Cohu’s recovery is real, but Q4 exposed the wrong bottleneck: mix and product pruning, not demand

Cohu did not miss the quarter on demand; it missed the quarter on earnings quality, and that distinction is the variant perception in this print. Consensus was effectively priced for a revenue event with $122.1 million expected, and the company delivered $122.2 million, a +0.1% surprise that should not change anyone’s view of test demand. What was not priced in was a -$0.15 EPS result against $0.07, a -314.3% surprise, and the reason matters more than the headline loss: the reported recovery is not yet translating into the margin structure investors associate with Cohu’s normalized model. The bull case can still work, but it has to pass through a very specific gate. Management is telling investors that $130 million a quarter can produce gross margin in the high 46% range, $150 million can approach just under 48%, and $160 million can produce 48%; the current quarterly history shows $122.2 million revenue and 34.1% gross margin in Q4 FY2025, followed by $125.1 million and 40.4% in Q1 FY2026. That is an improvement, but not yet the economics embedded in the recovery narrative.

The clean separation between priced-in and surprise is important because it prevents the wrong trade. Priced in was stabilization: revenue had already recovered from $94.1 million in Q4 FY2024 to $126.2 million in Q3 FY2025, and the Street’s $122.1 million estimate was basically at management’s own Q4 level. Actually surprising was the collapse in EPS relative to the Street, plus the persistence of margin leakage below the model implied by management’s revenue sensitivity. The quarterly path makes the point. Revenue troughed at $94.1 million in Q4 FY2024, rose to $96.8 million in Q1 FY2025, $107.7 million in Q2 FY2025, and $126.2 million in Q3 FY2025, before easing to $122.2 million in Q4 FY2025. Yet gross margin moved from 38.5% to 33.5%, 34.4%, 35.7%, and then 34.1% over those same quarters. A PM paying for recovery should demand operating leverage; this print delivered revenue stabilization but not the margin capture that normally follows.

The financial trajectory therefore reads less like a demand disappointment than a self-help transition with a still-unproven payoff. Jeffrey Jones gave investors the company’s own account of the gap when he said, “The Q4 gross margin of 40.8% was lower than guidance due to onetime inventory charges resulting from discontinuing certain product lines and consolidating offerings, which better align our engineering and support resources with customer requirements.” The quote earns attention because it commits the margin shortfall to product-line actions rather than broad pricing pressure, but it also raises the burden of proof for Q1. The data pack shows Q4 FY2025 gross margin at 34.1%, while the call excerpt states Q4 gross margin of 40.8%; those are different reporting bases, so they should not be blended. The investment question is the same on either basis: if the product pruning is genuinely one-time, Q1 FY2026 must show the reported gross margin recovery already visible in the history table at 40.4%, and then management needs to bridge from there toward the 46.7% to 46.8% range it tied to $130 million quarterly revenue.

That bridge is not just accounting; it is the fulcrum for EPS. Q4 FY2025 revenue was $122.2 million, essentially flat with the estimate, but diluted EPS was -$0.48 in the quarterly history and -$0.15 on the Street-comparison basis in the print. Q1 FY2026 in the history table shows $125.1 million revenue, 40.4% gross margin, and -$0.26 diluted EPS, which tells us that even a margin rebound of that size is not yet enough to restore positive earnings under the reported quarterly basis. Management’s Q1 call framework points in the same direction: operating expenses are expected to be about $50 million, after Q4 operating expenses of $49.8 million were in line with guidance. At this revenue scale, small changes in gross margin do not overcome roughly $50 million of quarterly operating expense quickly. That is why the Street miss is not just a backward-looking tax or inventory issue; it exposes how much revenue and mix Cohu still needs before the model can clear the fixed-cost base.

The capital structure changes sharpen that interpretation because they reduce balance-sheet risk while increasing the need for operational proof. Cash and investments increased by $286 million during Q4 to $484 million at year-end, while total debt is $305 million, including $288 million from the Q4 convertible debt offering. Jones described the transaction as “raising gross proceeds of $287.5 million at attractive rates, including 1.5% interest rate, 32.5% conversion premium and a five-year term.” The financing gives Cohu time, and the capped call matters because management says it purchased a 100% capped call to limit shareholder dilution until the stock price doubles and exceeds $41 per share. But time is not the same as earnings power. Q4 capital expenditures were $3.4 million, full year 2025 capital expenditures were $21 million, and 2026 capital expenditures are targeted at about 2% of revenue; the company is not spending like a business that needs a capacity build to recover. The limiting variable is utilization, product mix, and conversion of design wins into higher-margin systems and recurring revenue.

