ACLS - NASDAQ

Good day, ladies and gentlemen, and welcome to the Axcelis Technologies call to discuss the company results for the fourth quarter and full year 2025. My name is Victor, and I'll be your coordinator for today. I would now like to turn the presentation over to your host for today's call, David Ryzhik, Senior Vice President of Investor Relations and Corporate Strategy. Please proceed.

Thank you, operator. This is David Ryzhik, Senior Vice President of Investor Relations and Corporate Strategy. And with me today is Russell Low, President and CEO; and Jamie Coogan, Executive Vice President and CFO. If you have not seen a copy of our press release issued earlier today, it is available on our website. In addition, we have prepared slides accompanying today's call, and you can find those on our website as well. Playback service will also be available on our website as described in our press release. Please note that comments made today about our expectations for future revenues, profits and other results are forward-looking statements under the SEC's safe harbor provision. These forward-looking statements are based on management's current expectations and are subject to the risks inherent in our business. These risks are described in detail in our annual report on Form 10-K and other SEC filings, which we urge you to review. Our actual results may differ materially from our current expectations. We do not assume any obligation to update these forward-looking statements. Given the pending merger with Veeco, we will not be addressing questions related to the transaction. Please note that today's call is neither an offering of securities nor a solicitation of a proxy vote in connection with our previously announced transaction with Veeco. During this call, we will be discussing various non-GAAP financial measures. Please refer to our press release and accompanying materials for information regarding our non-GAAP financial results and a reconciliation to our GAAP measures. Now I'll turn the call over to President and CEO, Russell Low.

