All right, good morning. Welcome to the Axcelis Investor Event. Appreciate everybody coming today. A lot of you have talked to us two, three, four times this week, so looking forward to sharing some new information. I'm gonna just quickly run through sort of some logistics. So what to expect? First of all, we have the full management team. Russell's gonna introduce you to everybody in just a couple of minutes. The presentation is about 90 minutes long, and we'll have about 30 minutes of Q&A following that. Please hold your questions till after the presentations, if you would. As with our past Axcelis investor events, we have a little takeaway gift.
In this case, it's a cool little charging device with three positions, you know, to charge phones and watches and various things. But mostly to optimize your travel backpack and, you know, remind you, when you set it up on the nightstand, that Axcelis uses, you know, three all of our Purion products to optimize our customers' fabs and we're especially strong in power. Now, no, this is not silicon carbide, but it represents that, so. So anyway, thank you very much again. I have to remind everybody that we will be discussing mostly forward-looking statements in today's presentation.
I'm not gonna read this whole thing, just encourage you to look at our SEC filings and 10-Ks to understand the risks associated with our business and everything we're discussing today that's forward-looking is based on our current knowledge of the markets and situations. So anyway, with that, I'm going to turn it over to Russell Low, our President and CEO.
Great. Thanks, Doug. Oh, never got the, A little help? Okay, right, we're off and going. Okay, well, good morning, everybody. Really great to see you all. I wanted to introduce the presenters for today. So, Greg Redenbaugh, hopefully, you've had a chance to meet Greg. He's our head of marketing. He's gonna talk a little bit about the how do we get to the new model. Jamie, our CFO, will talk about, okay, so what's it gonna look like financially when we get to these models? And, Doug will be the moderator. I just wanted to mention, so I think many of you know that, Doug is retiring this month. He's been with Axcelis a long time, and you guys know him very, very well.
So naturally, we wish Doug the absolute best in retirement. And hopefully, you've actually had a chance to meet Dave Ryzhik, who is our new IR person, and he'll be taking over. He's been with us a couple of weeks now, so he's all ready to go. Oh, I realize it's the really big button with a forward arrow on it, isn't it? Okay, I see what's happening. So I will be taking you through some of the secular growth drivers of our business, how we are positioned within those markets, and then what we believe will be the opportunities for us. So, you know, this is our investment case. So, I'll take you through the presentation. I'll talk to you a little bit about how foundational ion implantation is. You really cannot manufacture a device without ion implantation.
It's a critical step. There's a number of really strong growth drivers. So, we'll be talking a little bit about electrification and, you know, artificial intelligence. We'll talk about the fact that these growth trends are gonna impact pretty much every type of device, and we'll talk a little bit about how we are positioned within those markets to take advantage of that. And then the last two pieces, Jamie will talk about what that means in terms of the financials. So that's kind of the outline of our investment case. So hopefully, at the end of this, you'll see all the pieces come together. So a little bit of an introduction for those of you that aren't as familiar with Axcelis as others. So we solely focus on ion implantation.
We design them, we develop them, we manufacture them, and then we also support them in the aftermarket once they've been shipped. We've been around for over 45 years. We have 1,600 employees globally, so everywhere where our customers are, we are, so we have a truly global infrastructure. And last year was our record revenue year. We achieved over $1.1 billion of revenue last year, and you can see that that's actually flowed very nicely through the P&L. So, you can see some of the financial metrics that we achieved. So we're actually pretty excited about that. So many of you are very familiar with the semiconductor, you know, industry. These are the basic building blocks, really, for building devices, whether either in the substrate or on top of the substrate, this is what you need.
And obviously, these devices are teeny, teeny, tiny, and we've got to do this with really high yield. So ion implantation is a critical step. Like I've said before, you really can't make any devices without ion implantation. I can think of maybe one device, like a pHEMT, that doesn't have too much ion implantation. And if any of you are saying, "What, what's a pHEMT?" Exactly. So, you know, and then when I talk about ion implantation, it's not just one size fits all. We have a platform of products, so when a customer is looking to develop devices, ultimately, they're gonna need a portfolio of implanters. And we'll talk a little bit more about that in a moment as well. So this is one of my favorite slides.
One, I have this dream that one day my kids are gonna ask me, "Hey, Dad, what's an ion implanter?" Hasn't happened so far. So, so we basically introduce material into substrates. So those substrates can be, you know, gallium nitride, they can be silicon carbide, they could be silicon, but we're basically introducing material into those substrates incredibly precisely. And, you know, what that does is it can change the, the electrical characteristics or the physical characteristics. You know, when I was a kid, we mostly did electrical changes. So, you know, to form the transistor with the, the source, the drain, the counterdoping, that was the device.
Fast forward, there's a whole bunch of material implants, where you're basically changing the physical characterizations, introducing stress and strain and amorphization for various reasons to optimize the performance of the device. So we really are implanting these tiny ions into substrates, and we can change the depth they go into, and we can also change the amount of dose we put into those wafers. These tools, I think I should mention, are about the size of a double garage. So that guy standing in front of that machine is probably a good 6 foot, and you can see the FOUPs in the front. Those are 300 millimeter wafer FOUPs. You can see how big this machine is. I kind of think that we have more in common with a litho tool than we do, say, a cluster tool.
