Camtek Ltd

NASDAQ:CAMT

Ladies and gentlemen, thank you for standing by. I would like to welcome all of you to Camtek's results Zoom webinar. My name is Kenny Green. I'm part of the Investor Relations team at Camtek. All participants other than the presenters are company muted -- following the formal presentation, I'll provide some instructions for participating in the live question-and-answer session. I would like to remind everyone that this conference call is being recorded, and the recording will be available on Camtek's website from tomorrow.

You should have all received by now the company's press release. If not, please view it on the company's website. With me on the call today, we have Mr. Rafi Amit, Camtek's CEO; Mr. Moshe Eisenberg, Camtek's CFO; and Mr. Ramy Langer, Camtek's COO.

Rafi will open by providing an overview of Camtek's results and discuss recent market trends. Moshe will then summarize the financial results of the quarter. Following that, Rafi Moshe and Ramy will be available to take your questions.

Before we begin, I'd like to remind you that the statements made by management on this call will contain forward-looking statements with the meaning of the Federal Securities laws. Those statements are subject to a range of changes, risks and uncertainties that can cause actual results to vary materially -- for more information regarding the risk factors that may impact Camtek's results, I would encourage you to review our earnings release and our SEC filings and specifically the forward-looking statements and risk factors identified in the 2024 annual results and such other factors discussed in our annual report on Form 20F as published on March 21, 2024.

Camtek does not undertake the obligation to update these forward-looking statements in light of new information or future events. Today's discussion of the financial results will be presented on a non-GAAP financial basis, unless otherwise stated. As a reminder, our detailed reconciliation between GAAP and non-GAAP results can be found in today's earnings release.

And with that, I would now like to hand the call over to Mr. Rafi Amit, Camtek's CEO. Rafi, please go ahead.

Okay. Thanks, Kenny. Hello, everyone. Camtek ended the fourth quarter and the old year with record results. Q4 revenue were $117 million, an increase of 32% year-over-year. The operating income was $36.3 million, 42% improvement year-over-year. Regarding the all years results, Revenue were $429 million, an increase of 36% year-over-year and operating income came at $130 million, 56% better than last year.

The distribution of revenue for the whole year was around 50% for HBC and 20% for other advanced packaging applications. The rest was divided between CIS, compound semi, front end and general 2D applications. We are starting 2025 with strong business momentum and expect 2025 to be a year of growth, well beyond the 5% WFE growth estimates.

The main growth driver is expected to be high-performance computing, or HPC, in which we are a key provider. The demand in HPC segment remains healthy. And overall, we expect the contribution of HPC to our business to be at least 50% in the first half of 2025.

Just this week, we received orders totaling over $10 million for HPC-related product which reinforce our assessment of continued growth in the demand for this segment. We identified a shift in the market regarding manufacturing and packaging of HPC modules, where these models are gradually start to be manufactured not only by IBM and foundries, but also by OSATs. This change in trend is a positive opportunity for us as we have a strong position with OSAT in the different regions.

From orders we have on hand, our pipeline and from discussion with customers, our revenue guidance for the first quarter of 2025 is between $118 million to $120 million, close to a 25% increase over Q1 last year. We also estimate that the demand for our systems for HPC-related products will continue into 2025 and beyond.

In our last call, we discussed the 2 new models that we have introduced to the market, the Eagle, G5. The Eagle G5 is currently in operation at multiple customers' production lines, delivering excellent performance that align with both our expectations and those of our customers. Meanwhile, the [ OAK ] has successfully completed qualifications with several customers, and we anticipate receiving more orders for it in all in the near future.

We plan to officially launch the ARC in Semicon Korea show next week. A few words about the OAK and it's in from the Eagle. The OAK is a new cutting-edge platform engineered for high-end applications such as detecting 100-nanometer defects, measuring several hundred of 1 million micro bump at pitch lower than 10 microns, all while performing high to -- these advance requirements are beyond the scope of the Eagle platform.

