Cohu, Inc.（COHU） · Electronic Test & Measurement

Cohu (COHU): A Cyclical Semiconductor Test Equipment Stock with Insufficient Margin of Safety

Jun 9, 2026·~22 min read

Quick ReadPlain-language overview · read this first

Cohu (COHU) is a company that supplies test equipment to chip manufacturers. The stance of this report is Watch: the business is understandable and not poor, but the current price does not justify buying. It belongs on a watchlist for now.

What does it mainly do? After chips are made, each one has to be tested to see whether it meets specifications. Cohu sells the machines, related parts, and software that do that work. Beyond selling full systems as one-off purchases, it also earns recurring revenue from consumables, spare parts, and maintenance. That recurring revenue has risen in recent years from a little over four-tenths of the business to six-tenths, making the company somewhat steadier than it used to be.

Are its profits dependable? That is exactly where the report is most concerned. Its industry is highly cyclical. In good markets, it can generate substantial cash; in weak ones, profits collapse. The clearest evidence is consecutive losses in 2024 and 2025. More importantly, the most profitable and highest-barrier segment of test equipment has long been controlled by two much larger competitors, so the richest economics are not in Cohu's hands. It has real core capabilities, but they are not broad.

Is it expensive now? As of 2026-6-9, the share price was about USD 52.51. The report's calculated ideal buying range is only USD 18 to USD 25, so today's price above fifty dollars is clearly higher than its value, and even the most optimistic scenario only barely reaches it. In other words, the current share price has already priced in a future recovery that has not yet happened. If that view proves wrong, there is almost no cushion. Its strengths should also be noted: it has more cash than debt, a solid balance sheet, and is unlikely to run into major trouble in a downturn. Still, the report is restrained. It suggests waiting until either the fundamentals become stronger or the price falls back before considering a purchase.

The above is only a plain-language explanation of this report and is not investment advice. The stock market involves risk; invest with caution.

Lead
Cohu is a global supplier of semiconductor test equipment, spanning handlers, test interface consumables, inspection and metrology equipment, and analytics software, with recurring revenue now at roughly 60%. The core thesis is that this is a cyclical equipment business with some aftermarket stickiness, not a high-certainty compounder with a wide moat, and the 2024 and 2025 operating losses show how hard the cycle still hits. Report rating Watch: the company is worth tracking, but the current price leaves too little margin of safety for a balanced, conservative long-term investor.

Full report

Prices in the article are as of publication; see the valuation band above for the live price.

Conclusion First

My preliminary conclusion is: at the current price, COHU deserves a rating of "Watch." This is not an incomprehensible business. In fact, its commercial logic is quite clear: it provides semiconductor manufacturing with test, interface, automation, inspection, and software analytics tools, then generates a portion of recurring revenue through consumables, spare parts, services, and software subscriptions. The issue is not whether the business can be understood. The issue is that this looks more like a cyclical semiconductor equipment company with some aftermarket stickiness than a high-certainty business with a wide moat that can compound capital easily over the long term.

If we think like owners for the next 10 years or more, rather than stock-price traders, COHU's central tension is clear: the long-term demand direction is sound, and the company is benefiting in the short to medium term from AI/HPC, HBM, SiC/GaN, and related areas; but the competitive landscape is intense, cyclicality is heavy, earnings sensitivity is high, and true cash flow fluctuates sharply across the cycle. This kind of company can offer favorable odds when expectations are low, valuation is depressed, and the cycle is weak. But after the market has already priced in a fairly strong recovery, and after the share price has moved far above the reference price range implied when management issued convertible debt last year, the margin of safety is clearly inadequate.

As of June 9, 2026, COHU traded at about $52.51, with a market capitalization of roughly $2.468 billion. Based on my valuation under conservative, base, and optimistic owner-earnings scenarios, the current price is very likely at a clear premium to the reasonable value range. It is only barely close to, or slightly above, even the optimistic range. For a balanced but conservative investor with a holding period of more than 10 years, this looks more like a company worth tracking than a price worth acting on now.

Item	Conclusion
Investment rating	Watch
Does the current price offer a margin of safety	No
Better suited investors	Investors familiar with semiconductor equipment cycles; cyclical investors
Less suitable investors	Conservative ordinary long-term value investors; not suitable at the current price
Greatest uncertainty	Whether AI/HPC demand can turn into sustainable cash flow, whether the moat is strong enough against Teradyne/Advantest and other formidable competitors, and whether the current valuation has already overdrawn expectations for recovery and growth

Business, Industry, and Competitive Landscape

How This Company Makes Money

In its 2025 10-K, Cohu describes itself as a global supplier of test, interface, automation, inspection and metrology, software analytics, and related services to the semiconductor industry. The company currently has one reportable segment: Semiconductor Test and Inspection. Its product lines include automated test equipment, or ATE, test handlers, interface products such as contactors, probe heads, and probe pins, inspection and metrology equipment, DI-Core/AI analytics software, and spare parts, kits, and services. Revenue essentially comes from two categories: system capital equipment and non-system or recurring revenue, with the latter including interface consumables, spare parts, services, software, and similar items.

Customers are mainly IDMs, OSATs, and fabless companies. Based on the language in the 2025 10-K, Cohu's customers span the global semiconductor ecosystem. The company emphasizes an active installed base of more than 25,000 systems, covering 108 customers, 31 countries, and more than 280 high-volume manufacturing facilities. This shows that Cohu is not simply selling a piece of equipment once and moving on. It is trying to keep selling contactors, kits, spare parts, maintenance, and software analytics services around installed equipment. For long-term investors, this matters because it means COHU is not a pure one-off capital spending business.

From a revenue mix perspective, COHU's "recurring revenue" share has indeed risen materially: 42% in 2022, 49% in 2023, 65% in 2024, and 60% in 2025. In Q1 2026, system revenue was $49.44 million, while non-system revenue was $75.68 million, meaning non-system revenue accounted for about 60.5%. This shows that over the past few years, the company has been moving the business away from a purer equipment-cycle revenue model and toward more stable post-installation revenue. Still, "recurring revenue" does not equal "high-certainty revenue." Interface consumables and services repeat, but the underlying demand is still tied to customer test volumes, yields, utilization, and product transition cycles.

The billing model is also fairly easy to understand. System revenue is typically recognized upon shipment and customer acceptance. Service revenue is recognized over time or as milestones are completed. Spare parts, contactors, and kits are generally recognized upon shipment. The company states clearly in its 10-Q that new products or first-time customer acceptances often lead to deferred revenue, which means quarterly profit can swing with acceptance timing. For long-term investors, that is not necessarily bad, but it does mean quarterly volatility should not be annualized mechanically into a trend.

On the cost side, COHU's costs include materials, assembly, test labor, and manufacturing overhead. The company also states clearly that gross margin is heavily affected by product mix, inventory provisions, capacity utilization, and support costs. Add to that the fact that part of its manufacturing is outsourced to third parties, with Jabil manufacturing most of its semiconductor test systems in Malaysia, and the picture is clear: this business is not mysterious, but it is also not an asset-light software model. It is an industrial technology business with engineering, manufacturing, supply-chain, and after-sales-service complexity.

If the stock market closed for 5 years, my view of this business would be: I could own it at the right price, but I would not want to own it at the current price. The reason is straightforward. The business itself is understandable, but its through-cycle earnings power is unstable, its moat is not wide enough, and the current price gives me no room for error. Overall, I score its "business understandability" at 4/5.