Thank you, David, and good afternoon, everyone, and thank you for joining us for the fourth quarter and full year 2025 earnings call. Beginning on Slide 4, we generated solid results in the fourth quarter with revenue of $238 million and non-GAAP earnings per diluted share of $1.49, both exceeding our outlook. Our better-than-expected results were primarily driven by stronger CS&I aftermarket revenue in the quarter, which had a favorable mix impact on our gross margins. Bookings in the fourth quarter improved significantly on a sequential basis, primarily led by power, general mature and memory, specifically DRAM. Before I provide more details on the trends we are seeing by market segment, I'd like to provide a brief update on our pending merger with Veeco. We continue to make meaningful progress towards securing all the approvals required to close the merger. We are pleased that shareholders of both companies voted in favor of the transaction at our special meetings on February 6. We have also received regulatory clearance in several important jurisdictions and are actively engaged with the regulatory authorities in China for the final approval needed to complete the merger. We continue to expect closing in the second half of 2026. In the meantime, we are working closely with Veeco on integration planning to help ensure the combined organization is fully aligned and ready to hit the ground running on day 1. As we collaborate more deeply with our Veeco counterparts, we are increasingly impressed by the clear alignment in our cultures, values and our shared commitment to innovation and customer satisfaction. The integration planning process has reinforced our confidence in the potential of this merger and in our plans to come together as one company to unlock even greater value for all of our stakeholders. Turning to Slide 5. In the quarter, as well as the full year, sales to mature-node applications accounted for the majority of our system shipments, in particular, power and general mature. Now on Slide 6, let me review our trends by end market. Within our power business, shipments to silicon carbide moderated slightly on a sequential basis. Consistent with our commentary over the past several quarters, customers continue to take a disciplined approach to capacity investments following the significant build-out a few years ago with many customers now prioritizing the technology transitions. As an example, a key driver of our CS&I strength in the quarter was driven by system upgrades for a number of our silicon carbide tools at a customer location where a customer converted their tools from 150 millimeter to 200 millimeter and chose to do this with our recently introduced Purion Power Series+ platform while staying within the same footprint. We also continue to see select customers in China expanding their silicon carbide device capacity and capabilities, while customers in other regions are focused on the next-generation technology investments such as trench and superjunction. We are seeing growing interest for our newly developed high-energy channeling capabilities, which are essential and necessary for superjunction development. Broadly speaking, while near-term ion implant demand for silicon carbide applications is anticipated to remain muted as customers absorb their capacity, we expect strong long-term demand through the cycles, driven by clear secular trends such as the growing penetration rate of silicon carbide into electric vehicles, particularly as we see more 800-volt models released, growing content of silicon carbide within EVs, the growing overall volume of EVs on a global basis and the growing adoption of silicon carbide outside of the automotive sector such as solid-state transformers, circuit breakers, solar battery inverters, HVAC systems, industrial motor drives, aerospace and defense, just to name a few applications. Adding this all up, we remain excited about the long-term demand profile for silicon carbide, and we believe we are well positioned in ion implant for this application. In our other power market segment, ship system revenue also moderated slightly on a sequential basis, primarily due to a muted demand trends in silicon IGBTs. In general mature, revenue improved sequentially as customers made select investments, particularly in our high current tools. Overall, customers continue to manage capacity investments amid stabilizing auto industrial demand. However, we noticed a continued improvement in implant tool utilization rates across multiple customers in the fourth quarter. While we are not yet seeing signs of a cycle recovery in CapEx spending for general mature process technologies, we are encouraged with the improved utilization rates. Turning to Slide 7. In advanced logic, we generated revenue on a follow-on order from an existing customer and we continue to work closely with customers on next-generation technology needs, including implant for backside power contacts as well as other material modification implant applications. Moving to our memory market. Demand improved sequentially for DRAM and HBM applications, and we expect this momentum to extend into 2026 as customers expand capacity to address growing demand for AI-related applications. I'm also pleased to say that we secured an order for a high current system from a leading North American memory manufacturer, which is an important customer win that broadens our presence beyond our strong position in Korea. Interestingly, we received this new order prior to the completion of an existing system evaluation, which we view as an endorsement of our technology. In NAND, customers remain focused on higher layer counts, which does not drive incremental ion implantation demand. As a result, we continue to expect demand for NAND applications to remain muted in the near term. That said, recent improvements in NAND bit demand and pricing are encouraging, and we believe we are well positioned once our customers resume wafer capacity additions, which we expect is a matter of when, not if. On Slide 8, let me wrap up my thoughts on 2025. I am pleased with our team's focused execution given the changing macro environment throughout the year. We navigated the cyclical digestion period across our markets exceptionally well and focused aggressively on product development and customer engagement. Early this month, we announced the introduction of the Purion H6, our next-generation high current ion implanter. The Purion H6 incorporates a series of notable technology advancements across the beam line, source, particle control and dosimetry subsystems. The system delivers industry-leading dose repeatability as well as significant advancements in purity, precision and productivity, supporting high current applications across advanced logic, memory and mature process technology nodes. As devices scale and architectures become more challenging, customers are demanding tighter implant control, lower contamination, higher uptime and lower overall cost of ownership requirements that Purion H6 was engineered specifically to meet. In the fourth quarter, we delivered our strongest quarter of high current shipments in 2 years and the introduction of Purion H6 builds on that momentum. We also delivered double-digit year-over-year growth in our CS&I aftermarket business, supported by our growing installed base and a deliberate strategic focus on upgrades and services both of which reached record levels in 2025. Despite overall revenue declining in 2025, our non-GAAP gross margins grew by 30 basis points. In combination with disciplined cost control, we delivered strong profitability and cash flow in 2025. Now on Slide 9, let me discuss our initial perspectives on 2026. We expect our memory business, led by DRAM to grow as customers invest in capacity to meet accelerating AI-driven demand. In the power and general mature markets, while utilization trends are improving, customers are continuing to manage existing capacity following a strong investment cycle over the past several years. As a result, we expect this to result in slightly lower year-over-year revenue in these end markets. However, over the long term, we anticipate power and general mature to be a key beneficiary of electrification and the increasing demand for efficient power generation delivery and use. Importantly, while the rapid growth of AI large language models has been a strong catalyst for advanced compute and memory demand, we expect the power semiconductor market to also benefit from the growing need for power associated with AI. And this includes silicon, silicon carbide and gallium nitride applications, distributing power all the way from the grid to the core. Moreover, we also anticipate the emergence of physical AI such as robotics, autonomous vehicles and many other devices on the edge to become yet another sector driver of power devices as well as general mature technologies such as MCUs, image sensors, analog and others. And finally, in advanced logic, we anticipate relatively similar revenue levels in 2025. We are making progress in our long-term strategy to penetrate this market, but it will take time before our evaluations translate into meaningful volume as we engage with customers on next-generation logic architectures. Taken together, we currently anticipate overall revenue to be relatively flat compared to 2025 levels. With that, let me turn the call over to Jamie for a closer look at our results and outlook. Jamie?

