Fewer Cancellations, Worse Signal

Second-derivative read on US battery facility status changes, Q4 2025 through Q1 2026

Period covered: October 1, 2025 – March 31, 2026. Facility universe includes cell, module, pack, cathode, anode, separator, and recycling across all sponsor types. Korean JV-specific narratives addressed only where they bear on the aggregate pattern. Sources graded inline: [A] = company disclosure, SEC filing, DOE record, WARN Act filing; [B] = trade press, local reporting, indirect inference.

New US battery facility cancellations and delays peaked in the December 2025–January 2026 window and have since decelerated. Read against the denominator, the deceleration looks like exhaustion: the pool of cancellable independent and startup projects is nearly empty. The negative status changes still propagating through the system are idlings, conversions, JV dissolutions, and ownership transfers inside the Korean supplier cohort that was supposed to constitute the reliable production core. The wave has changed character.

Denominator

Every cancellation rate requires a denominator, and the US battery manufacturing denominator is contested in ways that matter for any percentage cited against it.

CEA's tracker placed total announced US cell manufacturing capacity above 1 TWh as of mid-2025, with Korean and Japanese suppliers accounting for approximately 54% of announced 2026 cell production. Energy-Storage.news projected ESS cell manufacturing capacity reaching 50 GWh by end of 2026. These two figures measure different things: CEA measures share of a nameplate total; ESN measures a raw capacity sum including mixed EV/ESS facilities. Neither adjusts for utilization, qualification timelines, or the distinction between a facility already producing and one that has announced a start date.

The 1 TWh figure was always a theoretical ceiling. It stacked concurrent development assumptions, included first-phase announcements from companies that had not secured financing, and counted capacity expansions contingent on demand that did not materialize. When CEA reported 21 GWh of outright cancellations by mid-2025, the denominator against which that figure was measured had already been inflated by the same projects being cancelled.

For ESS specifically, CEA's most granular published figure: ESS-dedicated cell production capacity accounts for approximately 4% of total domestic cell production capacity in 2026. That 4% is measured against the same ~1 TWh announced denominator. If the denominator is overstated, the 4% understates the concentration problem by making the ESS-dedicated slice look like a small fraction of a large base rather than a thin absolute volume exposed to a handful of sponsors. What constitutes that thin volume is the subject of the rest of this memo.

Time-clustering: where the second derivative turned

Negative status changes within the six-month window are not evenly distributed. They cluster in two sub-periods with distinct causal signatures.

December 2025–January 2026. This concentrated the largest-magnitude events: Ford's cancellation of a $6.5B battery supply contract with LG Energy Solution [B], the dissolution of BlueOval SK (Ford–SK On JV) [B], GM's recognition of $7.6B in Ultium platform charges with concurrent idling of the Warren OH and Spring Hill TN Ultium Cells plants affecting over 2,000 workers [B], and Our Next Energy's January 6 layoffs following the loss of its largest EV customer. The timing reflects Q4 balance-sheet decisions catalyzed by the September 30, 2025 elimination of the 30D consumer EV credit under OBBBA and the widening gap between OEM EV sales trajectories and contracted cell supply commitments.

March 2026. Operational fallout from December decisions: 958 workers laid off at SK Battery America's Commerce, Georgia plant (approximately 37% of workforce, per WARN Act filing [A]), BlueOval SK Kentucky executing approximately 1,500 delayed layoffs [B], and Stellantis seeking exit from its StarPlus Energy JV with Samsung SDI.

February was quiet. The decisions had already been made. The second derivative turned negative most sharply in December 2025. By March the rate of new negative announcements was declining even as the operational consequences of prior announcements were still propagating.

Sponsor type and geography

Read across sponsor type, the pattern is unambiguous: startups and independents cancelled first (KORE Power and FREYR both in February 2025, pre-window), OEM-backed JVs dissolved or restructured second (December 2025–March 2026), and the Korean majors' own facilities absorbed workforce reductions last (SK Georgia, March 2026). Standard credit-quality waterfall.

