All right. Good morning, everyone. My name is Duksan Jang. I'm part of the US Semiconductors and Semis Equipment Research Team here at Bank of America. I'm very delighted to host James Coogan, Executive Vice President and CFO, as well as Douglas Lawson, Executive Vice President, Corporate Marketing and Strategy of Axcelis. Doug, Jamie, welcome.
Yeah, thanks for having us. Appreciate it.
Sure. Before we get started, if anyone in the audience would like to ask any questions, we are happy to take any questions. Just raise your hand, and we'll bring the mic to you. And with that, did you have any safe harbor statements that you had to go through, or?
Yeah, all the ifs, we's, shoulds, coulds, mights, all those should be construed just as what they are. And again, information that we disclose and talk about today is consistent with those other public statements regarding the performance of the business. And you should read our 10-Qs and 10-Ks for the relative risks.
Sounds good.
Sounds good?
Yep, yep. Well, now on to the fun stuff. Just so for those unfamiliar with the story, would you mind giving us a brief overview of what you do, who you serve, and maybe your competitive advantages?
Yeah, I'll start and pass it off to Doug as well. So Axcelis Technologies has been in existence for 45 years now. We design, engineer, manufacture, service, ion implant equipment, serving the capital equipment, sorry, the semiconductor market space. We serve all markets of the semiconductor capital equipment space, including memory, power device, other mature technology devices, image sensors, as well as advanced logic. We are geographically very diverse. We have a large customer base, global customer base, and we have a very large and global field service support team that performs aftermarket sales, aftermarket spares, consumables, and services, our tools out in the field. That puts us very close to our customers. We've got very long-tenured relationships with our customer base with some differentiated products in the market space that I'll have Doug talk about.
That allows us to maintain tight relationships with those customers and continue to advance the technology going forward. We're continually iterating across our technology sphere to keep up with the current needs and requirements of our customers. We invest approximately 9%-10% of our revenue annually into R&D. Last year, it was about $100 million or so worth of R&D spend. That gives us new product extensions as well as upgrade opportunities, which feed our aftermarket business. Ultimately, really, it's our three products: our high-energy tool, our high-current tool, and our medium-current tool. The differentiation we have in the marketplace relative to those tools that provides us an advantage.
Yeah, the only thing I'll add to it is, in talking about the aftermarket business, we have a very large installed base, having been in business for over 45 years. So that contributes quite a bit to our aftermarket business, which is margin accretive and very important to our customers and keeping the customers satisfied.
Sure. Let's focus on the near term for a little bit. You just had your Q1 earnings. Maybe could you provide us a summary of where you are in the cycle, what you're seeing in the market, and also, more importantly, the sentiment might have changed as of late, just given a lot of new investments out there. Maybe could you explain a little bit about what you're seeing as well?
Yeah. So entering the year, I'll maybe go back to sort of entering the year. As we entered 2024, there was a little bit of, I'll call, lower visibility relative to the timing of recovery. Axcelis, the semiconductor markets, have been in a very significant downturn for the past two years or so here. And as a result of that, we did not experience it at the same rate as other capital equipment suppliers did. We had developed a very strong position in the power market and power device segment, which actually helped propel 2023 to record revenues as well as record profitability for the organization.
As we entered into the year, we started to see some incremental, I'll call it, weaknesses in our general mature space, as well as some of the in the memory market continued to be weak or was expected to continue to be weak for us going into 2024. As we worked our way through Q1, we're starting to see very positive signs that indicate that those markets, which were providing a little bit of lower visibility to us, are starting to firm up in a nice way. So on our year-end call, we talked about some requests from customers to delay some deliveries. During the Q1 call, we talked about how we were starting to see those same customers come back to us and ask for those deliveries to be pulled back into the period.
We didn't see material cancellations out of our business at all, but we started to see actually increased quarter quoting activity, which is a precursor to incremental order activity that we expect later in the period. In addition to that, we ended the period for Q1 slightly ahead of our prior expectations. And although we guided our Q2 results to be roughly at a run rate of approximately $1 billion for 2024 for the full year, we do expect the second half performance to be stronger, given some of the signs that we saw that we talked about on our Q1 earnings call. Part of that is the recovery in the memory space. And so we do expect the memory markets to recover for us in a fairly meaningful way into 2025. And right now, we're projecting that recovery to begin for us in the Q4 time frame.
