New York's grid operator found no actionable reliability needs in its latest ten-year plan. It also said that result "should not be mistaken for long-term system adequacy." The New York Independent System Operator's 2025-2034 Comprehensive Reliability Plan describes a system where the margin for error is extremely narrow and most plausible futures point to significant shortfalls within a decade.
The biggest new variable is large loads. At the end of 2024, the NYISO interconnection queue held roughly 4,000 MW of large load projects, averaging 175 MW each. By September 2025, that had more than doubled to over 10,000 MW seeking service before 2031, with an average size of 285 MW.
Who is asking for power
The plan says most large load requests come from semiconductor plants and data centers other than cryptocurrency mining, and their developers say their operations are not flexible. Upstate New York is the main draw, thanks to its labor, land and promotion of job creation, the NYISO said.
The growth in requests has been fast. In its Power Trends 2025 report earlier this year, the NYISO said that six large load projects accounted for 1,045 MW of queue requests in 2022. By April 2025 there were 20 projects totalling nearly 4,400 MW, and forecasters expected as much as roughly 2,500 MW of that to be on the system by 2035. That marks a break with the past decade, when demand in New York was relatively flat as energy efficiency programs offset growth from homes and businesses. Power Trends also noted that forecasters have lowered their expectations for how quickly electrification will be adopted, leaving microchip fabrication and data centers as major drivers of load growth. The operator's 2025 Gold Book put expected demand from large load projects at 2,567 MW by 2035.
The reliability plan shows how far the queue has run ahead of the forecast. As of September 30, 2025, there were more than 8,000 MW of large load requests in the queue beyond those included in the baseline forecast used for the 2024 Reliability Needs Assessment.
"Though not every project will materialize, the speed and scale of these requests far outpace the development of new supply," the plan says. "Large loads can come online quickly, but the resources needed to serve them, generation, transmission, and storage, require years to plan, permit, and build."
Testing the extremes
Because it is impossible to know which projects will be built, the NYISO tested three bookends. One assumes no new large loads come into service, which could also represent a future in which all new large loads are flexible at peak. A second assumes all large loads, whatever their use, are inflexible and draw power under all conditions. The third assumes every large load request as of September 5, 2025 is in service at its full requested size.
In its baseline analysis, the NYISO assumed cryptocurrency mining and hydrogen production would be flexible during summer and winter peaks, consistent with earlier studies. Data centers and chip plants were not treated that way, because their developers have told the operator they cannot curtail during peaks.
Across the range of assumptions in the plan, the operator found that the factors affecting supply and demand could stack up to produce either sufficient surpluses or deficiencies of up to 10,000 MW.
An old fleet
The supply side is the other half of the problem. Roughly 25% of New York's generating capacity is fossil-fuel generation that has been running for more than 50 years, and 7% of the fleet is 70 years or older, the plan says. As these units age, they suffer more frequent and longer outages.
To reflect that risk, the NYISO identified about 3,000 MW of existing conventional fossil generation as likely to be unavailable by 2034, roughly 60% of it in New York City. Depending on demand growth and retirements, the system may need "several thousand megawatts of new dispatchable generation" over the decade.
The plan follows the 2024 Reliability Needs Assessment, which found a need in New York City starting in summer 2033 and growing to a deficiency of 97 MW for three hours on the peak day in 2034. Updates since then, including a 200 MW cut in the city's ten-year demand forecast, resolved that need for now.
"We continue to observe declining reliability margins while forecasting a dramatic increase in load," said Rich Dewey, the NYISO's president and chief executive, when Power Trends was released. "It's imperative during this period of transition that we maintain adequate supply to meet growing consumer demand for electricity."
Making large loads part of the answer
The plan points to market rules as one way to turn large loads from a risk into a resource. New York already has a participation model for behind-the-meter natural gas generation, which lets on-site generation serving a host load offer excess output into the wholesale market. That model assumes a firm host load that does not respond to market signals.
The NYISO said expanding the model to accommodate flexible host loads could let these resources supply more surplus power to the market. It also suggested allowing a broader range of technologies, including renewable energy and storage, within the same model. For data centers that build their own generation, that would create a route to sell power back to the grid when they reduce consumption.
The plan also notes that New York Power Authority's small gas plants provide fast-start flexibility and voltage support that intermittent resources cannot yet replicate, and that retaining those capabilities, or equivalent ones, would have meaningful reliability benefits in New York City and Long Island.
What to watch
The 2026 Reliability Needs Assessment will be the first full study to include the expanded large load queue. Its treatment of data center flexibility will matter as much as the headline megawatts. If developers continue to insist their loads cannot respond to grid conditions, the NYISO will have to plan for the inflexible case, and the need for new dispatchable generation upstate will grow accordingly.