庫力索法 (KLIC) 2025 Q1 法說會逐字稿

Greetings, and welcome to the Kulicke & Soffa Q1 2025 conference call and webcast.

(Operator Instructions) As a reminder, this conference is being recorded.

It's now my pleasure to turn the call over to Senior Director of Investor Relations, Joe Elgindy.

Please go ahead, Joe.

Welcome, everyone, to Kulicke & Soffa's fiscal first-quarter 2025 conference call.

Fusen Chen, President and Chief Executive Officer; and Lester Wong, Chief Financial Officer, are also joining on today's call.

Non-GAAP financial measures referenced today should be considered in addition to, not as a substitute for or in isolation from, our GAAP financial information.

GAAP to non-GAAP reconciliation tables are included within the latest earnings release and earnings presentation.

Both are available at investor.kns.com along with prepared remarks for today's call.

In addition to historical statements, today's remarks will contain statements relating to future events and our future results.

These statements are forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995 and are subject to risks and uncertainties that may cause our actual results and financial condition to differ materially from the statements made today.

For a complete discussion of the risks associated with Kulicke and Soffa that could affect our future results and financial condition, please refer to our recent and upcoming SEC filings, specifically the latest Form 10-K as well as the 8-K filed today.

With that said, I would now like to turn the call over to Fusen Chen for the business overview.

Please go ahead, Fusen.

Good morning, everyone.

Over the past several quarters, our General Semiconductor and Automotive end market have shown signs of inventory and capacity digestions, and we continue to anticipate a gradual improvement in fiscal 2025.

In parallel, we have continued to demonstrate technology leadership positions within the growing thermo-compression, and advanced dispense opportunities.

Visibility within our higher-volume Ball and Wedge markets is typically limited this time of year.

Regardless, our core business remains in the late stage of a market downturn.

While we continue to anticipate a return to broader capacity additions within the core Ball, Wedge, and APS businesses through fiscal 2025 due to improved field utilization rates, market trends and the reasonable industry growth expectations, we remain focused on what is within our control, primarily ongoing development, customer qualifications, and market adoption of our newest systems.

Additionally, industry momentum with thermo-compression technology, continues to broaden, and we are extending our leadership through new offerings and customer engagements.

Approximately three weeks ago, we shipped our latest Fluxless Thermo-Compression or FTC system in a new dual-head configuration to a key foundry customer.

This system provides nearly twice the throughput as our existing production-proven FTC system, which was already qualified and already provides an additional value proposition for advanced logic customers within its single-head configuration.

This new dual-head configuration will further extend value for advanced logic customers and also provide access into the high-bandwidth-memory market.

Beyond this new dual-head system, we are aggressively developing a future panel-based platform, which will further extend the value of FTC.

Of note, we are pleased to extend our FTC customer engagements, which now include a leading memory customer, in addition to our existing base of leading IDM, foundry, and OSAT customers.

This new memory engagement is supporting process development for future-generation HBM applications, leveraging our FTC leadership.

FTC is positioned to enhance the future HBM process. providing critical performance, form-factor, and efficiency enhancements by significantly reducing pitch and increasing I/O density for future AI and cloud computing workloads.

Last, the copper-to-copper TCB process is continuing to pick up momentum and is currently being reviewed by a leading IDM customer, in addition to our previously announced customer.

Our copper-first solution provides hybrid benefits such as zero-die gap, ultra-fine pitch, and direct copper-to-copper interconnects without the licensing fees, front-end production requirements, or yield issues, which are associated with initial hybrid bonding technologies.

These benefits are available today through our broadening FTC portfolio, which is well positioned to provide additional value for emerging advanced logic and advanced memory applications.

This technology leadership in FTC is enhancing our position within the broader TCB market as well.

We recently accepted orders from two new advanced packaging customers for several new APTURA systems.

We estimate the TAM for TCB in calendar year 2024 has exceeded $300 million, which represents a key market milestone for two significant reasons.

