The US grid in 2026 is running two timelines at once. On one, renewable construction is setting records, with solar, wind and battery storage on pace to account for nearly all new generating capacity added this year. On the other, federal reliability regulators are growing more cautious about the five-year outlook, warning that more than half of the country’s grid regions face resource adequacy challenges as demand growth outruns the pace at which new supply can actually get connected. Both things are true simultaneously. The gap between them is where most of the grid’s current stress lives.
The generation mix itself is still fossil-heavy. The EIA’s September 2026 Short-Term Energy Outlook forecasts 2026 shares of roughly 40% natural gas, 16% coal and 18% nuclear, with total US generation expected to hit a record 4,368 billion kilowatt-hours this year. Renewables are closing the gap faster than the annual average suggests, though: EIA data for the first third of 2026 put all renewables, including hydro, wind, solar, biomass and geothermal, at 30.0% of total generation, up from roughly 26% for 2025 as a whole. Wind and solar alone account for close to 19 percentage points of that renewable share in rolling 12-month figures, with hydropower and biofuels making up the rest.
That renewable share is rising because of what’s actually getting built. EIA’s 2026 capacity additions point to 43.4 GW of new utility-scale solar, a 60% jump over 2025, alongside 24 GW of planned battery storage, up from a record 15 GW the year before and concentrated heavily in Texas (53% of the total), California (14%) and Arizona (13%). Wind adds another 11.8 GW, more than double 2025’s pace and including about 4.2 GW offshore. Natural gas, by contrast, is adding only around 6.3 GW of new capacity in the same period, meaning the overwhelming majority of what’s physically being built right now is solar, wind and storage rather than new fossil generation.
The bottleneck isn’t construction, it’s connection. Berkeley Lab’s count of projects sitting in US interconnection queues stood at 2,061 GW of generation and storage capacity at the end of 2025. Even that figure is down 10% from 2024, mostly because more than 750 GW of requests were withdrawn during the year rather than because the backlog cleared through approval. Solar makes up the largest slice of what remains queued at 773 GW, with another 749 GW of storage waiting alongside it. Wait times explain the attrition: median time to an interconnection agreement runs past three years nationally. PJM, the regional operator covering 65 million people from Illinois to Washington D.C., averages closer to eight years from request to connection. A project that gets built and approved can still sit idle for the better part of a decade before it’s allowed to deliver power to the grid.
Demand is the other half of the squeeze. Data centers are the fastest-growing piece of it. Estimates of their eventual electricity share vary by source and methodology, but they all point the same direction: EPRI puts data centers at 4–5% of US electricity demand today, rising to somewhere between 9% and 17% by 2030, while Lawrence Berkeley National Lab’s narrower range projects 6.7–12% by 2028. Wood Mackenzie expects data center capacity itself to grow from roughly 24 GW in 2026 to about 110 GW by 2030, which alone would account for 68% of all US electricity load growth over that span. NERC has had to revise its own demand forecasts upward as a result, most recently raising projected data center peak demand to 132 GW, a 65% jump in a single year’s estimate.
NERC’s actual reliability outlook reflects that tension. Its May 2026 Summer Reliability Assessment found more than 58 GW of new generation added across North America since summer 2025, enough that every assessment area should meet normal summer peak demand, with only three subregions, New England, SaskPower’s territory and the Pacific Northwest, flagged for elevated shortfall risk under above-normal heat, plus a narrower, transmission-constrained risk in far-west Texas. The longer-term picture is less comfortable: NERC’s January 2026 Long-Term Reliability Assessment found 13 of 23 assessment areas face resource adequacy challenges within five years, with projected summer peak demand through 2035 potentially rising 224 GW, 69% higher than the prior year’s projection, as coal plants and aging nuclear reactors retire faster than firm replacement capacity comes online. Not every analyst agrees NERC has the balance right. Grid Strategies has argued the assessment understates how much of the interconnection queue is likely to actually connect and overstates underlying demand growth, which is a real methodological dispute rather than a settled number.
Key Numbers at a Glance
| Metric | Value |
|---|---|
| 2026 generation mix (EIA forecast) | 40% gas, 16% coal, 18% nuclear |
| Renewable share of generation (early 2026) | 30.0% |
| Record total 2026 generation (EIA forecast) | 4,368 billion kWh |
| New utility-scale solar capacity, 2026 | 43.4 GW (+60% YoY) |
| New battery storage capacity, 2026 | 24 GW (up from 15 GW in 2025) |
| New wind capacity, 2026 | 11.8 GW (incl. ~4.2 GW offshore) |
| New natural gas capacity, 2026 | 6.3 GW |
| Total interconnection queue (end of 2025) | 2,061 GW |
| Queue withdrawals during 2025 | 750+ GW |
| Median wait for interconnection agreement | 3+ years (PJM averages ~8 years) |
| Data centers’ share of US electricity, today | 4–5% |
| Data centers’ projected share by 2030 | 9–17% |
| Data center capacity growth, 2026 to 2030 | ~24 GW to ~110 GW |
| NERC areas facing 5-year adequacy risk | 13 of 23 |
| Projected 2026–2035 summer peak demand growth | +224 GW |
What ties these numbers together is a grid that is simultaneously building faster and falling further behind. Record renewable capacity additions are real, the interconnection queue clearing at even current, slow rates is real and NERC’s widening five-year risk window is real too. The constraint isn’t whether enough generation exists or gets proposed, it’s whether it can get approved, connected and brought online before load growth, driven largely by a data center buildout still in its early innings, catches up to whatever capacity is actually in service rather than merely queued.
Sources:
- EIA September 2026 Short-Term Energy Outlook
- EIA renewables near 30% of generation, Electrek
- EIA 80 GW of new solar, wind and storage capacity in 2026, Electrek
- Berkeley Lab, Queued Up: 2026 Edition
- EPRI, Data Center Load Growth in Context
- Data centers’ share of US electricity seen doubling by 2030, E&E News
- NERC 2026 Summer Reliability Assessment
- Key takeaways from NERC’s January 2026 Long-Term Reliability Assessment
- NERC overstates reliability risks, Grid Strategies critique via Utility Dive