That is why the order and end-market comments are more constructive than the EPS line, but still not enough to buy the stock without a margin checkpoint. Jones said, “For the full year 2025, orders increased 29% year-over-year,” and full year revenue of $453 million was 13% higher year-over-year. The order growth being more than the revenue growth supports the idea that the recovery has not fully flowed through the income statement yet. The market may be missing that the revenue base is not deteriorating after the Q3 FY2025 step-up, because Q4 FY2025 still grew +29.9% year-over-year and Q1 FY2026 grew +29.3% year-over-year in the quarterly history. But the market is right to penalize the stock if it treats the EPS miss as evidence that the recovery is lower quality than expected. The variant view is not “ignore the loss”; it is that the loss is survivable if, and only if, gross margin normalization follows product consolidation and the revenue floor holds near the current $122 million to $125.1 million range.

The mix detail helps explain why the debate is not just cyclical beta. On the call, Jones said fourth-quarter company-reported revenue of $123 million was up 30% year-over-year and split 40% systems and 60% recurring, while also stating that for Q4 2025 revenue was in line with guidance at $122.2 million. Those are company-account and guidance-context figures, not the Street-comparison basis, but the split is still useful. A 60% recurring mix should have made the quarter feel more defensible, yet the EPS result showed that recurring revenue alone did not offset inventory charges, tax reserves, and operating expense. Estimated test sterilization was up 1 point to 76% at the end of December, with computing at 78% and automotive at 75%; that says utilization is improving, but not uniformly enough to pull Cohu toward the $130 million quarterly level where management says gross margin should be in the 46.7% to 46.8% range. The computing strength matters because Neon exited 2025 at $11 million on the HBM market, and HBM-related demand is the cleanest path for Cohu to move from stabilization to mix-led expansion.

The read-through for customers is correspondingly specific rather than broadly bullish. For ASE Group and Amkor, both listed customers for Cohu test handlers and contactors, the useful signal is that Cohu had no customer above 10% of total sales for the full year 2025, while recurring represented 60% of company-reported fourth-quarter revenue and estimated test sterilization reached 76% at the end of December. That combination implies broad-based service and consumables pull rather than a single-customer capital spike, which is incrementally supportive of outsourced assembly and test activity at ASE Group and Amkor. The magnitude is bounded: Cohu’s full year revenue was $453 million, Q4 was $122.2 million on the guidance basis, and orders were up 29% year-over-year, so the read-through is for improving utilization and contactor demand, not a sudden capacity supercycle. There are no named suppliers in the data pack, so there is no defensible supplier implication to stretch.

The peer frame also argues against calling this a sector problem. In the test and assembly peer set, ATEYY posted +43.8% revenue YoY with 67.4% gross margin, DSCSY posted +12.3% revenue YoY with 70.8% gross margin, 6871.T posted +48.3% revenue YoY with 47.3% gross margin, and 6315.T posted +22.6% revenue YoY with 36.2% gross margin. Cohu’s Q4 FY2025 revenue YoY was +29.9%, which sits closer to the better growth cohort than the laggards, but its 34.1% gross margin sits below 6315.T’s 36.2% and well below 6871.T’s 47.3%. That comparison matters because the sector tape can reward test exposure while still refusing to pay a normalized multiple for Cohu until its gross margin proves the revenue recovery is not being absorbed by charges, mix, or underutilization. The company is growing like a recovery name, but reporting margins like a business still in restructuring.

The call delivery reinforces that split signal: management sounded more confident on guidance, but the Q&A quality did not improve enough to erase execution risk. The tone history shows guidance_tone rising to 0.49 in Q1 FY2026 from 0.33 in Q4 FY2025, while prepared_sentiment rose to 0.65 from 0.62. At the same time, qa_sentiment fell to 0.16 from 0.30, uncertainty rose to 82.1 from 76.6, and tone_confidence stayed low at 0.17 after 0.17. That is exactly the pattern investors should expect when a company can frame a recovery in prepared remarks but still faces pushback on the mechanics. The improvement in qa_evasiveness to -11.1 from 12.2 is a positive, but it conflicts with the higher uncertainty index of 82.1; management answered more directly by that measure, while the underlying range of outcomes widened. The right reading is not that the call was promotional or weak. It is that the script is ahead of the proof.

That proof has to arrive in the next quarter because management has given investors unusually concrete levels. Jones guided Q1 revenue to “approximately $122 million, plus or minus $7 million,” and also said the Q1 tax provision is expected to be about $5.5 million with diluted share count projected at approximately 48.5 million. The history table shows Q1 FY2026 revenue at $125.1 million and gross margin at 40.4%, which would satisfy the revenue floor but still leave the model short of the company’s own gross-margin sensitivity at $130 million. The critical watch item is whether revenue can clear the guided band’s center of $122 million and move toward the $130 million level management tied to 46.7% to 46.8% gross margin, without operating expenses moving above about $50 million. If revenue is stuck around $122 million and gross margin fails to hold the Q1 FY2026 40.4% level, the thesis breaks because Q4’s charges were not isolated enough. If revenue moves toward $130 million, gross margin approaches the high 46% range management described, and tax stays near the $5.5 million guide, the market will have to re-underwrite Cohu as a margin recovery story rather than an EPS miss.