First, it highlights TCB is comparable in annual revenue to the mature flip chip mass-reflow equipment market.

Considering flip chip has been in high-volume production for over three decades, this milestone was reached relatively quickly for thermo-compression.

Second, TCB adds significant incremental value beyond traditional packaging techniques due to its ability to efficiently stack die, increase package-level-transistor density, and simplify the wafer-fabrication process.

TCB technology has a long life ahead and is anticipated to grow significantly over the long-term.

Over the coming years, we anticipate the broad TCB market to grow at a compound annual growth rate of 20% to 25%, with FTC growth expected to be materially faster over the coming years.

During fiscal 2025, we anticipate additional customers to move into higher-volume FTC production.

The need for More-than-Moore-related packaging solutions are emerging rapidly, and we are excited to be leading this transition with our market-ready FTC solution.

This broad industry evolution to more advanced chiplet-based packages is still in an early stage and is anticipated to play out over the long-term, driven by emerging artificial intelligence, cloud computing, and edge-device requirements.

As explained over recent quarters, we have secured clear leadership positions in Fluxless Thermo-Compression, which will play an increasingly important role within future heterogeneous and chiplet-based applications.

Currently, most advanced logic applications are dependent on aging flip-chip technology, which we anticipate will continue to transition to traditional TCB or Fluxless Thermo-Compression applications.

We continue to demonstrate consistent progress to expand our TCB portfolio, customer base, and market access.

This highlights our ongoing leadership, industry focus, and long-term potential of this emerging technology.

Turning to our financial results for the December quarter, we delivered $166.1 million of revenue, 52.4% gross margins, and non-GAAP EPS of $0.37. GAAP EPS of $1.51 was largely supported by customer reimbursements associated with our fiscal second quarter 2024 impairment charge of Project W. Lester will provide additional details shortly.

From an end market standpoint, the December quarter is generally driven by seasonality within the General Semiconductor market, although we continue to anticipate broader industry growth and demand for our core solutions within fiscal 2025.

The General Semiconductor market continues to be largely in a state of capacity digestion, with Ball Bonder revenue sequentially lower from September as expected.

We continue to expect we are in a late recovery stage and remain positive on broader recovery through fiscal 2025.

Our Ball Bonder teams remain very active developing new features and platforms to support the evolving high-volume assembly market.

We look forward to sharing more information and broadening the Ball Bonder portfolio later in fiscal 2025.

Within Automotive and Industrial, we have seen the demand improve over the same quarter last year, related to EV and also power semiconductor demand.

During the December quarter, we shipped several sets of battery assembly systems to several customers, including a leading EV company and a promising solid-state battery manufacturer.

We continue to anticipate additional recovery in the power semiconductor market over the coming quarters.

Similar to General Semiconductor for Ball Bonding, Auto and Industrial is our primary market for Wedge Bonding.

Also similar to Ball, our Wedge team is aggressively developing new systems to expand our dominantly heavy-wire wedge position into thinner wire, pin welding, and also chip-attach markets, which will help us better support the rapid evolution of emerging automotive and power-semiconductor needs.

This transition is being driven by growing global demand for more efficient power delivery and storage.

We are playing a critical role leading the transition from lower conductivity aluminum interconnects, which are the standard in power semi applications, to copper interconnects.

As we are currently demonstrating within the copper-to-copper TCB process and have also led the transition from gold to copper in the high-volume Ball Bonding market over 10 years ago, we have inherent competencies in copper bonding.

In wedge, the benefits of more expensive material like copper are significant as it supports more efficient charging, energy generation, and high-power applications such as AI and cloud computing.

We are excited to support customers through this potentially significant long-term wedge transition and will provide additional information on these emerging opportunities over the coming quarters.

Finally, within memory, we remain focused on driving vertical-wire adoption for emerging applications within both DRAM and NAND.

Separately, vertical wire continues to be another emerging memory solution for several global memory customers, who are either requesting information, developing their process, and/or beginning to produce samples to drive market adoption for future stacked DRAM applications.

