What Happened to 44 Battery Manufacturing Commitments Made in 2022–2023

Six of forty-four measurable commitments were met as originally stated.

The cohort is twelve US battery manufacturing facilities: nine announced during 2022–2023, plus three earlier projects whose capital deployment and revision cycles fell inside the same window. Each made public commitments across four dimensions — start of production (SOP), nameplate capacity in GWh, job count, and cell chemistry or product type. Twelve facilities times four dimensions gives 48 cells. Four of those were never stated in the original announcement, leaving 44 that can be scored. Sixteen were missed or abandoned. Eight resist clean classification, and the accounting for those sits in the analytical section below.

The three pre-2022 entries are Ultium Warren (December 2019), Microvast Clarksville (February 2021), and BlueOval SK Glendale (September 2021). They are retained because their revisions all landed between 2022 and 2026. Toyota Liberty's cleanest 2022–2023 anchor is the October 2023 expansion, layered onto a 2021 site selection. The cohort was chosen to span the outcome spectrum, not to represent it proportionally, so the ratios below describe these twelve facilities and nothing wider.

Startups

Forge Battery — Morrisville, North Carolina. North Carolina's November 2023 announcement promised operations in 2026, 204 jobs, and lithium-ion cells. The state release gave no GWh figure and no specific chemistry; the 3 GWh NMC/silicon-graphite target and 2029 full-rate date surfaced later, in a DOE project sheet. Forge broke ground in August 2026, now targeting production in 2028 with 250-plus jobs. The deep audit in this publication's registry update carries the current detail. Score: SOP revised from 2026 to 2028; jobs revised upward, though as a target rather than verified employment. Capacity and chemistry were not originally stated and cannot be scored.

Our Next Energy — Van Buren Township, Michigan. ONE's October 2022 announcement committed to lithium iron phosphate (LFP) production beginning in 2024, 20 GWh by end of 2027, 2,112 jobs, and a vertically integrated campus making LFP and anode-free cells. Pilot LFP production began November 2023. In January 2026 ONE exited passenger-EV work, cut roughly 45% of its workforce, and retooled for defense, rail, and infrastructure LFP cells. Michigan's FY2025 report recorded 48 qualified jobs against 2,112 committed, with $70.24 million of a $200 million state grant disbursed. A January 2026 layoff notice under the WARN Act, the federal statute requiring advance notice of mass job cuts, covered 29 positions at the Novi headquarters rather than the plant. Score: SOP met at pilot scale only. 20 GWh not established. 48 of 2,112 jobs qualified, though the state milestone runs to December 2029. Product revised from automotive to defense and infrastructure.

Microvast — Clarksville, Tennessee. The February 2021 announcement committed to $220 million, 287 jobs, and cells, modules and packs. Microvast's 10-K later specified 2 GWh of NMC capacity with a Q3 2023 target. Construction was suspended in Q2 2024 for lack of funding, and the planned chemistry moved from NMC to 565 Ah LFP. The August 2026 SEC filing says only that the localized pack line is expected to begin initial operations by year-end 2026; cell-plant construction remains contingent on financing. No Tennessee WARN filing was identified, though an SEC-disclosed lawsuit alleges WARN violations. Score: SOP missed, 2 GWh not completed, 287 jobs not verified with workforce reductions instead. Product revised twice — NMC to LFP, and cell manufacturing to near-term pack assembly.

OEM-Owned

Ford BlueOval Battery Park — Marshall, Michigan. Ford's February 2023 announcement committed to production in 2026, approximately 35 GWh, 2,500 jobs, and prismatic LFP cells built on licensed CATL technology. Michigan's July 2024 board packet approved a formal scope reduction: $2.5 billion minimum investment against $3.5 billion, 1,700 minimum jobs against 2,500, and approximately 20 GWh against approximately 35. In June 2026 Ford reported 500-plus employees and production-intent LFP cells running through the full process. Trade reporting identified the output as D-sample cells: pre-commercial validation units, not product shipping to vehicle assembly. Score: SOP not verifiably met as commercial output. Capacity revised 35 to 20 GWh, jobs revised 2,500 to 1,700, LFP chemistry retained.

