Every month, EIA's EIA-860M lists the U.S. generators that developers plan to bring online, each with a planned month. I took every generator planned for the next month and checked whether the next month's file lists it as operating. I did this for every month from August 2015 to August 2026, 133 months in all.
The share fell. By count:
- 2016-2019: 1,980 of 4,223 (0.469) were listed as operating in their planned month.
- 2022-2025: 2,110 of 6,047 (0.349).
Every year from 2021 to 2026 is below every year from 2016 to 2020. Meanwhile the yearly number of generators planned for the next month grew from 951 in 2016 to 1,586 in 2025.
December stands out. In each of the ten years 2016 to 2025, December had the largest group of the year and a share above its year's. Leaving December out, the share still fell, from 0.431 to 0.328.
What this does not show: it counts generators, not capacity, and it does not split by technology, so it cannot say whether projects slip more often now or different kinds of projects are being planned. Generators at one plant move together, so I give no significance test. EIA has published a related measure for solar alone (the share of planned solar capacity reporting a delay); this one covers every technology.
To test the December pattern on a year nobody has seen, I fixed two predictions before EIA published any month after August 2026: December 2026 will have the largest group of 2026, and its share will be above the pooled share of January to November. EIA's December file is due in late January 2027.
The code, the data file, the note and a test that recomputes every number are here: https://github.com/munimtechnologies/eia860m-slip
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)