Cohu Inc Q3 FY2025 Earnings Call

Good day, and thank you for standing by. Welcome to Cohu's Third Quarter 2025 Financial Results Conference Call. Please be advised that today's conference is being recorded. I'd now like to hand the conference over to Jeff Jones, Chief Financial Officer. Please go ahead.

Good afternoon, and welcome to our conference call discussing Cohu's third quarter 2025 financial results and our outlook for the fourth quarter of 2025. I'm joined today by Luis Müller, Cohu's President and CEO. If you need a copy of our earnings release, it can be found on our website at cohu.com or by contacting Cohu Investor Relations. A slide presentation accompanying today's call is also available in the Investor Relations section of the website. Replays of this call will be accessible via the same page after the conclusion of the call. During this call, we will be making forward-looking statements that reflect management's current expectations concerning Cohu's future business. These statements are based on the information available to us at this time, but they are subject to rapid and even abrupt changes. We encourage everyone to review the forward-looking statements section of our slide presentation and the earnings release as well as Cohu's filings with the SEC, including the most recently filed Form 10-K and Form 10-Q. Our comments are current as of today, October 29, 2025, and Cohu does not assume any obligation to update these statements for events occurring after this call. Additionally, we will discuss certain non-GAAP financial measures during this call. Please refer to our earnings release and slide presentation for reconciliations to the most comparable GAAP measures. Now I'd like to turn the call over to Luis Müller, Cohu's President and CEO. Luis?

Good day, everyone. Thank you for joining our Q3 2025 earnings call. I'm pleased to share our latest results and provide guidance for Q4. First off, let's talk about some highlights. Recurring revenue continued to grow for the third consecutive quarter, driven by strength in interface solutions and test handler spares. Systems revenue improved sequentially for the fourth quarter in a row, though it remains below normalized levels. We had several notable events in the third quarter. We announced an offering of convertible notes on favorable terms, which closed just after the quarter end and will support future growth and strategic initiatives. Jeff will discuss this in more detail later. We welcomed Matthew Hutton, our new Vice President of Strategy focused on advancing our growth initiatives, including mergers, acquisitions, and partnerships. Prior to joining Cohu, Matt was Head of Corporate Development at AspenTech. We communicated repeat orders for Neon HBM inspection tools, raising this year's revenue forecast for these systems to between $10 million and $11 million. These systems are used for inspection in metrology of high-bandwidth memory devices, which are critical components in high-performance computing and generative artificial intelligence applications. We shipped our first system configured for HBM4 inspection, reinforcing our optimism for future market prospects and high-bandwidth memory. Our Eclipse handler equipped with proprietary active thermal control was selected for production test of next-generation AI processor devices by a leading U.S.-based semiconductor manufacturer. The Eclipse platform is designed to scale seamlessly across diverse power applications, providing the flexibility and operational efficiency required to support our customers' evolving high-performance processor road maps. This adaptability ensures that as processor technologies advance, our solution remains a reliable foundation for next-generation computing needs. Our current thermal solution ensures optimal device temperature control and test repeatability up to 3,000 watts power dissipation with ultrafast temperature ramp rates and tight thermal guard band, supporting demanding semiconductor test requirements. Now let's dive into the detailed results. Consolidated revenue reached $126 million with both systems and recurring revenue improving quarter-over-quarter. Revenue was split 45% systems and 55% recurring. Non-GAAP gross margin of 44.1% reflects the value differentiation of our products and the resilience of our recurring business model. Estimated test cell utilization remained stable quarter-over-quarter ending September at 74.5%. While systems orders moderated last quarter, growth in recurring revenue and new wins position us well for Q4 and beyond. We secured new business wins, including orders for our automated test equipment and automated optical inspection for high-growth markets. During the quarter, we secured roughly $1.7 million in new business, highlighted by our first Diamondx order from a long-standing Cohu handler customer. This order will support the testing of application-specific analog power integrated circuits, serving key automotive and industrial market segments. This customer win marks the continuation of Cohu's growth in the mixed signal test market with Diamondx as we push to diversify our test platform beyond RF and display driver IC test. We secured a new order of our Krypton system with a European customer, enabling advanced optical inspection of devices used by a prominent U.S. mobile phone brand. We booked a $2.3 million order for precision analog test contactors at a U.S. IDM and continue to diversify our test platform portfolio with this customer. We anticipate a seasonal slowdown in Q4, partially offset by ongoing market recovery and remain optimistic about long-term prospects, especially in computing and high-bandwidth memory inspection. As tariffs returned to the spotlight in recent news, I want to reassure everyone that Cohu's current exposure to China remains very limited. Revenue from customers based in China accounts for only a low single-digit percentage of our total consolidated results. Additionally, a substantial share of our business is generated outside of the U.S., further diversifying our global footprint. Thank you for your attention and continued support. I'll now turn it over to Jeff for a deeper dive into our financial results and Q4 guidance. Jeff?