As explained last quarter vertical-wire-based assembly, is positioned to support future high-volume, stacked-memory applications but also has significant potential to enable future high-volume, stacked-logic applications.

Similar to TCB, we have a significant technology leadership position with a growing base of engaged customers who are developing new vertical-wire packages.

Our process and development engagements have recently increased.

We are currently working with leading memory customers in Korea, the US, and China.

Over the coming year, vertical-wire connected memory applications are anticipated to move into higher-volume production; and longer-term, we expect this enabling technology to extend into higher-volume General Semiconductor markets.

As explained earlier, in addition to vertical wire, we are also supporting a major memory customer who is examining fluxless HBM alternatives in coordination with our highly capable FTC process development team.

The company continues to be at unique period times.

We are well positioned for several promising high-growth opportunities that are supporting long-term critical technology transitions in both leading-edge and high-volume semiconductor assembly.

While we have already experienced core market improvements, we continue to anticipate both Ball and Wedge will reach more normalized levels of demand within fiscal 2025.

Although we anticipate this broad and coordinated recovery to be imminent, we remain focused on what is within our control.

Our priorities are to maintain our aggressive cadence of development across all served markets while we continue driving customer acceptance for our new products and services.

I will now turn the call over to Lester for the financial update.

Thank you, Fusen.

My remarks today will refer to GAAP results unless noted.

As Fusen explained, we continue to anticipate a broader cyclical recovery for our Ball and Wedge businesses driven by coordinated improvement within General Semiconductor and Auto, Industrial end markets.

We remain very focused to support multiple development programs, product releases, timelines, and customer qualifications.

Looking back at our December quarter results, we generated $166.1 million of revenue and 52.4% gross margin.

This strong gross margin performance was partially related to revenue recognized in the December quarter for systems expensed in prior periods.

During the December quarter, we also recorded a gain of $71 million due to customer settlement associated with our fiscal second-quarter 2024 impairment charge.

Non-GAAP operating expenses, which excludes this item, were $68.6 million.

This was below prior expectations due to a favorable foreign exchange gain as well as an ongoing focus on operational and development efficiency.

During the December quarter, we booked GAAP tax expenses of $11.3 million, primarily related to the customer segment benefit but also related to our mix of profit and loss across entities during the quarter.

We continue to anticipate our effective tax rate will remain above 20% per quarter through fiscal 2025.

As announced on December 2, 2024, we also completed our previous repurchase program and began our new $300 million share repurchase program.

Repurchases for the first quarter represented activity from both programs and totaled $36.9 million, reducing shares outstanding by nearly 800,000 shares.

Turning to the outlook for the March quarter, we expect revenue of approximately $165 million, plus or minus $10 million, with gross margins of 47%.

Non-GAAP operating expenses are anticipated to be $70.5 million, plus or minus 2%, and we expect GAAP EPS of $0.03 per share and non-GAAP EPS of $0.19 per share.

As we await broader core market recovery, we remain very focused on key development, qualification, and market adoption across our growing portfolio of solutions.

We look forward to announcing additional success with these efforts over the coming quarters.

This concludes our prepared comments.

Operator, please open the call for questions.

(Operator Instructions) Krish Sankar, TD Cowen.

Hi, thanks for taking my question.

I actually had three of them.

Number one, Fusen, you kind of said that March quarter general semi should grow.

Is it fair to assume that sequentially, into June and September, your bonder revenue should grow sequentially?

Yes, okay.

So actually, just the past few weeks, [we actually revised] CY -- semiconductor FY25 growth from

[17% to 13%].

And we also see our Q2 actually due to Chinese New Year and also global political dynamics.

Some of our customers actually delayed their investment decision to up to Chinese New Year.

But despite the change, we still anticipate even 13% of semi growth in CY25 should benefit all of our business.

And downtime typically takes around six to seven quarters, and we are approaching 10 quarters.

So we believe we are in the last stage of the downtime.