Industry Attractiveness and Competitive Landscape

Long-term demand in semiconductor test and related equipment is not poor. In its autumn 2025 forecast, WSTS/SIA expected global semiconductor sales in 2026 to approach $1 trillion. SEMI expected global semiconductor equipment sales to rise to $126 billion in 2026. This means the long-term demand direction is upward, especially as AI data centers, advanced logic, HBM, and power semiconductors continue to drive capital spending.

But that does not automatically make this a "good industry." Cohu is quite candid in its 10-K: the industry is highly competitive, technologically fast-moving, cyclical in demand, and customer-concentrated. In 2025, the company explicitly named Advantest, Teradyne, Hon Precision, KLA, and other Asian manufacturers as competitors. Teradyne's public filings similarly list Advantest, SPEA, and Cohu as competitors in its semiconductor test business. In other words, COHU is not operating in a mildly competitive oligopoly. It is in a market led by strong players, with demanding engineering and service requirements.

More importantly, the core ATE battlefield, where technical barriers are highest and market share is most concentrated, has long been dominated by Advantest and Teradyne. Advantest has said in its official materials that its semiconductor test market share has risen to around 58%, and it has emphasized maintaining a majority share. Teradyne disclosed in both 2025 and 2026 that AI-related test demand significantly boosted its semiconductor test business, even saying in Q1 2026 that about 70% of company revenue was tied to AI demand. The implication for COHU is that it can certainly benefit from an industry recovery, but the richest part of the industry profit pool is not in its hands.

COHU's own position is more like this: it has a place in handlers, interfaces, thermal control, inspection, multi-product integration, and aftermarket services, and it emphasizes that it is a "global leading supplier" in handlers. But on the main ATE battlefield, the company also acknowledges that it faces two much larger U.S. and Japanese giants. In 2025, it defined its serviceable available market, or SAM, at about $3 billion, which means the company does have room to grow. But from a Buffett-style perspective, this looks more like a middling industry with long-term demand, but not necessarily persistently excellent returns on capital.

Is the industry vulnerable to technological disruption? The answer is that it is easy to reshape, but hard to eliminate. Testing will not disappear. It becomes more important as chips become more complex. However, test solutions, handler architectures, interface technology, thermal control schemes, and software analytics capabilities can all be reshaped by process technology, packaging, HBM, chiplets, power devices, and AI workloads. In other words, demand is stable, but the winners are not stable. That is not the ideal industry profile for COHU. My "industry attractiveness" score is 3/5.

Moat, Management, and Capital Allocation

How Wide Is the Moat

COHU does not have a classic consumer-brand moat, nor does it have network effects. The parts that may form a real moat mainly come from four sources: the installed base, process and application know-how, recurring sales of interface products, consumables, and services, and certain proprietary thermal control, inspection, and software technologies. The company highlights technologies such as active thermal control, visual inspection, machine-learning process control, and precision test instruments. Its test interfaces also need to fit customers' specific chip designs and be replaced frequently, which naturally creates some stickiness.

The second potential moat is switching cost. In semiconductor high-volume manufacturing, test handlers, contactors, thermal control systems, and inspection equipment all require long validation. Once they enter a customer's volume production line, switching suppliers is possible, but it creates risks of revalidation, yield fluctuation, and production-line downtime. Cohu has also mentioned high customer retention in the semiconductor equipment industry in acquisition accounting, which supports the view that switching costs exist. Still, this moat is more of a moderate engineering and validation barrier than an unshakable structural barrier.

The third factor is scale and channel reach. Cohu's global service network, direct sales organization, and 25,000-plus installed base help it follow up after sales, sell spare parts, sell upgrades, and sell software. The company explicitly emphasizes that its global service team and customer portal can provide 24/7 support. This kind of moat is real in industrial equipment.

But I want to be clear: these moats are not enough to form a "wide moat." Cohu does not have ASML-like exclusive equipment status, software-platform-style network effects, or the pricing mindshare of a consumer monopoly brand. It operates in a B2B high-technology industrial market where customers are powerful, competition is ample, and both price and performance matter. Gross margin fell from 47.6% in 2023 to 42.7% in 2025, which looks more like margin driven by product mix and cycle than a rock-solid structural excess return.

Therefore, my judgment on COHU's moat is: it exists, but it is not wide; it has improved somewhat in recent years as recurring revenue has risen, but it is still far from wide and steadily widening. Competitors would not need more than 10 years or unbearable capital to replicate part of its capabilities. What is truly hard to copy is long-term field experience, customer validation relationships, and multi-product coordination, rather than any single patent. As for pricing power in an inflationary environment, I think interface consumables, software, and services are better than whole-system equipment; pricing power on the system side is limited. In a downturn, the company can survive thanks to net cash and recurring revenue, but it may not remain profitable, as the consecutive losses in 2024 and 2025 already show. My "moat strength" score is 2/5.

Can Management Be Trusted, and Is Capital Allocation Rational

On "honesty and governance structure," I do not see an especially obvious red flag in the public filings. The company is audited by Ernst & Young. Recent annual filings did not disclose restatements due to error corrections. Seven of the nine board members are independent directors. The company has a Code of Conduct, an insider trading policy, and disclosed executive share ownership guidelines. In terms of formal governance, this is a well-behaved public company.

But Buffett-style analysis looks beyond form. It asks whether management is truly oriented toward per-share intrinsic value. On this point, my assessment is neutral with reservations. On one hand, most of COHU's acquisitions have been small bolt-on deals, such as MCT and EQT in 2023 and Tignis in 2025. The Tignis transaction had a total price of about $36.6 million, not a large amount, and the 2025 earnout was not achieved, requiring no additional payment. At least on the surface, it does not look like a typical overpriced expansionary acquisition. On the other hand, about $33.69 million of the Tignis deal was recorded as goodwill, which means the company is using cash to buy a growth option. If software synergies fall short, that value may still be written down economically.

The company's cash uses in recent years have mainly been small acquisitions, repurchases, debt repayment, and investments in short-term securities. Repurchases were about $23.64 million, $27.00 million, and $8.586 million in 2023, 2024, and 2025, respectively. As of Q1 2026, $22.80 million of repurchase authorization remained, but the company did not continue repurchasing shares in the first quarter. Judging by outcome, the 2025 average repurchase price was roughly below the current market price, which is not the worst result in price terms. But from an intrinsic-value perspective, continuing to repurchase shares while free cash flow and earnings remain weak does not prove optimal capital allocation.

The incentive structure deserves more attention. In 2025, executive short-term incentives were mainly based on one-year sales growth, four-year rolling sales growth, and non-GAAP pretax profit. They did not put ROIC, free cash flow, or per-share value growth directly at the center. For a highly cyclical equipment company that needs rational capital allocation, I would prefer to see metrics more closely aligned with shareholder returns. The good news is that the company has executive share ownership requirements, and all NEOs met those requirements as of the end of 2025. The bad news is that management and directors together owned about 3.05% of shares, which is not trivial, but also not a deep owner-operator alignment. CEO Luis A. Muller owned about 464,700 shares.

Overall, my judgment on management and capital allocation is: there is no obvious integrity problem, governance is basically proper, and acquisitions have been small and relatively restrained; but capital allocation is not yet excellent, and the incentive design is not especially Buffett-like. Score: 3/5.

Financial Quality and Cash Flow

What Recent Financial Performance Shows

Start with the big numbers. COHU's revenue path is highly cyclical: $636 million in 2020, $887 million in 2021, $813 million in 2022, $636 million in 2023, $402 million in 2024, and $453 million in 2025. This is not the revenue curve of a business that compounds steadily. It looks like the typical cycle of semiconductor equipment. Q1 2026 revenue grew 29.3% year over year to $125.1 million, showing that the business is indeed recovering, but it is still far from the profitability levels of peak-cycle years.