Although the Eagle models, especially the Eagle G5 provide excellent performances in a variety of applications. The OAK system has high capabilities but also a high price tag. So there is a room for both models. Going back to our opportunity in the HPC segment. A major part of our business support HPC-related hardware. The growing demand for HPC hardware has placed Camtek in a leading position and allowed us to grow significantly in 2024 much beyond the market growth rate.

The use of AI capabilities in large organization is causing rapid growth in demand for HPC and we are seeing significant investment in server farms around the world in recent years and expect this trend to continue. Looking into the future, we expect to see AI capabilities penetrate edge computing devices such as automotive, robotics, PC and eventually mobile phone which require the development of powerful dedicated hardware to be implemented in these edge devices. When this happens, it will be a great opportunity for us.

In summary, our primary growth engine for the upcoming years will be advanced packaging, particularly in high-performance computing, HPC. We are strongly positioned in this sector. And with the introduction of our 2 new advanced systems, we will reinforce our leadership position.

And now Moshe will review the financial results. Moshe?

Thank you, Rafi. In my financial summary ahead, I will provide the results on a non-GAAP basis. Reconciliation between the GAAP results and the non-GAAP results appear in the tables at the end of the press release issued earlier today. .

Fourth quarter revenues came in ahead of our guidance, a record $117.3 million, an increase of 32% compared with the fourth quarter of 2023 an increase of 4% from last quarter. This is the fifth consecutive record quarter in revenues. For the year, we ended 2024 at $429 million versus $315 million last year, which represents 36% increase year-over-year. The geographic revenue split for the quarter was as follows: 92% Asia and 8% the rest of the world.

Gross profit for the quarter was $59.3 million. The gross margin for the quarter was 50.6%, an improvement from the 49.2% reported in the fourth quarter of last year, and similar to the third quarter of this year. Operating expenses in the quarter were $23.1 million compared to $18.2 million in the fourth quarter of last year, and similar to the $22.9 million in the previous quarter.

Operating profit in the quarter was $36.3 million compared to the $25.5 million reported in the fourth quarter of last year and $34.2 million in the third quarter. The increase is mostly due to the increase in revenue levels and accordingly, in the gross profit.

Operating margin was 30.9% compared to 28.7% and 30.4%, respectively. Financial income for the quarter was $6.2 million, an increase from the $5.7 million reported last year and similar to the $6.4 million in the previous quarter.

Net income for the fourth quarter of 2024 was $37.7 million or $0.77 per diluted share. This is compared to a net income of $28.2 million or $0.57 per share in the fourth quarter of last year. Total diluted number of shares as of the end of Q4 was $49.5 million for the year in a whole, we recorded net income of $139 million a 45% improvement over last year.

Turning to some high-level balance sheet and cash flow metrics. Cash and cash equivalents, including short- and long-term deposits and marketable securities as well December 31, 2024, were $501.2 million. This compared with $488.7 million at the end of the third quarter. We generated $16.2 million in cash from operations in the quarter.

And for the whole year, we generated $122 million. Inventory levels increased to $123.1 million from $116.3 million. The increase over previous quarter is to support the anticipated sales growth in the coming quarters. Accounts receivables increased to $99.6 million from $71 million in the previous quarter, mostly due to the timing of collections. We have already recorded strong collections since the beginning of 2025, and we expect accounts receivable level at the end of Q1 to be in line with revenue. DSO of the end of Q4 was 77 days, down from 90 days a year ago. With respect to guidance, as Rafi said before, we expect revenue of between $118 million to $120 million in the first quarter and that we look forward to a year of growth in 2025.

And with that, Rafi, Ramy and I will be open to take your questions, Kenny?

Thank you, Moshe. [Operator Instructions] So as we have a lot of people on the call, we will wait a few moments to pull for your questions. So our first question is going to be from Charles Shi of Needham. Charles, you may go ahead and I meet it ourself. .

Yes, thanks. The congrats on the good results and the solid guide. I think my first question is about the guidance on the HPC contribution in the first half, 15, I think you guys mentioned about 50% -- but from the commentary, it sounds like you're seeing HPC demand probably broadening more to OSAT where you have the historical strength I recall like last year, I think that you guys were providing a little bit more details into the HPC part of that is HBM. Part of that is chiplet. So assuming OSAT is probably taking up more of the chiplet related demand from the leading foundry is the mix between HBM versus chiplet kind of shifting mortgage tile, at least through the first half of 25, I know you guys don't give that details, but directionally, can you comment on that? .