So I wish I answered your question.

And we still anticipate the transition to be below normal -- are below normal to more normalized level of ball bonding later this year.

So I hope I answered your question, Krish.

Yes, that is helpful.

Just to follow up on that point, normalized level for fiscal '25, is that a $550 million run rate?

Or how do you think about what is normalized core demand?

Okay.

Okay.

So Krish, I will answer this way.

So typically, our second half is always stronger than first half.

And we expect the normalized level probably will reach later part. end of -- like our FY25.

So it's not unreasonable to expect our second half would be 20%, 30%, or even more than 30% higher than 1H, first half.

And I'll give you an example, 20% will lead to about 730, and 30% will lead to 760 just for the revenue.

So how I define the normalized?

I'll give you an example.

At the peak cycle, our ball bonder revenue is at the $1 billion.

But [average] of FY23, '24, actually, we only see just over $300 million.

So we believe our normal year, just ball bonder revenue itself, should be around $500 million to $600 million.

So that is really our expectation.

Industry price will enter a normalized level, maybe end of our fiscal '25.

And when we enter '26, we expect probably full year will be a normalized year.

Got it.

Got it.

Very helpful.

And then just a quick follow-up for Lester.

The December quarter margin strength, you said rev rec on systems that are shipped before.

Were this TCB, and how many tools were there?

No.

So Krish, these are tools that is associated with Project W. The customer ordered these tools prior to the cancellation of the project as part of the impairment, which we took in Q2 FY24.

The cost of these tools were included in the impairment.

Now that we have settled with the customer on Project W, these machines have been recognized in the quarter, which obviously helps strengthen the gross margin.

Got it.

Awesome.

Thanks, Lester.

Thanks, Fusen.

Craig Ellis, B. Riley Securities.

Thanks for taking the questions, guys, and congratulations on some of the progress in the business, like what you're seeing with high-bandwidth memory DRAM.

Fusen, I wanted to go back to where Krish started and just see if I could get a follow up.

So really like the potential for more normalized core business revenues and Ball and Wedge bonding.

But from all the observable data points, I think we and others see PC, smartphone, other high-volume demand in auto industrial is really very anemic.

And clearly, there are some good company-specific things going on in AMI right now.

But what are your customers telling you about their need for incremental capacity as they move into that second-half fiscal period for you and need to add capacity to meet demand, which seems more bouncing along the bottom than anything else.

Okay.

So I think I mentioned even in the Q2, we see customer investment decision actually pushed out after that decision will be delayed until Chinese New Year and also because of the concern about the global dynamic.

But the second half typically is really our stronger half, right?

So as I mentioned, I think a downturn already take about 10 quarters rather than traditionally six quarters.

So we do believe there will be also some investment.

I think a mature node -- mature node capacity will come up, particularly in China for 28-nanometer and above.

So we are actually, at this moment, is quite -- are still quite bullish about our second half.

Also, when we enter later part of second half and into '26, so we believe the run rate for ball bonder can reach to $500 million to $600 million.

$500 million actually is not high.

The peak level actually is $1 billion.

And at this moment, it's only $300 million, right?

So $500 million is just to go back to a pre-COVID level.

And the whole industry actually, semiconductor content continues to increase.

And the downturn takes very long, and we do believe it's a very late stage of a downturn, right?

So when we enter '26, I think we also -- not only ball bonder have a good potential.

We also see wedge bonder, we can take market shares from new products like pin weather and clip attach.

These are for high-power semi.

And we also see the momentum of vertical wire, our AP and also ADS.

So we shall give you the [tip for it] for the next couple of quarters.

That's really helpful color, Fusen.

Thank you.

The next question I had was a follow up on deck slide number 3.

Very intriguing point made with a growing pool of high-growth, AI-related opportunities for the company.

The question is, can you quantify what the value of those is now, either in the recently recorded first quarter what you would expect in the first quarter.

And as we go through '25 and '26, how big can those opportunities become for the business?