Thank you, Russell, and good afternoon, everyone. I'll first start with some additional detail on our fourth quarter and full year results before turning to our outlook for Q1. Starting on Slide 10. Fourth quarter revenue was $238 million, with systems revenue at $156 million and CS&I revenue reaching another quarterly record of $82 million, both above our expectations for the quarter. Our better-than-expected CS&I revenue was primarily driven by strong demand for upgrades as customers look to optimize their technology within the same footprint. In addition, the quarter also benefited from some pull-in activity given improving utilization rates. We are pleased with our execution in CS&I and our aftermarket offerings are resonating with customers. For the full year, CS&I revenue grew 14% on a year-over-year basis, led by strong growth in upgrades and services revenue. A key driver here is the deliberate strategic initiatives we've undertaken over the past few years to drive better adoption of upgrades and service contracts. Moving to consolidated revenue. From a geographic perspective, China decreased sequentially to 32% of total revenue, down from 46% in the prior quarter, which was consistent with our expectations as customers continue to digest the robust investments they've made in mature node capacity over the past few years. Turning to the other regions. We saw revenue in Europe at 15%, the U.S. at 14%, Korea at 13%, Japan at 9%, Taiwan at 3% and the rest of the world at 13%. For the full year 2025, our revenue from China was 42% of total revenue, while U.S. came in at 16% and Korea at 13%, Europe at 11%, Taiwan and Japan at 5% and the rest of the world at 7%. As Russell mentioned, bookings increased notably on a sequential basis to $128 million, and we exited the fourth quarter with a backlog of $457 million. Now turning to Slide 11. I'd like to share some additional detail on our GAAP and non-GAAP results. GAAP gross margin was 47% in the quarter. And on a non-GAAP basis, gross margin was 47.3%, above our outlook of 43%, primarily due to a higher mix of CS&I as well as a more favorable mix of upgrades within CS&I. GAAP operating expenses totaled $76 million. And on a non-GAAP basis, operating expenses were $62 million, above our outlook of $56 million, primarily due to higher variable compensation expenses associated with the better-than-expected fourth quarter performance. Our non-GAAP results exclude transaction-related expenses associated with the pending Veeco merger, along with other typical non-GAAP adjustments such as share-based compensation and restructuring charges. In this context, the fourth quarter reflects an expense from a onetime voluntary retirement program, and we recorded a portion of that expense in the period. As a result, our GAAP operating margin was 15.2%, while our non-GAAP operating margin was 21.1%. Moreover, in the fourth quarter, we delivered adjusted EBITDA of $55 million, reflecting an adjusted EBITDA margin of 22.9%. We generated approximately $4 million in other income with the sequential increase primarily due to foreign currency. Our tax rate was approximately 14% in the fourth quarter, both on a GAAP and non-GAAP basis. Our weighted average diluted share count in the quarter was 31.1 million shares. This all translates into GAAP diluted earnings per share of $1.10, which was higher than our outlook of $0.76. Non-GAAP diluted earnings per share was $1.49, also exceeding our outlook of $1.12. The higher-than-expected non-GAAP diluted earnings per share was primarily due to better-than-expected revenue and a favorable mix. For the full year, we delivered GAAP gross margin of 44.9% and non-GAAP gross margin of 45.2%, a 30 basis point increase year-over-year despite the lower revenue. This was due to favorable mix and a continued focus on cost control. For the full year, adjusted EBITDA was $177 million, reflecting an adjusted EBITDA margin of 21.1%. We generated approximately $19 million in other income this year, and our tax rate was approximately 13% for the full year on a GAAP and a non-GAAP basis. This all translates into GAAP diluted earnings per share of $3.80 or a non-GAAP diluted earnings per share of $4.88. Moving to our cash flow and balance sheet data shown on Slide 12. In the fourth quarter, free cash flow was negative $9 million, driven by the timing of our sales, which were skewed more to the month of December. In turn, this increased our days sales outstanding. In addition, our cash flow was negatively impacted by approximately $5 million in cash transaction expenses associated with the pending Veeco merger. For the full year 2025, however, we generated robust free cash flow of $107 million. And despite the decline in our revenue, this reflects strong cash generation of our business. Turning to share repurchases. In the fourth quarter, we repurchased approximately $25 million in shares and have $110 million remaining under the share repurchase program previously authorized by the Axcelis Board of Directors. For the full year, we repurchased approximately $121 million of shares. We exited the fourth quarter with a strong balance sheet, consisting of $557 million of cash, cash equivalents and marketable securities on hand. It's important to note that this includes $182 million of long-term securities. With that, let me discuss our first quarter outlook on Slide 13. All measures will be non-GAAP with the exception of revenue. We expect revenue in the first quarter of approximately $195 million. The sequential decline is expected to be across both systems and CS&I. In systems, we have seen a few pushouts due to the timing of available cleanroom space. And in CS&I, there's some seasonality at play, which typically results in higher volume in the fourth quarter given customers like to manage their budgets into year-end. In addition, as Russell noted, we saw a large customer upgrade in the fourth quarter that we do not expect to recur in the first quarter. And finally, a portion of the CS&I upside we saw in the fourth quarter was pulled forward from the first quarter. We expect non-GAAP gross margins of approximately 41%. The sequential decline is primarily due to less favorable mix and lower volume relative to the fourth quarter. More specifically, we anticipate a higher volume of sales into the memory market, which typically carry lower gross margins due to the nature of the systems they are buying. In addition, we anticipate a sequential decline in CS&I revenue, combined with a lower mix of upgrades, which typically carry higher gross margins. And finally, to a lesser extent, we anticipate a modest incremental impact from tariffs as a greater mix of our inventory includes the tariff impact and older inventory is cycled out. We expect non-GAAP operating expenses of approximately $59 million in the first quarter. Adjusted EBITDA in the first quarter is expected to be approximately $26 million. And finally, we estimate non-GAAP diluted earnings per share in the first quarter of approximately $0.71. As Russell indicated, we anticipate full year 2026 revenue to be approximately flat compared to 2025 levels. We expect revenue to be second half weighted. We anticipate growth in our memory market to offset a decline in our power and general mature markets. We currently estimate full year 2026 non-GAAP gross margins of approximately low to mid-40%. The year-over-year decline is primarily due to the anticipated stronger mix of memory business in 2026 relative to our other markets. In addition, as I referenced earlier, we anticipate a modest year-over-year impact from tariffs, which we estimate at less than 100 basis points. As we think about operating expenses for the year, we expect to continue to balance investments in product innovation to drive organic growth in our business while remaining disciplined on cost. As a result, we expect quarterly non-GAAP operating expenses for the balance of the year to remain relatively similar to anticipated first quarter levels. And finally, we anticipate our tax rate to be approximately 15% for the full year. In summary, we're pleased with how we performed in 2025 despite a softer demand backdrop in our power and general mature markets. While systems revenue declined on a year-over-year basis, we delivered strong growth in our CS&I business. We grew our gross margins, and we delivered robust free cash flow while opportunistically increasing our pace of share buyback. We enter 2026 in a solid financial position and are excited to close our pending merger with Veeco, which we believe will enable us to unlock the full potential of the combined company and drive long-term value creation for our shareholders. With that, let me hand the call back to Russell for closing remarks. Russell?

