For years, artificial intelligence has operated in the cloud, a virtual space that exists without a physical location or neighbors. But the cloud is no longer abstract—it has landed in the real world, occupying vast areas of land and appearing on household electricity bills. In New York, a single AI project requires 500 megawatts of power, enough to supply 400,000 homes. The United States, often seen as the birthplace of the cloud, is now realizing that its digital infrastructure has a tangible cost. Governor Kathy Hochul of New York has taken notice. In July, she suspended, for one year, the approval of new large data centers. Two months later, she introduced a more comprehensive plan to manage the growing impact of these facilities. The first part of Hochul’s plan focuses on taxation. She wants to end the sales tax exemption that has been granted to large data centers since 2000. The second part addresses energy use. The state’s regulator is being asked to require data center operators to either generate their own electricity or pay higher rates for grid power. In addition, Hochul advises local governments to demand one million dollars per megawatt of power used. For a 50-megawatt site, this would add 50 million dollars in costs. On October 7, Hochul, along with Congresswoman Alexandria Ocasio-Cortez, emphasized that these large companies should no longer seek tax breaks. In the Hudson Valley, the town of East Fishkill acted before the state government, passing a three-year moratorium on new data centers in June. Taxpayers are resisting the idea of subsidizing machines that increase their utility bills. The financial pressure is growing. Take the PJM network, which provides power to 67 million Americans across 13 states and the federal capital. The cost of capacity for 2024/2025 was about 29 dollars per megawatt per day. It has now surpassed 320 dollars. For three consecutive auctions, the price reached the regulatory limit. According to Monitoring Analytics, an independent market observer, data centers are responsible for 63% of the increase in the 2025/2026 auction. That amounts to 9.3 billion dollars passed on to consumers. Artificial intelligence was expected to lower production costs, but instead, it has raised household expenses. Across different counties, opposition is growing. In the first quarter of 2026, at least 75 projects totaling 130 billion dollars were either blocked or delayed. This figure comes from Data Center Watch, an observatory run by 10a Labs, a firm that supports AI companies. However, the data has its limitations. It adds up announced investments rather than actual costs and includes both abandoned projects and those delayed for various reasons. The information is drawn from public sources like local news and petitions, and details about each project are available only upon request. In effect, the 130 billion dollars figure reflects more of a political risk than a concrete loss. This risk is growing, as the number of recorded opposition groups has risen from 396 at the end of 2025 to 833 in March, across 49 states. According to Gallup, 71% of Americans oppose the construction of data centers near their homes. The movement has brought together people of different political views—Republicans criticizing tax breaks and Democrats concerned about water use and climate impact—all united by the rising cost of energy. The industry has taken notice. In March, major tech companies like Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI signed a non-binding agreement at the White House, promising to fund their own energy and infrastructure. Amazon has gone further, pledging over one billion dollars over five years to support local communities. However, the company’s cloud boss attributes the opposition to misinformation.