Cohu Inc

NASDAQ:COHU

Good day, and thank you for standing by. Welcome to Cohu's Fourth Quarter 2024 Financial Results Conference Call. [Operator Instructions]

Please be advised that today's conference is being recorded. I would now like to turn the conference over to Jeff Jones, Chief Financial Officer. Please go ahead.

Good afternoon, and welcome to our conference call to discuss Cohu's fourth quarter 2024 results and first quarter 2025 outlook. I'm joined today by our President and CEO, Luis Muller.

If you need a copy of our earnings release, you may access it from our website at cohu.com or by contacting Cohu Investor Relations. There's also a slide presentation in conjunction with today's call that may be accessed on Cohu's website in the Investor Relations section. Replays of this call will be available via the same page after the call concludes.

Now to the safe harbor. During today's call, we will make forward-looking statements reflecting management's current expectations concerning Cohu's future business. These statements are based on current information that we have assessed which, by its nature, is subject to rapid and even abrupt changes. We encourage you to review the forward-looking statements section of the slide presentation and 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 speak only as of today, February 13, 2025, and Cohu assumes no obligation to update these statements for developments occurring after this call.

Finally, during this call, we will discuss certain non-GAAP financial measures. 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 Muller, Cohu's President and CEO. Luis?

Hello, and welcome to our quarterly earnings call. Full year 2024 revenue was about $402 million as Cohu's main market segments are crossing the chasm of a downturn. On the other hand, the full year non-GAAP gross margin of 45% demonstrates the value of selling differentiated products and optimizing our manufacturing cost structure.

Fourth quarter revenue was within guidance range, but gross margin was impacted by an inventory reserve charge of $2.1 million. Jeff will get into more details on these later.

Revenue for the quarter was split 62% recurring in the balance systems. Systems revenue increased sequentially in computing, industrial and consumer segments, although offset by declines in automotive and mobile where customers are working through ongoing inventory correction.

Estimated test cell utilization at the end of December increased 1 point quarter-over-quarter to 73%, driven by OSATs that improved 2 points to 76% and while IDMs closed the quarter at 70%. We see our traditional IDM customers at different stages in the cycle and OSATs benefiting from the strength of fabless customers aligned with data center and network infrastructure.

As mentioned last quarter, Cohu entered the memory and silicon carbide power semiconductor markets with orders for HBM inspection and die level burn-in. We shipped our first HBM inspection system and received a repeat order early in first quarter that we expect to ship in the middle of this year. Growth prospects in HBM look positive with the potential to deliver $7 million of revenue this year as we support HBM3 and a new HBM4 product ramp later in the year.

We're working hard to align new Cohu products to compute applications, mainly positioning the company to capitalize on the growth in the data center market and later opportunities with AI at the EDGE. Our interface product team had a design win for testing 800G switches used in next-generation data centers and cloud computing. These are high-bandwidth devices that accelerate AI/ML workloads with stringent signal performance requirements in test. We're excited to see our products gaining traction across various data center applications spanning memory, network infrastructure and compute.

Turning to our software platform. We are strong believers in the value AI will bring to semiconductor manufacturing with opportunities to optimize yield and productivity. Just last month, the CFO of TSMC made a comment during a public interview that 1% productivity improvement has the potential to drive $1 billion of value for their business.

Putting this a bit in context, the global semiconductor process control market is believed to be about $2.6 billion today. We estimate that the potential total available market for data analytics for process control, data visualization, connectivity and predictive applications in the back-end semiconductor manufacturing segment to be about $600 million, which accounts for third-party and in-house developed solutions.

We're pushing to establish Cohu as a main player in this area, focusing primarily on back-end manufacturing and our current customers' spending on the software stack. In the fourth quarter, we qualified Cohu's DI-Core and received the first PO to optimize visual inspection yield for a U.S. headquartered semiconductor manufacturer with operations in Asia.

Customers continue to subscribe to our software solutions, adding to our critical recurring revenue stream in an area that we expect can generate substantial growth over time. In line with this strategy, last quarter, we announced a definitive agreement and early this year confirmed closing the acquisition of Tignis, a provider of artificial intelligence, process control and analytics-based monitoring software. We believe there is an opportunity to grow Cohu's software revenue at an annual rate of 50% or more over the next 3 years as the industry seeks solutions to optimize yield and productivity using AI power process control and data analytics solutions.