The margin trend tells the same story. Gross margin fell from 47.2% in 2022 and 47.6% in 2023 to 44.9% in 2024 and 42.7% in 2025, then recovered to 46.3% in Q1 2026. Operating margin moved from 6.8% in 2023 to -17.8% in 2024 and -15.4% in 2025, and was still -8.9% in Q1 2026. This shows that the company does have product capability, but it is extremely sensitive to end-market conditions, capacity utilization, and system revenue mix. If earning power depends on everything going right in a strong cycle, this is not the kind of good business Buffett would most prefer.

On cash flow, COHU's profile is: free cash flow is decent in normal or up cycles, but cash generation deteriorates sharply in weak cycles. Operating cash flow in 2020 to 2023 was about $49.88 million, $97.92 million, $112.9 million, and $101.5 million, respectively. It plunged to $2.78 million in 2024 and recovered to $31.69 million in 2025. Rough free cash flow after PP&E purchases was about $31.22 million, $85.92 million, $98.09 million, $85.42 million, -$7.86 million, and $10.73 million from 2020 to 2025. In Q1 2026, operating cash flow was $10.31 million, capital expenditure was $2.00 million, and single-quarter free cash flow was about $8.31 million. This means profit is not always real cash profit, but cash flow is also not showing a long-term fraud-like divergence; it is highly cyclical.

The balance sheet is one of COHU's current strengths. In Q1 2026, the company had about $488.7 million of cash and short-term investments, total debt of about $304.6 million, and net cash of roughly $184.1 million. Shareholders' equity was about $769 million, current assets were $754 million, and current liabilities were $117 million. This means that even with temporarily weak earnings, the company has no obvious survival crisis. That said, the $287.5 million 1.5% convertible notes issued in September 2025 had an initial conversion price of about $27.18. Based on the disclosed conversion ratio, if the excess portion is settled in shares in the future, potential dilution could reach about 10.58 million shares, equivalent to more than 20% of current shares outstanding. This is a capital-structure issue that must be tracked continuously.

On working capital, year-end 2025 accounts receivable were $108.8 million, and inventory was $129.0 million. In Q1 2026, receivables declined to $101.5 million, while inventory rose to $130.8 million. First-quarter operating cash flow turned positive partly because of receivable collection and higher customer prepayments, but inventory still rose. For an equipment company, this is normal. For a value investor, it means one should not become overly optimistic about a single quarter's cash flow improvement.

Key Financial Table

The table below shows COHU's through-cycle financial profile more directly.

Metric	2020	2021	2022	2023	2024	2025	Q1 2026
Revenue	636.0	887.2	812.8	636.3	401.8	453.0	125.1
Gross margin	42.7%	43.6%	47.2%	47.6%	44.9%	42.7%	46.3%
Operating margin	0.5%	22.7%*	15.4%	6.8%	-17.8%	-15.4%	-8.9%
Net income	-13.8	167.3	96.8	28.2	-69.8	-74.3	Not separately cited
Operating cash flow	49.9	97.9	112.9	101.5	2.8	31.7	10.3
Capital expenditure	18.7	12.0	14.8	16.1	10.6	21.0	2.0
Rough free cash flow	31.2	85.9	98.1	85.4	-0.8	10.7	8.3
Ending total debt	Not separately cited	119.3	79.7	8.8	8.8	305.1	304.6
Ending cash + short-term investments	Not separately cited	376.5	376.1	300.7	262.1	484.0	488.7

2021 operating margin included the gain on sale of the PCB Test business and should not be treated as fully sustainable operating profit. Note: 2020-2022 data mainly comes from the 2022 10-K, 2023-2025 data comes from the 2025 10-K, and Q1 2026 data comes from the 2026 first-quarter report. Free cash flow is my rough calculation of operating cash flow minus PP&E purchases.

Based on these numbers, my judgment is: COHU does not look like a company with a strong smell of financial fraud, but it is also far from the typical high-quality compounder that earns more cash as it grows. It is more like an industrial technology company that can produce a lot of cash in an up cycle, sees profit collapse quickly in a down cycle, but is unlikely to run into trouble because the balance sheet is still acceptable. For conservative investors, this profile is "acceptable," not "excellent."

Owner Earnings and Intrinsic Value Estimate

How to Estimate Owner Earnings

Buffett-style owner earnings focus on the cash a business can truly distribute to owners, rather than accounting profit itself. For COHU, I would not mechanically use the 2025 net loss of -$74.27 million to conclude that "the company does not make money," because 2025 also included substantial depreciation and amortization, restructuring, and drag from low utilization during a weak cycle. Operating cash flow had already recovered to $31.69 million in 2025, showing that the company was not completely bleeding cash in a weak cycle.

But it is also wrong to simply add back all depreciation and amortization and call it "free cash." There are two reasons. First, COHU is an equipment and interface company, and maintaining its technology and manufacturing capabilities is not costless. Second, its working capital is heavily affected by the cycle, and a single year's release or absorption can distort "true earnings." Therefore, I use a more conservative method: anchor on operating cash flow first, then deduct a reasonable maintenance capital expenditure assumption. Total capital expenditure in 2025 was about $20.96 million. Considering that some of it supported expansion and product development, I conservatively estimate maintenance capital expenditure at $12 million to $15 million. This gives 2025 conservative owner earnings of roughly $17 million to $20 million.

Using the 2021-2023 three-year average free cash flow as a reference, COHU's rough FCF in a mid-to-strong cycle could reach $85 million to $98 million. This shows the company has materially higher cash-generation capacity in a normal cycle. The problem is that this is not a "steady state." It is the result of "good years." So in valuation, the more reasonable method is to use three through-cycle owner-earnings scenarios: conservative, base, and optimistic, instead of treating the best year as normal.

Combining the actual 2025 results, the Q1 2026 recovery, and cash-flow performance in prior strong-cycle years, my owner-earnings starting assumptions are: $35 million conservative, $60 million base, and $90 million optimistic. These three figures correspond to low recovery, normal recovery, and strong recovery. Note that none of them is the actual 2025 value. They are my judgment of future through-cycle "sustainable cash capacity." The current market capitalization of about $2.468 billion means the market is effectively valuing COHU at very high owner-earnings multiples: about 70x / 41x / 27x market-cap multiples under the conservative/base/optimistic starting points. If measured by my stricter 2025 conservative owner earnings of $17 million to $20 million, the current price equals 123x to 145x owner earnings. For a cyclical company with a weak moat, that is clearly not cheap.

Value Ranges Under Three Valuation Methods

Owner-Earnings DCF

I use a 10-year DCF, a 10% discount rate, a 3% terminal growth rate, and the latest net cash of about $184 million as an additional safety cushion. Based on the owner-earnings starting points above, the resulting ranges are:

Scenario	Starting owner earnings	10-year growth	Discount rate	Terminal growth	Estimated equity value
Conservative	35 million	4%	10%	3%	About $15-20/share
Base	60 million	6%	10%	3%	About $24-32/share
Optimistic	90 million	8%	10%	3%	About $38-48/share

These figures are not a "precise answer." They are a valuation framework based on public data and conservative assumptions. The key message is simple: the current price of $52.51 is generally above my base-case value judgment, above most conservative value estimates, and not cheap even under the optimistic scenario.

Relative Valuation

Based on the latest market capitalization and share price, COHU's headline valuation metrics do not look attractive: negative PE, P/B of about 3.2x, P/S of about 5.4x, and P/FCF above 200x based on 2025 FCF. On an enterprise-value basis, COHU's EV is about $2.284 billion, implying 2025 EV/Sales of about 5.0x. The issue is not just that the stock is "expensive." It is expensive while current fundamentals have not yet delivered.