Charles, this is Ramy. So in general, yes, we continue, let me try and answer it -- I'll give you 2 views about the HPC market. So I think what differs us for probably some of the competition is that we have a strong position in all the big manufacturers. And as a result, we are less and we are performing several inspection and metrology steps at each and continually winning these steps. So this makes us a little bit less sensitive to a specific customer. And as you mentioned, the OSATs are starting to open up, and this is definitely -- this is reflected in our forecast for the first half. Regarding the ABM. So yes, we do not give the exact numbers. But what I can tell you that we continue to see demand and we have significant orders on hand for the first half of 2025 that we plan to deliver.

Any early view on the second half of this year at this point. .

It's very typical, obviously. We have a lot of discussions we are starting to see a backlog building up through the second half. We, I would say, the discussions with our customers are positive. But still, -- it's too early to give something more solid from a guidance point of view. .

Got it. Got it. So maybe my last question, what's the China revenue contribution last year? I recall you said is somewhere around 30% to 35%. Is that number still the case based on the actual results? And any expectation for China revenue this year is the percentage going up or going down or going to be flattish versus 24 .

So Charles, this is Moshe speaking. With respect to the geographical split for the year, Indeed, China went down a bit from 2023. In 2023, we reported 47% from China. This year in 2024, we expect China to contribute around 30%. It's a bit early to say about 25%, but we see demand coming also from China. So we expect somewhat in the range of anywhere between 30% and 35% coming from China in '25.

Yes. I would like -- if you don't mind, I would like to go back to previous question and try to expand a little bit the view of HPC in general. As you know, the HPC includes 3 major components. We talk about the HBM, we talk about all the GPU. We talk about the interposers. There are more, but they are the major 3 components. Now if 1 of them suffer of any issue of capacity or missing capacity, definitely, it could affect the all other -- because if you cannot put your component on the interposer, nobody will make any extra HBM or CPU, if there are no room to put them on the interposer. .

So you have to look always not just to the final demand of the end user, the server farm, but also in the capacity in their production capacity. And TSMC made announcement, they said that they don't have enough capacity. They build extra. And hopefully, until mid of this year, they will make catch-up. So you asked about the second half. I would say that this is 1 of the major issue. If TSMC can make the catch-up and expand its capacity and production capacity, definitely, it may open the other player to provide the components or all the module could be built. So this is something that probably you will see very soon.

Thanks, Rafi. I appreciate the insight. .

Thanks, Charles. Our next question is going to be from Tom O'Malley of Barclays.

This is Kyle on for Tom O'Malley. So I wanted to start off asking about the move to hybrid bonding. You guys have talked about in the past about how your inspection steps actually goes up versus current generations. So I was kind of wondering like if you could refresh us like what your expectation is on the timing for hybrid bonding and what like the delta is at the number of steps you could address there versus in current solutions now.

This is Ramy. So definitely, hybrid bonding is starting to make the first steps in the market. I think in our segment, we are starting to see initial production. We have already installed machines at several sites that are being used as we speak for several steps in the hybrid bonding. The OAK is definitely going to play a major role in this segment. Although we also have machines from the Eagle family as well there. Definitely, we are seeing activities there. But I think when we think about hybrid bonding and the contribution to production high volume production, it's still a couple of years away, I would say, starting '27, '28, we will start to see volume production there. What else I can mention that definitely, there are going to be some metrology steps where we will take part of it. So all in all, as I said in previous calls, we view the hybrid bonding as an additional opportunity in our market space.

And I would like -- I have 1 more comment on that. As far as we understand from customers who develop this process, it suffer a very low yield because it's very complicated and very high risk process. So as far as we understand, the major application for this go for the very high-end application. It means that the hybrid bonding is not going to replace all the connection or the current connection method that we have today.

Probably it's focused on very, very high-end applications and definitely cannot replace all the current application. That's our, I would say, observation at this point.

