Cohu’s miss was not demand weakness, it was mix and timing hiding an early recovery signal

The variant perception in this print is that Cohu’s headline revenue miss is less informative than the composition of the recovery now forming underneath it. What was priced in was a weak quarter: the Street expected revenue of $114.5 million and EPS of -$0.02, which implied investors were braced for continued under-absorption and no immediate earnings leverage. What actually surprised was split: revenue of $107.7 million missed by -5.9%, but EPS came in at $0.02, a swing from an expected loss to a profit. That combination matters because the miss did not come with a management retreat on cycle language. Jeffrey D. Jones put a stake in the ground on utilization, saying, “Our estimated test cell utilization increased by 3 points quarter-over-quarter to 75%, which historically indicates the industry is entering a recovery cycle.” The number is not large enough to declare a clean upcycle, but it is specific enough to challenge the bear case that Q2 was simply another demand disappointment.

The reason to lean into the recovery interpretation, rather than fade it on the revenue miss, is that Cohu’s reported trajectory has moved off the trough while earnings remain optically ugly on the GAAP series. Revenue has been pinned near the bottom of the cycle since Q1 FY2024, but Q2 FY2025 finally put up $107.7 million after the trough at $94.1 million. The company’s own street-comparison print still disappointed versus $114.5 million, yet the sequential shape is now positive rather than just stabilizing. The income statement table shows the damage clearly: gross margin fell from 48.1% early in the downturn to 34.4% in Q2 FY2025, so the market’s skepticism on operating leverage is not irrational. The point is narrower and more actionable: when revenue inflects before GAAP gross margin repairs, the first debate should be whether orders and utilization are pulling forward, not whether last quarter’s absorption already proves the model has healed.

That financial trajectory explains why the EPS beat carries more signal than it first appears. On the Street basis, $0.02 versus -$0.02 is not an earnings level that changes valuation by itself, but it changes the burden of proof because it came on $107.7 million of revenue rather than the expected $114.5 million. In other words, Cohu did not need the full Street revenue base to avoid the expected loss. The company’s own account was also careful to separate reported revenue from non-GAAP margin: Jones said, “Our revenue for the second quarter of 2025 was just under $108 million with a non-GAAP gross margin of 44.4%.” That quote earns attention because the non-GAAP gross margin is much higher than the 34.4% gross margin in the quarterly history, which means investors must avoid mixing bases when assessing leverage. The defensible takeaway is not that margins are fixed; it is that the P&L contains enough adjustments and mix effects that the Street revenue miss alone is a poor proxy for cash earnings capacity.

The Q3 guide is the bridge from “stabilization” to “recovery,” because management did not merely say conditions were better, it attached revenue mix and timing to the improvement. Cohu guided Q3 revenue to approximately $125 million, plus or minus $7 million, with systems expected to be approximately 47% of revenue. That matters because recurring revenue is expected to be flat quarter-over-quarter, so the implied growth is not a broad-based consumables rebound, it is systems-heavy and tied mainly to test automation systems for the mobile market. A systems-led bounce can be lower quality if it is one-off, but it is exactly what an early-cycle test equipment recovery often looks like before recurring revenue follows installed-base activity. The market may be mispricing Q2 as a miss against a stale estimate, while Q3 guidance says the business is already being pulled into a higher revenue band than the trough quarters.

The risk to that thesis is that operating expenses are not yet giving investors much room for error. Q3 operating expenses are projected to be about $50 million, and management said the post-restructuring target is approximately $49 million per quarter when revenue is approximately $130 million. Those numbers frame the earnings debate cleanly: Cohu is guiding near the revenue level that should support the reset cost base, but it is not there with much cushion. The $2 million of variable R&D product development prototype materials in Q3 also means near-term expenses are rising with product activity rather than falling mechanically with restructuring. This is why the right view is not a generic “buy the recovery” call. It is a call that the market is overly focused on the -5.9% top-line miss and underweighting evidence that Cohu is entering the part of the cycle where systems revenue returns before GAAP margins normalize.

The customer evidence adds substance to the recovery claim because the wins are not just macro commentary. Cohu landed $3.5 million in new customer wins in Q2 across testers, handlers and inspection systems, and management also identified a revenue stream opportunity of approximately $20 million in precision analog tied to qualification of the Ultra-S contactor at a leading IDM customer. The latter is more strategically relevant than the former because contactors sit closer to recurring and application-specific attach than a generic tester win. For customers ASE Group and Amkor, both listed as users of Cohu test handlers and contactors, the implication is not immediate demand acceleration across their whole outsourced assembly and test base. The sharper read-through is that handlers and contactors are where utilization-sensitive spending first appears, and Cohu’s 75% test cell utilization datapoint gives those customers a concrete utilization threshold to watch for capacity pull-through.