Compared with peers, Teradyne and FormFactor also do not look cheap on headline metrics: Teradyne's latest PE is about 69.5x, while FormFactor's is about 142.8x. But at least they are currently profitable. FormFactor had 2025 revenue of $785 million, operating cash flow of $115.4 million, and no material leverage pressure. Teradyne is benefiting more strongly from AI test demand, publicly disclosing that about 70% of its Q1 2026 revenue was AI-related. Therefore, "peers are also expensive" does not automatically mean COHU is cheap. The more accurate statement is: the entire semiconductor test chain is being priced at high valuations by the market, and COHU is one name whose profit recovery has not fully materialized while its share price has already rerated substantially.

Asset or Liquidation Value

COHU is not a typical liquidation-value investment, so liquidation value is not the core investment logic. The main asset floor it can provide comes from net cash and investments. In Q1 2026, cash and short-term investments were $488.7 million, total debt was $304.6 million, and net cash was about $184 million. This means the company is not a risk asset crushed by debt. On the other hand, Q1 2026 shareholders' equity was about $769 million, while current market capitalization was $2.468 billion, giving the market a valuation of about 3.2x PB. For a company that has posted consecutive losses for the past two years and has unstable ROIC, this PB level does not provide a margin of safety.

Combining the three methods, my price framework is:

Range	Price judgment
Conservative intrinsic value range	$15-20/share
Reasonable intrinsic value range	$24-32/share
Optimistic intrinsic value range	$38-48/share
Ideal buy price range	$18-25/share
Acceptable holding price range	$25-35/share
Clearly overvalued price range	Above $40 looks more overvalued; above $50 requires very optimistic assumptions to make sense

Margin of Safety, Risks, and the Opposing Case

Is There Enough Margin of Safety Now

My answer is clear: no.

The most fragile assumption embedded in the current valuation is not "will COHU fail?" It is "how much is the market willing to pay in advance for the possible profit recovery over the next few years?" If the company merely returns to a base-cycle level instead of entering a multi-year, high-profit AI supercycle, today's price is unattractive. In other words, COHU now looks like a cyclical recovery stock that has already priced in the recovery, rather than a value stock trading clearly below intrinsic value.

If future growth falls short of expectations, or gross margin recovery disappoints, does the investment still work? My judgment is: it is hard to make it work. You are not buying a weak-moat company at 15x or 20x owner earnings. You are buying a company that may be above 100x owner earnings under a conservative lens and still lacks stable GAAP profitability today. If valuation multiples compress, investors may suffer poor long-term returns even if the "business has not broken."

This is the classic "good company, bad price" problem. More precisely, COHU may not even be a "great company among good companies." It looks more like "a decent cyclical company in a market-favored phase." In this situation, waiting for a better price is entirely rational. For long-term capital, not buying is not necessarily missing an opportunity. Often, it is simply preserving capital for better odds.

Main Risks and the Strongest Opposing View

I would rank the most important risks by the logic of "permanent capital loss."

First is competitive risk. The key profit pools in COHU's market are controlled by larger players such as Advantest and Teradyne. Cohu itself clearly acknowledges that the ATE market is dominated by two major suppliers, while the interface market is highly fragmented. If customers continue to concentrate higher-value test steps with the giants, COHU's profit recovery may fall short of market expectations.

Second is technology substitution and share-pressure risk. Testing will not disappear, but test solutions, temperature control, inspection, and analytics software will continue to evolve. If COHU fails to win enough share in HBM, advanced packaging, AI processor thermal control, or software analytics, its "growth story" could be disproved quickly. Especially when Teradyne has already clearly benefited from the AI test wave, if COHU still cannot convert AI opportunities into higher margins and cash flow over the next few years, the market rerating could reverse.

Third is operating leverage and cycle risk. The consecutive losses in 2024 and 2025 show that the company struggles to remain profitable when demand weakens. Even with around 60% recurring revenue, that recurring revenue is not as stable as software maintenance fees. It still moves with customer utilization and test volumes. For long-term holders, this means you are not owning a business that can "remain profitable even in a recession."

Fourth is capital structure and dilution risk. The $287.5 million convertible note issued in 2025 carries a low cost, which is not necessarily bad by itself. But with the current share price already far above the $27.18 initial conversion price, the instrument may affect per-share value in the future through share dilution or cash consumption for settlement.

Fifth is supply-chain and geopolitical risk. The company relies on Jabil in Malaysia to manufacture most of its test systems, while its revenue has significant exposure to Asian markets. In Q1 2026, China accounted for about $13.51 million of revenue. In the semiconductor equipment industry, export controls, tariffs, supply-chain interruptions, and regional political conflict can quickly change delivery and profit timing.

The strongest opposing view is simple: you may be mistaking a "cyclical equipment stock in a recovery phase" for a "long-term stable compounder value stock." Bulls see installed base, a rising recurring revenue share, AI/HPC/HBM/SiC opportunities, net cash, and small bolt-on acquisitions. Bears see a weak moat, pressure from industry giants, excessive valuation, convertible dilution, and poor earnings quality over the past two years. At the current price, I find the latter view more persuasive.

The following facts would overturn my cautious judgment, meaning these are the triggers that would make me admit I was wrong:

The company can stabilize gross margin at 45%+ over the next two to three years, while restoring operating margin to 10%-12% or higher, without relying on a one-off product-mix benefit.
Recurring revenue remains above 60%, and truly shows cash-flow resilience in a weak cycle instead of being nominally recurring but still volume-sensitive in substance.
AI/HPC/HBM, software analytics, and inspection and metrology businesses can continuously bring higher ROIC, rather than just a revenue story.
The company handles the convertible notes smoothly without materially diluting shareholders and continues to maintain net cash or low leverage.

Comparison, Checklist, and Final Investment Conclusion

Does It Deserve Capital Compared with Other Opportunities

If COHU is compared with its strongest competitors, I would rank them this way: Teradyne has a stronger industry position and more direct AI exposure; FormFactor's profit recovery and balance-sheet quality in its own niches are also clearer; COHU's advantages are smaller size, greater sensitivity, and improving post-installation revenue, but its disadvantages are a narrow moat, unstable earnings, and a current valuation that is also unfriendly. Teradyne is not cheap either, but at least its high valuation is backed by stronger market position and more visible AI monetization. COHU looks more like a case where "imagination has moved ahead of fundamentals."

Compared with a broad market index, I do not think COHU at the current price is clearly superior to buying the index. The reason is not that the index must be cheap. It is that COHU requires investors to judge the industry cycle, competitive landscape, technological evolution, capital allocation, convertible-note settlement method, and valuation-compression risk all at once. In my base scenario, its expected annualized return over the next 10 years is only in the low single digits to mid-single digits, which clearly does not offer enough risk compensation. If I had only 5 core portfolio positions, I would not put it in.

For a precise comparison with the risk-free rate or high-grade bond yields, I did not supplement this report with same-day official rate data, so I will not hard-code a specific figure. But the principle is simple: when a weak-moat cyclical stock has no obvious margin of safety, it should offer potential returns meaningfully above sound fixed income before it is worth taking single-stock risk; current COHU does not offer that compensation.