All right. That's very helpful. And then for my follow-up, I wanted to ask on the HCM market. So of like the 3 or the 3 big customers, I know you can't talk like by customer specifically, but 1 of them ends up having a slowdown in spend or how trouble getting qualed on leading production. Do you expect like other customers to be able to pick up any sort of gap from 1 of the big players? Or I'm kind of just trying to figure out like -- I know you talked about the first half visibility being very strong for HBM, just what it could look like in the long term across those customers. .

So this is a commodity product at the end of the day. So definitely, if there is a slowdown in 1 customer. The other customer will pick it up. I believe there are some that are doing better than others. But as we said, we have a very strong position in all 3. So from our point of view, we expect to continue and play a major role in this specific market.

I want to reiterate, we will ship machines in the first quarter, and I think also, of course, beyond. Rafi made a very important statement just a few minutes ago that really this high-performance computing is spilt out of a few building blocks. It's the GPOs, it's the HPM, it's the interposer and other things. So basically, as we understand, the TSMC are going to double the capacity next year, this is going to free additional capacity to other players to enter this market and there are more places that the OSAT.

So definitely, we see this market developing, growing at very high rate, the growth rate of the cohorts and cars like technology, the number of packages, the growth, the CAGR is in the range of 50% over the next few years. So definitely, this is a market that is going to draw a lot of HBMs to it, and we feel very comfortable with the long-term future of this market.

Our next question is going to be from Matt Prisco of Cantor. Matt, .

I think it would be great if we could talk about the product traction across Eagle G5 and Hawk, -- maybe an update on orders and early customer reception? And what type of market opportunities do you view these opening for Camtek. Is this kind of going to expand your current SAM? Or is this more to address future interest rate .

So first of all, there are 2 aspects here. So first of all, there is 1 aspect of being more competitive. So the Eagle G5 or fifth generation that is a much faster, more accurate brings in more capabilities will make our position more competitive and we expect to take for a market. This is on the G5. The Hawk will open new markets for us. markets that today are beyond the capabilities of the Eagle product line. And there are quite a few. I'll give you an example. It's the high number of micro bumps on wafers or logic devices. You are talking about 500 million devices at a single wafer.

That's an application that is starting to develop that is beyond the capability of the current product line. So definitely, the Hawk will increase our total available market and it's a couple of, I would say, at least $200 million and potentially more. It will free up some front-end applications and so forth.

I think I won't be able to give you more details in this call. But definitely, it will increase our available market. It will take us to places that our current projects today just cannot perform. Although, we have an excellent product line for our existing market. About the orders, we made an announcement last November that we received orders for over $50 million. We also said in that announcement that we expect to get additional orders and in this, this is the case, and we plan to ship these machines this year or the G5, we already received a comparatively large number of orders that came, we already started to ship them. actually in the fourth quarter of last year, we shipped the initial machines are already in production, and we are starting to ramp the G5 into production. The Hawk is also finished already qualification. And so we are in a very good shape there as well.

Okay. I would like to add again 1 more comment because you asked if this new product is for the near future, maybe longer. I would say that the Hawk in general has 2 major capability. Number 1 is this capability for end application. Probably we will see more and more by the end of the year next year because while R&D running production and it's moved to production, it takes some time. .

But the other aspect of the Hawk is very, very high throughput, even in current application. So if customer -- the throughput is very important for him and is willing to pay the price tag. Definitely, it may choose the Hawk. So the oak definitely can serve customer that the main issue is higher throughput. But when we designed the Hawk, it was designed mainly for high-end applications. So it can answer for both of them.

Helpful. And then for my follow-up, as the industry transitions to technology such as HPM and Camtek, -- how does that impact Camtek opportunity or maybe process control intensity in general? .

So in general, it's a positive, I would say, it's a positive trend -- but what you call process control, the intensity will probably grow as you're putting more and more dies in the same module, so you would like to make sure that none of the prices any defect. So I -- we -- in our view and our discussions with customers, the have -- this will have a positive effect on the quality control. .

Yes. But the main application is 100% inspection and metrology. Our customers use our machines to inspect and measure each wafer. Now when we go to metrology, in metrology, they do maybe more sampling. But in inspection, most of them do 100%.. Matt, that answers all your questions. .

