In February, EIA said developers planned to add 24 GW of utility-scale battery storage in 2026. Its monthly generator inventory (Form EIA-860M) had 24.28 GW of batteries in the December 2025 report's 2026 book. I followed every one of those 263 batteries to the August 2026 report, read exactly the way I had already read solar, and asked one new question: when a battery sits at a plant that also has 2026 solar, does it share the solar's fate? Four checks and six predictions were published before the job ran. All four checks passed; four of the six predictions held.
- The same quarter left the year as for solar. By August, 6.0 GW (24.8%) of the December battery book had left 2026: 3.65 GW moved to 2027 or later, 1.88 GW dropped out of the inventory, and 0.5 GW is listed as operating with an online date outside 2026. 4.5 GW joined.
- Batteries are further along than solar. 39.4% of the December battery book was online by August, against 28.8% for solar, and only 15.0% slipped to 2027+, against 17.5%. I predicted batteries would do worse than solar on both counts, and both predictions missed.
- Batteries at solar plants waited for their solar. 78.6% of battery capacity at a plant with 2026 solar was in the same state as that solar in August. Every one of the 19 such batteries that slipped to 2027 or later did so together with its solar, to the same month. None slipped alone. This last point was found after the run, so it is a description, not a tested prediction.
- EIA's four largest named 2026 batteries (2,139 MW): none online by August. Lunis Creek (621 MW) left the inventory after the April report. Clear Fork Creek (600 MW) moved to March 2027 with its solar. Bellefield 2 (500 MW) and Tehuacana Creek 1 (418 MW) are planned for December 2026, the latter while its 837 MW solar side has moved to May 2027. All four were checked against the raw records.
Before the September inventory comes out, I have written down two more predictions: the battery book shrinks again, and at least 0.5 GW of batteries comes online in that report.
The tables, the per-generator file, the pre-registration and the code are here: https://github.com/munimtechnologies/eia860m-slip/blob/main/docs/storage-flows.md . Data: Catalyst Cooperative's PUDL copy of the EIA-860M changelog (CC BY 4.0). A generator's state is what the inventory says, not what happened on site, and "no longer listed" is not a confirmed cancellation.
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)