As our core automotive, industrial and mobile customers continue to navigate an inventory correction, we're committed to growing Cohu in 2025 with customer design wins and SAM expansion driven by new product investments targeting data center and EDGE AI applications.

Let me now turn it over to Jeff for further details on last quarter results and next quarter guidance. Jeff?

Thanks, Luis. Before I walk through the Q4 results and Q1 guidance, please note that my comments that follow all refer to non-GAAP figures. Information about the non-GAAP financial measures, including the GAAP to non-GAAP reconciliations and other disclosures are included in the accompanying earnings release and investor presentation, which are located on the Investor page of our website.

Now turning to the Q4 financial results. Revenue for the quarter was within guidance at $94.1 million. Full year 2024 revenue was $401.8 million. Recurring revenue, which is largely consumable driven and more stable than systems revenue, represented 62% of total revenue in Q4 and 65% of full year 2024 revenue.

During the fourth quarter, one customer in the automotive market accounted for more than 10% of sales. However, no customer accounted for more than 10% of sales for the full year 2024.

Q4 gross margin was 41.8%, about 220 basis points lower than guidance due to a $2.1 million charge to our inventory reserve for old, slow-moving, customer-specific inventory. Excluding the impact of the inventory reserve, gross margin was in line with the guidance. Full year 2024 gross margin was resilient and higher than modeled at 45%.

Operating expenses for Q4 were lower than guidance at $45.3 million, driven by lower labor costs due to replacement and new hire delays as well as higher vacation utilization than forecasted. Fourth quarter non-GAAP operating loss was approximately $6 million.

Q4 interest income net of interest expense and a small foreign currency gain was $2.3 million.

Q4 pretax income consists of foreign profits combined with a loss in the U.S. The Q4 tax provision of $3.4 million reflects tax expense on foreign profits but no tax benefit from the U.S. loss due to our valuation allowance against deferred tax assets.

Non-GAAP EPS for the fourth quarter was a $0.15 loss. The $2.1 million inventory charge accounts for approximately $0.04 of EPS.

Moving to the balance sheet. Overall, cash and investments decreased by $7 million during Q4 to $262 million due to $2 million used in operations, debt repayment of $2 million and fourth quarter CapEx of approximately $3 million.

CapEx for full year 2024 was approximately $11 million, lower than prior years and primarily driven by facility improvements in the Philippines and Germany, supporting operations for our interface and automation businesses.

Cohu had 0 share repurchase activity in Q4. Through the end of Q3 and fiscal 2024, we had repurchased approximately 915,000 shares for $27 million, which exceeds our goal to offset share dilution from our equity compensation plan of approximately 500,000 shares per year.

Overall, Cohu's balance sheet remains strong, supporting investment opportunities to expand our served markets and technology portfolio in line with our growth strategy and returning capital to shareholders through our share repurchase program.

Now moving to our Q1 outlook. Recent customer requests to delay Q1 shipments to later in 2025 have impacted our initial view of first quarter revenue. As a result, we're guiding Q1 revenue to be approximately $97 million plus or minus $7 million.

First quarter gross margin is forecasted to be approximately 44%, benefiting from Cohu's differentiated products and our stable high-margin recurring business, which adds resilience to profitability and provides consistent cash flow through industry cycles.

We expect gross margin to increase again when our revenue recovers with a broader semiconductor device market recovery and with better absorption of our factories infrastructure costs.

Q1 operating expenses are forecasted to be approximately $49 million, about $4 million higher than Q4. Nearly half of the quarter-over-quarter increase is due to the addition of Tignis, a provider of artificial intelligence process control and analytics-based monitoring software. The balance of the increase is due to higher labor costs quarter-over-quarter, including the implementation of a delayed merit increase across the employee base and the typical reset of employer payroll taxes that are unique to Q1.