We are also seeing some solidification in the customer expectations for requirements in the general mature market space, which was what we needed to see in order for us to have a higher level of confidence going into higher second half performance over what we're seeing today through the first half of the year. So all of those signs are, as we look at 2024, starting to shape up for us in a nice way. Again, still very meaningful performance for Axcelis relative to our historical financial performance and really positioning, we'll talk about some of the macro trends here, positioning us to have incremental organic growth opportunities very broadly across the space as we look beyond 2024 into 2025 and 2026.
Yeah. Well, probably, I'm sure Duksan has a few questions about power. So we'll add some more on power as you get to those kind of questions.
Sure. Just one more question before we get to power. I mean, you're definitely seeing this near-term recovery come through, but there are also concerns that mature node in general is slowing down. I don't want to name specific customers, but we've heard from a couple others based both in the U.S., and in China that they're cutting CapEx. So would that impact your outlook at all?
Yeah. So it doesn't impact our outlook. At this point, we started to see the beginnings of that essentially last year. The consumer spending started to come down. That brought the consumer side of the mature market down. Middle of year last year, industrial started to slow in response probably to consumer. And then towards the end of the year, automotive started to slow, which was still in a probably would have slowed earlier, except for the fact the pandemic and supply chain issues sort of kept that moving a little longer. As we entered this year, as Jamie just explained, that's sort of what had tailed off. Power continued to be strong in the mature market. We expect that that market does start to recover. So despite some of our customers potentially changing CapEx plans, the long term looks still very strong for the mature markets.
We expect there to be a third wave of IoT kind of activity related to AI. So the first wave of IoT, which brought the mature markets back, happened back in the 2015 time frame. Second wave happened with 5G, which enabled just a higher density of endpoints for IoT-type devices. AI is very data hungry. And to feed that beast, they're going to want a lot of source data. So all the information coming from our thermostats and our cars and our refrigerators and all the industrial applications feeding information is going to be important. All of those devices are coming out of the mature markets, whether they're image sensors, analog-type devices, RF related to it, and so forth. So while there might be some near-term corrections, we don't expect that to be long. We expect it to be a very strong market.
Yep. Got it. Now on to the highlight of the business, power devices. Roughly 60% of your sales today. You've also been one of the pioneers kind of in the supply chain. But we're also hearing that the demand sustainability might not be there, especially given the end market EVs are slowing down. China, there are some oversupply concerns as well. You mentioned record sales last year. So how can we reconcile these differences? What gives you this confidence of visibility? That'll be helpful.
Yeah, there's a couple of things. So as we think about power devices, right, I think they've been generally tied very explicitly to growth in EV, right? Ultimately, though, what power devices are is it is silicon carbide as well as silicon IGBT. And so if we start with electric vehicles first, there may be some portfolio rationalization going on amongst the automakers to introduce hybrid, plug-in hybrid models into the portfolio. That is going to increase the quantity of power devices that are currently utilized as all of those vehicles have some level of power management, power device inside of them from the inverters, converters, and others. That will be some combination of silicon IGBT, given the price point and the range characteristics of those. It may over time migrate over to silicon carbide.
Those provide us an opportunity to continue to push our products into the power segment, given the implant intensity of those types of power devices that are going into those vehicles. And I think the important thing there is that those vehicles are not necessarily taking share away from electric vehicles. They're still continuing to take share away from the internal combustion engine cars. And so that's sort of the sort of U.S., perspective relative to that. Globally, electric vehicle adoption is still continuing to grow. And as we think about where China is today and their sort of view of self-sufficiency in order to be able to support the electric vehicle market that they have in country, we still think that they're in the very early innings of being able to build out the capacity to meet their internal demand characteristics.