Your question, you were talking about our AP -- PCB?

Yes, and just the products that were referred to in deck slide number 3, lower right, when you talked about a number of AI-related opportunities.

Okay.

So I will give you -- so this is our forecast.

We haven't changed it, but we believe we still have an upside.

I think this year, when we define advanced packaging, our revenue for advanced packaging, total packaging, including our multi-die chiplet, SiP, vertical wire, and TCB, right?

So 2024, actually, the total AP is $220 million, and we believe the '25, we are aiming at $275 million to $300 million, right?

And we do believe the next couple of months, we might add a little bit more strong TCB forecast after long-term forecast we will receive from some other customers.

Got it.

That's helpful.

Thanks for the color on those items, Fusen.

I'll get back in the queue.

David Duley, Steelhead Securities.

Thanks for taking my questions.

I got a couple.

Could you just talk a little bit more about -- help us understand how many customers now are using your thermo-compression bonding tool.

You listed, I think, an OSAT, an IDM foundry.

Just I got -- I didn't -- I wasn't able to collect all the information.

So if you could just review that again and kind of let us know which ones will be the biggest growth drivers in the near term.

Okay.

Actually, we actually have multiple customers.

We have foundry.

We also have IDM.

I think we are in OSAT.

Actually, I don't have a total number.

I think it will be quite significant.

So I think, Dave -- this is Lester.

I think between customers who are using it in high-volume production, who's about to use it in volume production, plus qualifying, I think it would be close to 8 to 10 customers who are looking at it or who will soon receive shipments of our TCB bonders.

Okay.

And then I think last conference call, I think you highlighted the thermal compression bonding opportunity in 2025.

I think it was $40 million.

And that was part of this $220 million going to $275 million to $300 million, I think.

Could you maybe just break out the pieces of that $275 million to $300 million amongst your advanced packaging pieces?

Okay.

Actually, I don't think I have a breakdown over here.

This is including -- as I mentioned, this is like a multi-die chiplet, SiP, this free chip vertical wire and TCB.

But TCB alone, I think I mentioned in the last call, I think our '24 is $55 million.

And this quarter, I think we are aiming at $75 million to $100 million.

And the total TCB market, I think, this year for total TCB is about $300 million.

So we are probably -- this year, we are aiming probably about 30% of our total market shares.

And we do believe we can grow market shares from here.

Okay, thank you.

And then what, as far as the vertical wire solution, would you expect to see revenue from that in 2025?

And maybe help us understand if you do have revenue from how many customers.

Okay, so I think we are very excited.

Almost every memory customer, we are working with that, also recently including China, so also Korean and also a major US IDM.

We do believe this is very beneficial.

A lot of the first application is going to for the LP DDR5.

This is for the sale.

But in the future, we believe this vertical wire, not only DRAM.

We are also working with a NAND customer and founding actually a lot of new applications.

In the future, we will also go to logic, right?

So we are quite excited.

So in terms of customer, working with the customer, with the revenue, I would say, three at this moment.

This is the initial production.

We do believe '25 will be initial production.

Some of them just start to run their process.

And so including NAND -- including this, I think, we are working probably with seven or eight customers, right?

So we believe '25, we have a low revenue, maybe, I would say, below

[$20 million].

And we are looking for '26 to be bigger.

And '27, I think, will take off to be much, much bigger.

So '26 -- if you want me to pick a number, and I can give you a number, roughly, say, $50 million, '26.

And this year, probably below $20 million.

And hopefully, this will take off after that.

And we do have a strong belief this will be a game changer for the next couple of years.

Okay.

And final question for me has to do with the dual-head tool for the HBM market.

I think you've made mention in the past that you need to have a higher throughput tool in this market to win business.

And maybe help us understand what your throughput advantages are versus the competition.

Or what are some of the parameters or key metrics that will determine your success in winning business with an HBM customer?

Well, I think the HBM, let's take two prospects.

I think we are engaging for our next generation.