Toyota Battery Manufacturing — Liberty, North Carolina. The December 2021 state award covered $1.29 billion, 1,750 jobs, and production beginning in 2025. Toyota's October 2023 expansion added roughly $8 billion and about 3,000 jobs, bringing the campus to $13.9 billion, 5,100-plus planned employees, and 30-plus GWh through phased line launches by 2030. Toyota began shipping batteries in June 2025 and formally opened in November 2025. Hybrid production started; battery-electric and plug-in hybrid lines remain on the phased schedule. North Carolina reported 1,500 hires facilitated as of October 2025. The state agreement requires at least 4,500 jobs and $4.7 billion for Phase IIa, with a repayment trigger if Toyota holds fewer than 1,488 positions on December 31, 2029. Score: 2025 SOP met, the only unrevised SOP in the cohort. 30-plus GWh not yet due, 1,500 hired against a target not yet due, hybrid product met with BEV and PHEV staged.

Established-Company Joint Ventures

L-H Battery — Jeffersonville, Ohio. Honda and LG Energy Solution's October 2022 announcement committed to mass production by end of 2025, approximately 40 GWh, 2,200 jobs, and pouch-type EV cells for Honda. L-H declared mass production on July 2, 2026 — of stationary storage cells routed to LG Energy Solution Vertech, not Honda EV cells. Local reporting put employment at approximately 1,000. Current GWh output, chemistry, and yield were not disclosed. This publication's earlier audit established the gap between L-H's production declaration and what is publicly verifiable about available output. Score: EV SOP missed, storage production declared roughly six months past the original date. The 40 GWh figure was an EV number and does not transfer to the storage route. Jobs approximately 1,000 of 2,200. Product replaced.

Ultium Cells — Warren, Ohio. GM and LG Chem's December 2019 announcement committed to 30-plus GWh, 1,100-plus jobs, and EV cells, with LG targeting mass production in 2023. The plant began production in August 2022, ahead of schedule, and paused in January 2026. An Ohio WARN notice covered 1,334 workers — 850 temporarily, 484 indefinitely — citing slower EV adoption. The 850 were recalled by mid-August, though the union account had production ramping only after training. DOE's $2.5 billion Ultium loan was repaid in full in May 2025. Score: initial SOP met. Nameplate capacity built, but post-pause output is unverified. Jobs historically surpassed, then WARN-affected, with no current total disclosed. EV-cell product retained.

Samsung SDI–GM — New Carlisle, Indiana. Indiana's June 2023 announcement committed to operations in 2026, 30-plus GWh, approximately 1,700 jobs, and nickel-rich prismatic and cylindrical EV cells for GM. Samsung SDI announced in August 2026 that it was acquiring GM's stake and initially redirecting the plant to stationary storage. Construction had already paused with the building exterior complete. No revised SOP, GWh figure, chemistry, or hiring schedule was disclosed. County agreements require $3.5 billion of investment and 1,360 jobs — 85% of 1,600 — by December 31, 2027, and no extension was located. This publication's earlier audit covers the transaction mechanics. Score: 2026 SOP abandoned, the 30-plus GWh EV promise retired with no storage quantity stated in its place, jobs not met, product replaced.

BlueOval SK Battery Park — Glendale, Kentucky. Kentucky's September 2021 announcement committed to two 43 GWh plants — 86 GWh combined — production beginning in 2025, 5,000 jobs, and EV batteries for Ford and Lincoln. One plant began initial production in August 2025 with 1,450 jobs reported. Ford and SK then dissolved the JV; Ford closed its acquisition of both plants in May 2026, assuming a $3.805 billion DOE note. Ford now plans at least 20 GWh of LFP prismatic cells and containerized grid storage systems, first deliveries targeted for late 2027. A Kentucky WARN filing covered approximately 1,500 workers. The state amended its $250 million loan to Ford with a replacement commitment of at least 2,100 jobs. Score: SOP partially met, one plant of two. 86 GWh replaced by 20 GWh of storage capacity. 5,000 jobs abandoned, terminations filed, 2,100 substituted. Product replaced.