Thank you, Jamie. As we look ahead, we remain focused on disciplined execution and advancing our technology road maps that differentiate Axcelis across our end markets. The progress we made in 2025 from product innovations to acceleration in our CS&I business to operational excellence positions us well as we enter an important year for the company. With respect to our pending merger with Veeco, we anticipate significant long-term opportunities as a combined company. Together, we expect to be even better positioned to capitalize on the secular tailwinds driven by AI and electrification and expect to leverage the complementary strengths across our portfolios and our people to deliver greater value for all stakeholders. I want to thank our customers, employees, partners and shareholders for their continued support and trust in Axcelis. With that, operator, we are ready to take questions.

[Operator Instructions] Our first question will come from the line of Jed Dorsheimer from William Blair.

Congrats on Q4. I guess a question for me is just in the memory market, congratulations on getting the high current into the U.S. memory manufacturer. I'm wondering if you might be able to provide a bit more color is to whether or not this was more of a second sourcing opportunity? Or I just look at the Q4 to Q1 transition, and I know you said that memory is going to be stronger. Do you anticipate that, that new customer is ramping up on that capacity expansion? Or how to think about that? And then I have a follow-up question.

Jed, it's Russell. So when I think about memory, obviously, I'm talking about DRAM specifically, and the demand is going up and up, pretty much driven by AI. What we're seeing right now is that basically, it is cleanroom space limited. So we're seeing many, many customers trying to slide an extra couple of tools in to increase their capacity, but you won't really see large numbers of tools until you actually open up the new fabs that are coming through. So I think you've seen them in the press. Each one of the really large DRAM manufacturers have plans to open new factories in the near term. So I think what you'll see is that, that's -- we'll see DRAM significantly up in 2026 compared to 2025, although somewhat of a lower base in '25. But we're looking to see that momentum continue into 2027 as the -- basically the bottleneck, which is cleanroom space starts to dissipate and there'll be more tools coming out. And that will be across all the DRAM manufacturers.

That's helpful. And then just on this CS&I and specifically the silicon carbide, if you look at your base that's out there, what percentage has converted from the 150 mm to 200 mm already? And where I'm going with this is just to try and understand sort of that conversion premium in the CS&I business that you saw in Q4, what's left out there to kind of run through or capture?

That's a great question, Jed. So I guess the question is, what is the market potential for a 6-inch to 8-inch transition in terms of upgrades?

Yes. Yes.

So I would say that only a very small part of our installed base has gone up to 200-millimeter. It's still very much in that transition phase and kind of the leaders are trying to get there. And I'd say the leaders typically outside of China are trying to get there first. I think that's where they're going to have the biggest cost advantage. So I think there's ample opportunity to continue to provide these really high-value upgrades into fill them. I think I should just clarify that, yes, there was a large customer going from 6-inch to 8-inch, but they also upgraded the system as well to our Purion Power Plus Series, which was kind of an upgrade on top of an upgrade. So that really did reduce the cost of ownership. And like Jamie said in his prepared remarks, that was within the same footprint of the tool. So this really is a field upgradable system that gives -- continues to give value.

