New York Just Froze Big Data Centers — And Every Other State Is Watching New York Governor Kathy Hochul signed an executive order on July 14 imposing a year-long moratorium on new permits for data centers drawing more than 50 megawatts of power, making New York the first US state to enact such a pause. The order aims to give regulators time to develop environmental, grid, and water-usage rules for the AI infrastructure boom, potentially reshaping where hyperscale data centers are built nationwide. New York Just Froze Big Data Centers — And Every Other State Is Watching New York Governor Kathy Hochul signed an executive order on July 14 pausing new permits for data centers drawing more than 50 megawatts of power for up to a year. The first-of-its-kind US state moratorium gives regulators time to write environmental, grid, and water-usage rules for the AI infrastructure boom. Governor Kathy Hochul signed an executive order on July 14 making New York the first US state to pause new permits for data centers drawing more than 50 megawatts of power. The year-long moratorium gives regulators time to write environmental, grid, and water-usage rules for the AI infrastructure boom. The order, reported Tuesday morning, exempts hospitals, universities, and smaller commercial facilities. It doesn't touch existing data centers or projects already permitted. But any new hyperscale facility — the kind AI companies need to train frontier models — now faces a minimum 12-month wait in New York. Hochul paired the moratorium with a push to repeal sales-tax exemptions for large data centers. Empire State Development has 60 days to publish what the order calls a "Community Interest Framework" — essentially negotiation guidelines for localities weighing data center proposals against competing land and power uses. The Scale Problem A 50MW data center is roughly the size of a large suburban shopping mall in power terms. It's enough for about 15,000 to 20,000 homes. The AI industry's trajectory points toward facilities drawing 500MW to 1GW — the equivalent of a small city. Meta's Hyperion campus in Louisiana is targeting 5GW. New York isn't the only jurisdiction struggling with the math. Ireland's grid operator effectively banned new data centers in Dublin until 2028. Singapore imposed a three-year moratorium in 2019 and only recently began approving new projects with strict efficiency requirements. Northern Virginia — the world's largest data center market — faces growing community opposition over noise, water use, and transmission lines. Hochul's order is significant because New York is the first US state to impose a blanket pause rather than negotiating project-by-project. It signals that even in a country where AI infrastructure is treated as a national competitiveness issue, local concerns about grid stability, water consumption, and land use can override the tech industry's appetite for compute /glossary/compute . The Timing The order comes as AI infrastructure spending accelerates toward what SoftBank CEO Masayoshi Son called a $5 trillion annual requirement. Microsoft, Google, Amazon, and Meta are collectively spending more than $250 billion on data centers in 2026 alone. New York's moratorium doesn't change those numbers — but it does redirect where the spending lands. Virginia, Texas, Ohio, and Louisiana are the most likely beneficiaries. Each has available land, relatively cheap power, and state governments more welcoming to hyperscale development. The question is whether Hochul's move is an outlier or the start of a pattern. If California or Massachusetts follow, the geography of US AI infrastructure starts shifting in ways that matter for latency, workforce, and political influence. FAQ Q: What exactly does the order prohibit? A: New state permits for any data center drawing more than 50MW of power. Existing facilities and already-permitted projects are unaffected. Hospitals and universities are exempt. Q: How long does the pause last? A: Up to one year, while New York regulators write environmental, grid, and water-use rules. The clock starts immediately. Q: Why does this matter for AI? A: Training /glossary/training frontier AI models requires massive compute clusters housed in hyperscale data centers. If large states restrict where those can be built, it reshapes the infrastructure map and potentially increases costs. Q: Will other states follow? A: Likely some will, particularly those with strained grids California or strong environmental movements. States actively recruiting AI investment — Texas, Ohio, Louisiana — probably won't. Related Articles Get AI news in your inbox Daily digest of what matters in AI.