EIA publishes two numbers for how much utility-scale solar the U.S. will add this year. One is the developers' schedule in the monthly generator inventory, Form EIA-860M. The other is the Short-Term Energy Outlook (STEO), EIA's own forecast. In January 2026 they were far apart: 43.4 GW scheduled against 32.1 GW forecast. Both are restated every month, so I followed them through the year to see which one would give way.
I wrote down three checks and five predictions before running anything. For each EIA-860M report from November 2025 to August 2026, the "implied 2026 total" is the solar already online in 2026 plus the solar still planned for a 2026 month. For each STEO edition from January to October 2026, the 2026 figure is the end-2026 capacity minus the end-2025 capacity in its table, and I read all twenty figures again by hand. Each edition is paired with the report two months earlier, the latest one available when it was finished.
- Neither number moved much. The developers' implied 2026 total stayed between 41.0 and 43.7 GW in all ten reports (41.1 GW in August). The STEO stayed between 29.1 and 32.0 GW (30.7 GW in October). The gap was 9.3 to 12.8 GW in every pair, and it is 10.4 GW now.
- The year is back-loaded. By August, 14.7 GW (36%) was online and 26.5 GW was still scheduled for September to December. The October STEO implies about 16 GW in those four months, roughly 60% of what is scheduled.
- Missed months move, the year does not. In no report was more than 0.21 GW of 2026 solar still planned for a month that had already passed. A project that misses its month gets a later 2026 month, so the year's total barely changes while the year runs out.
- Wind is different. The STEO's 2026 wind figure stayed within 1.2 GW of the developers' wind schedule in all ten pairs. The discount is a solar discount.
- Two of five predictions held. I expected the developers' plan to shrink toward the STEO by at least 4 GW and the gap to halve. It fell by 2.5 GW, and the gap went from 12.2 to 10.4 GW. I also expected 40% of the year's solar to be online by August; it was 36%.
With four months left, one of the two numbers will be off by at least 5 GW. Before the September inventory comes out, I have written down four more predictions. The September report will hold the implied total within 1.5 GW of 41.1 GW, and at least 16.0 GW will be online by then. EIA's first count for 2026, from the December report, will come in below 35.9 GW (closer to the STEO than to the plan) but at least 28.0 GW. The third one is a close call, and I have said so in the write-up.
The month-by-month table, the pre-registration, the raw job output and the code are in the repository: https://github.com/munimtechnologies/eia860m-slip/blob/main/docs/plan-path.md . Data: Catalyst Cooperative's PUDL copy of the EIA-860M changelog (CC BY 4.0) and EIA's archived STEO editions. An implied total takes every plan at face value; neither number is right or wrong until EIA's final 2026 figures in late 2027.
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)