E2's tracker counts 54 YTD cancellations, closures, and downsizings across US clean energy manufacturing, representing $18.2 billion in lost investment through June 2026. The aggregate captures genuine losses at the sector level. At the level of battery procurement decisions, it obscures a category that changes what domestic supply actually produces.
Underneath the cancellation count sits a category that E2's published methodology does not appear to separate: facilities where standing assets changed ownership, chemistry, end-market application, or all three. The GWh survived, but what those GWh serve changed fundamentally. No pre-existing tracker distinguishes these conversions from cancellations. This piece constructs a sample from five publicly sourced US facility cases identified through SEC filings, company releases, DOE project pages, and named reporting on ownership, chemistry, or application changes disclosed during 2025–2026. It reads the sample for directional clustering and asks what the accumulated pattern implies for the composition and fungibility of domestic supply. The sample is small. The directional signal within it is unambiguous.
Defining the Boundary
KORE Power's Buckeye, Arizona facility and FREYR's Coweta County, Georgia project establish what conversion excludes. At KORE, a conditional DOE commitment sat unresolved for approximately twenty months before the plant was cancelled and a retrofit strategy announced. At FREYR, construction never began and the state sought grant clawback. No physical or commercial asset survived to be repurposed in either case. These are cancellations. They belong in E2's count.
Conversion requires a standing asset, built or commercially committed, that changes what it serves while retaining its physical identity. Five US facilities meet that definition based on disclosed changes during 2025–2026.
The Five Cases, Read for Pattern
These are listed to build the pattern claim, not to re-audit each facility. The relevant facts are ownership structure, chemistry direction, and application shift.
Ford Energy Battery LLC (Kentucky). Ford's May 2026 8-K confirms Ford acquired SK's interests in both Kentucky battery plants, assuming obligations under a $3.8B DOE note whose product definition encompasses battery energy storage products, systems, and services alongside vehicle applications. Subsequent reporting indicates Ford Energy targets at least 20 GWh of storage products annually from late 2027, with an EDF agreement for up to 20 GWh of 2028 delivery. The ownership conversion is documented. The application conversion is reported but not yet confirmed by primary disclosure. JV dissolved. Direction: EV toward ESS.
StarPlus Energy (Kokomo, Indiana). Samsung SDI's March 2026 release states that part of StarPlus Energy's Kokomo lines have been converting from EV to ESS batteries since Q4 2025, with LFP mass production expected Q4 2026 alongside existing high-nickel NCA. The second Kokomo plant was announced at 34 GWh annual capacity. Samsung SDI does not quantify the converted share. Simultaneous application and chemistry conversion on disclosed lines. Direction: EV/NCA toward ESS/LFP.
LGES Lansing (Michigan). GM sold its stake in the nearly completed Lansing Ultium Cells plant to LG Energy Solution in December 2024. By mid-2026, the DOI summit release disclosed a $4.3 billion Tesla-LGES supply agreement for LFP prismatic cells from Lansing, targeting Tesla Megapack 3 production launching 2027. A facility originally financed as an NCMA pouch-cell plant for GM EVs is now an LG-controlled LFP prismatic plant serving Tesla's stationary storage business. Ownership, chemistry, customer, and application all changed. The DOE project page still describes the original configuration. Direction: EV/NCMA toward ESS/LFP. JV dissolved.
SK Battery America (Commerce, Georgia). SK opened the $2.6 billion Commerce plant in January 2022. By March 2026, AP reported 958 layoffs with approximately 1,600 workers retained and the company stating it was pursuing future customers including the BESS arena. No GWh figure or line-level conversion is disclosed. This is the weakest case in the sample: workforce contraction with stated BESS intent but no confirmed line conversion. I include it because the plant is built and operating, the EV customer base contracted, and the company named ESS as a direction. It may prove to be a cancellation-in-progress rather than a conversion. Observable by year-end.
Ultium Cells Spring Hill (Tennessee). WIRED reported in July 2025 that GM would convert cell lines at Spring Hill from NCMA to LFP for the updated Chevrolet Bolt EV, expected late 2027. This is a chemistry conversion (nickel-rich to LFP) that remains EV-directed. It complicates the application-clustering claim while reinforcing the chemistry-clustering claim.
Panasonic's Kansas and Nevada plants did not surface a disclosed conversion in the checked sources. Panasonic Kansas opened at 32 GWh for automotive cylindrical cells with Tesla as primary customer. It may be converting quietly. I cannot source it and will not assert it.
| Facility | Sponsor change | Chemistry direction | Application direction | Status |
|---|---|---|---|---|
| Ford Energy (KY) | SK → Ford (JV dissolved) | Implied LFP (BESS) | EV → ESS | Reported 20 GWh BESS target, late 2027 |
| StarPlus (Kokomo, IN) | JV retained | NCA → NCA + LFP | EV → EV + ESS | LFP mass production expected Q4 2026 |
| LGES Lansing (MI) | GM → LGES (JV dissolved) | NCMA → LFP prismatic | EV → ESS | Tesla Megapack 3 supply, launching 2027 |
| SK Commerce (GA) | No change | Undisclosed | EV → stated BESS intent | 958 layoffs; weakest case |
| Spring Hill (TN) | No change | NCMA → LFP | EV → EV | Bolt EV, expected late 2027 |
Three Clusters and a Partial Exception
Five cases constitute an inventory. The directional signal within that inventory is still worth stating plainly.