And so there are two important parts.

One is really the process.

We are working for the future.

And our goal is target to ship the system end of the year.

And so the process, I think, is very, very important.

We do believe we are able to create a copper-to-copper.

And you have so many there.

I think we will provide very good electrical performance.

And then I think it's going to be the throughput and productivity.

We do believe our tool with the twin head will have an advantage or slightly better than our current competition, I think, at this moment.

But we do believe the performance and reliability is what we are aiming at and with a bit of productivity compared to competitors.

Thank you.

Charles Shi, Needham & Company.

Good evening.

Fusan.

My first question is about the PCB.

This dual-head system versus single head, you already shipped and qualified.

I wonder, when this customer goes into volume production, do you expect the volume production tool to be single head or dual head, or it's going to be a mix?

And maybe a related part of this question is this shipment of the dual-head system sounds like it's an evaluation system; and if that's the case, when do you think you can get the qualification of this tool?

So Charles, you asked if this is an evaluation system.

No, this is not an evaluation system.

This is one of the PO we have.

So we do believe from one chamber to two chambers, each will go smoothly.

We have developed our engineering capacity.

The tool right now is used for pilot production and also for the new customer qualification.

After maybe in a couple of months, I would say maybe three months, around that, we hope to receive a long-term outlook, business outlook just from customers.

But do you think the volume production to be dual head or single head?

Yes, that's correct.

I think it's going to be dual head.

Will be dual head, okay.

Thanks.

Maybe the second question, also on plug-less TCB, I think you mentioned about the engagement with a leading memory customer, HBM.

Can you provide a little bit of details on what's the engagement right now that -- what will be the next milestones, et cetera?

Okay, so I think the first two years, we're actually very focused on logic, right?

So I can tell you, it's really for the next-generation HBM.

And we do believe we did a successful demo.

And our goal is really to ship the system by the end of the year.

Hopefully, if everything is successful, we can catch high-volume production already within 18 to 24 months.

That's really our goal.

Got it.

Okay.

So lastly, a question about the VFO.

Thanks for the color you provided to David earlier.

I want to ask you, the DRAM wire bounding, I think traditionally, it's not your market, right?

It's a Japanese competitor who had that market.

So for the top DRAM customers, What's your expected market share in VFO?

Because I would assume that the incumbent will also have some solutions.

And what would you believe, from where you stand today, the potential market share in VFO could be?

So Charles, I don't know why you make a comment about Japanese.

I think the total market shares, Just all ball bonders, I think we have close to 75% to 80%.

And so the NIM, I think we probably have 90-some percent market shares.

And for the DRAM, I think DRAM, in terms of a ball bonder -- okay, DRAM, I think at this moment, I would say maybe 60% is a free chip.

And then probably ball bonder -- traditional ball bonder, actually, I would say maybe 30%, 40%.

I think we still have a very, very high market share.

We are still number one, not only in the NAND, also in the DRAM.

So go back to the VFO.

I think the first application is going to be DDR5.

And this probably can reduce our total packaging form factor about 30%.

So that's why it's going to be useful, the

[cellphone].

And it's a kind of replaced TSV process with a better processing performance and also cost.

I hope I answered your questions.

Yes.

I mean, do you expect to have a VFO position at all three leading DRAM customers a few years down the road?

That's actually my question.

Okay, yes.

The answer is yes, we do expect that.

Okay, all right.

Thanks, Fusen.

Yes, thank you.

Thank you, Charles.

Tom Diffely, D.A. Davidson.

Okay, thank you.

I appreciate the chance to ask a couple of questions here.

Fusen, just maybe some questions about the general semi market or the core market.

What are the end markets where you're seeing the most excess capacity?

Or maybe said a different way, which of the end markets do you think will recover first from a ball bonder utilization rate point of view?

Well, I think, actually, the recovery in my mind right now is both Auto and also General Semiconductor is leading by China.

So like a front-end wafer equipment, I think it's always, in past couple years, always at the quite high level.