That is perfect.

Okay. Thanks. Our next question is coming from Brian Chin of Stifel. .

Maybe just for the stick on the HPM 4 topic for a moment here. How large of an increase are you expecting on the micro bumps per die per wafer for HBM 4 versus 3? And do you expect your system shipments to be stronger really next year versus this year? And is this -- are there any applications there that are better suited for Hawk relative to Eagle? .

So let's talk about the HBM. Those application range today for you take a wafer, it's over $100 million. And it's kind of stay in this range, at least for the foreseeable future. It's not going to go to $300 million. It is not in the near future. So from that point of view, definitely, I would say that the workout of the industry is the Eagle product line. And I expect that in most cases, it will continue.

And I think as Rafi alluded to, there will be some customers that are very, very sensitive to the space and may prefer to take the Hawk in order to get a double throughput for the same space of the machine. So -- this is a potential. This is something that we will see over time. But from the capability of the Eagle to respond and definitely the G5. It is a very good machine that can probably meet all the requirements of the HP. Now as we said in previous calls, definitely -- and today, we are doing, in most cases, 100% inspection and the metrology 3D metrology, and there are additional metrology steps that are done in different areas. So definitely, the intensity of the quality control is going to stay or improve in certain areas that don't do 100% inspection and metrology in as we go into the HB M4 and beyond, definitely, the opportunity for Camtek is going to grow, but you need to couple it with the larger number of cohorts and covers like packages that are going to grow by a CAGR of about 50% in the foreseeable future. So the combination of the market expansion together with the complexity of the die and the wafers and the need for higher intensity of quality control is definitely this is the ingredient for better business for Camtek .

Okay. Great. And relative to the cohos opportunity. I think there's a fair bit of transparency about maybe what the size of that could be in terms of expansion this year. Can you help us size maybe the planned capacity footprint for some of these OSATs who kind of work in conjunction with the lead customer there, a sense of sort of how much capacity could they be planning to build out this year, next year to kind of help to alleviate the overall industry constraint and also your position -- reinforce your positioning in that OSAT?.

It's very hard now to give numbers. What I can tell you from what we see -- and what we see is that we sell, I would say, I would say, a significant number of machines go to these ones. Now the point with the OSAT that you don't really know all the applications that they are doing. But definitely, we are seeing all such buying machines from us for these applications. .

So yes, the market is broadening. Now you need to take in account that are all kind of fabless companies that you know are just making the entire design in sales some of the silicon, they be designed by the cell, the cheaper they design by themselves and then they buy the -- or by the and have the OSAT to do the full integration of the full manufacturing of the module for them. So that's something that is very hard today to track down the capacity, but I think it's going to be a significant capacity in the next couple of years. .

Great. Maybe just 1 last question to maybe frame the market. There's been some discussion or uncertainty whether the HBM TAM for inspection metrology will increase this year versus last year. If you're not comfortable maybe providing that update. Overall, do you -- it sounds like you expect the overall TAM for AI packaging or HPC packaging, inclusive of cohos, et cetera, to increase this year versus last year? .

We definitely expect it to increase, and you can see the capacity. I mean TSMC are going to double the capacity in there are going to be a number of homes that are going to jump into this area, but actually already manufacturing some of them. So this market is going to grow. One of the building blocks create HBM and the more complex application becoming they're going to become more and more complex you're going to see more and more memory that is required. We've seen from history that whatever memory you provide for your products and the requirement is higher. .

So this is a trend that we have seen before. I think it will continue. And definitely, we adjusted the beginning of the AI era and see how fast is catching up a lot of organizations are starting to adapt these capabilities. So I think whatever the numbers are, the numbers seem to be solid.

And as you can hear here, that there is not enough capacity in the market to support the requirements of all the companies that are going to go into it. So our view that the market is going to grow, there is going to be requirements for more HBM, the available market or the market for these products is going to grow and we believe that we will grow with it in the foreseeable future.

Our next question will come from Vivek Arya from Bank of America.

