In April 2023 General Motors and Samsung SDI Co., Ltd. (KRX:006400) announced they would build a battery cell plant together in Indiana. The venture was finalised in 2024 at a headline cost of $3.5bn, construction started in 2025, contractors were laid off, and in May 2026 building stopped altogether.
On August 11 the board resolved to end the joint venture agreement. Samsung SDI will acquire GM's 49.99% of SDI-GM Synergy Cells Holdings LLC and run the entity alone, renaming it Samsung SDI Synergy Cells Holdings LLC. The New Carlisle site becomes its first wholly owned battery plant in North America.
The interesting question is not why GM left. It is what Samsung SDI thinks it just bought.
The filing gives one number and it is more useful than the $3.5bn headline. Samsung SDI's cumulative capital contribution to the venture before this decision was ₩275,283,991,858, which it scores at 1.2% of consolidated equity and 0.7% of total assets on the FY2025 balance sheet.
Against equity of ₩26.84tn at June 30 that's about one percent of the company. Against the project's announced cost it's a small fraction, because a plant of that size is funded over years and largely with debt as construction proceeds. Since construction stopped in May, most of the money was never spent.
That matters in two directions. It means GM's exit costs Samsung SDI relatively little in sunk capital, and it means the asset being taken over is closer to a permitted, partially built site than to a finished factory. A shell is easier to repurpose than a commissioned production line. It is also further from producing revenue.
What the filing does not say is what Samsung SDI is paying GM for the 49.99%. That figure is absent, and it is the single most important unknown here. A token price would say GM wanted out badly. A price near book would say the two sides parted on level terms. Nothing in the disclosure lets you tell.
Item two of the filing states the plan: operate as a wholly owned entity in order to respond to demand across various application markets "including ESS and electric vehicle batteries."
Energy storage is listed first. In a Korean regulatory filing that ordering is not accidental, and press reporting confirms the reading: Samsung SDI intends to add an energy storage system line to the plant now that it controls the design.
The logic writes itself. GM's original demand case was electric vehicles, and Samsung SDI's own statement cites slower-than-expected EV demand growth as the reason for the ownership change. Meanwhile grid-scale storage demand in the United States has been running hard, driven by renewables integration and by data centres that need firm power. A cell plant designed for automotive prismatic cells is not trivially convertible to storage duty, but it is far more convertible than it is saleable.
You can see the same pivot in the company's own numbers. Samsung SDI posted an operating profit of ₩203.8bn in the second quarter of 2026, its first positive operating quarter in six, after operating losses of ₩434.1bn, ₩397.8bn, ₩591.3bn, ₩299.2bn and ₩155.6bn in the five before it. Gross margin went from 8.8% in Q2 2025 to 24.9% in Q2 2026. Revenue in the quarter was ₩3.77tn against ₩3.18tn a year earlier.
A margin move of sixteen points in four quarters is a mix change, and the mix that changed is storage.
There is a coda that is easy to miss and that follows a familiar Korean pattern. Alongside the termination, the two companies signed a separate agreement to jointly develop next-generation prismatic cells for possible future electric vehicle use.
So the binding joint venture becomes a development agreement. Hanwha Aerospace did the same thing this summer, ending a four-year eVTOL supply contract with Vertical Aerospace and replacing it with a memorandum of understanding on the same day. The convention preserves the relationship, avoids the appearance of a rupture, and commits nobody to anything.
Whether it means more here is genuinely uncertain. GM still buys batteries and still needs a second source outside its LG venture. Samsung SDI still wants a US automotive customer. A prismatic development programme is a reasonable place to keep that option alive. It is not a purchase order.
Three concerns deserve airing.
First, the capital. Samsung SDI cut capital expenditure from ₩6.27tn in FY2024 to ₩3.07tn in FY2025, and spent ₩1.13tn in the first half of 2026, down about 40% year on year. It has been conserving cash after a year in which it lost ₩1.72tn at the operating line and raised roughly ₩2tn of new equity, lifting share capital from ₩356.7bn to ₩415.8bn in the second quarter of 2025. Taking sole ownership of a paused $3.5bn plant reverses the direction of travel, and the filing says explicitly that investment plans will change without giving a number.
Second, the customer. A joint venture with GM came with an implied offtake. A wholly owned plant does not. Samsung SDI now has to fill an Indiana factory with orders it has not disclosed, in a storage market where Chinese LFP cells set the price.
Third, the timing risk on US policy. A US battery plant's economics depend heavily on production tax credits, and the value of those credits to a foreign-owned, wholly owned facility making storage cells rather than automotive cells is a policy question rather than an engineering one. Nothing in the filing addresses it.
The case for the decision is straightforward and I think it's the stronger one. A half-owned plant whose partner has lost interest is the worst possible asset: you cannot repurpose it, you cannot sell it, and you keep funding it. Buying out a reluctant partner at what is likely a modest price, in order to redirect an early-stage site toward the one battery market that is actually growing, is the correct response to a bad situation. The alternative was to keep a stalled project on the books indefinitely.
A follow-up disclosure with a number in it. The filing commits to further announcements under the timely disclosure rules once investment plans firm up, and Korean rules require a separate filing for a new facility investment once the board approves one. Until that arrives, "we will run it alone and add ESS" is intent without cost.
The second marker is Q3, reported in late October, and specifically whether the operating profit holds. One positive quarter after five negative ones could be a mix effect, a one-off credit, or a genuine turn. The company also recorded an unusually low tax charge of ₩7.6bn on ₩479.2bn of pretax income in the second quarter, which flattered net income of ₩471.6bn considerably. Two consecutive clean operating quarters would settle whether the storage business is now large enough to carry the company while the EV business waits.
kstock reads DART every morning and writes up what moved — the contract, the buyback, the number that does not add up. The daily post and a Saturday roundup, by email.