US battery project cancellations exceeded new announcements for the first time in 2025, $11B cancelled against $8B announced. As a procurement signal, this number lacks a denominator.
A raw cancellation count cannot distinguish a 2022 commitment that cancels pre-construction from a 2023 commitment that converts from EV to ESS cells. For a team qualifying suppliers against stated timelines, these carry different implications for domestic supply availability and schedule risk.
The promise ledger groups projects by announcement year and tracks each cohort's promised GWh against dated outcomes: what share hit original SOP, what share slipped and by how much, what share converted application, what share cancelled before breaking ground. The result is a reliability rate per vintage, telling a procurement reader whether a supplier's 2023-vintage commitment carries the same credibility as a 2021-vintage one.
Neither the Clean Investment Monitor nor E2 segments this way. Both track cancellation value without separating schedule drift from application conversion from outright abandonment, and neither provides a denominator by announcement year. The Log's registry is structured to support that segmentation. The four cases at right show what it surfaces: SOP drift of 18 months to two years appears across every vintage sampled, and application conversion turns up more frequently than aggregate cancellation figures would suggest.
Promise drift by vintage
- Lansing (LGES), 2022 vintage: 50 GWh, late-2024 SOP → >35 GWh, Aug 2026 SOP. Two-year slip, ~15 GWh nameplate reduction.
- Forge Nano, 2023 vintage: 3 GWh, 2026 SOP → groundbreaking Aug 2026, production 2028. Two-year slip.
- NOVONIX Riverside, 2024 vintage: Panasonic production early 2026 → H2 2027. Stellantis promise extinguished.
- L-H Battery Ohio, 2023 vintage: ~40 GWh Honda EV cells → ESS production for Vertech, July 2026. Application converted, not cancelled.
Context
- 55 GWh operational (2021) → ~1,000 GWh projected by 2030 (DOE, Jan 2023)
- US battery investment down ~38% from late-2024 peak (CSIS, April 2026)
- Most facilities take >5 years from SOP to nominal capacity (IEA GEV Outlook 2026)
- No published vintage-cohort reliability study for US battery projects currently exists