Investment Checklist

Checklist	Conclusion	My judgment
Can I understand this business	Pass	The commercial logic is clear, and the products and billing model are transparent.
Does it have long-term stable demand	Pass	Semiconductor testing will exist for the long term, but the cycle is highly volatile.
Does it have a durable moat	Fail	It has some stickiness and an installed base, but the moat is not wide enough.
Does it have pricing power	Fail	Clearly limited on systems; somewhat better in consumables/services.
Can it generate stable free cash flow	Fail	It can earn money through the cycle, but annual volatility is high.
Are returns on capital excellent	Fail	Good at cyclical peaks, unstable over the long term, and weak in the past two years.
Can management be trusted	Pass	Governance is proper, with no obvious integrity red flags.
Is capital allocation rational	Uncertain	There is restraint, but it is not outstanding.
Is the balance sheet solid	Pass	Net cash and strong liquidity provide enough survivability.
Is valuation below intrinsic value	Fail	The current price is above my reasonable range.
Is the margin of safety sufficient	Fail	Almost none.
Would I feel comfortable holding it long term	Fail	The price is not comfortable, and neither is the industry.
What facts would make me sell	Already listed	Share loss, failed margin recovery, inefficient use of cash, and convertible/acquisition damage to per-share value.
Am I tempted only because the share price has risen or sentiment is strong	Watch carefully	The current valuation very likely embeds market optimism.

Final Investment Conclusion

【Final Rating】 Watch

【One-Sentence Investment Thesis】 COHU is a semiconductor test equipment company that is understandable and may make money in a cyclical recovery, but it looks more like a weak-moat cyclical business, and the current price leaves almost no margin of safety for a conservative long-term investor.

【Core Bull Case】

Recurring revenue has risen from 42% in 2022 to 60% in 2025, strengthening the installed base and aftermarket revenue.
Q1 2026 revenue grew 29.3% year over year, and gross margin recovered to 46.3%, showing that the cycle and product mix are improving.
The latest balance sheet has net cash, very strong liquidity, and decent survivability in an extreme downside scenario.
The company does have exposure to AI/HPC, HBM, SiC/GaN, thermal control, inspection, and software analytics. It is not purely an old-product installed-base company.

【Core Bear Case】

The industry's core profit pool is controlled by stronger players such as Advantest and Teradyne, and COHU's moat is not wide.
The consecutive losses in 2024 and 2025 show that the company cannot maintain stable profitability in a down cycle.
At the current price, P/B, P/S, P/FCF, and owner-earnings metrics are all clearly not cheap.
The convertible debt issued in 2025 creates meaningful dilution risk if the excess portion is settled in shares in the future.

【Key Assumptions】

Test demand growth from AI/HPC/HBM/power semiconductors can continue, rather than being only a short-cycle pulse.
The roughly 60% recurring revenue share can translate into more stable margins, rather than merely looking better as a revenue category.
Management will not destroy per-share value through acquisitions, dilution, or inefficient repurchases.

【Ideal/Fair Buy Price】 $18-25/share. Rationale: this range is determined jointly by conservative-to-base owner-earnings DCF valuation, the current weak-moat industry profile, a cyclical discount, and the need for at least a 30%-40% margin of safety.

【Target Holding Period】 If bought at a lower price, at least 5-10 years; but the premise is continuous tracking of the cycle position and share changes, rather than buying and then "blindly holding forever."

【Expected Annualized Return】 Buying at the current $52.51, roughly estimated over a 10-year horizon:

Conservative scenario: -10% to -5%
Base scenario: 0% to 4%
Optimistic scenario: 6% to 10% This is a rough projection based on the owner-earnings path and terminal multiple discussed above. It is not a price forecast.

【Maximum Downside Risk】 If AI-driven demand falls short, the industry enters another down cycle, margin recovery fails, and the market valuation reverts from "optimistic recovery stock" to "ordinary cyclical equipment stock," it would not be surprising for COHU's price to move back toward my base or conservative value ranges. From the current price, long-term capital loss risk is not small, and a 40%-60% drawdown is not unimaginable. This risk is not bankruptcy. It is paying too high a price for a cyclical company that should not have been held at a high valuation in the first place.

【Tracking Metrics】 I will continue to watch the following indicators:

Whether recurring revenue stays at >=60%.
Whether gross margin can stabilize at 45%+.
Whether operating margin returns to double digits.
Whether operating cash flow and free cash flow improve consecutively.
Whether backlog continues to expand and converts into revenue.
Whether major-customer concentration rises again.
Potential dilution from the convertible notes and the cash-settlement strategy.
Whether acquisitions keep adding goodwill rather than adding per-share free cash flow.
Whether COHU narrows the gap with Teradyne/Advantest in share and technology position.

【Signals That Would Trigger Reassessment】

The company remains unable to restore positive and sustainable free cash flow for more than two consecutive years.
Gross margin fails to return to around 45%, suggesting that the claimed "higher-value product mix improvement" has not materialized.
Recurring revenue falls to around 50% or below.
Large acquisitions or aggressive repurchases materially weaken the net cash buffer.
Convertible-note handling clearly damages per-share value.
Major customers or major product lines show signs of share loss.

【Final Recommendation】 For a more-than-10-year, balanced but conservative investor, my recommendation is: put COHU on the watchlist rather than buy it into the portfolio now. This is not a bad company, and it is not a company that cannot make money. But under a Buffett-style framework, you either buy a "high-quality business" or a "very cheap price." COHU currently does not stand out enough on either dimension. The most restrained and rational approach is to keep tracking its margin recovery, cash-flow quality, and valuation pullback window, and wait for at least one of two things to happen first: "the company becomes stronger" or "the price becomes lower."

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.

Mentioned Tickers

TER 6857 FORM KLAC JBL ASML

Baillie Framework · Ten Questions for Growth Investing

Hunting ten-year five-baggers among great growth stocks — pressing the upside question: "Can it get much bigger?"

How large is its market ceiling? Is it expanding an existing pie, or creating an entirely new market?4/10

Conclusion: COHU's market ceiling is not so small that there is no room to grow, but it is mainly expanding share and increasing test content per chip within the existing pie of semiconductor test, automation, interface consumables, inspection and metrology, and software analytics. It is not creating an entirely new market. The macro semiconductor industry is large. SIA/WSTS even expects global semiconductor sales to exceed $1.5 trillion in 2026; but for Cohu itself, the addressable market is much narrower. The company put its SAM at about $3 billion in its 2025 10-K.

The positive side of this ceiling is that testing will not disappear, and the more complex chips become, the more important test, thermal control, interface, inspection, and yield analytics become. Cohu already has an installed base of more than 25,000 systems, and recurring revenue accounted for 60% in 2025. This shows it is not only selling one-time equipment, but also selling interface consumables, spare parts, services, upgrades, and software around existing customers. In 2026Q1, the company also reported net sales of $125.1 million, about 60% recurring, and raised its AI-driven compute addressable market to about $750 million and its FY26 HPC revenue outlook to about $80 million-$100 million. This indicates that AI/HPC/HBM is indeed raising its local ceiling.

But the Baillie framework asks about "fivefold upside in ten years," not simply whether there is a cyclical recovery. On that point, COHU's ceiling is moderate. With 2025 revenue of $453 million, if it can capture more of the $3 billion SAM and make AI/HPC and recurring revenue real, revenue rising to $800 million-$1 billion is not completely unimaginable. But to support a fivefold increase in today's market cap of about $2.468 billion to about $12.3 billion, "the industry getting bigger" is not enough. It also needs to lift market share, margins, and cash-flow quality materially.

So my judgment is: COHU is increasing content within an old market that will continue to grow, not opening up a new market with no competitors. Its ceiling is enough to support a decent cyclical repair, and may also support moderate growth; but it does not yet show the very large upper bound that NVIDIA or ASML enjoy from a new demand paradigm or a near-monopoly link in the chain. For Baillie-style growth-stock screening, the answer to Q1 is: there is room to grow, but it is not a rare creator of a large new market.