This is Michael Mani on for Vivek. To start, could you give us a sense for between Hawk and Eagle which 1 will be a bigger contributor to growth this year? And then maybe a little further out, as we think about the competitive landscape, especially as it pertains to what opportunities -- how can present to you guys -- it seems like these tools should help you gain traction in the 2D market where there's already a strong competitor who participates here.

But based on your previous experience of product cycles and how your customers make allocation decisions -- what sort of time line do you expect for any share gains in this market? Does it -- could it happen more quickly just because of how rigorous the new technology apartments are for some of these upcoming inflections like HBM 4? Or do you expect these share gains to unfold more gradually over time?

So you can see -- there is a limited information that I can provide here. But I think that from the order size that we receive from the Hawk and you understand that we are getting more orders -- so you can understand that it's going to be a substantial number for the Hawk. And I can tell you that it's going for the G5 to be also a substantial number. However, the Hawk ASP is higher -- so it's here, I would say, the comparison is not on one.

But in general, I believe that we will see a comparatively high ramp into production in getting these 2 products. And the reason for that, that they provide a better cost of ownership to our customers, they understand it. That's on 1 side. On the other side, on the Hawk side, it open markets and applications that today we don't have and are needed by our customers, that will be the second reason of buying these products.

And for the G5 definitely, the Eagle is a very, very popular regime. I would say, it's the workhorse of the industry in many of the metrology, but also in many of our -- there are many 2D applications. In fact, we sell more machines eventually the 2D than 3D. So we are very strong. Yes, there is competition in this market and an excellent competition, but we are doing pretty well and certain applications we dominate and we expect to have even a stronger position in the 2D market as a result of the introduction of the Hawk and the G5. So yes, in fact, -- it's going to be a pretty, I would say, steep ramp into production.

We over the over the first year. And to many of our customers, and we have thousands of Eagle machines from previous generations. Some of those customers will not want to change anything other production lines, they are very satisfied with what they have. We'll still continue to buy the regular Eagle, which is a very good machine. And that's how this industry behaves. Certain customers are more conservative, of sensitive to making changes. Others are willing to make those changes faster. So this is something that we will see as we move along.

Great. And just a question on the other part of your business, compound semi, CMOS image sensors. I think last quarter, you indicated that there were some green shoots unfolding there. Just did those kind of persist into this quarter? What are your kind of expectations for that market this year? And if you do have a constructive view, what's really driving the growth there just because as we look at all the various end markets that it's exposed to, right, and demand seems to be pretty weak right now. So just any more color there would be appreciated. .

So I think when we look at the, I would say, the non-HPC market. So definitely, I would say the business is stable there. I think most of the applications are stable. There isn't now an application that is growing very fast. And the reason for that is the consumer goods in general are not growing. There isn't an application there that is growing very fast.

We do see some improvement on the CMOS market. Apparently, there are new sensors -- and so there is, I would say, some demand coming from the CMOS market, the CMOS image sensors market, if I want to compare it to last year. I think when you look at the silicon carbide market, it's stable. The reason, I would say, huge growth there. There's a lot of capacity out there that I think will take some time to digest it. And we'll continue to see some business, but I don't think there will be a major growth. I would say that the non-advanced packaging, which is about 30% of our business will grow next year very much in line of what the industry expectations are. But most, I would say, the growth is going to come from the high-performance computing and other advanced packaging, I would say, the more traditional advanced packaging application such as fan-out and others that continue to grow. Michael, if that answers all your questions. .

Our next question will be from Craig Ellis Craig or Stacy, B. Riley please go ahead.

This is Stacy asking for Craig. And I was wondering if you can discuss a little bit about the digestion risk and also the magnitude of it in HBM and/or cost -- and if it's related to foundry or OSAT or kind of both?

Well, let me Stacy, this is Ramy. Let me try and answer and maybe Rafi and Moshe will jump in. In general, when we look at the HPC market, -- we see a very solid business going into next year. We have, as we said in the prepared notes, we see solid of our business is going to go into this market. The overall specifically regarding the HVM yes, we are going to see to the HBM segment significant orders that we already have on hand for the first half of '25, and we will deliver and install those machines. So all in all, we don't see a digestion -- on the contrary, understanding what TSMC is going to double the capacity with all the increased capacities in the different on sets, and we are actually shipping machines for to OSATs and we will ship, including in the first half of this year.

