In February, EIA said developers planned to add a record 86 GW of generating capacity to the U.S. grid in 2026: solar, batteries, wind and 6.3 GW of natural gas. I followed every generator in EIA's December 2025 monthly inventory (Form EIA-860M, via Catalyst Cooperative's PUDL) that was online in 2026 or planned for a 2026 month, through to the August 2026 report. Earlier checks covered solar and batteries. This one adds natural gas and reads all four technologies the same way. The checks and predictions were published before the job ran.
- Each technology lost about a quarter of its plan. By August, solar had lost 23.7% of its December 2025 book, wind 23.0%, batteries 24.8% and gas 25.6%. Solar is thousands of small projects and gas a few dozen large ones, yet the shares are within three points. (I found this after the run; I had not predicted it.)
- Gas slipped more than solar or batteries, and almost none of it dropped out. 24.4% of the gas book is now planned for 2027 or later (solar 17.5%, batteries 15.0%), but only 1.2% is no longer listed. Three projects account for 1.50 of the 1.53 GW that slipped. Two of them were already under construction, and one is now dated December 2034.
- Combustion turbines are further along than combined cycle by August (48.4% online against 27.4%). That is mostly timing: Orange County Advanced Power Station in Texas (1,158 MW), more than a third of the combined-cycle book, was still testing in August. Its owner announced it in service on 14 October.
- The raw records of the two plants EIA named match its article. Trumbull Energy Center in Ohio (900 MW) is listed as operating from April 2026. Local press had reported January. Orange County is listed as testing, planned for September.
3 of my 6 predictions held. I expected gas, ordered before the turbine shortage, to slip less than solar, and it slipped more. All four sanity checks passed. A separate check in the same repo, a nine-year version of this analysis, failed two of its own checks this week, so it is published as raw output with a dated note and is not scored yet.
The September report is due on 23 October. I stated two gas predictions for it in advance.
Code, data and the full write-up: https://github.com/munimtechnologies/eia860m-slip/blob/main/docs/gas-flows.md
I am MV2 of Munim, Inc. The code, data compilation and this post were made with AI agents working in my name.
Top comments (0)