Jun 9, 2026
Can its revenue at least double over the next five years? Will growth mainly be driven by volume, price, or new businesses?4/10

Conclusion: doubling is possible, but it should not be treated as a high-confidence base case. Starting from 2025 revenue of about $453 million, a five-year doubling would mean about $910 million, essentially just returning to the 2021 peak of $887 million shown in the report. That means "doubling" is not mathematically absurd, partly because 2025 was a trough. But from 2025 to 2030 it would still require roughly 15% compound annual growth, which is an optimistic scenario for a highly cyclical test equipment company, not steady-state compounding.

Cohu does have evidence of near-term repair. The company officially reported 2026Q1 net sales of $125.1 million, about 60% recurring, and GAAP gross margin of 46.3%, and guided to Q2 sales of $144 million plus or minus $7 million. It also raised its AI-driven compute addressable market to about $750 million and its FY26 HPC revenue outlook to $80 million-$100 million. If AI/HPC, HBM, and SiC/GaN-related orders continue to materialize, and test capacity utilization recovers as well, revenue returning to the $700 million-$900 million range is conceivable.

Breaking down the drivers, the first driver is volume, not price. System equipment, handlers, interface consumables, services, and spare parts ultimately relate to customers' test volume, installed base, capacity utilization, and product refresh frequency. About 60% recurring is not a rigid software SaaS subscription; it is repeat activity from consumables, kits, services, and software after installation. It improves resilience, but the underlying business is still affected by cycles and customer utilization.

The second driver is new business/content, especially AI/HPC test, thermal control, inspection and metrology, and software analytics; but it has not yet proven that it can carry a five-year doubling by itself. Software assets such as Tignis/DI-Core, AI process control, and aftermarket recurring revenue can improve gross margin and customer stickiness, but the Tignis transaction was only tens of millions of dollars, and even if the FY26 HPC outlook reaches $100 million, it is still only part of company revenue, not a fully formed second curve. As for price, on the system side Cohu faces Advantest/Teradyne and powerful customers, so pricing power is limited. Interface consumables, services, and software have somewhat better pricing power, but it is more likely to show up as mix and gross-margin improvement than as revenue doubling from price increases alone.

So my answer is: COHU has a path to doubling revenue in five years, but not high certainty. The most realistic route is "an upcycle in semiconductor test + higher AI/HPC/HBM content + amplified consumables/services from the installed base," not a pure new-market breakout curve. If revenue cannot sustainably move above $600 million-$700 million over the next two or three years, gross margin cannot stay above 45%, and recurring revenue cannot truly translate into cash-flow resilience, then a five-year doubling is more likely a cyclical-stock rebound narrative than the ideal long-term growth compounding in the Baillie framework.

Jun 9, 2026
Five years from now, what will take over as the next growth engine? Does this "second curve" exist today?4/10

Conclusion: if COHU has a "second curve" five years from now, it should not be a simple equipment-cycle recovery, but a handoff led jointly by high-complexity chip test demand from AI/HPC/HBM and software analytics, interface consumables, and service revenue on top of the installed base. It already has an outline today, but it cannot yet be called a sufficiently large and independent growth curve.

There is evidence. In 2026Q1, the company disclosed quarterly revenue of $125.1 million, about 60% recurring, raised its AI-driven compute addressable market to about $750 million, and lifted its FY26 HPC revenue outlook to $80 million-$100 million (see Cohu's 2026Q1 earnings release). Relative to the $453 million revenue base in 2025, this is no longer a completely negligible business. Add the 2025 10-K disclosure that recurring revenue already accounted for 60% and includes interface products, spares, software, and services (see Cohu 2025 Form 10-K), and it is clear the company is indeed moving from "selling equipment" toward "continuously monetizing around the installed base."

But I would separate this second curve clearly: AI/HPC/HBM/SiC/GaN test looks more like the high-end upgrade of the first curve, while software analytics and aftermarket recurring revenue look more like the true second curve. The former can drive demand for systems and test cells, but it is still affected by semiconductor capex cycles, customer order timing, and strong rivals such as Teradyne/Advantest. If the latter succeeds, it is what can improve revenue stability, gross-margin structure, and customer stickiness. Cohu's acquisition of Tignis also follows this direction. The 10-K describes Tignis as a provider of AI process control and analytics-based monitoring software, intended to strengthen DI-Core software. But the same 10-K also disclosed that Tignis's 2025 revenue was not material to Cohu and that the earnout target was not met. So for now this is still "buying a growth option," not a second engine that has already broken out.

Therefore, from the Baillie perspective, my judgment is: the second curve exists today, but it is still an observable seedling, not a trunk strong enough to support a fivefold gain in ten years. To turn this line from narrative into fact, at least three things need to happen consistently: AI/HPC revenue must not be a one-off cyclical pulse; recurring revenue must stay at 60%+ and produce more stable free cash flow; and software plus inspection and metrology must contribute visible incremental profit, not only add R&D, goodwill, and adjusted-profit definitions. Otherwise, COHU's "growth engine" five years from now is more likely to be the next semiconductor equipment cycle than a true second curve.

Jun 9, 2026
What is its core competitive advantage? Will this moat widen or narrow over the next three to five years?5/10

Conclusion: COHU's core competitive advantage is not a monopoly technology that cannot be replicated, but moderate engineering-validation barriers, post-installation recurring revenue, interface consumables, and a global service network. Once semiconductor production test equipment enters a customer's production line, switching suppliers requires requalification and creates yield-volatility and downtime risks. At the same time, interface products, spare parts, services, upgrades, and software continue to generate revenue alongside the installed base. Cohu's 2025 10-K disclosed an active installed base of more than 25,000 systems, covering 108 customers, 31 countries, and more than 280 high-volume manufacturing facilities. That is a real barrier.

But this moat is not wide. COHU does not have ASML-like exclusivity, nor software-platform network effects. The markets it serves, including semiconductor test, automation, interface, inspection and metrology, and software analytics, are highly competitive. The company's own 10-K lists Advantest, Teradyne, Hon Precision, KLA, and other Asian manufacturers as major competitors, while Advantest's official annual report shows its CY2024 semiconductor tester / ATE market share at about 58%. This indicates that the industry's core profit pool leans toward the large players, while COHU is more often fighting for position in handlers, interfaces, thermal control, aftermarket services, and niche applications.

Over the next three to five years, I lean toward this judgment: the operating moat may widen slightly, but the investment-relevant moat remains narrow. The reasons for possible widening are a higher recurring-revenue mix, higher testing difficulty from complex chips such as AI/HPC/HBM/SiC, and the potential for software analytics capabilities such as DI-Core/Tignis to increase customer stickiness. In its official 2026Q1 release, Cohu also disclosed net sales of $125.1 million, about 60% recurring, and raised AI-driven compute TAM to about $750 million and FY26 HPC revenue outlook to $80 million-$100 million.

The real risk is that these growth points may not translate into pricing power and high ROIC. If the high-value parts of AI/HPC test continue to be captured by Teradyne and Advantest, and COHU's "recurring revenue" is mainly consumables and services that fluctuate with customer utilization, the moat will not widen materially and may even narrow because of price pressure and technology iteration. The report's gross margin decline from 47.6% in 2023 to 42.7% in 2025, plus operating losses in both 2024 and 2025, also remind us that this company has stickiness, but has not proven it owns a wide moat. My judgment is: a slight widening is possible, but the evidence is not enough to show a transformation into a wide moat.

