Data centre moratoriums have moved from the fringe to the mainstream of American policy in little over a year. Moratorium activity rose from about 1% of local data centre decisions in early 2025 to 28% in 2026, with 313 measures reaching local governments in 138 jurisdictions.
After the midterms, roughly 40% of planned US data centre capacity is in states whose incoming leaders want to pause or restrict it. The justification is always the same: electricity bills, grid reliability and water.
The problem is that moratoriums address none of these three challenges and, in several cases, make them worse.
The case for a moratorium assumes that data centres exhaust the energy available on the grid. The data suggest something different.
US data centres used about 4.4% of national electricity in 2023, and Lawrence Berkeley National Laboratory projects a range of 9.5% to 15.3% by 2030.
That is significant growth, but the binding constraint is capacity during a small number of peak hours. The Department of Energy has noted that providers can currently meet data centre demand on about 350 days.
A blanket moratorium treats a peak capacity problem as if it were a structural energy deficit. The more efficient tool is a flexible interconnection system, under which a facility connects more quickly in exchange for reducing its draw by using on-site generation or storage during grid stress events.
Poorly designed tariffs
The second argument claims that households subsidise the infrastructure built for hyperscalers. That risk is real when tariffs are poorly designed, but it has a regulatory solution that already works.
AEP Ohio’s data centre tariff, approved in July 2025, requires new facilities above 25 MW to pay for at least 85% of contracted demand for up to twelve years, post collateral, and pay an exit fee equal to 36 months of minimum charges if they leave early.
National evidence points in the same direction. Lawrence Berkeley National Laboratory finds that rising data centre load was associated with slightly lower residential retail rates, by about 0.4% for every 10% increase in load between 2015 and 2024.
Water is the most emotive argument and technically the weakest
Where bills have risen, the cause is a supply side that is unable to respond. A moratorium freezes demand and leaves that bottleneck untouched.
Water is the most emotive argument and technically the weakest. On-site water use depends almost entirely on cooling design. The water footprint of computing is mainly a function of the generation mix.
Any jurisdiction concerned about water should set a water usage effectiveness standard and favour dry-cooled or low-water generation. A ban on the buildings does neither.
A regressive tax on innovation
Restricting supply in a market already short of capacity has evident consequences. Compute prices rise, incumbents with operating capacity capture scarcity rents, and smaller firms, universities and startups that rent capacity bear the cost.
A moratorium works like a regressive tax on innovation, collected by those who already own the infrastructure.
The strongest counterargument is that projects simply relocate.
Relocation still has a cost. In this industry, time to power is the scarcest variable, and a project that moves loses its interconnection position, its permits and often its land option, restarting a process that lasts several years.
Delay raises financing costs on assets whose hardware depreciates within a few years.
China’s intelligent computing capacity grew 146% year on year to the end of September 2026
Relocation also tends to push load towards grids with less spare capacity or a more carbon-intensive mix, which undermines the environmental case for the restriction in the first place.
Worst of all, part of the investment leaves the country. China’s intelligent computing capacity grew 146% year on year to the end of September 2026, under a planning model that builds generation and computing capacity together.
Every gigawatt delayed in Ohio, Texas or New York narrows a technological lead that took decades to build.
The alternative that already exists
The alternative to moratoriums already exists and is being implemented. Large-load tariffs with minimum demand charges, collateral requirements and exit fees protect households.
Policy should be judged by its results, not its intentions, and on that test, moratoriums fail completely - Daniel Lacalle
Flexible interconnection unlocks existing headroom. Bring-your-own-generation rules, including gas, nuclear and storage, add supply rather than ration demand.
Water-usage standards address local resources directly. Reforming transmission permitting and the interconnection queue tackles the bottleneck that actually pushes prices up.
Each of these tools aligns the cost with the party that causes it and keeps investment flowing. A moratorium does the opposite: it postpones the problem, increases the cost of solving it, and delivers no improvement in bills, reliability or water.
Policy should be judged by its results, not its intentions, and on that test, moratoriums fail completely.