Thank you, Luis. Before reviewing the third quarter results and providing fourth quarter guidance, please note that my comments refer to non-GAAP figures. Details about non-GAAP financial measures, including GAAP to non-GAAP reconciliations and other disclosures are included in the earnings release and investor presentation on our website. For Q3 2025, revenue exceeded guidance and reached $126.2 million. Recurring revenue, which is primarily driven by consumables and is more stable than systems revenue, accounted for 55% of total revenue for the quarter. During the third quarter, three customers, one in the mobile segment and two in the automotive segment, each represented more than 10% of our sales. The Q3 gross margin was in line with guidance at 44.1%. Operating expenses for the quarter were $48 million, which is $2 million lower than guidance. This reduction was mainly due to the timing of R&D material now scheduled for receipt in Q4. Net interest income after accounting for interest expense and a small foreign currency loss was approximately $1.1 million for Q3. The tax provision came in about $3.5 million lower than forecast at $11.7 million, resulting from the reversal of tax reserves following the completion of a jurisdictional tax authority audit. Moving to the balance sheet. Cash and investments decreased by $11.2 million during Q3. This was primarily due to cash used in operations to support a 17% growth in sales quarter-over-quarter and to fund a $33 million increase in accounts receivable. No stock repurchases were completed during Q3. Since the inception of our share repurchase plan, we have repurchased around 4 million shares for approximately $117 million, leaving about $23 million available for additional future repurchases. Total debt stands at $18 million, unchanged from the previous quarter. Q3 capital expenditures were $4 million, mainly for facility improvements. We're maintaining our 2025 capital expenditure target of approximately $20 million, which includes the $9 million Melaka facility purchase completed in Q1. In late Q3, we announced a strategic convertible notes offering. In early Q4, we completed the upsized offering, raising gross proceeds of $287.5 million at attractive rates, including a 1.5% interest rate, 32.5% conversion premium, and a 5-year term. We purchased a 100% capped call to limit shareholder dilution until the stock price doubles and exceeds $41 per share. The repayment structure of the notes is net share settlement, meaning Cohu will repay the principal of $287.5 million in cash and has the option to settle any in-the-money amounts in cash, shares or a combination of both. This structure, combined with the up 100% capped call limits shareholder dilution. The net proceeds will provide additional liquidity to strengthen our balance sheet and support strategic initiatives. Looking ahead to Q4, as Luis noted, we anticipate a seasonal slowdown for systems, which is partially offset by a continued market recovery. Overall, we expect Q4 revenue to be about $4 million or 3.5% lower than Q3, driven by systems revenue. Our resilient recurring revenue is forecasted to increase for the fourth straight quarter and should represent about 60% of total Q4 revenue. Our guidance for Q4 revenue is approximately $122 million, plus or minus $7 million. The gross margin for Q4 is projected at approximately 45%. Operating expenses are expected to be about $50 million, including around $2 million for variable R&D product development prototype materials. Total operating expenses are consistent with the restructuring plan targets implemented in late Q1 of this year. Once the full impact of the restructuring plan is realized at the beginning of 2026, we anticipate quarterly operating expenses to be approximately $49 million when revenue is around $130 million per quarter. Q4 interest income, net of interest expense and foreign currency impacts is projected to be approximately $1.7 million at current interest rates. The Q4 tax provision is expected to be about $4 million, and the diluted share count for Q4 is projected to be about 47.1 million shares. That concludes our prepared remarks, and now we'll open the call to questions.

Our first question comes from Brian Chin with Stifel.

So I guess, first question, nice to see the improved system revenue momentum, particularly from the mobile segment these past few quarters. Based on the customer broadening metric you shared and the uptick in utilization, is that the main area of improved near-term revenue visibility for the company? And how much confidence does this give you on sustaining some top line momentum kind of moving beyond the seasonal period into the first half next year?

Yes, you're correct. A significant portion of the momentum in the third quarter was tied to a customer purchasing the Eclipse handler, along with HBM alongside the Neon system. These are the two main highlights of the quarter. We have qualified the Eclipse with another computing customer, which we announced ahead of SEMICON West in the third quarter. Additionally, we have several other customers evaluating the system, including one for a new GPU application set for 2026, and two others related to data center network communication and an ASIC accelerator. Regarding our confidence for 2026, we are optimistic about the progress of the HBM business, as we've received a couple of repeat orders since the start of the fourth quarter. We're also developing a relationship with another customer to discuss requirements for qualifying one of our inspection tools. Several customers are in various stages of evaluating the Eclipse for data center applications. We recently saw another customer announce their successes in the data center market, for which we are involved with the Eclipse in their inference data center device partnership. Overall, we are confident that we are expanding our business beyond traditional sectors such as automotive, mobile, consumer, and industrial, moving towards AI applications, including GPU and network processing. We expect to see positive results from this shift in 2026 with the Eclipse and our inspection systems.