Established Cell Company

Panasonic Energy — De Soto, Kansas. Kansas's July 2022 announcement committed to approximately $4 billion and up to 4,000 jobs. Panasonic specified 30 GWh, mass production by end of March 2025, and 2170-format cylindrical EV cells. It declared mass production on July 14, 2025, roughly 3.5 months late, and raised the future annual-capacity target to approximately 32 GWh. The 4,000-job figure remains a target rather than a disclosed headcount. A June 2026 report indicated plans to convert some lines to data-center batteries. Kansas's 2025 APEX review recorded $377.25 million of investment tax credit earned against $3 billion-plus of certified capital. Score: SOP 3.5 months late. Capacity revised upward, 30 to 32 GWh, with current output undisclosed. Jobs not verified at 4,000. 2170 EV product met, with a partial future conversion announced.

Materials

NOVONIX Riverside — Chattanooga, Tennessee. NOVONIX's December 2022 update targeted 10,000 tonnes per year of synthetic graphite anode material, with roughly 3,000 tonnes per year beginning in 2024. The target was later raised to 20,000 tonnes per year. The July 2026 SEC filing reported that mass-production equipment for Panasonic had been installed and commissioned and a C-sample delivered, but formal customer validation remained pending and commercial production was expected in H2 2027. A Stellantis qualification effort stalled over specifications. The IRS certified Riverside's $103 million Section 48C manufacturing tax credit allocation in April 2026, conditional on the first 11,000 tonnes per year being placed in service before April 2028. Score: 2024 commercial production missed, now H2 2027. Capacity revised upward while remaining noncommercial. Jobs were not stated in the original expansion. Synthetic graphite retained as the product, with customer routes changed.

Gotion — Big Rapids, Michigan. Michigan's October 2022 announcement committed to $2.364 billion, 2,350 jobs, 150,000 tonnes per year of cathode material, and 50,000 tonnes per year of anode material. No defensible SOP date was published; the agreement used staged deadlines running through 2030–2031. Michigan found Gotion in default in September 2025 after 120-plus days of no eligible activity, and demanded return of $23.7 million in January 2026. The Sixth Circuit quoted Gotion acknowledging the project was "no longer viable." Michigan's FY2025 report recorded zero qualified jobs and zero disbursement of the $125 million company grant. Score: no SOP to score. Capacity, jobs, and product abandoned.

What the Cohort Shows

Four commitment dimensions across twelve facilities yield 48 cells. Four were never stated in the original announcement — Forge's capacity and chemistry, NOVONIX's jobs, Gotion's SOP — leaving 44 measurable.

N/S = not stated in original announcement (excluded from the 44 measurable cells).

Met as originally stated: 6. Toyota's 2025 SOP and hybrid product. Ultium's initial SOP, met in 2022, well before the January 2026 pause, and its EV-cell product. Ford Marshall's LFP chemistry, retained through the scope reduction. Panasonic's 2170 EV-cell product.

Formally revised: 9. Forge's SOP and jobs; Ford Marshall's capacity and jobs, via the Michigan board filing; Panasonic's capacity; NOVONIX's SOP and capacity; BlueOval SK's capacity, 86 to 20 GWh, and jobs, 5,000 to 2,100.

Missed or abandoned: 16. All four of Samsung SDI–GM's measurable cells and all three of Gotion's. Microvast on all four. ONE's capacity. L-H's SOP, capacity and product. BlueOval SK's product, where EV cells were replaced entirely by grid storage.