For the rest of 2025, operating expenses should be approximately $48 million per quarter when revenue is approximately $100 million. As revenue grows to about $130 million, operating expenses are projected to increase to approximately $50 million per quarter.

We're projecting Q1 interest income, net of interest expense and foreign currency impacts, to be approximately $1.3 million at current interest rates.

The Q1 non-GAAP tax provision is expected to be approximately $3 million because of tax on foreign profits without benefit from the U.S. loss. Until the markets recover, we expect a similar tax provision profile as we navigate through this cycle.

The basic share count for Q1 is expected to be approximately 46.6 million.

And that concludes our prepared remarks, and now we'll open the call to questions.

[Operator Instructions] Our first question comes from the line of Craig Ellis with B. Riley Securities.

Yes, and appreciate all the detail on the new businesses that are coming into the model. I wanted to start my questioning there. So Jeff, you said that Tignis, and I apologize if I pronounced it wrong, but you said that would be about $2 million of OpEx in the quarter. The question is, is there any revenue associated with that OpEx? And if not, when would we expect that business to be earnings per share breakeven as we think about its ability to grow with that 50% CAGR you outlined?

Yes. So Craig, it's Tignis, T-I-G-N-I-S. And last year's revenue was sub-$1 million, likely to be the same this year, are expecting healthy growth rates over the next few years, but at the moment, probably right around $1 million or sub $1 million for 2025. So it's going to take a few years for it to get to the breakeven point.

Got it. And then, Luis, if I wanted to hone in on the revenue impact this year of some of these incremental drivers, I think in the deck we identified that the high bandwidth memory opportunity could be around $7 million.

If we look at software all-in, Tignis and DI-Core, how big could that be? And then the silicon carbide market expansion, the single-chip burn in capability that you've identified recently. How big could that be this year? How big are these new drivers as we think about '25 and '26?

So if I take '25, Craig, like I said, HBM, about $7 million, the silicon carbide at about $5 million. So tallying up to $12 million. We think, as Jeff mentioned here, software about [ 113 ]. And then I think you got to top it off with the Diamondx win at an automotive customer that we talked about in the last 2 quarters, that is likely to be $10 million to $15 million. So I think if you put it all together, we're looking at sort of $25 million, $30 million incremental this year from these new either SAM expansion or design wins. And I would expect those to accelerate in '26, but I don't -- I wouldn't have a number to give you at this time on that.

Yes. And if we look at those together, Luis, and thanks for the color on that. If we look at them all in, is that a mid-40s gross margin that we should be thinking about? Would it be higher than that? I think down the road in a couple of years, it would be higher with software growing. But how do we think about the gross margin on that incremental business this year?

Yes. Craig, I would say high 40s but ultimately growing to 50 as software grows.

Okay. Nice. And then lastly, if I could, guys, as we look at utilization levels across some of the end markets, A&I is the closest to that, somewhat magical 80% threshold at 75%. The question is, is A&I on its way up towards 80%? Or is it actually drifting lower, just given some of the comments we've heard from bigger U.S.-based analog companies over the last month?

Sorry, what do you -- Craig, clarification here. What do you mean by A&I?

Auto and industrial. Yes.

Okay. Thank you. Yes, you're correct. Auto industrial continues to be the one that's sitting at the highest plateau. Nevertheless, it continues to be the customer base, particularly in the analog space that has the highest inventory correction yet to be digested. Some of these customers are talking about another 2 quarters of digestion, another one seemingly 3 quarters, if you look at the numbers.

In the meantime, while mobile is sitting at a lower level, there's some dynamic happening between Android and iOS that I think is going to drive some incremental business. I guess I'm not going to pick a side here to say it. But in one of those in particular that we expect to see some leverage in the business for us starting in the middle of this year.

Our next question comes from David Duley with Steelhead Securities.

Could you just remind us, you just referred to the Diamondx win of $10 million to $15 million in 2025. Could you just elaborate on the application again? And how you won? What was the key parameter for you winning that business?

Sure, Dave. Well, starting from the end here, the key parameter to winning that business is ultimately cost of test. It's being able to do their portfolio of devices, which spans power semiconductor applications, microcontrollers. And there is also a PMIC group that's separate from power. In the moment, we're addressing 2 of the 3 groups, hoping to be able to extend it to the third group. So it's sort of a cost of test differentiation. And I guess, like I said, 2 of the 3 groups is what we're addressing today.