Next question comes from the line of Denis Pyatchanin from Needham.

So regarding the strong bookings that you guys are seeing, can you tell us maybe a little bit more about which 2 or 3 segments are driving this growth?

I'd say that it's actually -- so if you look at our revenue history and where those segments have been, I'd say that our bookings have pretty much matched our prior. So it's nothing unusual. And while we're seeing a bump up in memory, I just want to mention that memory typically books and ships in the same quarter. So you don't typically see a lot of memory in our backlog. There's a little bit there depending on where the quarter ends. But ultimately, it's still general mature and power are the main parts of our bookings.

Understood. And given the outlook for approximately flat revenue year-over-year for 2026 versus 2025, and given the strength that we're seeing in DRAM from both HBM and DDR5, can we assume that both general mature and both segments within power are showing some weakness throughout the year?

So I think we've kind of said that general mature is slightly down but we are seeing the utilization rates coming up. So obviously, our customers are seeing hopefully a rebound in revenue but that's because they're now getting to utilize the tools they already have. Once those tools are up to high utilization rates, that's when they'll start to buy more tools. So certainly a very positive sign is just that we are further down the supply chain. So I would say that we're slightly down year-over-year with raising utilization rates. Regarding power, I'd say power is slightly down, although it's still a large part of our revenue and is still doing, I'd say, relatively well. And when I look at power, I think about the long-term secular trends, electrification, for example, all the way from generating efficient energy, transporting efficient energy and using less of it, I really do think that there's a great opportunity in the long-term secular trends for power. So while they're taking a bit of a pause, and I say sort of slight down, you're seeing the slack being picked up by memory. And like I said before, memory, right now, the bottleneck is the cleanroom space.

Understood and...

Go ahead, Denis.

I was just going to say, is there any way you could provide a little bit more color on like whether power SiC or other power is looking a little bit stronger over the next few quarters or even throughout 2026?

Yes. I would say our current view is it's like what we're really talking about is a handful of systems, Denis, right, in probably both of those segments, right, lower than the year-over-year prior year. And as we look out, what we're trying to just be cautious on is calling the timing of a recovery, right? We're seeing those encouraging signs, as Russell said, in utilization rates. That's helping to contribute to some of the incremental CS&I volume that we are seeing as well. So we are getting some beneficial financial performance from those improved utilization rates. And similar to our approach last year with calling a memory recovery, we wanted to kind of see it in the data before we kind of -- we're willing to put that out there. What I'd suggest is we're seeing those encouraging signs, but we're not yet seeing those -- the order rates pick up at that rate just yet for us to call that a slight recovery in the cycle for those markets.

Our next question will come from the line of Christian Schwab from Craig-Hallum.

I just want to get back to memory for a second. So your largest customer in Korea is taking some capacity off of NAND for DRAM for high-bandwidth memory plus we really need a substantial increase of wafer starts in DRAM specifically, not NAND, given layer count increases in order to get supply and demand more in balance. So are you suggesting that you think others in the space are talking about a super memory CapEx cycle since you really only have one competitor and are leading supplier to the largest. Are you suggesting that you don't have the orders in hand for the second half of '26 yet? Or are you suggesting that '27 should be a fantastic year?

So a couple of things there, Christian. So I think what we're saying is that typically, we build to forecast, which is why we often get the PO, i.e., the booking in the same quarter, the tool shifts. So just to kind of clarify where we are with 2026. We believe that we're going to have significant improvements in our DRAM business in 2026, and that's based upon staying very close to our customers and understanding what their needs are. However, I think it's also clear that while they're kind of begging and borrowing space, they have made public announcements of really significant increases in their capacity. So if you think about the Yongin cluster, for example, with SK hynix, it's going to be 4 mega fabs of 200,000 wafer starts a week and that's where they've been investing a lot of money for quite some time. So they've got the CapEx going into building the fabs and then they'll be looking in, I think, early 2027, for example, for taking systems to then start ramping up that factory. And I think you've seen in the public disclosures multiple companies saying they're looking to build more capacity. But right now, that's what you're seeing with DRAM is that the prices are spiking because the supply is very constrained.

Right, right. Can you -- on the capital intensity, can you remind us how much ion implant equipment is needed for every 100,000 wafer starts?

Yes, sure. So I think NAND and DRAM are slightly -- are mostly similar in terms of the requirements, but the mix with inside that number would vary between high energy, high current and medium current. So if I go with DRAM, for example, typically, it's about $150 million to $200 million worth of capital for implants for every 100,000 wafer starts that are bought online. And that, I would say, is there's a little bit of high energy in there. There's -- but a large amount of that is high current. So when we talk about growth in memory, we have a very good position in high energy and we're looking to get -- to build upon our good position in high current as well. So that's why we've been positive about high current.