Not yet due: 5. Toyota's phased capacity and jobs. Panasonic's workforce target. ONE's job milestone, which runs to December 2029 under the state agreement. NOVONIX's product — synthetic graphite was retained, but the facility has not produced commercially, so there is nothing yet to score.

Resist clean classification: 8. These are the cells where company language and observable evidence diverge enough that a binary score would mislead. Ford Marshall's SOP: the company says production-intent cells are moving through the process, and trade reporting says D-samples, which are validation units rather than commercial shipments. ONE's SOP: pilot production satisfies the literal "beginning production" language, with 48 of 2,112 committed workers on site. BlueOval SK's SOP: one of two announced plants started. Ultium's capacity and jobs: nameplate was built and the original job target surpassed, but the January 2026 pause and the WARN filing leave current operating levels unverified. Panasonic's SOP at 3.5 months late reads as a revision or an approximate hit depending on the scoring rule. L-H's jobs, at roughly 1,000 of 2,200, are measured against a target tied to an EV program that no longer exists at the facility. ONE's product retained LFP chemistry while changing application from automotive to defense and infrastructure — a revision or a replacement, depending on how narrowly "product" is defined.

Four things the cohort data supports.

First revisions appeared 17 to 45 months after announcement, clustering mostly between 27 and 39. Ford Marshall's formal scope reduction at 17 months — February 2023 announcement, July 2024 Michigan board filing — is the fastest documented revision here. Microvast, ONE, Samsung SDI–GM and Gotion all reached first public revision or default between 35 and 39 months. BlueOval SK is the long-end outlier, with JV dissolution roughly 54 months after the September 2021 announcement, though one of its two plants did start production at the 47-month mark. Ultium Warren, having already met its original SOP, first revised at roughly 73 months when production paused. That spread is too wide to serve as a predictive rule. But an announcement surviving 18 months without a public revision has not thereby been validated; most revisions in this cohort arrived later than that.

SOP slips first. Capacity and job revisions follow with a lag. In most cases the first public revision was to the production timeline. Capacity reductions and job-count amendments were formalized afterward, frequently through state incentive amendments rather than company statements — Ford's July 2024 Michigan filing, Kentucky's April 2026 BlueOval loan transfer. Where both a SOP revision and a formal capacity or job revision are documented, the lag between them ran roughly 6 to 18 months. The state incentive amendment is often the first verifiable document carrying the revised numbers.

Product and application revise most severely. Five of twelve facilities changed intended end market or cell chemistry: L-H from EV to storage, Samsung SDI–GM from EV to storage, BlueOval SK from EV to LFP storage, ONE from automotive to defense and infrastructure, Microvast from NMC to LFP. Panasonic has announced a partial future conversion to data-center batteries. The EV-to-storage pivot concentrates in JVs where the OEM partner's EV program weakened or the JV dissolved outright. GWh announced in 2022 as EV NMC may, if it materializes at all, arrive in 2027 as stationary storage LFP.

Outcomes sort by sponsor type. The two OEM-owned facilities, Ford Marshall and Toyota Liberty, sit closest to their original commitments; Toyota is the only entry that met its SOP without revision and has not changed its product. Three of four JVs saw product conversion, JV dissolution, or both, and the fourth, Ultium Warren, held both product and structure but paused production for seven months. All three startups missed their SOPs, and two of the three, ONE and Microvast, revised their products — Forge's original announcement named no specific chemistry to measure against. Both materials facilities carry the same timeline risk as cell plants with an added dependency: customer qualification. NOVONIX cannot ship commercial product until Panasonic completes validation, and no amount of capital shortens that. My assessment, marked as such rather than derived from the table: the sorting reflects who could absorb a delayed ramp on their own balance sheet, which is not a variable the four commitment dimensions capture.

One thing the cohort cannot support is an industry-wide cancellation or success rate. As this publication's earlier analysis set out, announcement-vintage promise performance is not interchangeable with an industry cancellation rate, because public trackers use different universes, time windows, and definitions of reversal.