The North American Electric Reliability Corporation's annual 10-year outlook, released in January, has a simple headline: demand is rising faster than the resources planned to meet it, and data centers are the main reason. The 2025 Long-Term Reliability Assessment forecasts that summer peak demand across its assessment areas will rise by more than 224 GW over the next decade, 69% higher than last year's 10-year growth projection of 132 GW. Winter peak demand is expected to grow by 245 GW. NERC finds that 13 of its 23 assessment areas face resource adequacy challenges over the next 10 years.
For PJM, the assessment is especially stark. NERC classifies the region as elevated risk from 2026 to 2028 and high risk from 2029 onward.
What NERC says about PJM
According to the assessment, demand for electricity in PJM is growing at its fastest pace in years, driven primarily by data centers, followed by electrification and manufacturing. PJM expects its summer peak to grow by 56 GW to 210 GW in 2035, and its winter peak to climb by 62 GW to 198 GW by winter 2034 to 2035. Annual net energy for load is projected to grow at an average of 4.8% a year over the next decade, up from 2.3% in last year's projections.
On the supply side, NERC describes an extreme and rapid tightening of capacity in the near term because of generator retirements and project delays. A large share of PJM's new interconnection requests are from variable resources, about 40% of them solar, while dispatchable resources are leaving the system faster than other dispatchable technologies can replace them. Combined with PJM's limited ability to import power, with maximum total transmission interchange capability under 2% of internal generation capacity, NERC projects that PJM's anticipated reserve margin may fall below its installed reserve requirement in 2029.
NERC notes that PJM has about 30,000 MW of generation in its transitional interconnection queue to be processed in 2026, and that the new cycle process opening in April offers one- to two-year review timelines. Projects that won expedited interconnection under PJM's Reliability Resource Initiative were not far enough along to be included in this assessment's risk analysis.
The rest of the map
PJM is not alone. NERC finds ERCOT elevated risk for 2026 to 2028 and high risk from 2029. ERCOT forecasts summer peak total internal demand rising from 94,650 MW in 2026 to 154,077 MW in 2035. NERC's table for Texas assumes that much of that growth can be met by demand response, which climbs from 13,346 MW in 2026 to 63,321 MW in 2035, leaving net internal demand near 90,756 MW at the end of the period. The anticipated reserve margin on that basis falls from 28.2% in 2026 to 22.8% in 2035. NERC finds MISO becomes a high-risk area beginning in winter 2028. NERC says data centers are the largest contributor to MISO's accelerated demand growth, with 18 GW of data center load projected by 2035, while MISO's accredited thermal capacity has fallen by 8.8 GW. For its 2029 study, MISO assumed 14 GW of generator retirements that are uncertain to occur. NERC notes that MISO had more than 54 GW of mostly solar nameplate generation in development as of July 2025, a figure that rose to more than 70 GW by December 2025.
The expedited resource programs that the Federal Energy Regulatory Commission approved in late summer 2025 for MISO, PJM and SPP could change the picture. NERC says MISO's Expedited Resource Addition Study is expected to add more than 20 GW of summer on-peak capacity by summer 2030, but its timing meant those additions were not included in this year's modeling.
Fewer retirements, more demand
One trend runs in the right direction. NERC says confirmed and announced retirements over the next 10 years remain high, at more than 105 GW of peak seasonal capacity, but that is about 10 GW lower than last year's projection. Growing demand, market signals and resource plans have highlighted the need to keep plants online longer than previously expected.
That is the logic behind the capacity price signals PJM has sent in recent auctions. The 2027/2028 Base Residual Auction in December cleared at the cap and fell short of the reliability requirement for the first time, with forecast peak load up 5,250 MW from the previous auction. NERC's numbers suggest that holding on to existing plants buys time without closing the gap.
How firm is the data center forecast
NERC addresses the uncertainty directly. It says load-serving entity forecasts are based on interconnection information and agreements with large customers, and that to be counted, data center projects must have moved from speculative stages into development commitments. Still, large loads add volatility, because project timelines can change with construction, permitting, grid development and owner decisions.
NERC also notes that ERCOT and PJM have each prepared revised load forecasts since the assessment's data collection period, which were too late for this report. Both indicate that some large-load projects have slowed or failed to materialize in the near term, while interconnection requests for later years continue to grow. NERC says downward revisions could shrink the projected shortfalls.
What NERC wants done
The assessment's recommendations are aimed at both sides of the equation. NERC wants new resources capable of supporting voltage, frequency, ramping and dispatchability, and it urges grid operators and FERC to continue work on how large loads, including data centers, connect to and behave on the bulk power system. It also highlights the need for early identification of generators seeking to retire when they are still needed for reliability.
For PJM's states and customers, the message is that the next three years are tight and the years after that are tighter. Data center demand is the variable that matters most. If it arrives on the schedule utilities have filed, NERC expects a shortfall by 2029. If some of it slips, the region gains time. Either way, the region needs more firm capacity than it is currently on track to build.