Important note, Christian, is that all this commentary really is around DRAM at this point in time, right? And our expectations are for NAND to continue to be relatively muted as we go through '26. That could free up as we go into '27, depending on how those NAND -- what those NAND producers need to increase their capacity. So...

Our next question will come from the line of Craig Ellis from B. Riley Securities.

Guys, I was hoping to get a better understanding of what you're seeing in the mature foundry business with some insight geographically. We're hearing from companies that they're seeing foundry utilization approaching 90% in China in the 80% area in Southeast Asia, I suspect it's much lower than that in mature in the U.S. and Europe. But as you're engaged with your many customers in mature, can you help us understand where we're getting closer to levels where they need to really start adding more capacity versus areas where there may be less very near-term pressure?

Craig, it's Russell. So yes, so utilization rates do obviously vary by customer. I'd say that high utilization will then drive to high CapEx or use of CapEx. And you've seen some of our customers state that their CapEx may not be that high this year. I think there's been a couple of very vocal customers about that. So obviously, they've got to fill up what they have and they may be slightly behind where they would like to be and then they'll start spending. But are you focusing on China? Yes, they certainly want to provide more chips for domestic use. So this is very much about the self-sufficiency. So they're going to -- and I believe they're quite a long way behind their goal. So they're going to continue to buy equipment and ramp up factories. And since these are quite demanding technologies, it may take them a while to get the cost down as they kind of get the yields up and ramp up the factory. But yes, you're seeing that in China, there are certainly still high utilization rates and also a demand for additional growth. The other thing is, I think we said this many times before, I mean, our Chinese customers behave like any of our other customers. They absolutely want the best technology. They want world-class service and support. And we've got great relationships in China. We spent over $100 million last year on R&D, and that goes to innovation and our Chinese customers are looking for innovation as much as anybody else. They're looking to do basically they need cutting-edge technology to give them the benefit in their own businesses that they're looking for to survive.

Russell, that's really helpful. And just sticking with the China theme versus the 32% of revenues in the fourth quarter or if we want to baseline off of the 42% for 2025. And I know the company is not giving full year guidance, but can you help us understand where you feel the demand intensity would be from that region in 2026, given in part that we've heard a number of companies suggest trends could be flattish rather than down year-on-year. I know you have over 20 customers, but as you look at what's possible in China, what does it look like for 2026?

Yes. We see it to be relatively flat to down slightly, right? And I kind of maybe stress the word slightly there, Craig, to some extent. When we start to think about the ASPs and the number of units we ship, right, you're not talking about massive number of, I'll call it, quantity-wise down. We do expect China to continue to remain a really durable part of the overall market and business for us as they build out that capacity. As Russell said, they're still really far away from their self-sufficiency goals and targets and the conversations that the team is having over there with those customers continue to sort of stress the expansion that they continue to plan with more fabs coming online in order to meet those self-sufficiency goals and targets. So we're prepared to continue to support that ramp in the market. And as they work through that, we call this digestion of capacity, we really have seen that more as a mixed bag outside of China, more so than really inside of China.

Very helpful. And if I could sneak in one more. Jamie, with regard to the 41% gross margin guide for the first quarter, clearly, CS&I less of a tailwind than that spectacular fourth quarter gross margin number. But are there any one-offs impacting gross margin is coming in a little bit lower than I would have otherwise expected?

Yes. No, it's a good question. And sort of we dissected the sort of the dip we're anticipating here. I think that, again, the strength of that margin in Q4, right, was driven by really record levels of CS&I volume that we saw in the period as well as the mix within the upgrades, right, upgrades being a larger mix of CS&I load. So on a normalized basis, right, I would say that, that margin decrease is not as material as it might otherwise look from Q4 to Q1. Within that, we do see a higher mix of memory systems making up the systems volume in the period as well as sort of the moderation of CS&I and expectation for moderation of CS&I volume given that there's some seasonality that happens from Q4 to Q1 as our customers will use up some of their budgets to buy incremental CS&I in the period. And then I don't want to overstress this, but we are seeing a Q1. There's a little bit of a tariff impact in Q1, just given the way our inventory flows out from -- we procure inventory last year, which gets used in Q1 of this year. That is now at a slightly higher cost. We anticipate that to be a little less than 100 basis points for the full year, but the impact in Q1 is higher and we expect that to moderate over the course of the year as we sort of catch up with our sort of Q4 COGS levels.

Our next question will come from the line of Jack Egan from Charter Equity Research.