So on a global perspective, we do see electric vehicle adoption rates continuing to be very strong. It's a question of, is it a slowing growth, right, versus a retraction? And I think the final piece is really, given the volume that we've seen now with electric vehicles, hybrid vehicles, and others, specifically in silicon carbide, we've been able to, our customers have been able to reduce the price point of these devices now to where you're opening up industrial applications. And when we think about the power needs, and this is where another kind of call it ancillary impact associated with AI and data centers, there's a lot of reports out there on how power hungry these data centers are and the power requirements of those.
Silicon carbide and power devices really do provide a significant benefit to the power management associated with AI data centers and meeting sort of the requirements of those data centers going forward. That's going to be in things like clean tech, clean energy, right? Microsoft recently announced a $10 billion investment to use clean tech and clean energy to support their data center requirements. You've got other reports of turning on coal-fired power plants in order to sort of meet the data requirements of data centers today. And so the data needs and the investments that are going to be needed to be able to support these requirements are ultimately going to be very significant.
A couple of proof points that we've seen over the last week or so is Infineon, one of our customers, announced a sort of 400-volt low voltage silicon carbide trench-designed power device that is going to be focused and targeted towards the AI data centers. That was countered just yesterday by onsemi announcing another power family using the trench design there. Why do we mention the trench design? Because that actually is right in the, I'll call it our wheelhouse as an ion implant company, specifically in the high-energy space. The trench design really needs the types of high-energy and high-current tools that we offer into the power market today. And those are very implant-intensive devices that are being manufactured. And so ultimately, we would expect as there's increased proliferation of power devices into the industrial space that we would see incremental needs for our systems to support that. Yeah.
Would you say there are any metrics that you look at to maybe calculate any potential oversupply or overutilization as your customer double ordering?
Yeah. Double ordering for ion implant, we don't really experience. The concerns relative to your China question and so forth, we're not really seeing it. If we look at this past year, we said 40%-60% of our business is expected to come from China in any given quarter. The percentage of power is much higher than the mature markets right now. That's actually a good sign relative to any oversupply or issues. It means that the mature foundries like SMIC or HH Group, they're operating based on the economy, right? So no different than a GlobalFoundries or UMC. When the market slowed down, they slowed down as well. The power market, on the other hand, there's a couple of things to note, and Jamie commented a little bit, but the Chinese EV market is at about 30% penetration, expected to move up towards 50%.
There's a lot of activity in China pushing to increase the domestic content of those vehicles from around 10%-25% over the next couple of years. So we're seeing on the power side a couple of trends. One is the car companies in China are moving from silicon IGBT to silicon carbide. So there's significant investment there to make that migration. The fact that they're supplying such a small percentage of their needs right now suggests that they're not in an overcapacity position. They've got a lot of catching up to do. The second piece is the Chinese automakers are beginning serious export activity of their electric vehicles to sort of, we'll call it unprotected areas. So probably won't find a BYD dealership in California, but you do in many other parts of the country, of the world. And so that's something that's part of their national plan.
And again, they want to self-source as much as possible. Second thing, if we look down their supply chain in silicon carbide and in silicon IGBT, is they do want to be a global low-cost supplier, both of substrates and devices and probably inverters as well. And so again, that sort of goes to sort of the unprotected areas. A lot of times in the U.S., we see the protections that are put in place. We see that in parts of Europe, Japan, U.S., but that doesn't include Africa and South America. And there's a lot of parts of the world that we don't think about when we think about those protections. Those are all places that are going to be looking for the low-cost part. So we see a significant growth period continuing in China, especially in the power market.
What would you say about the competitive dynamic today? You said Chinese or China increased their local self-sufficiency. Would that also apply to ion implant tools themselves? Are you seeing any threats from local vendors there?
Yeah. So there are a couple of implant companies in China. They've been around for more than 20 years. And one came from a combination of two of the government research labs. About 10 years ago, they put together and created a company. Another one was more of a private enterprise. Both companies have produced ion implanters that are in some of our customers. They're pretty low quality and not capable. They don't have the capabilities that we or our primary competitor have in terms of building these more sophisticated devices. And so we watch it, but we don't expect them to catch up quickly. Implant is a very complex tool. We would liken it to the complexity of a lithography tool versus a chamber-based basic etch or dep system.
