.[1]

The debate over the physical infrastructure of AI also touches on broader societal anxieties regarding automation and economic displacement. The Sanders-Ocasio-Cortez legislation explicitly cites the potential for AI to hollow out the American workforce, referencing projections that automation could displace tens of millions of jobs over the next decade. By tying the physical construction of data centers to broader demands for labor protections and equitable wealth distribution, the bill attempts to use infrastructure permitting as a lever to force a national conversation about the economic consequences of artificial intelligence.[1][2]

Even if the Artificial Intelligence Data Center Moratorium Act fails to pass both chambers of Congress, its introduction serves as a critical signaling mechanism. It provides a comprehensive legislative blueprint for state and local governments that are currently fighting zoning battles on a case-by-case basis. By clearly defining what constitutes an "AI facility" and outlining specific environmental and economic safeguards, the bill offers a standardized framework that state legislatures can adapt as they negotiate with hyperscalers.

Ultimately, the clash over data center construction highlights a fundamental tension in the AI era: the frictionless, digital nature of artificial intelligence relies on a massively resource-intensive physical reality. As the demand for computing power continues to accelerate, policymakers at all levels of government are being forced to balance the promise of technological supremacy against the tangible realities of strained power grids, rising utility bills, and community pushback. The proposed moratorium represents the opening salvo in what is likely to be a decade-long legislative battle over who bears the physical costs of the AI revolution.[3]