On March 25, 2026, Sen. Bernie Sanders (I-Vt.) and Rep. Alexandria Ocasio-Cortez (D-N.Y.) announced the AI Data Center Moratorium Act (S.4214 in the Senate): an immediate federal moratorium on new AI data centers in the United States until national safeguards are in place, paired with a ban on exporting American AI computing infrastructure to countries that lack comparable protections.
The timing is aimed squarely at the buildout frenzy. The same week, OpenAI was reported to be pursuing a $122 billion raise at an $852 billion valuation, with most of that money bound for compute. Rather than regulate models, the two progressive lawmakers are targeting the physical layer — land, water, and the electric grid — making this the most direct legislative challenge to the AI construction boom Congress has produced so far.
What the Bill Demands
According to Sanders’ press release, the moratorium lifts only when three national safeguards are all in place: AI must be safe and effective for the public; the economic gains must flow to workers and families rather than being captured by “a small group of wealthy owners”; and AI development must not raise electricity rates for households and small businesses, harm communities, or destroy the environment.
This is not a technical-standards bill. It writes social preconditions into law: prove the benefits are distributable and the externalities controllable, then talk about speed. The regulated object is not the model but the infrastructure the model runs on.
Why Data Centers Are the Target
Every unit of AI expansion eventually becomes steel and megawatts: training and inference need GPU clusters, and clusters need land, water, and firm power. The release notes that more than 100 local communities across the country have already enacted their own data center moratoriums, and that proposals for statewide moratoriums are moving in 12 states. Local resistance has already crystallized; the federal bill is an attempt to institutionalize it.
Sanders’ statement does not hedge: “The American people do not want Big Tech to move at breakneck speed just so CEOs can make a nickel more,” and the Trump administration “has no plans” to deal with the disruption AI could bring to workers and families. AOC framed it as a question of distribution: “The AI future doesn’t need to be an inevitability — it needs to be a choice,” arguing that healthcare, housing, education, and a livable planet are fundamental rights that must not be subordinated to profit-driven algorithms.
The Export Control Extension
The bill’s second leg is discussed far less: it bans US exports of AI computing infrastructure to any country that has not enacted comparable safeguards. That extends a domestic pause into a foreign-policy instrument, structurally similar to chip export controls — compute infrastructure is treated as a strategic asset whose export depends on the recipient’s regulatory regime.
If anything like this clause ever lands, the AI infrastructure supply chain picks up a new political risk premium: overseas site selection, cross-border equipment procurement, and GPU delivery schedules all get repriced.
Local Backlash and Shifting Lab Rhetoric
The release also charts how industry rhetoric has moved. In 2023, more than 1,000 industry leaders and scientists — including Elon Musk, Yoshua Bengio, and Stuart Russell — signed the call for a six-month pause on AI training. In December, Musk said he had seen “a lot of AI nightmares” and would “certainly slow down AI and robotics.” In January, DeepMind CEO Demis Hassabis said he would support a pause if other major companies did the same. In February, Anthropic CEO Dario Amodei said he would be “absolutely in favor of trying” to slow down.
So “pause” is no longer the vocabulary of fringe advocates; frontier lab leaders now publicly keep the option open. The remaining fight is over who operates the brake, and through which mechanism.
What It Means for Developers and Product Teams
Realistically, this bill is unlikely to pass in the current Congress, but it changes the risk math. First, data center siting, energy procurement, and capacity planning now sit squarely inside policy risk; spreading capacity across jurisdictions is a reasonable hedge. Second, if “electricity rates shall not rise” becomes a legislative template, energy cost graduates from a finance problem to a licensing problem. Third, the export-control clause is a reminder that the geopolitical map of the AI supply chain is still expanding, and inference capacity deployed overseas may face new scrutiny.
Sources
- Sanders, Ocasio-Cortez Announce AI Data Center Moratorium Act — U.S. Senate
- S.4214 Artificial Intelligence Data Center Moratorium Act — Congress.gov
- Sanders, Ocasio-Cortez push bill to impose AI data center moratorium — AP News
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.