So we are seeing the market expanding. The business is healthy. There is a lot of interest. There might be some for other people, at least we don't experience it. There could be some customer-related concerns. But when we look at the entire industry as we serve all the customers we feel very good about this market as we enter 2025. Rafi, you want to add something? .

No. I think you'll covered it very well.

Yes. And if I can add a follow-up is can you also talk about the -- maybe the backlog levels through 4Q and 1Q to date? -- and the composition of those by end usage and, if possible, some degree of visibility maybe throughout the year 2025.

Okay. So -- what we can comment on the backlog in general is that we have a very strong backlog for the first half of the year. The second half is starting to build up it's very -- and for the first half, let me just complete that. We said that 50% will go to HBC. For the second half is starting to build up it is much too early to say to talk about the composition of the backlog or to give a feeling on the actual numbers in the second half of this year, and this is very typical to our business, is the lead times are 3 to 6 months. So still, we're starting to build up. I think we will be in a position to understand the third quarter in our next call, okay?

But I think at this stage, I think we feel that next year -- this coming year, the '25 will be a growth year. We feel very well about the business, but still the second half to really try and give more color on it. It's a little bit too early.

Thanks, Stacy. Our next question is going to be from Patty [ Scott ] from Evercore.

The first 1 I have is how should we think about your revenues versus the core capacity that's being added. So if I go to as CapEx on advanced packaging, it's doubling from $3 billion to $6 billion. Can that -- does that mean your revenues from cohorts double as well as we go into 2021? Is that a fair interpretation?

Okay. So let me -- let's talk about how the business builds. Definitely, the capacity that was installed is now serving the market. So part of the growth in '25 is built on the capacity that was built -- that was already built in '24. And this will go on. So the capacity that we are going to install in the first half, I assume will still be installed and serve the growth in the second half of '25.

What will be shipped in most of the second half will probably affect the tail end of this year and really contribute in '26. And this is how this industry works. It's true for the collars capacity. It's true for the HBM, -- it's true for most of the building blocks. So we, I think, in my mind, should look at the second half of this year, what will be shipped out, a lot of it will be served or the first half of 2016. And I think the additional capacity of what TSMC are adding in the OSAT. Some of it will still end up in this year, again, will go to '26.

Understood. And then for my follow-up, as you ramp Eagle G5 and how products, how should we think about gross margins from here? So .

I think as we are ramping these products, they are definitely accretive from day 1. And specifically about the Hawk, I think we discussed it in previous calls, we definitely expect the Hawk to have a positive growth impact on our gross margin in 2016. .

So maybe just to clarify with Rafi. Overall, the Gen 5 has a positive impact on the gross margin, and they are accretive to the gross margin. However, as we are ramping the production this year, most of the impact to the gross margin level, you will see it all in 2026. This year, we are kind of ramping the production, streamlining the process -- and next year, you will see the benefit and the contribution to the gross margin.

Our next question is going to be from Gus Richard of Northland.

Thanks for squeezing me in here. You walk through the AI ecosystem. A couple of major players that are doing their own chips, have moved to chiplets, this generation. The hyperscalers, the revenue was constrained by their capacity and their data centers. And when you look at it, it looks like the new process COAs is what's sort of limiting the output of chips and servers, et cetera.

And so my question is, what's slowing the ramp? Is it facilities? Is it somebody else's equipment? Is it OSATs coming up? Can you kind of help us understand what's going on there?

So for us, it's very hard to know who is the limiting factor. But at least from our view, things are starting to pick up. We see there's a lot of additional capacity. So I think they will catch up pretty quickly. We expect them to catch up in these areas. In the first half of this year, I believe they will catch up with most of the capacity, and we will start to see, hopefully, the ramp for these guys in the second half of '25, as Rafi alluded to in his comments or early in '26, but definitely, all of these guys are going to add more and more. We need more and more capacity that will definitely increase the potential opportunity for us.