That supply-chain read-through is also important because Cohu has no named suppliers in the data pack, so the second-order signal is demand-side rather than upstream inventory restocking. ASE Group and Amkor are the relevant names because Cohu’s exposure to test handlers and contactors maps directly to outsourced test cell activity, and the 3 point quarter-over-quarter utilization increase suggests incremental consumption of interface hardware before broad capex declarations. The magnitude investors should attach is bounded by the company’s own figures: $3.5 million of Q2 new wins is helpful but not enough to move Cohu alone, while the approximately $20 million precision analog opportunity is material relative to Q2 revenue of $107.7 million. If that opportunity starts converting into recognized revenue, it would validate the idea that the current recovery is product-led rather than merely a cyclical bounce off depressed utilization.

The peer context reinforces the same caution: Cohu is recovering, but it is not yet keeping pace with the stronger parts of test and assembly. The peers table shows several subsector companies posting much higher latest-quarter revenue YoY growth, including ATEYY at +43.8% and 6871.T at +48.3%, while Cohu’s Q2 FY2025 revenue YoY was +2.8%. That gap argues against treating Cohu as a fully de-risked cyclical winner today. At the same time, Cohu’s setup is different because the base is still depressed and the Q3 guide points to an acceleration from $107.7 million to approximately $125 million. The comparative point is therefore not that Cohu has already matched peers; it is that Cohu may have more operating controversy left to resolve, which can be attractive if the guide proves demand rather than timing.

The call delivery supports the idea that management became more specific without becoming euphoric. In the tone history, Q2 FY2025 sentiment rose to 0.31 from 0.07 in Q1 FY2025, while uncertainty fell to 76.9 from 100.6. That is the kind of tone shift that matters because it coincided with concrete operating guideposts: 75% utilization, approximately $125 million Q3 revenue, and systems at approximately 47% of Q3 mix. Prepared sentiment at 0.56 was far higher than Q&A sentiment at 0.10, however, which means the confidence was more controlled in scripted framing than in investor interrogation. That split is not a reason to dismiss the recovery, but it is a reason to demand confirmation in bookings, systems shipments, and tax-adjusted EPS rather than simply accepting management’s cycle label.

The tone chart also helps separate this Q2 event from later call patterns in the data. Across the broader tone history, Q1 FY2026 shows guidance_tone at 0.49 but uncertainty at 82.1, a combination that says better guide language can coexist with more question marks. For the Q2 FY2025 event, the useful feature is not pure optimism; ai_optimism was 0.52, below the Q1 FY2025 reading of 0.85. The useful feature is that uncertainty fell while management attached dollar amounts and mix to the next quarter. That is why this print should not be read as a promotional narrative papering over a revenue miss. It is better read as a company moving from “trough management” to “early-cycle qualification,” with the caveat that Q&A sentiment at 0.10 leaves room for investor pushback on durability.

Balance sheet and capital allocation further support patience through the messy margin bridge, but not because buybacks alone can carry the stock. Cash and investments increased by $8 million during Q2 to $209 million, driven primarily by $16 million of cash flow from operations. Total debt was $18 million, so the company has flexibility to fund product programs and restructuring without leaning on external capital. The repurchase authorization has $23 million remaining after approximately 4 million shares were bought for $117 million from inception through Q1 2025, but that is not the central thesis. The balance sheet matters because it gives Cohu time for the utilization and systems guide to translate into revenue without forcing defensive actions at the bottom of the cycle.

The most tangible near-term complication is tax and share count, which can obscure whether operating progress is showing up in reported EPS. Management guided the Q3 tax provision to approximately $15 million, and Q4 effective tax rate to the 30% to 35% range. That means investors should be careful not to overreact if Q3 EPS fails to mirror the revenue step-up, because tax can swamp the operating signal. The operating signal should instead be judged against the approximately $125 million revenue guide, the systems mix of approximately 47%, and the ability to keep Q3 operating expenses near about $50 million. If those numbers hold, a messy EPS bridge would not break the recovery thesis; if revenue slips below the guided range while recurring revenue is flat, the systems-led recovery case weakens immediately.

What to watch next is therefore concrete. For Q3 FY2025, the first confirmation is revenue near approximately $125 million, plus or minus $7 million; a result below that range would reclassify Q2’s miss as demand weakness rather than timing. The second confirmation is whether systems land near approximately 47% of Q3 revenue while recurring revenue stays flat quarter-over-quarter, because that is the mix architecture management gave for the recovery. The third is whether operating expenses stay about $50 million despite around $2 million of variable R&D prototype materials, since leverage cannot emerge if product investment keeps resetting the cost line. For Q4, watch whether the company recognizes the about $12 million it said it would ship and recognize in each Q3 and Q4, and whether the effective tax rate moves into the 30% to 35% range. The thesis breaks if utilization fails to build from 75%, if the approximately $20 million precision analog opportunity does not begin to look like revenue conversion, or if the Q3 guide proves to have been a systems timing pocket rather than the start of a broader test automation recovery.