Got it. Can you help us understand what is driving the success of the Eclipse handler with the active thermal T-Core subsystem? Is it due to the increased wattage of some of the new AI processors coming out? And does your platform continue to scale? It seems like with each passing year, more advanced and higher-performing chips are being introduced. Are you replacing existing competitors with that tool?

Yes. Yes, you basically hit the nail on the head. We talked here about 3,000 watts of power dissipation. I mentioned that in my prepared remarks, but that's sort of the current state of what would be shipping for production needs in 2026. With that said, the requirements continue to go up, and we're already working on the next-generation thermals that will support 2027 and different sets of applications coming up later next year as well. So your question is pretty much the answer to it. It's all about the thermal power dissipation and power densities. We never talk about power densities, but power density per square inch of silicon as well, the size of the dies, how delicate and the amount of force you have to apply. So it's all related to that complexity of dynamically controlling heat dissipation on very complex semiconductors actually doing tests.

Maybe just one last quick follow-up. In terms of revenue contribution, is that sort of first half next year? And just compute, I think, has not historically been in recent years like a double-digit exposure segment, but do you feel pretty comfortable thinking that could be double digits next year for the year?

Double digit, you mean double-digit growth, double-digit from a revenue contribution. If you look from a revenue contribution, I would say, I would expect computing to be sort of in the low teens. We always talk about systems and recurring, so not counting recurring in the mix here. I'm thinking it would be sort of the low teens percentage contribution of revenue going into 2026.

Our next question comes from David Duley with Steelhead Securities.

To follow up on Brian's question, it seems that based on your Eclipse win at a major AI processing company and another recent win or tool of record with the CPU company, it’s reasonable to assume that any of these APUs, CPUs, XPUs, GPUs, or whatever the term is for networking processors will need to be thermally controlled and tested. Therefore, is it fair to assume that the total addressable market for this is quite large, considering there are many significant customers you have not yet reached?

Yes, that's absolutely correct, Dave. The power dissipation levels vary significantly. We currently have some inference processors that are around 600 watts of power dissipation. Our high-end GPUs are approaching 3,000 watts and we have network processors currently being qualified at 1,200 to 1,400 watts. Overall, the power dissipation levels range from hundreds to a couple of thousand watts and continue to rise. The roadmap indicates this upward trend.

So the higher the wattage, the higher the heat, and so that becomes more and more important. And so I guess it's a fair assumption that going forward, you might have a little bit more exposure on the GPU side with this product, and networking the hyperscaler custom ASIC customers?

That's correct. That's correct. The higher the wattage, the more complex these things are getting, the more it lends itself to expertise that we have at Cohu. We're being asked by quite a few customers now to address some of their requirements. They're very difficult. I mean, as you can imagine, when you're approaching 3,000, 4,000 watts of power dissipation, this is fairly complex. There aren't that many people out there that have the engineering and the technology to do this. So we're working pretty heavily on it. Needless to say, we're pivoting the business more towards AI applications.

Okay. When you consider overall AI exposure in 2025, could you share your thoughts on the revenue stream from the HBM inspection product, the Eclipse, and some other initiatives? What do you anticipate your revenue dedicated to AI will be in 2025, especially since it was nearly nothing in 2024?

Yes, I would say it's nearly zero in 2024. When discussing AI, it's important to be cautious, as many associate it primarily with data centers. However, there's actually a combination of processors that are already functioning with some level of language model integrated. Looking ahead to 2025, I would estimate around $40 million in system revenue this year related to edge AI or data center-related AI, and we anticipate that will grow into 2026.

Our next question comes from Robert Mertens with TD Cowen.

This is Robert on behalf of Krish Sankar. I guess just the first one, with the recent convertible raise, how are you thinking about the best use of cash between developing some of the new areas of expansion, be it investment in the software business or high bandwidth memory versus historically completing a number of smaller tuck-in M&A deals to bolster the technology portfolio. And then maybe I'll just add in your views on using cash for share repurchases and I know that been on pause for the last few quarters?

