Every January, EIA's monthly generator inventory (EIA-860M) lists the generators their owners plan to bring online that year. I followed each January plan from 2017 to 2025 to what became of it, and every generator that came online back to where it stood in January. The checks were fixed before the run, and all of them passed.
- 62.3% came online in the planned year. Of 391.6 GW planned across the nine years, 243.8 GW came online in its year. 118.2 GW (30.2%) was still planned for a later month, and only 7.9 GW (2.0%) dropped out of the listings.
- Stage matters more than technology. Plans already under construction or testing delivered 0.718 of their capacity in the year, and early-stage plans 0.271. The five largest technology groups range only from 0.557 (batteries) to 0.823 (gas combined cycle).
- A quarter to a third of each year's additions were not in January's plan. Of 345.3 GW built, 104.8 GW (30.3%) was not planned for that year in January, and most of it (74.6 GW) was not listed at all.
- Two 2018 plants used to be counted twice. TVA's Allen and Duke Energy Florida's Citrus County combined-cycle plants changed plant id while under construction. A tested re-key rule now links them (11 links, 3.3 GW in total).
- An open question. For 2017, EIA's own count says 86% of planned onshore wind came online, while EIA-860M gives 62.5%. A pre-registered test of whether the annual EIA-860's operating dates explain the gap is queued to run next.
The full tables, the raw job output and a test that checks every number against it are here: What January's plan delivered, 2017-2025
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)