A call from Sen. Bernie Sanders for a moratorium on new artificial intelligence data centers drew swift pushback on The Bottom Line, where Heritage Foundation senior economist Peter St. Onge framed the idea as government overreach. The exchange highlighted a growing national debate over how to balance AI growth with concerns over energy use, jobs, and community impact.
Sanders, an independent from Vermont, has argued that the rapid buildout of AI infrastructure should pause while policymakers assess its effects. St. Onge warned that such a move could slow innovation and weaken U.S. competitiveness during a global race to deploy advanced computing.
What the Proposal Seeks
The concept of a moratorium would halt approvals for new AI-focused data centers for a set period. Supporters say a pause would give time to study safety risks, labor impacts, water use, and grid strain. They argue that rules have not kept up with the scale of AI computing and the resources it consumes.
Backers also point to the pace of recent development. New facilities are being announced across the country as technology firms seek more computing power for large models and services. Community groups in several states have pressed officials to secure stronger protections on siting, noise, and resource use before more projects move forward.
Critics Warn of Overreach
On The Bottom Line, St. Onge said Sanders’ approach hands too much power to Washington and risks freezing private investment.
St. Onge called the proposal a “power grab” that could stall growth and send projects overseas.
He argued that the market is already adjusting to pressures on the grid and that utilities, regulators, and companies can coordinate without a nationwide halt. Industry groups often contend that speed matters and that delays could shift new capacity to other countries that are competing for AI jobs and capital.
Energy, Water, and Local Infrastructure
Data centers have long drawn concern for electricity demand and water use. The International Energy Agency has warned that global data center electricity consumption could roughly double by the mid-2020s as AI workloads rise. Researchers estimate data centers account for around 2% of U.S. electricity use, with higher shares in clusters such as Northern Virginia.
Utilities say they face a surge of interconnection requests tied to AI and chip manufacturing. That creates planning challenges for power plants, transmission, and substations. In some regions, local officials are weighing setbacks, design rules, or temporary pauses to match growth with available infrastructure.
Jobs, Automation, and Competitiveness
Sanders has raised alarms about the effect of AI on workers, calling for stronger guardrails and training. Labor groups want assurances that new investment produces stable, well-paid jobs and that AI adoption does not erode wages. Business groups counter that AI can raise productivity and open new roles, especially in construction, grid upgrades, and operations tied to data centers.
Economists split on the balance of gains and risks. Some see near-term reshuffling of tasks but long-term growth if firms deploy AI responsibly. Others warn that concentrated benefits could widen inequality without policy action on education, social insurance, and bargaining power.
By the Numbers
- Global data center electricity demand could double by the mid-2020s, according to the International Energy Agency.
- Data centers use an estimated 2% of U.S. electricity, with rapid growth expected in AI-heavy regions.
- Water needs vary widely by site and cooling design, prompting calls for stricter reporting and local review.
What Comes Next
Any national pause would face legal and political hurdles. Congress would have to define which projects qualify as “AI data centers,” how long a pause lasts, and what standards must be met to lift it. States could also move on permitting, siting, and reporting rules that shape how quickly new capacity comes online.
For now, the debate is sharpening. Sanders and allies want time to evaluate risks to workers and communities. St. Onge and other critics argue that a sweeping freeze would slow innovation and cede ground to rivals. The next phase will likely hinge on concrete bill text, utility planning studies, and whether industry offers measurable commitments on energy efficiency, water use, and job quality.
The public should watch three signals: whether Congress takes up a defined pause with clear criteria, how grid operators plan for AI-driven demand, and whether companies adopt more efficient hardware and cooling. The outcome will shape where the next wave of AI infrastructure lands and who benefits from it.