So we've found that they've been more successful in those chamber-based systems where the chemistry and the recipe is the more difficult part and the machine itself is relatively simple. An ion implanter, the beamline and all that are very sophisticated and very difficult to just go in, as an engineer would say, and take calipers to it and measure everything and try to copy it without understanding why you did certain things. And then even maybe more important is the algorithms that we use to tune beams and to run the system and get the ion from the source down to the wafer with high level of purity, angle control, dose control. And those are very difficult to master. They're 30+, 45+ years of learning. That's all in know-how. There's not a lot of IP that somebody can just go to the patent office and copy.
We find that that's the difficult piece. The recipe for an implanter is pretty simple. You dial in a dose and energy and an angle, tell it what ion to implant and press the go button. But the rest of it is really complex, so.
Would you be able to talk about your U.S., and Japanese competitors then? I bring this up because Gartner's latest WFE data shows that they gained a little bit of ion implant share last year despite you guys doing so well, record revenue, huge sales increase in power devices. Are they gaining share from you or is it just the other markets?
Yeah. So we do a detailed calculation of market size and TAM. We are a global company with a pretty significant implant footprint. So our calculation says the implant TAM is about $2.7 billion. The Gartner number you're referring to was closer to $3.5 billion. And so we have a great deal of respect for Gartner. But in this particular case, we find it hard to believe that there's $800 million worth of implanters that we didn't know about. And so I'm not sure what their methodology is for this past year. What we do know is Axcelis is the only company that publicly discloses our exact system sales. And so where the rest of the numbers come from for that number, don't know. There are other research firms out there that are more aligned with where Axcelis is. And we've had discussions with them.
So it's not our belief that they haven't gained share. The way the market is divided up right now, Axcelis has very high market share in power markets and in the mature markets. They have very high market share in the advanced logic space. We split the market in memory between the two companies. And on the whole, we estimate that we're based on this 2.7 number. We estimate that we're in the 30s% in terms of market share and they're in the 50s% with our two Japanese competitors making up the other 15%-20%. Yeah.
Makes sense. Let's talk about memory then. You said a recovery is likely to begin in Q4 for you guys. The rest of the industry, I'll say the end markets, the DRAM vendors, also some of the other equipment vendors, they've been seeing the recovery a little bit earlier. So would you say this timing difference is just because of the nature of your tools, or is it because we're seeing lower memory implant intensity? Is this more of a structural change?
Yeah. It's all starting, but I can add in, I'm a non-technologist, but HBM at the end of the day is a form of DRAM. And so ultimately what's happening here is the die size is bigger, about 40% or so bigger. Yields are a little bit lower. And so as the memory makers are pushing wafers through the process to develop HBM, they're using up capacity today, right? And so that capacity utilization is great for some of the capital equipment providers. Specifically, if we think about advanced packaging and inspection technologies that are necessary to actually create the stacking and ultimately create the HBM. So what we really need for us to see recovery is wafer starts.
And so what we're kind of, as we look at the market very broadly, what we see is the first step of this phase is how do we start to chew up some of the excess capacity that exists today? HBM is doing that at likely a faster rate than you would have otherwise seen in a traditional DRAM recovery. In addition to that, with the higher ASPs on that and the increased profitability, we're also seeing the DRAM manufacturers sort of recover the lost profit from the last downturn at a faster rate than they otherwise would, given the profitability of those HBM systems or tool or devices to them. And then the last bit is we think about what, again, this is sort of an ancillary impact to AI and what AI could ultimately mean.
We look at the sort of trends in terms of what baseline requirements may be for, I'll call it AI PCs and others, and those are being specced anywhere from 16GB-32 GB all the way up to potentially 64 GB worth of RAM versus the traditional 4GB-8GB that we would see in a baseline PC today. So we think about the need to now increase potential capacity to satisfy a growing DRAM need in the future while utilizing the existing DRAM capacity to be able to manufacture the high bandwidth memory. And so ultimately what we need to see is wafer starts. Wafer starts are ultimately where we're going to see incremental benefit from an implant technology as it relates to intensity.