And we need to be ready for the next couple of years because of advanced packaging investment and also China.

China is expanding a lot of legacy fab capacity.

And after they finish our front-end facility, I think the back-end investment will be needed.

So the investment of back end and the front end, not necessarily going very, very well.

But I can -- actually, we can tell you, I think, at this moment, what are really leading General Semi and also Auto, including ball bonder and also wedge bonder, actually is happening in China.

Okay.

And maybe just to dig in on one piece of the market, so a few of the other players in the semi cap world have talked about some NAND strength recently.

It sounds like it's more just kind of on the technology side.

Curious if you think you're going to get a boost from that on your NAND business?

Or do you have to wait for true capacity buys, excess capacity, or incremental capacity to come back into the marketplace before you see a pickup in that business?

Okay, so what I can tell you is I think NAND pretty much is really is a ball bonder.

And for the NAND, actually, we have, if not above, close to 90% of market shares for the NAND.

Yes.

So wondering about the health of that and if you're going to see an uptick in that over the next couple of quarters, like guys like Lamar on the equipment side, or if it's going to be kind of a delayed recovery for you versus those peers.

Sometimes I think the investments are probably not in a well-coordinated line.

So I actually didn't know which customer you referred to.

For example, the [edge company], the NAND, there are many more layers, probably [edge development] and the capacity is more important.

And for us, then you finish your product and then it's a stack die.

Actually, we can connect a stack die.

So what I can tell you is when this finish building in the front end, I think we will benefit because NAND packaging is quite simple.

Yes, okay.

Thanks for that.

And then maybe just a quick follow up for Lester.

Lester, when you see the core markets come back in 2026 and kind of get back to that normalized level, what does that do to the margin structure, if anything?

I assume that those traditional core markets might be a little lower margin than some of your newer stuff.

So I was curious if there's anything we need to think about from a gross margin point of view.

Well, Tom, from a gross margin standpoint, we're still aiming towards 50%, right?

And we think as we head to '26, Even our traditional core products like Ball and Wedge, we are introducing new products which gives us higher margins for both Ball and Wedge.

And also, I think we continue to focus on cost reduction, again, both in ball bonder and wedge bonder.

So we think actually the margins will continue to improve as we roll into '26.

And also, obviously, with higher volume, our margins does get better because our factory utilization goes up.

Great.

Okay, thanks.

Appreciate it.

Craig Ellis, B. Reiley.

Yes, thanks for sneaking in the follow-up question.

Fusen, I wanted to go back to some commentary around Charles' question around when we get the volume shipment intercept with high-bandwidth memory on fluxless TCB.

I think you said at the end of this year, that's possible.

And so the question is, is the enabling development the industry's transition to HBM4?

Is it as industry goes to higher stacked and likely more volume 16 stack at that time or something else that's creating that window of opportunity for you?

Thank you.

Yes, next generation right now is 3E.

So I just want to make sure we have a right expectation.

Before us trying to penetrate through a DRAM market, there's already a machine provider from Korea and also a lot of other customers.

So this one, I think when we say we will ship a system, this will be actually after customers have confidence, after the demonstration is good.

We actually intend to ship a system maybe later part of the year.

And we still need to go through a lot of development, right?

So to really win the high-volume production, I think probably successful, qualified -- probably this already big milestone, we are targeting about 18 months.

Okay.

Got it.

So commercial revenue really would be sometime in calendar '26 rather than something that would be exiting '25.

That is quite correct, yes.

Thank you.

We've reached the end of our question-and-answer session.

I'd like to turn the floor back over for any further or closing comments.

Thank you, Kevin, and thank you all for joining today's call.

Over the coming quarter, we'll be presenting at several conferences and roadshows.

As always, please feel free to follow up directly with any additional questions.

This concludes today's call.

Have a great day, everyone.

Thank you.

That does conclude today's teleconference and webcast.

You may disconnect your line at this time and have a wonderful day.

We thank you for your participation today.