COHU INC (COHU) stock intrinsic value fair value estimate & owner earnings analysis

Cohu designs and sells the machines that test and handle semiconductor chips during manufacturing: test handlers, test contactors, automated test equipment, inspection systems, plus spare parts, kits and service. Demand rises and falls with chipmakers' capital spending. Revenue was $583M in 2019, $887M at the 2021 peak, $636M in 2023, $402M in 2024 and $453M in 2025. The ten largest customers were 60% of 2025 revenue. Production is in Malaysia, the Philippines, Singapore, Japan and the U.S., partly through a Jabil plant in Penang. About 13% of 2025 sales ($61M) went to customers in China; the company is U.S.-based. How we measured earnings. We use cash from operations, remove the swings from receivables, inventories, payables and other working capital, and deduct stock compensation and the capital spending needed to keep the business running (growth spending was estimated at 2.3% of the revenue increase and set to zero in years when revenue fell). On that basis owner earnings per share were: 2019 -$0.40, 2020 $0.75, 2021 $3.09, 2022 $2.60, 2023 $1.05, 2024 -$0.65 and 2025 -$0.86. Adding back interest paid (the cash the business produces before its lenders, which is what we value) gives -$0.04, $1.14, $3.22, $2.67, $1.12, -$0.64 and -$0.85. We measure from 2019 because the Xcerra acquisition (October 2018) changed the company (revenue rose from $353M in 2017 to $583M in 2019), and 2019 to 2025 covers a full up and down cycle. The median is $1.12 per share (about $53M a year), and that is our normal level. Amortization of acquired technology ($37.5M of $50.7M depreciation and amortization in 2025) is a non-cash charge and is added back; the technology it represents is kept current by research spending of $92M a year (20% of sales), which is already charged against earnings. For 2023 to 2025, the gap between simple net income plus non-cash charges and our figure is explained by capitalized software and other long-term asset spending, lease payments, and working capital timing; the cash flow statement rebuilds exactly. Growth path check. Owner earnings per share fell in 2025 (-$0.65 to -$0.86). Our base case still starts year 1 at $0.40, because the newest quarterly report shows a turn: first-half 2026 sales were $274.1M versus $204.5M a year earlier (+34%), the second quarter had an operating profit of $0.3M versus a $17.2M loss a year earlier, and the order backlog was $165.1M at December 27, 2025 versus $138.0M a year before. We did not raise the normal level on this; it takes two or three more quarters of corroboration, and the first-half working capital detail was incomplete. Balance sheet. At June 27, 2026 cash and short-term investments were $498.2M. We treat $100M as needed to run the business, leaving $398.2M as excess. Debt at December 27, 2025 was about $305M: $287.5M of 1.50% convertible notes due January 15, 2031 (conversion price about $27.18; conversions are settled in cash up to the principal, and the company bought capped calls) plus about $17.6M of bank and construction loans. Because the notes cost 1.5% versus our 6% yardstick for their remaining 4.3 years, we reduce the debt figure by the present value of the difference, about $47.7M, giving adjusted debt of $257.3M. Excess cash less adjusted debt is a net $140.9M, or $2.98 per share on 47.3M diluted shares. The 2031 maturity is covered by cash on hand in every case below. Cases (each is business value plus that $2.98). - Bear, $2.96 per share. The business is worth about -$1M (essentially nothing), the company owes $257M against $398M of excess cash, a net of about $141M. The story: the 2024-2025 pattern of losing about $0.75 per share a year lasts three more years (about $35M a year of cash burn, funded from the cash), a partial recovery to only $0.35 per share follows (about a third of normal) because the ten largest customers are 60% of sales and the company reports price pressure in mature product lines, and this fades to zero over five years. The recoveries after the weak 2019 year (2020 and 2021) are why we allow any recovery at all. - Base, $12.55 per share. The business is worth about $453M ($9.57 per share), the company owes $257M and holds $398M excess cash, a net of about $141M. The story: $0.40 in year 1, $0.80 in year 2, then the $1.12 normal level held flat through year 10, with no growth credited, then a straight-line fade to zero over ten years, because nothing we can name and evidence sustains the products beyond that. - Bull, $20.30 per share. The business is worth about $820M ($17.32 per share), the company owes $257M and holds $398M excess cash, a net of about $141M. The story: $0.80 in year 1, $1.60 in year 2, then $2.00 a year (about $95M) for eight years, which equals the 2020-2023 average, followed by a ten-year fade to zero. This uses the current backlog and first-half sales recovery and a level the company has already delivered. What we left out. We give no credit to market expansion from AI, high-bandwidth memory, or silicon carbide chips, to Tignis (the AI software business bought in January 2025, whose 2025 sales were immaterial), to margin gains from the Poway restructuring ($10.1M of 2025 restructuring cost), or to share gains, because none is yet contracted or shown in earnings. Growth beyond a return to the normal level is not credited at all; the recovery from the current trough to that level is a return to earnings the company has already delivered, not growth. Compared with our earlier estimate (bear $5.00, base $13.50, bull $37.25): we now use one fixed 6% discount rate for every company, credit no growth, value the business before its lenders, hold back $100M of operating cash, and rely only on levels the company has already reached. This lowers the bear and bull cases and leaves the base case close to before. The prior bull case also assumed 4% growth forever, which we cannot evidence; ours fades to zero. What would make this wrong. Upside: a sustained upcycle that lifts revenue well above $636M. Downside: revenue stuck near $400-450M with continued cash burn, or the loss of a top-ten customer, would erode the excess cash; net cash is about a quarter of the base value and nearly all of the bear value. The margin-of-safety price of $7.50 is roughly 40% below the base case because the business is cyclical and concentrated, and the bear case is mostly balance-sheet cash.