NY Bill Mandates Data Centers Get 90% of Power from Renewables by 2040

For states that have legally binding climate targets, the artificial intelligence revolution isn't just a technological breakthrough—it's a direct assault on their ability to decarbonize. The pending law in New York, demanding that massive data centers run on 90% clean power within the decade, is the most concrete policy response to an energy surge that could lock in decades of new gas-fired emissions. How regulators, utilities, and utilities' corporate customers navigate this conflict will shape whether the U.S. can reconcile the digital economy with a livable climate.

The artificial intelligence boom's insatiable demand for electricity is triggering the largest construction wave of natural gas-fired power plants in history, but a counter-movement is intensifying among renewable energy advocates and climate-focused policymakers. At the center of the clash is pending legislation in New York that would require large data centers to meet escalating renewable energy benchmarks, reaching 90% clean energy by 2040. The bill, authored by State Senator Kristen Gonzalez and now on Governor Kathy Hochul's desk, exemplifies a multi-state push to prevent the AI expansion from derailing climate goals. Tech companies, pressed by shareholders and their own net-zero pledges, are also working regulatory levers to compel monopoly utilities to integrate more wind and solar, even as the breakneck pace of data center construction vastly outstrips what renewables can currently supply.

“

At the center of the clash is pending legislation in New York that would require large data centers to meet escalating renewable energy benchmarks, reaching 90% clean energy by 2040.

The energy math is stark. A single AI-training data center can now consume more electricity than a mid-sized city, and the pipeline of such facilities is swelling. This has revived the fossil fuel sector: utilities are filing plans for new gas-fired generation, plant operators are delaying coal retirements, and the federal government has shown openness to extending the life of carbon-intensive assets. Renewables, while cheaper per megawatt-hour in many regions, face permitting, interconnection queues, and intermittency challenges that make them too slow to meet the immediate, around-the-clock power demands of massive computing loads.

New York's legislation directly confronts this tension by imposing a timetable that begins in 2030, with the 90% target just a decade later. Senator Gonzalez argues that the world's wealthiest companies, which are ready to pour billions into data center construction, have the resources to build out the renewable infrastructure needed to power them. The framing is both a moral and economic rebuttal to the argument that AI growth must default to fossil fuels. In Michigan, Oregon, and Minnesota, similar efforts are percolating, suggesting a patchwork of state-level mandates could emerge rather than a single federal standard.

For corporations with public clean-energy commitments—think Amazon, Google, Microsoft—the regulatory push is a double-edged sword. On one hand, mandates align with their stated goals and can create market certainty for green power purchasing. On the other, they risk creating compliance costs and limiting site selection in states where utilities are reluctant or unable to deliver renewable capacity quickly. Environmental groups, meanwhile, are leveraging public utility commissions and integrated resource plans to force disclosures and demand that grid planning account for the full climate impact of serving AI loads.