So the first is on power and general mature. I just kind of want to make sure I understand your guidance correctly. So because based on -- if you look at a lot of the commentary from the big analog and power companies in -- just from this quarter, it seems spending will come down pretty materially across the Board. But then as you mentioned, there are some tailwinds like the higher utilization and some onetime upgrades to larger wafers or in other areas and then some build-outs by newer Chinese customers and I. So I mean, does that kind of encapsulate the situation where I think you said it's going to be down slightly year-over-year, and it's mainly capacity investments being down and then offset a bit by some of those other idiosyncratic factors.

Jack, it's Russell. So I think the question was we're saying it's kind of like flat to slightly down, you're saying what are the factors in there. And I think it's -- I think China is still looking to add capacity, but they're adding it carefully, I'd say. They've got to absorb the got and make sure they ramp up effectively with costs in control. And I think you're finding some of the non-Chinese people are also -- the utilization rates are definitely going up. And it matters which customer you're talking about, the utilization rates are going up. And since we're a little bit further down the supply chain, they need to get their utilization rates up before they start thinking about new tools. So I think that's kind of like we'll see a lot of customers this year start to recover their capacity. And then that's what we'll then see hopefully as an opportunity going into 2027.

Okay. Sure. That's helpful. And then as you've said earlier for memory, those orders come in with a really short lead time, but something that we're seeing like at the memory companies is that their customers are kind of becoming more willing to sign these unusually long LTAs just given the severity of the shortage. And so I was just kind of curious on is any of that trickling down to Axcelis, just maybe in kind of your customer discussions with customers, have they indicated that maybe we'll give you longer lead times? Or is it largely the same as it has been for Axcelis in recent quarters?

Yes, Jack. So whenever I hear about long-term contracts being signed, it always makes me a bit worried because I've kind of seen these happen before. And I think somebody in the newspapers [indiscernible] as kind of the demand for RAM is so, so high. And you're absolutely right, you're seeing people buying up large amounts of RAM in long-term contracts. So yes, the price is spiking up. I think that you're definitely seeing that. But I also -- we're talking about RAM, right?

I think it's important we do maybe just like a process check on how our customers order with us just to clarify. Sorry you talked about short order cycles it's PO placement, Jack. But what we do get from our customers in this space is we get a long-range forecast that can cover sort of multi-quarters. So we're getting multi-quarter forecast. Our historic experience has been that those forecasts are very sound, right, in terms of what the needs and requirements are. And sometimes we actually will see requirements go above what is in that initial forecast as they are able to free up incremental capacity in space. That activity is occurring, continues to occur with these customers. This is why when we talk about our confidence of memory recovery, right, it's founded in those discussions that we're having relative to our customers' needs for the period. What follows short behind that is I'll call it, the legal framework for which we enter into the purchase order. That is just the timing of that purchase order can be a week, a month and maybe if we're lucky, a quarter ahead of time.

Yes. That's exactly what I'm sorry. I think [indiscernible] last week, that's the jet lag kicking in. And I think [indiscernible] talking to our customers about what they see going on. And yes, Jamie is absolutely right. We work on forecast. The forecast, as you can imagine, are quite positive. But -- and we're making sure that we're ready for that demand.

Yes. It would be interesting to see, Jack, is that as they free up their constraints, how quickly they can free up their constraints if those buying patterns accelerate, right? But as of right now, that's why we're forecasting an acceleration of memory into '27 as we'll see higher volumes here and then we'll see acceleration into '27. We're going to watch pretty closely our customers' expansion as we move through the course of the year as they try to pull in some incremental fab space. So we want to position, leverage the balance sheet and position ourselves to take advantage of incremental opportunities for '26 if they're able to free up some incremental space.

Next question will come from the line of Duksan Jang from Bank of America Securities.

I have a question on the Q1 guide of $195 million sales. So I think that's about a $20 million delta from your -- what you alluded to last quarter. I know you talked about some pull-ins. But at the same time, I think you said memory has gotten much stronger in your bookings, which is more of general mature and power, that has also improved a lot. So I'm curious what the puts and takes are into that $195 million guide? How much is pull-in versus how much is real end demand?

Yes. So it's actually a combo factor. And Duksan, thanks for asking the question. I think it's important to get it clear. So we did see some incremental volume in the fourth quarter for CS&I that probably would have floated into Q1 had it not come into -- not overly material, but enough to sort of move our expectations for Q1 relative to that. On top of that, we saw some pushout of systems from Q1 into later in the year, given fab readiness for our customers and not necessarily focused on any one specific market. We just have these things happen. So as we look ahead, timing, right, of system delivery is always one that we manage very closely with our customers. Again, we don't want our systems sitting in parking lots or warehouses, we want the fabs to be ready. We want our installation teams to follow closely thereby. So what we really saw was some pushout of systems, coupled with some modest pull-in activity on the CS&I front.

