CPS TECHNOLOGIES CORP/DE/ (CPSH) stock intrinsic value fair value estimate & owner earnings analysis

CPS Technologies makes metal matrix composites (aluminum silicon carbide baseplates and heat spreaders), hermetic electronic packages and related armor and thermal-management products in Norton, Massachusetts. Nearly everything it ships is custom-built to a customer's design. About 81 percent of 2025 revenue came from commercial uses such as rail traction, wind and other power electronics, and 19 percent from defense. Customer concentration is the defining feature: three customers accounted for 64 percent of 2025 revenue, with the largest at 39 percent, and the remaining 36 percent came from about 43 smaller accounts. The company sells on purchase orders, not long-term supply contracts. Backlog was $26 million at December 27, 2025, mostly due to ship in 2026, and customers can reschedule or cancel it. Earnings record. We measure owner earnings as cash from operations, with year-to-year swings in inventory, receivables, payables and customer deposits taken out, less stock compensation and the capital spending needed to keep the plant running. Capital spending has averaged about 2.5 percent of sales. On that basis the ten years from 2016 to 2025 added up to a combined loss of about $1.4 million: four straight negative years from 2016 to 2019 (a combined loss of about $3.9 million while revenue dropped from $21.8 million to as little as $14.6 million), then $1.1 million in 2020, $0.5 million in 2021, a peak of $3.0 million in 2022, $1.7 million in 2023, negative $4.5 million in 2024 when gross profit went below zero on a revenue drop of about 23 percent, and $0.8 million in 2025 on record revenue of $32.6 million and a 16 percent gross margin. Taking out working-capital swings matters here, for example the $2.5 million of customer deposits run down in 2023 and the $1.3 million inventory build in 2025. Our previously published figures were prepared on a different basis and are not comparable. We judge the business as it has been since 2020, at more than $20 million of revenue and including the 2024 trough. The median of those six years is about $1.0 million a year, or roughly 5 to 7 cents a share. Interest is negligible because the company has no borrowings, so cash before lenders is the same as owner earnings. The first half of 2026 was weak: revenue of $15.3 million was flat with the year before, there was an operating loss of $0.8 million, and operations used $2.5 million of cash, largely because inventory grew by $3.1 million. In 2025 owner earnings swung from a large loss to a small profit, so no meaningful growth rate can be stated. Our first forecast year therefore assumes a loss, below what was actually achieved in 2025. Balance sheet. At June 27, 2026 the company held $15.4 million of cash and $3.8 million of Treasury bills, about $19.2 million in total, after raising equity in late 2025 and $9.0 million more in the first half of 2026. It has no financial debt. The only obligation is the Norton facility lease, which runs to February 2028 and is a normal operating cost. We set aside $3 million as working cash for a business this size, which leaves about $16.2 million, roughly 83 cents across about 19.5 million diluted shares. Share count rose from 14.5 million in 2024 to 19.4 million by July 2026, so owners have already paid for the losses through dilution. The company has said it is considering a move to a larger facility near Norton and warns that the move could be costly, disrupt production and require customers to requalify parts. The landlord of the current building went into receivership in December 2025. Bear case (about $0.61 a share). One downturn of the kind this business has already been through, sized from its own 2016-2019 and 2024 record: a loss of $1.5 million in the first year, $3.5 million in the second as the relocation coincides with a soft order book, then $1 million losses in each of the next two years, followed by six years at about $0.5 million a year, with no value after year ten because nothing in the record shows durable earnings. The business is worth about negative $4.2 million in this case. With no debt and $16.2 million of surplus cash, the equity is worth about $12 million, or $0.61 a share. Shareholders are protected by the cash, not by the operations. Base case (about $1.04 a share). A loss of about $1 million in 2026, consistent with the first half; roughly break-even in 2027 after about $2 million of relocation cost; then owner earnings of about $0.9 million a year, slightly below the post-2020 median through year ten, fading to nothing over the following decade. We credit no growth. The $26 million backlog supports revenue staying near current levels, but the company has never kept up a rising level of cash earnings, and the expansion the new facility is meant to serve is a plan, not a contract. The business is worth about $4.1 million in this case. Adding the $16.2 million of surplus cash and subtracting no debt gives about $20.3 million, or $1.04 a share. Bull case (about $1.83 a share). The larger revenue base produces cash on the 2022-2023 pattern: about $0.3 million in 2026 and $0.5 million in 2027 net of the move, then about $2 million a year, below the $3.0 million achieved in 2022 on lower revenue, held through year fifteen and then fading over the following nine years. The business is worth about $19.5 million in this case. With the surplus cash added that is $35.7 million, or $1.83 a share. We left out any defense or new-product wins because none is under contract. What could make this wrong. The largest single risk is the 39 percent customer. Losing it or seeing it shrink, as happened in 2024, would put the business back into steady cash burn. A relocation that goes badly, with parts that fail requalification, could do the same while the cash pile shrinks. A long run of losses funded by further equity raises at low prices would reduce value per share below the bear case. On the other side, the decade-long record of break-even owner earnings may understate a business that is now 50 percent larger than in 2023. Margin-of-safety price. We are not confident in the earnings level. Ten years of cash earnings net to a small loss, and the six-year median of about $1 million depends on a single good stretch in 2020-2023. The worst plausible outcome is the largest customer leaving during the move. The business has no debt, so it would survive, but the cash would be spent on losses and moving costs instead of reaching shareholders. If we are wrong about the median earnings level and about the move, the shares are worth roughly the cash that remains after a few years of losses, somewhere around 50 to 60 cents. The business pays no dividend, so any return in the cases has to come from the price moving toward value. The price assumes a holding period of about three years, and at a 20 percent annual return that means the starting price has to be well below where the base case ends up. The margin-of-safety price is $0.65. That price is less than the company's cash and Treasury bills after a working reserve, with the operating business carried at nothing, and the base case implies about the target return if it is recognized within roughly two and a half years. From that price the bear case implies only a small loss.