Jun 9, 2026
If its core business is disrupted, does it have the genes for self-reinvention? How does it handle mistakes and bad news?4/10

Conclusion: COHU has the genes for incremental self-repair, but I do not yet see strong genes for actively disrupting itself and rebuilding the profit pool. It has extended from traditional test/handler equipment into interface consumables, inspection and metrology, thermal control, software analytics, and services. That is indeed turning one-time system sales into steadier post-installation revenue. The report shows recurring revenue rising from 42% in 2022 to about 60% in 2025, and 2026Q1 still had about 60% recurring. The issue is that the company still has only one reporting segment, Semiconductor Test and Inspection, and in 2025 it had revenue of $453 million and a net loss of $74.27 million. If the core test, handler, or interface business is squeezed by giants and Asian manufacturers, COHU can adjust its mix, but it has not proven it can use an entirely new curve to replace the whole core business.

Its most likely reinvention path is "small M&A + product-line pruning + aftermarket/software-ization," not a sweeping strategic rebirth. Tignis is a typical example: in 2025, Cohu used cash to acquire this AI process control and analytics-based monitoring software provider, a direction that helps DI-Core, software analytics, and yield optimization. But the 2025 10-K also disclosed that Tignis's 2025 revenue was not material to Cohu. This means the second curve exists, but it is still a reinforcement, not the main engine that can support a fivefold-in-ten-years narrative.

On handling mistakes and bad news, I would score it neutral to positive: at least it is willing to put bad news into the financial statements instead of only telling an AI/HPC story. In 2025, the company acknowledged that gross margin fell from 44.9% to 42.7%, with reasons including system product mix and the one-time impact of stopping production of certain products. The same 10-K also disclosed about $10.1 million of restructuring charges in 2025, and defined the actions as cost reductions during a semiconductor test and inspection equipment downturn. These actions show management can acknowledge cycle, inventory, and product-line problems, and is willing to cut, integrate, and move manufacturing to lower-cost regions.

But the boundary is also clear: COHU's communication of bad news is still the typical "compliant transparency" of a listed industrial company, not aggressive owner-operator self-criticism. The Q1 2026 release disclosed a GAAP net loss of $12.1 million, period-end cash and investments of $488.7 million, and Q2 sales guidance of $144 million plus or minus $7 million, while also emphasizing that AI-driven compute TAM was raised to about $750 million and FY26 HPC revenue outlook was $80 million-$100 million. This is reasonable disclosure, but it can also lead the market to weigh the "repair story" more heavily than the fact that profitability is still not stable.

So the Baillie judgment for question 5 is: COHU can adapt to the environment, cut weak products, and use small acquisitions to strengthen its technology stack. It does not appear to hide bad news, and its financial-statement and restructuring disclosures are relatively candid. But its self-reinvention is currently limited to "surviving the cycle better" and "making the equipment business more recurring." It has not proven that it can create a sufficiently large new profit pool if the core business is disrupted. For a fivefold-in-ten-years framework, this item is only moderate and does not add much.

Jun 9, 2026
Does management, especially the founder, have a long-term view and interests deeply tied to the company? Is it willing to sacrifice current profits for five to ten years from now?4/10

Conclusion: neutral with reservations; this is not the founder owner-operator profile that Baillie likes most. COHU is a long-established industrial technology company founded in 1947. Current CEO Luis Müller is a professional manager who joined Cohu in 2005 and has served as CEO since 2014, not a founder or family controller. The governance structure is standard, but that is not the same as "a founder treating the company as a life's work and willing to sacrifice short-term profits countercyclically."

There is interest alignment, but it is not deep. The company's proxy shows that directors and executives together own about 3.05%, and CEO Müller holds about 464,700 shares; at the same time, 7 of the 9 board members are independent. This indicates transparent governance and a complete oversight mechanism, but the equity tie looks more like standard public-company incentives than deep owner mentality. In the Baillie framework, this item is "no red flag," not a strong positive.

The main evidence of a long-term view lies in capital allocation: in 2025, the company used cash to acquire Tignis, strengthening AI process control and analytics-based monitoring software, which matches the report's themes of software-ization, recurring revenue, and the AI/HPC second curve. In 2026Q1, the company again emphasized its software strategy and, while cash and investments reached $488.7 million, did not repurchase shares during the quarter. At minimum, this suggests management did not mechanically buy back stock when the share price was high.

But the other side is also clear: 2025 short-term incentives mainly looked at one-year sales growth, four-year rolling sales growth, and non-GAAP pretax income, and 2026 short-term incentives are still composed of sales growth and non-GAAP pre-tax income, with no direct link to ROIC, free cash flow, or intrinsic value per share. Add the consecutive losses in 2024 and 2025 and still weak free cash flow in 2025, and the evidence is not strong enough to show whether management can keep bearing current profit pressure for software and high-end test opportunities five to ten years out.

So my judgment for question 6 is: management is credible, governance is sound, and capital allocation has some long-term orientation, but it lacks founder-like deep alignment and a clear per-share value orientation. This is not a disqualifying issue, but it is also not a core reason to support a "great growth stock that can rise fivefold in ten years" thesis.

Jun 9, 2026
If it disappeared tomorrow, how much would customers miss it? Is its growth model sustainable and not dependent on harming society or regulatory arbitrage?5/10

Conclusion: customers would miss COHU, but most likely in the sense that "production-line engineering teams would have a major headache," not that "the industry could not function." Once its handlers, test interface, thermal control, inspection, and services enter mass-production lines, replacement requires requalification and may affect yield, throughput, and downtime risk. The report's 25,000+ active installed base, 108 customers, 31 countries, 280+ high-volume manufacturing facilities, and about 60% recurring revenue in 2025 show that this stickiness is real. The official 10-K also disclosed that recurring revenues accounted for 60% in 2025, and included interface products, spares, software, and services.

But "miss it" is not the same as "cannot live without it." Cohu itself acknowledges in its 10-K that it operates in markets with intense competition, rapid technological change, customer concentration, and strong competitors such as Advantest, Teradyne, and KLA. So if COHU disappeared tomorrow, customers would be disrupted in the short term because of existing platforms, interface consumables, field services, and process know-how. Over the medium and long term, they would shift new projects to Teradyne, Advantest, KLA, or Asian suppliers. Under the Baillie framework, this looks more like "an industrial supplier with switching costs" than "irreplaceable infrastructure that customers would deeply miss."

On sustainability, COHU's growth model generally does not depend on obvious harm to society or regulatory arbitrage. It sells tools that improve chip testing, yield, reliability, and production-line efficiency. In its official Q1 2026 release, the company also positioned itself as an equipment and services provider that optimizes semiconductor manufacturing yield and productivity. Rising test complexity from AI/HPC, automotive, industrial, and power semiconductors is a relatively legitimate source of demand. The issue is that regulatory risk is not zero: Cohu's 10-K explicitly notes that its products and services must comply with U.S. and overseas export controls and trade requirements, and specifically points out that export restrictions related to China's advanced computing/semiconductor manufacturing may affect sales. The company is also subject to environmental, hazardous-substance, conflict-minerals, and climate-related compliance requirements.

So my judgment is neutral: customer stickiness and social license both pass, but they do not reach the strength of a great growth stock whose customers cannot replace it and whose growth is almost immune to social and regulatory backlash. Its growth becomes more sustainable only if it turns recurring/software/HPC test opportunities into long-term high-margin services, rather than relying on one-time system orders, sales to geopolitically sensitive regions, or regulatory gray zones.

Jun 9, 2026
What are the unit economics of this business, including gross margin and incremental returns? Do they improve or deteriorate as scale grows? Where does the money it earns go?4/10

Conclusion: COHU's unit economics "look decent, but do not yet have the certainty of a great growth stock that becomes lighter and more profitable as it gets larger." Gross margin can reach the 42%-47% range, which shows that test, interface, services, and software do have engineering value. But operating losses in both 2024 and 2025, and rough 2025 free cash flow of only about $10.7 million, show that incremental returns are highly dependent on the cycle, product mix, and capacity utilization. This is not a stable high-ROIC machine.