Okay. And then in aggregate, I think you've had a few years now of declining year-over-year revenue. When would you expect there to be some sort of turn either driven by new products or recovery in the end markets? What's your best guess at this point for your recovery?

Dave, guessing the market is the toughest part of it all, as you understand, and people seem to always indicate that it's 6 months away is the turning point. So we're not at a position of necessarily guessing the market nor sitting in either one. So what we've been working on is to expand our penetration in segments that have, at the moment, higher growth, which would be particularly things associated with data centers that we haven't had a lot of exposure in the past. HBM is one example. We have other activities that hopefully we can talk through in the coming quarters this year.

The software investment is perhaps more of a long-term play than necessarily a material 2025 investment, but we see many of the factories driving automation and optimization, whether it's yield or productivity, that could be benefited by predictability, which in essence, becomes the use of AI models or machine learning models sometimes. They're different, but in both cases, software.

So our focus right now is pivoting to areas that have both near-term opportunity for growth but as well as long-term opportunity for growth. In the meantime, the market is going to do what the market is going to do and the primary analog segment of the market, which is mostly automotive, industrial continues to be depressed. There's still inventory digestion happening. We keep looking at our customers' reduction of inventory quarter-over-quarter and where are they relative to trend lines. So it's very likely that at the current pace, we'll see another 2 quarters before our customers in that segment turn the corner from a general market perspective.

Okay. And then as far as when you start to see recovery in the revenue line, you have some nice numbers, Jeff, about what we would expect for operating expenses going forward at higher revenue levels. What would you expect maybe at those same revenue levels the gross margins to be?

So when we're hovering around $100 million, gross margin will be somewhere in the 44% to 45% range. Then as we migrate up to $130 million, we're expecting to be about 46.5% on gross margin.

Our next question comes from the line of Ross Cole with Needham.

I was wondering if you could verify the different segment revenues for '24 and then provide your latest assumptions on the segment revenues for 2025 and maybe rank order them, if possible?

So Ross, are you talking about all of 2024 and projecting all of '25?

That would be great, if possible. Yes.

Yes. We don't have -- we've got the historical data on 2024, just we don't have that perspective yet on 2025. So let's see. We look at systems revenue by market. So I'm going to give you those percentages for 2024. And we were 9% was automotive, 6% was industrial, 11% mobile, 4% consumer, 3% compute and 2% for IoT.

Great. And I was wondering if you could possibly rank order these in terms of your best guess for 2025?

Yes. I think at the moment, as Luis said, auto and industrial has probably the most inventory to work through at the moment. But if we're looking at second half, I believe the indications or the projections at the moment would be that automotive and industrial begin a recovery followed by mobile.

Great. And if I can ask one more quick question. I remember previously you had shipped a couple of tools for HBM and [ ASIC ] wafer test. Could you confirm if your plan on recognizing the revenue for these in the first quarter or if that's more of a second quarter opportunity?

The HBM is a Q1 revenue recognition.

First tool.

Yes, it's the first tool. One tool In Q1. Die level burn-in is the other one you referred to, which is further perhaps second half of 2025.

Our next question comes from the line of Robert Mertens with TD Cowen.

This is Robert Mertens on for Krish Sankar. I guess first, I believe you had mentioned earlier that some of the strong order flows in the September quarter sort of gave you confidence to look beyond the December guide into March, expecting that quarter to be up sequentially, maybe around 10% or so. So just trying to parse out that commentary in the new low single-digit guide. Is this new outlook primarily just because of some of those customer order pushouts that you mentioned earlier on your prepared remarks? And if so, would that be something you would expect to come back in the June quarter? And then I have one follow-up.

Robert, the answer to your first part is, yes. It was due to a recent pushout of tools out of Q1. So we have the orders. It's the actual shipment date that was -- has been delayed. Now it's not necessarily Q2. It's throughout 2025. It's about $7 million. So again, it's spread through the balance of the year.