Application: four of five move toward ESS. Ford Energy, StarPlus, Lansing, and Commerce all involve disclosed or reported movement from EV cell supply toward stationary storage. Spring Hill is the partial exception: it converts chemistry while remaining EV-directed. No case in the checked source set moves the other direction, and no US facility disclosed a 2025–2026 conversion from ESS to EV.
Chemistry: four of five move toward LFP. StarPlus is adding LFP alongside NCA. Lansing moved from NCMA pouch to LFP prismatic. Spring Hill is converting NCMA lines to LFP for EVs. Ford Energy's DOE loan product definition permits both vehicle and storage applications, but the reported customer pipeline is BESS, and the chemistry direction implied by the storage application is LFP. SK Commerce has not disclosed a chemistry shift. No counter-directional case was found.
Ownership: two of five involve confirmed JV dissolution toward single-sponsor control. Ford acquired SK's Kentucky interests. GM sold its Lansing stake to LG. Stellantis sold its 49% NextStar stake to LG Energy Solution at the Windsor, Ontario plant, which I exclude from the US sample but which follows the identical pattern. The JV model that characterized the 2021–2023 US battery buildout is unwinding in resolved cases, though StarPlus remains a JV. In the Lansing and NextStar cases, the surviving sponsor is the battery maker, not the automaker. Ford Kentucky is the exception: the automaker retained the asset and the battery maker exited. What persists across cases is that the original JV structure proved unstable, regardless of which partner walked away.
Facility type: all five are cell-level conversions. No component-level conversion (cathode, anode, separator) surfaced in the checked source set. Whether this reflects actual stability in the component base or lower disclosure visibility for component facilities cannot be determined from available sources.
What Cannot Be Claimed
The sample does not support a rate claim. I cannot say conversions are accelerating because the five cases span Q4 2025 through mid-2026 without enough temporal density to establish a trend line, and no baseline exists against which to measure acceleration. E2's tracker, the closest thing to a systematic registry, does not appear to distinguish conversion from cancellation in its published methodology. A facility that changes product or owner but retains physical capacity may appear in E2's cancellation count, its active count, or neither.
The sample also cannot tell you whether these conversions will succeed operationally. Converting a nickel-rich pouch line to LFP prismatic production changes equipment, formation protocols, quality systems, and supplier qualifications. The announcement-to-production gap that plagues greenfield projects applies to conversions too, compounded by legacy equipment and workforce skills optimized for the old product.
What the Pattern Does to 2027–2028 Supply Composition
If the directional clustering holds, and nothing in the available evidence contradicts it, then a material share of US cell capacity originally justified, financed, and in some cases DOE-supported as EV supply will be producing ESS cells by 2028.
Using Ford Energy's reported 20 GWh annual BESS target as the only quantified anchor, and adding undisclosed volumes from Lansing (Tesla Megapack supply) and StarPlus (partial line conversion of a 34 GWh complex), a rough lower bound is at least 20 GWh and plausibly 30–40 GWh of US capacity converting from EV to ESS application. This range is an inference from available cases, not a sourced aggregate.
The fungibility consequence follows directly. LFP prismatic cells sized for stationary storage are not interchangeable with nickel-rich pouch cells qualified for EV programs. They serve different formation specs, different pack architectures, different thermal management requirements, different customer qualification cycles. Nameplate GWh that converts from one application to the other does not simply change the demand curve it serves. It becomes non-fungible with its original use case. The capacity still exists, but redirecting it back would require another full conversion cycle.
These facilities will price against Chinese ESS imports, where landed costs face tariff and FEOC headwinds, and against each other, in a domestic ESS market that is growing but has never absorbed this volume of domestic cell supply. For the automaker who expected these GWh to be available as domestic EV cell supply, the capacity migrated to a different market.
The procurement lead who needs nickel-rich EV cells manufactured domestically by a non-FEOC supplier faces a thinner field than the nameplate numbers suggest. Meanwhile, the strategy director modeling domestic ESS supply sees a pipeline that barely existed twelve months ago. Fewer joint ventures, more single-sponsor Korean-controlled facilities, more LFP, more storage. The aggregate GWh number may hold through 2028. What those GWh produce, for whom, in what chemistry, under what ownership, has already shifted.
- DOE project pages lag: DOE's BlueOval SK page still describes the original $9.63 billion three-plant EV-battery loan framing, months after Ford's 8-K documented the Kentucky ownership transfer and obligor change.
- MACR proof burden tightens: IRS Notice 2026-15 requires supplier certifications signed under penalties of perjury and retained for six years, turning 45X eligibility into a documentation exercise that converted facilities must rebuild from scratch.
- Samsung SDI's cathode sourcing: Samsung SDI's KRW 1.6 trillion L&F cathode deal starting 2027 is a PFE-mitigation signal for StarPlus LFP production, but no public MACR calculation or facility-level 45X claim has been disclosed.
- SK Commerce workforce watch: AP reported SK Battery America retained about 1,600 workers after cutting 958, and whether the BESS pivot materializes as line conversion or remains stated intent should be observable by year-end.