The official numbers support this judgment: Q1 2026 net sales were $125.1 million, about 60% recurring, GAAP gross margin was 46.3%, and GAAP net loss was $12.1 million. The strength in this set of figures is the high recurring-revenue mix and recovered gross margin; the weakness is that decent gross margin has not yet flowed through to GAAP profitability. The 2025 10-K also clearly states that gross margin is affected by product mix, support costs, inventory provisions, and capacity utilization, and that 2025 gross margin fell from 44.9% in 2024 to 42.7%. This does not look like a business with very strong pricing power whose margins naturally expand once scale increases.

Will the business improve as scale grows? Conditionally, yes. If AI/HPC, HBM, and SiC/GaN test demand drives system shipments, while the 25,000+ installed base brings more interface consumables, spare parts, services, and software revenue, fixed R&D, sales and service networks, and manufacturing support costs can be spread over a larger base, creating upside for gross margin and operating margin. Conversely, if growth mainly comes from low-margin systems, customer bargaining, inventory provisions, or competitive order-taking, larger scale will not necessarily bring better returns and may even consume more working capital. The report's free cash flow of $85 million-$98 million in 2021-2023, followed by near-negative free cash flow in 2024, shows exactly that this is cyclical operating leverage, not stable compounding leverage.

The money it earns mainly does not turn into a stable dividend. It goes to four places: first, R&D and product iteration, with Q1 2026 R&D expense of about $26.4 million; second, working capital such as inventory, receivables, manufacturing, and service networks, with Q1 2026 inventory still at about $130.8 million; third, small acquisitions and software reinforcement, such as Cohu's cash acquisition of Tignis, an AI process control / analytics software company; fourth, buybacks and capital-structure management, with repurchases of about $23.6 million, $27 million, and $8.6 million in 2023-2025, respectively, but no repurchases in Q1 2026. Under the Baillie framework, this is an "industrial technology business with improvement potential," but it has not yet proven that every dollar earned can be rolled into larger free cash flow per share with high certainty.

Jun 9, 2026
What conditions need to hold simultaneously for it to rise fivefold in ten years? Are those conditions realistic? What expectations are embedded in today's share price?2/10

Conclusion first: for COHU to rise fivefold within ten years, "semiconductor-cycle recovery" is not enough. It would need to transform from a cyclical test equipment stock into a growth stock with AI/HPC incrementality, stable recurring revenue, high margins, and a high valuation multiple. Based on the U.S. market close of about $52.51 on 2026-06-08 and a market cap of about $2.468 billion, a fivefold gain would imply a share price of about $262.55 and a market cap of about $12.3 billion. At a 30x net-income valuation, that would require about $410 million of net income; at 20x, it would require about $620 million of net income.

To get there, at least four things need to happen simultaneously: first, revenue must not only recover from the trough to the previous cycle peak, but rise from 2025 revenue of $452.96 million and a net loss of $74.27 million to roughly more than $2 billion, or even higher; second, AI/HPC, HBM, SiC/GaN, software analytics, and aftermarket recurring revenue must truly take over from the system-equipment cycle, allowing gross margin to stabilize at 45%+ and operating margin to return to double digits or even approach 15%-20%; third, Cohu must win enough high-value share against strong competitors such as Advantest and Teradyne, rather than only picking up peripheral equipment and service revenue; fourth, convertible debt, M&A, and buybacks must not dilute per-share value, because "the company getting bigger" is not the same as "shareholders' value per share rising fivefold."

These conditions are not completely unrealistic, but their realism is low. The positive evidence is that the company reported Q1 2026 net sales of $125.1 million, about 60% recurring, and a GAAP net loss of $12.1 million, and management gave Q2 guidance of $144 million plus or minus $7 million, AI-driven compute TAM of about $750 million, and FY26 HPC revenue outlook of about $80 million-$100 million. This shows that recovery and AI/HPC opportunities do exist. The problem is that these numbers are still far from the fivefold market-cap requirement of more than $400 million in annual net income. The report also repeatedly notes that COHU's historical revenue and cash flow are highly cyclical, that it posted losses in both 2024 and 2025, and that its moat is only moderate, not a wide-moat company that becomes steadier and compounds naturally as it grows.

Today's share price already embeds fairly optimistic expectations: the market is not pricing COHU as an ordinary cyclical stock, but is paying in advance for a combined scenario in which AI/HPC orders keep materializing, recurring revenue stays above 60%, gross margin returns to high levels, operating margin repairs materially, and the valuation multiple does not collapse. The report's bull range was only $38-$48, and the current price of $52.51 is already above the top of that optimistic range. So the answer to question 9 is that a fivefold gain in ten years requires a series of high-quality deliveries to happen together, while the current price already reflects a large portion of them and leaves a very thin margin of safety.

Jun 9, 2026
Why has the market not realized all this yet? Is it because the market does not understand it, looks down on it, or cannot look far enough? What could become the "narrative inflection point"?2/10

Conclusion: the market is not completely unaware of COHU's AI/HPC, aftermarket recurring, and software-ization story. More precisely, it has already paid for part of it, but is still unwilling to treat COHU as a great growth stock that can rise fivefold in ten years. The reason is not "does not understand it," but "looks down on it + cannot look far enough": it looks down on COHU because it is not a leader in the core profit pool like Advantest/Teradyne, and it cannot look far enough to treat 60% recurring, AI/HPC, and Tignis/DI-Core software analytics as a second curve that will change the long-term margin profile. More importantly, the current price of about $52.51 and market cap of about $2.468 billion are already above the report's bull range of $38-$48. That shows the market has not ignored the story; the story has moved ahead of profit delivery.

The market's skepticism has factual support: COHU posted operating losses in both 2024 and 2025, with 2025 revenue of $453 million, a net loss of $74.27 million, and rough FCF of only about $10.7 million. Although 2026Q1 recovered to net sales of $125.1 million, GAAP gross margin of 46.3%, and about 60% recurring, GAAP net loss was still $12.1 million. The company's official Q1 materials did raise AI-driven compute TAM to about $750 million and FY26 HPC revenue outlook to about $80 million-$100 million, but the market is looking not at TAM, but at whether these opportunities can continuously turn into high-margin revenue and cash flow.

The real narrative inflection point is not "telling the AI story again," but several consecutive quarters of numbers that rewrite COHU from a cyclical recovery stock into a structural growth stock: revenue recovers, gross margin stabilizes above 45%, operating margin turns positive and moves toward double digits, free cash flow improves continuously, and orders from AI/HPC/HBM, interface consumables, and software analytics are not just one-time system deliveries but can drive higher-margin recurring, software, and service revenue. If COHU can prove that 60% recurring can support profits even in a weak cycle, and turn AI/HPC opportunities from TAM into visible backlog, revenue, and FCF, the market may re-rate it.

Conversely, the narrative can also be disproven easily. If the AI test profit pool continues to flow mainly to Advantest/Teradyne and COHU only receives limited equipment orders alongside the industry recovery; or if convertible debt dilution, gross-margin decline, and lagging FCF appear, then the supposed perception gap will become valuation overreach. From the Baillie perspective, the real problem is not that the market is "too slow," but that COHU has not yet produced enough evidence to make investors believe it can move from a small, cyclical test equipment vendor onto the high-quality compounding path required for a fivefold gain in ten years.

Jun 9, 2026

Ask about this report

Members can ask about this report; once answered it appears under "Reader Q&A" on this page. You can also highlight a passage in the text to ask about it directly.