Geographically, Georgia absorbed the sharpest concentration of negative events: FREYR's Coweta County cancellation and SK Battery America's Commerce layoffs eliminated both an independent ESS project and a major Korean operational facility in the same state. South Carolina holds the largest paused investment (AESC Florence, $1B sunk). Michigan saw both ONE's pivot and the termination of Gotion's Green Charter Township project with incentive clawback. Kentucky is the one state where the net signal is not negative: AESC Bowling Green remains under construction and Canadian Solar Shelbyville is on track for December 2026 SOP. Whether Kentucky's relative resilience reflects better project selection, different state incentive structures, or simply timing is not yet distinguishable.

Beyond cell manufacturing

The non-cell supply chain produced fewer headline cancellations in this window but showed parallel stress. Redwood Materials, the largest US battery recycling operation, executed two layoff rounds: approximately 5% of staff in November 2025 and approximately 135 employees (~10% of workforce) in early 2026, with COO Chris Lister's retirement announced in April [B]. NOVONIX's Chattanooga anode facility delayed mass production for its lead customer Panasonic Energy to H2 2027, from a previously indicated 2026 start [A]. Microvast's planned separator plant in Hopkinsville, Kentucky was cancelled in 2023 after DOE loan rejection, and no replacement US separator project has been announced in the window.

Module, pack, and system assembly operates on different capital intensity and shorter qualification cycles than cell manufacturing. Tesla Megapack (Lathrop, CA) and Fluence are system integrators sourcing cells, not cell producers. Assembly and shipping throughput at these integrators is ample. FEOC-compliant cell supply feeding those operations is scarce.

Facility status registry

Status as of March 31, 2026. Source grade in brackets.

Sponsor	Facility	Ann. GWh	Primary App.	Status Q1 2026	Key Signal
Panasonic	De Soto, KS	32 (full build)	EV (2170)	Operational, partial. Ramp delayed indefinitely	1 of 8 lines confirmed at opening; post-Dec line count unsourced [B]
Panasonic	Sparks, NV	~42 (incl. expansion)	EV (2170)	Operational. No Q4–Q1 negative signal	NOVONIX anode supply pushed to H2 2027 [A]
AESC → Fixx Energy	Smyrna, TN	3	ESS (LFP)	Operational. Sold to Fixx Energy March 31 (FEOC compliance)	New owner: no disclosed financial history [B]
AESC	Florence, SC	30 (Phase 1)	EV	Construction paused since June 2025; Phase 2 cancelled Feb 2025	$1B invested; BMW cell supply gap unresolved [B]
AESC	Bowling Green, KY	30–40	EV	Under construction; SOP target 2027. No negative signal as of June 2025	No named customer for a $2B, 30+ GWh facility [B]
Gotion	Manteno, IL	4	ESS (LFP) — pivot from EV	Operational, 5 lines; ~300 employees Nov 2025	FEOC-ineligible for 45X; Richardson partnership May 2026 [B]
Microvast	Clarksville, TN	4	ESS	Delayed indefinitely; ~50% complete; $150M financing gap	No public update since April 2024 [B]
KORE Power	Buckeye, AZ	12	ESS (LFP)	Cancelled Feb 2025; site for sale	DOE $850M conditional loan never materialized [B]
FREYR / T1 Energy	Coweta County, GA	34 (Phase 1)	ESS (LFP)	Cancelled Feb 2025; site sold $50M	Complete exit from cell manufacturing [B]
Our Next Energy	Van Buren Twp, MI	1	ESS → defense/rail	De facto cancelled for commercial ESS; 45% workforce cut	MEDC stopped disbursements at $70.2M of $237M [A/B]
Canadian Solar	Shelbyville, KY	3 (Phase 1)	ESS	Under construction; SOP December 2026 per Q3 2025 6-K	Most credibly-timed new ESS cell facility [A]
Kontrolmatik	Undisclosed US	3	ESS (LFP)	Delayed twice in 2024; no Q1 2026 update	Public record gap
ABF	Tucson, AZ	7.2	ESS	Delayed; no Q1 2026 update	Public record gap

AESC Bowling Green warrants a harder read than its "no negative signal" status suggests. A 30–40 GWh facility under construction with a 2027 SOP, no named customer, a parent company that simultaneously paused its South Carolina plant, and a corporate structure undergoing FEOC-driven divestiture at Smyrna is carrying more risk than the absence of a cancellation headline implies. The public record is silent. Silence at this stage of construction, with this parent company profile, is its own kind of data.

