What happens to a gigafactory built for 670,000 cars a year when the cars stop arriving?
Samsung SDI answered on Oct. 26 with a contract. Its Michigan subsidiary, Samsung SDI America, will ship 1 GWh of energy storage systems to Alpha Omega Power by Dec. 31. Add an earlier order signed late last year and the two agreements total 1.5 GWh. Every cell and module comes from the StarPlus Energy plant in Kokomo, Indiana.
Kokomo was financed as a vehicle factory. The Department of Energy closed a $7.54 billion loan there in December 2024, sized for roughly 67 GWh a year and about 670,000 cars. The loan page still lists electric vehicle batteries as the product.
The chemistry was already right
The product going to AOP is the Samsung Battery Box 1.7, a 20-foot integrated unit built on high-nickel NCA prismatic cells. Samsung SDI says it stores 17% more energy than the previous model in the same footprint, is designed for 25 years of continuous use, and carries NFPA 855 and UL 9540 certification. Proprietary Enhanced Direct Injection puts fire suppressant directly into individual modules.
Those certificates are not decoration. Developers now settle fire safety and siting questions before permitting begins, and a pre-certified enclosure shortens the path from signed contract to energization. A storage order that clears those gates is worth more than one that restarts them.
Storage volume under the two AOP deals
Half of the total was contracted late last year; the new 1 GWh tranche ships by Dec. 31, 2026. · Samsung SDI, Oct 2026
AOP is a Houston developer and independent power producer owned by Fengate. It is not a speculative buyer. Its 100 MW / 400 MWh Caballero project in California reached full operation in June 2025, and a 200 MW Vertus battery in Galveston County began operating this month, Energy Capital HTX reported. The Kokomo systems go to projects in the ERCOT and WECC markets that begin construction in 2027.
That is the strongest version of the first position. Kokomo already had the cell chemistry, the module line and the safety file. The EV slowdown left it with capacity. Storage demand arrived with a reason to pay for domestic origin. Idle capacity met a buyer with a deadline, and both sides could price the trade.
The unit economics of a 20-foot enclosure help here. Storage buyers at grid scale compete for land, interconnection and time; energy per container decides how much of all three a project needs. A 17% density gain is a smaller number of containers per megawatt-hour, fewer truck movements, and a shorter build sequence at the substation.
Samsung SDI goes further, describing itself as the only non-Chinese supplier of prismatic storage batteries in North America. The company cited no market data for that claim. The release also does not say whether AOP's systems will carry supplier certifications, or what share of their costs those certificates cover.
The premium is a threshold, not a moat
Federal storage credits now exclude projects that take material assistance from a prohibited foreign entity. The IRS explained how to calculate that ratio in Notice 2026-15, published Feb. 12 under the One Big Beautiful Bill Act. Storage that begins construction in 2026 must show at least 55% of direct manufactured costs from sources other than prohibited foreign entities. Five points a year, then 60% for 2027.
AOP's projects are 2027 vintage. The cells ship in 2026 and get installed under the stricter test a year later. That timing is the whole margin question.
| Test | Construction starts 2026 | Construction starts 2027 |
|---|---|---|
| Domestic content floor | 55% of direct costs | 60% of direct costs |
| Safe harbor | Notice 2026-15 tables | Tables due Dec. 31, 2026 |
| Substantiation | Supplier certification | Supplier certification, penalties for false claims |
Thresholds for storage under Notice 2026-15 and the One Big Beautiful Bill Act
Now the second position sharpens. The premium for a Korean cell is set by a statute and a threshold, not by a manufacturing edge that rivals cannot copy. Korean and Japanese suppliers can build the same line. A US plant owned by a compliant entity can undercut it. Safe-harbor tables due by Dec. 31, 2026 will fix the arithmetic for every buyer in the queue, and the arithmetic is public.
Two numbers are missing from the announcement. The contract value and the project names. Neither company disclosed them. AOP chief executive Paul Choi said the second agreement strengthens the company's domestic supply chain. That is a supply-chain statement, not a price.
The earlier deal at the same plant points the same way. Energy-Storage.News reported a KRW 1.5 trillion, or about $1 billion, four-year supply agreement with an undisclosed buyer. Undisclosed buyers and undisclosed values are normal in this market. They also mean headline capacity says nothing about margin. That is the gap.
What a 60% floor does to a project finance file
If the ratio clears 60%, the credit survives and the offtake holds. If it does not, the project either re-sources cells or loses the credit, and the offtake gets re-cut.
Watch: whether AOP's filing names the cell source and attaches the certification, or whether both stay behind the contract.
Origin rules work like technical specifications. As we wrote in October, ERCOT turned grid-forming capability into a spec, and that rewrote battery economics in the state where AOP wants to build. A domestic-content floor is the same move. A procurement filter becomes a financing condition, and the filter decides who gets built.
What the 2027 vintage will prove
If the deals get re-cut once the safe-harbor tables land, or if storage pricing moves toward LFP and away from nickel-rich NCA, the premium was rent on a threshold.
Three items will settle it. First, whether AOP's projects appear in ERCOT and WECC interconnection queues with a named cell source and supplier certification attached. Second, whether the February tables shift the effective floor for battery systems, which would re-price every 2027 contract signed before them. Third, whether StarPlus reports storage revenue separately from vehicle cells. A factory with two disclosed product lines is a different asset from one that reports a pivot.
The chemistry argument is real, and so is the policy argument. Both cannot be decisive at once. A supplier that wins on both does not need to explain the difference.
Kokomo's cells were specified for cars. In 2027 they will be judged on a ratio. The plant did not change. The buyer did, and so did the test.