Okay. That's helpful. And then just another quick follow-on question. In terms of the new Neon and die level burn and test opportunity with high bandwidth memory. Congrats on securing another follow-on order. With the growth in this emerging business, will you be primarily driven just by scaling that success with this initial customer? Are you currently working on getting other tools, potentially full acceptances with the other major memory suppliers? Or do they tend to go internally or have sort of incumbent suppliers themselves that would make it harder to gain business there?

Yes. Robert, it's kind of both. The numbers that I quoted to the first question in terms of what you expect for 2025, we're essentially scaling with the first customer orders that we got for these new tools. But we absolutely are working on advocating for these tools and promoting them, selling them to other customers that could benefit from what these tools can do. But none of that is accounted for in the numbers that I referenced on the first question today.

Our next question comes from the line of Brian Chin with Stifel.

This is Denis Pyatchanin on for Brian. In light of the ongoing inventory control in the industry and fab utilization cuts that we've seen recently, how stable do you believe the $60 million run rate is for your services and [indiscernible] recurring revenue?

It is stable. It's proven to be stable over the -- historically. It's not immune to the downturn, and we can see that in the numbers, but it's probably has about 1/3 of the volatility of systems. So the utilization of equipment in customers' facilities has remained pretty steady. So we would expect our recurring revenue to also remain fairly steady.

Great. I think you were previously somewhat more positive on a mobile and RF test. Has this changed? I think you mentioned a little bit earlier on this call that you think mobile will recover after industrial automotive. Did I understand that correctly?

The general market could be after, there are some dynamics, as I mentioned earlier, between sort of Android, particularly in the Android market space on certain customer share gain and transitions of devices that could be beneficial for us in the second half of the year, sort of starting in the middle of the year so to speak. But overall, the mobile market is still projected to be sort of a low single-digit growth this year. I think more of the dynamics in between customers could be more interesting.

Great. And then as my last question, it looks like industrial saw a little bit of strength that the systems revenue might have finished at a high for 2024 in the fourth quarter. Can we expect industrial to stay at this level? Or what kind of bounced around a little bit around this $8 million to $9 million mark?

So industrial has the potential to actually come up ahead of automotive as we see it today. But we'll see how that plays out in the next 2 quarters.

Our next question comes from the line of Craig Ellis with B. Riley.

Guys, I wanted to go back to the software business and just understand how you plan to execute that in a little bit more detail. Can you focus on the selling motion that you plan to implement in that business? Is DI-Core and Tignis, is that going to sell out as a complement to new systems? Does it sell into the installed base as an enhancement to existing systems? Is it both? Just help us understand a little bit more how you plan to go to market and therefore, how we can expect you're going to start to accrue sales and grow that high-margin business?

Yes. Good question, Craig. The DI-Core that we do today, it really has 2 branches, okay? It has a vision inspection branch on the software that is using a newer network for vision technology. And that branch will stay, pretty much will stay as is. I don't think of Tignis getting much involved on that branch at the moment or at least not the initial plans, okay?

There is a second branch in DI-Core that is using machine learning for fault detection and predictability for equipment maintenance. So the intention there is to leverage on Tignis' artificial intelligence software to give continuity to the DI-Core to products outside of the Cohu umbrella. So we would like to expand our DI-Core predictive maintenance capability to serve more than just Cohu products and the broader semiconductor back-end space. So that is an area that Tignis should be able to augment, what we're currently doing with DI-Core, and go much broader.

Separately, Tignis itself, it has a solution that is being used for front-end manufacturing that has nothing to do to Cohu current markets. We don't intend to shut that down. In fact, we intend to see Tignis continue to evolve there and sort of back up what they're pursuing at the moment and provide the necessary infrastructure and resources that they need to be successful in -- more in the general process control market.

So to summarize, Tignis' own growth factor is going to stay, and we intend to back it up. And at the same time, we plan on leveraging Tignis' technology on our core PdM tools for the back end so we can go broader and beyond Cohu own equipment. I hope that was clear.

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

Well, thank you for joining today's call. We really appreciate your participation, and we look forward to speaking with you soon. Have a good day.

Bye now.

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