And I would like to add 1 more comment on that. You have to consider that also the core in second generation, try to improve the technology they realize that the size of the chiplet is too big. So they have to think about maybe some using more combination of organic and silicon. So there are also a lot of R&D involved in this product because they want to be more efficient and getting better yield an answer to the density and other capability. So it's not easy for us to understand how long it takes to make this process to be mature process with high yield and move to high volume. .

But definitely, we can see that the investment, we can see the new building and the construction. And it's amazing if you just go and walk in some places, you see the amount of building, well, it's a big wow to see this investment. So definitely, it will be converted to production and high volume. If it takes 6 months or 3 months or 9 months, we really cannot estimate right now.

Okay. And then just looking beyond AI and HPC, there's some evidence that maybe in 2026, some of the mobile guys might move to chiplet rather than move to 2-nanometer. I'm wondering if that's another driver of growth for you into '26 or it's not something you have visibility into? .

We don't have a visibility, but I think any move to chiplets is very positive for us. .

Got it. And then just the last 1 for me. If I was your most favored customer needed a system as soon as possible. Where would you slap me in at this point? .

Well, gas, it's a different question because you are close to us. But so no, I would say it depends really on the configuration and the specific models. So if you go from a regular in, we will do it sometimes even in weeks. If you come to the G5, it will probably be 2 months if somebody really needs a very content machine. .

On the Hawk, it will take longer. And it really is the complexity of the machine, the configuration. There are a lot of aspects are coming to -- so -- but yes, we will always find a slot for 1 of the important customers that need a machine in ASP will turn the world around to make it happen. But obviously, this is very small number of machines. When you go to the large orders that 3 to 6 months in year time. .

Got it. Perfect. That was very helpful. .

Thanks, Gus. Our next question is from Shahar Cohen of Lucid Capital.

I want to ask about the HBM or the mini HBM opportunity in mobile. We have seen what is called LFW DRAM -- we have seen both Hynix and Samsung adopting this mini HBM and we heard maybe some of the high-end phones may include this kind of stuff in 2026, anticipating the phone, which will have some requirement of better bandwidth between the memory and the CPU. So -- can you speak about this opportunity for you guys? And how should -- is it more like end of 2025 or 2026 and just the significance of that opportunity.

I think in specific DRAM, it's still more under production. It will take some time until it will turned into, I would say, a real opportunity for us. From our point of view, it's not -- there is nothing here that we need, especially to develop for it.

This is a kind of device -- it will have the micro bumps on it. It's a really typical, I would say, advanced packaging. Yes, it's a challenging one, but it's definitely with the inter capabilities of our machines. But in general, I think the -- if you ask me about mobile phones, I think this is more '26 and capabilities beyond -- and still, we will see this kind of an application ramping up. I don't think it is -- at least in my mind, it's not a big.

And if it will be 2026, it's going to be significant.

It depends on the number of cell phones, they will adopt it. And so I think it is too early to name -- it's still early in the game. But look, in general, I think we discussed it in our prepared notes. Today, the application is very, very centered around servers. And therefore, it's a lot of hardware in it, and you can see the size of the business, it is very significant.

As AI will go into cars, we're going to robotics, and these are probably the next 2 things that we will see these technologies and then go to PCs that it's really a factor more and eventually go to cell phones. So the more applications for us it would create a very big opportunity. So long term, yes, all of these things will adopt these technologies 1 way or another. And we definitely believe and see that this is an area the HBM as part of the colas as part of the model AI model is going to be a very, very big business for us. Thank you. .

That ends our question-and-answer session. Before I hand back to Rafi for his closing statement, I would like to let you all know that in the coming hours, we will upload the recording of this conference call to the Investor Relations section of Camtek website at camtek.com. I would like to thank everybody for joining this call, and we will see you next quarter, and I would now like to hand back to Rafi for your closing statement. Rafi, please go ahead. .

Okay. I want to express my gratitude to all of you for your ongoing interest in our business, especially things goes to our employees and management team for their outstanding performances. To our investors, I appreciate your long-term support. I look forward to our next conversation in the upcoming quarter. Thank you, and goodbye.