Every major US battery capacity tracker sorts facilities into some version of operational, under construction, delayed, or cancelled. None of them sort by buyer. A cell plant that converts from automotive to stationary storage keeps its GWh in the aggregate. E2's methodology is explicit: changes that do not reduce production capacity are excluded from cancellation counts. Rhodium's Clean Investment Monitor tracks production state, not application.
So the headline number can hold while the addressable supply underneath it redistributes. CSIS estimates roughly 40 percent of operating and planned US capacity was tied to automaker partnerships in 2025, and reports that manufacturers find EV-to-grid switching "financially and operationally feasible, though not without friction." When that switching happens, cancellations reduce the total GWh figure but conversions merely reroute it. The trackers capture the first movement. They are structurally blind to the second, and so is any procurement model built on their output.
What "operational" doesn't tell you
-
Application: EV, grid storage, and data-center cells serve different buyers with different qualification cycles, warranty terms, and compliance files. A converted line isn't immediately addressable to the new market
-
Chemistry shift: LFP's rise in the US pipeline (~200 GWh by 2030 per CSIS) partly signals application conversion, not just chemistry preference
-
Yield vs. nameplate: IEA notes most facilities take 5+ years to reach output close to nominal capacity. "Operational" GWh ≠ saleable GWh
-
The gap no tracker covers: No public source records when a facility changes which market it serves. The GWh persists in the aggregate while the buyer pool shifts underneath it, and procurement models built on the aggregate figure carry that distortion forward without marking it