Got it. That sort of leads into my second question, which is on your 2026 outlook of being flattish. When you say it's going to be second half weighted this year, is that mostly coming from memory? Or is it also more broad-based? And what's giving you really the confidence into that visibility?

So I think...go ahead for it Jamie.

Go ahead, Russell.

No, it is really...

Yes. Both of those things are occurring, right here for us. At the end of the day, we're seeing some incremental memory volume at a higher rate than we expected coming in the back half of the year. So that when we talk about those encouraging signs relative to memory and some of our ability to seeing the recovery occur, it is that memory volume coming in that we anticipate going higher. We are also anticipating just based on the timing of customer needs, we're seeing systems volume weighted towards the back half of the year. And then we are forecasting sort of similarly a little bit of CS&I uptick over the course of the year as we introduce upgrades and upgrade activity that can occur as customers take those things on.

Next question will come from the line of David Duley from Steelhead Securities.

I guess, first off, if I listen to you carefully about your memory business, it sounds like you're going to have higher growth in memory in 2027 than you're going to have in 2026, driven by the cleanroom space constraint. But in 2026, do you think this business can get to like 12% or 15% of revenue? I'm just kind of curious how much growth you think there will be in '26?

Yes. So we're not giving a specific number, Dave, just yet on that. But I would say, again, we're coming off a fairly low base in 2025, and we do expect the volumetrically to see an increase here to kind of keep the revenues, call it, flattish as we see some downward trajectory in the power and general mature space overall. We do anticipate acceleration of memory activity as we move into 2027. And obviously, the growth rate from '25 to '26 to '27 is all going to predicate on the volume that we're able to deliver within 2026. But as of right now, I would say the growth rate in '26 is a healthy number.

And do you think 2027, since that's probably going to be the higher growth year and the higher dollar amount of memory systems shipped, do you think you can get back to your historical range? I think that was 18% to 20% a long time ago when there was a memory cycle?

Sorry, let me -- okay. So Dave, I think if you go back to historic records, it's probably better to do it in absolute dollars than it is market share because I think as a company, we've progressed significantly. I mean, when I joined 10 years ago, we were selling tools to Korea, right? And I think since then, you've seen us getting into silicon carbide power, silicon power, got a nice business in general mature, image sensors. So I think it's more about the absolute numbers. Our last highest memory year, I think, was like -- I want to say, Jamie, it's like $120 million, but that was NAND and DRAM. So you then have to pare it out to what was the highest amount of revenue we ever made within a year on DRAM alone, that will probably get you close to where we would be wanting to get plus some, right? I mean I'm always the optimist and I'm always looking to take on more market share. And I think you've also got to look at the size of the market. And I don't think I've ever seen the DRAM market in such sort of supply before. So I think as ever, you're going to start to see a large amount of supply come on. And we know how the cycle works, but I think we're looking to get a piece of that action. And like we've said, we're very pleased with the fanning out of our memory into a large North American manufacturer. I think that gives us an even bigger footprint and bigger opportunity going forward.

I think Russell, that's really important is that last DRAM high was $90 million, right? And that was before the introduction of these systems into the North American memory manufacturer. And then as we think about the number of wafer starts, right, that predicated the $90 million build-out, right? I think the projections for that are higher. But again, we're going to -- again, we'll be sort of moderate in our tone here because, again, we need to see those wafer starts occur. We need to see the customers expand their fabs. We need to see the orders and the forecast come in. But all the signs right now are encouraging for that, that '27, we will accelerate out of '26 into '27.

Okay. And then final thing for me is, I think you've already answered, but I just want to make sure, I think you said you had record demand for high current systems or tools in the quarter. Is that just because the memory business is turning on? Or maybe you could elaborate a little bit about that.

So we do have -- so there's a couple of things. Yes, memory business is turning on. But like I say, it's still on a pretty low base in 2025. I'd say that we're actually moving out. So the goal was always, Dave, is to get in on an application that was critical that really had value and then start to fan out within a customer and then also take on new customers where we know our products will have significant value like we did with the North American memory manufacturer. So we are seeing a broader base in memory, a broader base in general mature and power. And I think that's also a testament to the value that our tools bring to our customers. So we are seeing an uptick in our high current market share.

David, just to add to that, that's the highest we've had in the 2-year time frame, not an all-time record for high current. But so over the -- if we look over the last 2 years, this was a record for high current product.

And this concludes the question-and-answer session. I would now like to turn it back over to David for closing remarks.

Thank you, operator. I want to thank everyone for joining the call and your interest in Axcelis. Operator, you can close the call.