Yes. Good question. And really the answer is we want to pursue both paths. And in order to pursue acquisitions of any meaningful size, we needed to go to the financing market, we needed capital, which basically drove our decision on the convert, to strengthen the balance sheet and have more flexibility when it came to growing through acquisition. And so we're going to continue to focus on organic development in the areas that Luis has been talking about. But clearly, we want to be opportunistic as well when it comes to M&A. And of course, with the recent hire of Matt, it's a priority for us. That's really the main driver for the convert. Now with respect to buyback, that's a sort of a Board decision. And yes, we're on pause for now. Should the stock valuation go to the point where it is more compelling, I think we would again get back into the game. But the objective for 2025 on the share repurchase was to offset dilution from our equity compensation plan. And so we've essentially done that in Q1. I suspect it will be similar for next year.

Got it. And then thanks for the color on the latest Eclipse system. Maybe just going back to that, in terms of the areas where that's focused, is that sort of more of a broad-based system or any sort of end market and you're just seeing more traction on the compute side, the power and the heat requirements and just end demand in that end market today versus sort of your traditional auto and industrial? Or is that something that auto and industrial customers could start to look into more once their end demand picks back up?

Yes, Robert, the Eclipse is not really a traditional product for industrial applications or other applications. So you could say we can use it for consumer products. We can use it for RFIC test. We can use it for general mobile applications. But we've been really more selective here with our engineering resources and putting them more around these complex thermal requirements that we see in general AI processor needs. I think they’ve shared a collection of letters here that people using AI today, right from XPU, TPU, NPU, APU, and GPU. And so we are really focused on that. We're really focused on, look, if it is AI-related, whether it's training or inference mode or network mode, backbone network connectivity, that's interesting because it applies or it lends itself well to our thermal technology. It lends itself to where we can differentiate. It lends itself to where we can bring value to the table. So we're being quite selective on where we are deploying the Eclipse right now and the bandwidth that we're deploying against customers that have those challenges. So the product could be used for a variety of other things, not traditionally your industrial auto use in this case. And so we're being more focused on AI end use cases.

Our next question comes from Denis Pyatchanin with Needham & Company.

So even with the recent uptick in Q3, mobile system orders year-to-date versus year-to-date last year seemed to be lagging somewhat behind other segments kind of even in light of utilization recovery there. Why have system purchase in the segment lag up somewhat? And are you perhaps expecting strength in mobile into next quarter even with systems have guided down?

No, not exactly, Denis. If you look at our Q3 revenue, mobile was actually our largest segment, tied closely with automotive. We mostly completed the mobile-related shipments in the third quarter. As we move into the fourth quarter, we anticipate more shipments in the auto and computing sectors. Regarding mobile, I was thinking more about our test handlers. We will see some mobile demand in RF testing in the fourth quarter, which will contribute some revenue. However, I don't believe mobile will be our largest segment in the fourth quarter. I don't expect that to happen again.

Great. So one more, so for automotive and industrial, the cyclical recovery continues to be somewhat muted. What are you seeing in these markets in terms of recovery? So I think you're saying there's going to be some strength into Q4, but is there any visibility beyond that?

Yes, there's some puts and takes. You're right. This has been sort of an elusive recovery both in auto and industrial. I think we have had a quarter where we had some green shoots in auto in Q2 I want to say, and then it had some green shoots in industrial. We're having now is more talks from customers that are saying that they are back to the mode of needing initial capacity in the auto and industrial segment, talking about some initial demand in Q1 of next year, into Q2 of next year. Nothing dramatic yet, but the talks are starting to improve. We're also seeing an increase in spare sales to our handlers in the auto and industrial segment, basically supporting the fact that they're taking systems that have been put aside, so underutilized segment for test and bringing those systems back online. I think, like I said, we had three consecutive quarters now of recurring business improving and continue to project the fourth quarter recurring business to improve again sequentially. This applies both to, like I said, spares for our test handler systems, which is a very good indicator as well as improvement in our test interface business.

Great. And then briefly, could you discuss the gross margin strength sequentially into Q4, even with revenue being down a little bit? What's driving that?

There's a mix component to it, Dennis. And as Luis just mentioned, we've got increasing recurring revenue, which has gross margins in the mid-50s. And so we're expecting the recurring revenue to be about 60% of the total revenue, it was 55% in Q3. I think that's the main driver of that increase in gross margin quarter-over-quarter.

That concludes today's question-and-answer session. I'd like to turn the call back to Jeff Jones for closing remarks.

Thank you. And before we sign off, I'd just like to note that Cohu will be attending several investor conferences over the next 3 months, The Stifel Midwest Conference on November 6 in Chicago. The New York City CEO Summit Conference on December 16 and the Needham Virtual Conference on January 15 of next year. If you plan to attend any of these conferences, please reach out to your conference contacts or contact us directly to arrange a one-on-one meeting. Thank you for joining today's call, and we look forward to speaking with you again soon.

This concludes today's conference call. Thank you for participating. You may now disconnect.