Conversions are not preserved capacity

The EV-to-ESS conversion wave is being cited, particularly in January 2026 trade coverage, as evidence that the US is approaching ESS cell oversupply. The reasoning: Korean EV cell plants are underutilized, ESS demand is growing, therefore conversion of idle EV lines closes the gap. Approximately right about direction and likely wrong about timing by a margin that matters for 2026 sourcing decisions.

A facility converting from NCM pouch cells for automotive to LFP prismatic cells for ESS requires new cathode supply qualification (LFP cathode sourcing from non-FEOC suppliers is itself constrained), new cell format tooling, new customer qualification cycles that typically run 6–18 months for utility-scale ESS, and new UL/safety certifications. The two operational LFP-for-ESS producers as of January 2026, LGES Holland MI and AESC Smyrna TN, both retooled existing EV lines over the course of 2025 [B]. I do not have month-specific data on when either conversion was initiated versus when first ESS cells shipped; "the better part of a year" is inferred from the 2025 calendar-year framing in trade reporting, not from disclosed conversion milestones.

The announced-but-unspecified conversions are less advanced. Ford's "Ford Energy" pivot for BlueOval SK Kentucky has no accompanying production commitment, SOP timeline, or ESS chemistry specification in the public record as of March 31, 2026. SK Battery America and Samsung SDI Kokomo have stated ESS production intentions for 2026 but have not disclosed customer qualifications or cathode sourcing. Stellantis is seeking to exit StarPlus Energy, not convert it.

A conversion announced is not a conversion completed. A conversion completed is not a qualification secured. A qualification secured is not volume production. Each transition takes months. Collapsing them into a single step produces the oversupply narrative. Tracking them sequentially puts credible converted ESS volume well into 2027.

What the trajectory implies for near-term ESS cell availability

FEOC-compliant, domestically produced ESS cells in volume for 2026 remain concentrated in LGES and, to a lesser extent, the Fixx Energy operation at Smyrna. Canadian Solar Kentucky, if it hits its December 2026 SOP [A], adds 3 GWh of Phase I capacity that will take additional months to ramp and qualify. Everything else is cancelled, delayed, converting, or FEOC-ineligible.

The six: KORE Power (12 GWh), FREYR (34 GWh Phase 1), ABF (7.2 GWh), Microvast (4 GWh), Kontrolmatik (3 GWh), and ONE (1 GWh). Two fully cancelled. One pivoted to defense. One half-built with no public update in two years. Two with no current public status.

Strip these and what remains of the ESS-dedicated base is: AESC Smyrna (3 GWh, now Fixx Energy), Gotion Manteno (operating but FEOC-ineligible for 45X), and Canadian Solar Kentucky (not yet producing). Apply three criteria simultaneously — operational, FEOC-compliant, financially established — and every facility fails at least one. Fixx Energy's Smyrna is operational and FEOC-compliant but owned by a startup with no disclosed financial history. Gotion Manteno is operational and financially backed but FEOC-ineligible. Canadian Solar Kentucky is financially backed and not itself a foreign entity of concern, but not yet producing. Whether the cells produced will meet FEOC material-sourcing thresholds under the MACR rules is not publicly described [A — research gap].

LG Energy Solution's stated ambition of 50 GWh North American ESS production capacity by year-end 2026 is the largest single number in this space. It is also a capacity target from a company that reported a 207.8 billion won operating loss in Q1 2026 on revenue of 6.6 trillion won, with AMPC credits falling to 189.8 billion won from 332.8 billion won in Q4 2025 [A]. Excluding AMPC, the operating loss was 397.5 billion won. LGES's US manufacturing economics are deteriorating in the quarter it is supposed to be scaling ESS capacity. The 50 GWh figure was stated during an earnings call. It is a capacity target, not a contracted production schedule.

The cancellation wave is decelerating because the startup and independent cohort has been substantially cleared out. The negative status changes now moving through the system are qualitatively different: JV dissolutions, workforce reductions at operational plants, FEOC-driven ownership transfers, indefinite ramp delays. These do not register as "cancellations" in most trackers. They register as capacity that exists on paper, may exist physically, but is producing cells at nothing close to nameplate rate. The announced capacity figures have not caught up to this reality because the trackers were built to count announcements. Announcements are the one thing this sector has never been short of.