This week we review the September EIA-860M release and what has transpired across 2026 so far. The data center boom is reshaping the generation project queue, headlined by a resurgence in natural gas power generation projects. Below we look at four questions: whether more projects are being added, when and where they plan to come online, how much the schedule is slipping, and what is being canceled.
January–September 2026 added 1,082 new pipeline generators totaling 89.0 GW, the highest on record. That beats the same period in 2025 (671 generators, 79.2 GW) and is roughly 3.5× the 2018 level.
The March 2026 release alone added 38.5 GW, driven by the multi-GW Project Matador filing in Texas. Strip that out and 2026 runs closer to the 2024–25 pace.
New gas reached 32.2 GW through September, up from 21.7 GW in 2025 and under 7 GW per year in 2021–24. Solar is flat at about 30 GW, batteries are slightly lower (17.8 GW vs. 21.1 GW), and wind keeps shrinking (3.7 GW).
August and September 2026 combined added more generators (147) but less capacity (8.5 GW) than the 13–15 GW seen in the same two months of 2023–25. That points to smaller projects.
Takeaway: Additions are running at record levels, and the mix has shifted decisively toward gas, consistent with data center load growth and reliability planning. The latest two releases skew toward smaller projects, so watch whether that cooling persists into Q4.
Of the 2026 additions, 36.7 GW plans to come online in 2026–27, peaking at 25.4 GW in 2027 and led by solar (11.3 GW) and batteries (8.2 GW).
Gas is backloaded. Planned online dates for gas climb from 3.8 GW in 2027 to 7.9 GW in 2029 and 11.6 GW in 2030. Project Matador’s nuclear capacity (2.2 GW per year) appears in 2031–32.
32 GW is under construction and 3.5 GW is already operating, concentrated in 2026–27. Another 42.9 GW is still early stage (planned or permitting), and that bucket dominates 2029 onward. The later the date, the less certain the capacity.
ERCOT leads with 28.9 GW, of which 15.3 GW is gas and 4.5 GW nuclear, mostly Matador and much of it timed for 2030 or later. MISO follows with 18.9 GW (8.1 GW solar, 7.2 GW gas), then the non-ISO West with 16.3 GW (solar and storage), SPP with 8.0 GW, CAISO with 6.8 GW (storage-led) and PJM with 6.0 GW.
PJM is a story of many small projects: it added 228 generators but only 6 GW, mostly small solar. NYISO and ISO-NE added under 1 GW each.
Takeaway: The 2026–27 additions are largely real, under-construction solar and storage. The gas and nuclear additions behind the headline growth are early-stage and dated 2029 or later, so they carry high execution risk, especially the ERCOT mega-filing.
Delays dominate. 289 projects (21.0 GW) moved their online date later, against only 19 projects (1.1 GW) that moved earlier. That is roughly 19 GW of delayed capacity for every 1 GW pulled forward.
228 of the delays are 1–3 months, and they carry 17.2 GW, about 82% of the delayed capacity. Only 10 projects slipped more than 12 months, and they are small (0.27 GW combined).
By technology, solar PV slipped 10.7 GW and batteries 5.5 GW. Gas combined cycle (2.9 GW) and combustion turbines (1.0 GW) are smaller totals, but they are concentrated in a few large units.
This is largely a Texas story. Texas slipped 11.4 GW, more than half the total: 6.0 GW solar, 2.9 GW gas and 2.3 GW battery. Arizona (1.8 GW) and California (1.5 GW) are a distant second and third.
The longest slips are sub-100 MW projects, led by Stevens Mill Solar in North Carolina (+37 months) and Burt County Wind in Nebraska (+26 months). The largest projects, such as Danish Fields (600 MW) and Cider Solar (557 MW), slipped only 1–2 months.
Takeaway: This month is a broad, short rolling slip of near-term Texas solar and storage, not a wave of large projects being pushed out by years. Expect 2026–27 commissioning to shift 1–3 months to the right.
September 2026 canceled only 2 generators (136 MW), well below the 2016–26 monthly median of 0.36 GW. August was 924 MW.
The year as a whole is another matter. 136 generators and 17.1 GW have been canceled in 2026, the most for any January–September on record. That is 28% more than the same period in 2025 (13.4 GW) and already close to full-year 2025 (17.4 GW). Most of it came in February–June, peaking at 5.7 GW in February.
Cancellations are concentrated in Texas and the West. ERCOT canceled 6.6 GW (4.0 GW batteries, 2.6 GW solar) and the non-ISO West 3.7 GW, followed by PJM (2.3 GW, mostly solar). Solar (8.1 GW) and storage (7.9 GW) account for almost all of it; gas is only 0.8 GW.
Low-ranked projects were the ones canceled. 73% of canceled GW carried a Vulcan rank of 2–3, against 49% of the active pipeline. Only 0.8 GW ranked 4 or higher was canceled (Walleye Solar and Ripley Energy Center), and nothing ranked 5 or higher. The median canceled rank is 2.67 versus 2.94 for active projects.
Takeaway: Attrition is high this year and mostly hits early-stage Texas and Western solar and storage. Vulcan ranks pointed the right way: low-ranked projects were canceled far more often than high-ranked ones.
The 2026 queue is bigger than ever and more gas-heavy than it has been in a decade, but the near-term build remains solar and storage, slipping a few months at a time. The gas and nuclear capacity that grabs the headlines sits years out and largely in early stages, and this year’s record cancellations show how much of the early-stage queue never gets built. Separating what is real from what is paper is exactly where Vulcan’s construction monitoring and ranking earn their keep.
Vulcan offers weekly research and insights such as this piece. We are more than a technology or data provider: we have subject matter expertise that gives you key insights. Being a client gives you access to this in-depth knowledge. Reach out to David Bellman at dbellman@synmax.com to become a Vulcan client.