This isn't necessarily, as Doug had noted, we're not necessarily shrinking the size of DRAM going forward, and we're not changing the implant intensity as a result of HBM coming online either. So ultimately, as we look ahead, it is going to be those wafer starts. We are starting to see a little bit of opportunity in the marketplace relative to some announced expansions with SK hynix and the M15 expansion that they announced. They pulled that into a 2025 timeframe. There could be some other indications coming out of Samsung that they may be looking now to try to recapture market share. So all of these things lead us to believe that maybe wafer starts are sooner rather than later, but they do position the memory market to be very strong, potentially going into 2025 and ultimately into 2026.
Yeah. Let me just add a little bit to what we see as potentially the strength coming out. So traditionally, in a downturn, the memory makers do a shrink, right? So they've got excess capacity. They shrink. That even makes more excess capacity, right? So more bits per wafer. And then wait for demand to absorb that capacity. This is the first time in my memory that we've had a larger die size that basically is carrying the same number of bits. And so coming out of this, we're going to see HBM with the larger die size probably counteract the shrink that has happened to some degree. And so we're probably going to see more substantial DRAM recovery as a result of that.
So at our peak for Axcelis, we've had roughly $130 million or so of revenue from memory during the past couple of peaks, split pretty evenly between DRAM and NAND. We would expect we haven't lost any market share there. We would expect that if the market recovers stronger, then we would have good opportunity in that market.
Got it. I did want to save some time for the financials. You laid out a 2025 model. I assume the sales portion might have been pushed out a little bit, but you've been pretty much on target with gross margins. But with this lower sales run rate, how should we think about OpEx? What's going to help you get to your OpEx potential target?
Yeah. So maybe I'll reference gross margin. We finished last year at about 43.5% gross margins, very strong performance for the organization and our goal to really be at plus 45% as we look ahead into the long range periods. For Q1, we actually delivered a 46% gross margin benefited by the aftermarket mix that we saw. We saw a higher proportion of our sales weighted towards our higher margin and accretive CS&I aftermarket business providing some uplift, as well as the systems mix. So our high energy tools made up a larger proportion of our system sales, and those typically carry higher margins for us in the period.
I think the promising thing about that is it shows that the efforts we've taken over the last number of years to drive out cost inside of the organization, both from a cost of sales perspective through finding alternative sources and low-cost supplied components to continue to support the growth of the business, to the investments we've made in our Asian operations center to have units and systems assembled and put ourselves a little bit closer to local supply in Korea, as well as some investments we're making now in our logistics center to reduce the cost of handling the material before we get it to the shop floor for assembly. All of those things are starting to pay off in our cost structure, which is going to drive improved opportunities in gross margin. On the OpEx side, it's a scale, right?
We've made investments in new technologies in order to be able to grow with the business as it grows without necessarily having to add incremental heads. We've been very thoughtful about making sure that as we add new positions, despite the growth, we're trying to be very thoughtful about making sure that we don't add for the sake of adding, right? That we're adding in areas where we see real opportunity for incremental investments and for the value of those investments to pay off, while at the same time making sure that we continue to feed the business the resources it needs to hit our long-term growth goals.
I know we're running out of time, but if I may squeeze one last one. You're having an Investor Day here in San Francisco next month. We wouldn't want a full spoiler for the show, but if you could give a trailer, that would be helpful.
Yeah. So the goal there is we've got a long-range model that is roughly six months out at this point. So our goal here is to reset the long-term model, which we believe has incremental organic growth opportunities inside of it. We want to talk about the end markets and where we see those end markets going and the opportunity for us within those. We're going to talk about some incremental technologies that we're working on today to be able to serve those end markets. And we're going to provide a little bit of context around, specifically as we think about how we expand our presence today in areas like advanced logic in Japan, where we're currently kind of underserving those markets relative to the potential opportunity that would be provided for us given the capabilities of our tool sets. So we're going to have information around all of that.
It's going to be, as we said, July 11th here in San Francisco. So not at the St. Francis, but at the St. Regis. So I guess we're nearby. And looking forward to seeing some of you or all of you or most of you there, right? Depending on what that looks like.
Sounds good.
Yeah.
Well, thank you, Jamie. Thank you, Doug.
Yeah. Thanks, Duksan.