So, you know, this is really large amounts of cool hardware that's driven by physics. You know, the recipes are relatively trivial: dose, energy, species, but the hardware is really complex. On the other side, you say a cluster tool, the hardware is relatively simple. You can put your arms around it, right? But the recipes are incredibly complex, right? That's just like pure black magic chemistry going on there. So, I think of those are kind of differences. Although, obviously, one very different aspect between an ion implanter and a litho tool is that we are precise and fast. So that's an unusual combination. So you can be really productive and not precise, so think of a, like a batch furnace, or you can be incredibly precise and slow, so think about ALD, but it's unusual that you can be incredibly productive and precise.
So that's kind of why the ion implanter gets used around the fab to solve other issues that weren't. You know, that's how we moved from electrical to physical. We, you know, we could solve issues around the fab that required material modifications, and that was great for an implanter because it was so precise. And it'd be remiss for me to say that we, like I mentioned, there are multiple flavors of ion implanters. You know, so if you're gonna build a fab and start a device, you will need a portfolio of products. We've built our portfolio on the Purion platform. The Purion platform is, you know, commonality across the systems, right?
There's the same control software, ion source technology, end-station wafer handling, and that has, you know, huge advantages for our customers, and it also has huge advantages for us as a company, and we'll talk a little bit more about some of the derivative products we've been developing and how those come from this, this, this platform. So, so where are these devices that we help fabricate? They're everywhere. They absolutely are everywhere. So, you know, we like to say our ions are everywhere. So whether they are in the Internet of Things, you know, I don't actually have a refrigerator that talks to me yet, but I think it's coming.
My doorbell doesn't do that yet, but every single pump in my house, having had been flooded recently, I now have water sensors and pumps everywhere, and they're constantly telling me what's going on with my house. Obviously, if I'm in California, I can't do anything about it. Automotive, we talk a lot about the electrification of autos. They, they are everywhere, and we'll talk about those certainly in a few minutes. Mobile, all of your phones, all of your, all of your PCs, advanced compute, all the way through to clean energy, and we'll talk a little bit about clean energy in a moment as well, but our ions are everywhere. So I'll talk a little bit about the growth now then. So, so we have multiple initiatives across, all of the markets. So, you know, clearly, power devices is a growing opportunity.
We believe it's in the first innings, and you know, we wanna hold on to all of the customers we have and add all the new customers that get into this area, whether they be GaN, silicon carbide, you know, this is an area that we have significant opportunity, while also having significantly differentiated products. A number of you have heard us talk about geographic expansion. So, you know, clearly, we have a very strong business in China, and China continues to buy a lot of chips. We are looking to get a bigger footprint in Japan.
Japan is, you know, getting into the power markets, and that's really good for us because we have the full portfolio, allowing us to sell to the Japanese customers where their local suppliers are unable to do that. And then, as ever, once we get into these accounts with a product, where we're demonstrating a unique value with the product, we then start to fan out to taking all the other applications, and then we try and bring our other products in as well. So we kind of, you know, we penetrate, and then we fan out. So that's kind of the growing footprint. So, you know, we are in every major customer, and, you know, once we've.
As you know, we are the market share leader for high energy, and we've always used that as our kind of opportunity to get in and then start bringing out the rest of the portfolio of our product, particularly high current and medium current. But along the way, as we're kind of like working with our customers to find that application where we can add significant value, sometimes we find new implant applications, and that typically comes through interacting with the R&D people, either through TRMs or evaluations, and Greg will talk a little bit more about that in a second. Very simple slide with a really strong message. The next few slides are gonna focus on two big, somewhat related long-term growth drivers, namely electrification and artificial intelligence, reviewed specifically from the power market.
So I don't really think at this stage we have much choice about reducing greenhouse gas emissions. I think Mary Puma, our prior CEO and board chairman, chairperson of the board, said, "You know, it's no longer climate warming, you know, global warming, it's global boiling." I think there's something to be said for that. So we really do need to have cleaner electricity, cleaner transport of that electricity, and then use less of it. So we'll be talking a little bit more about, you know, generating that clean energy, transporting it, and using it sparingly. Clearly, renewables and the use of power devices, particularly silicon carbide power devices, can certainly go a long way in driving us along that path to generating clean energy and using less of it. AI.
So as this graph shows, data centers already use an increasing amount of energy. You know, that's the blue part of this. And compute-intensive applications, such as crypto mining and now AI data centers, are accelerating that demand for power. So AI, in addition to being a compute hog, and it wants the best and fastest chips, to being a data hog, it wants all of your information, is also, you can see, a power hog. The power needs are putting additional strain on our abilities as a nation to support the power generation. I mean, power efficiency is absolutely key here.
So even though, say, NVIDIA and AMD are talking about chips that are gonna be less power hungry, they're still using a huge amount of energy, and we need to reduce the amount of energy. So, I think, you know, recently, for example, Infineon and onsemi announced silicon carbide chips, like 400-volt chips, specifically aimed at data center. So I'm not talking about inside the boxes, I'm talking about the power infrastructure. So, you know, a huge amount of power is coming into these data centers and being used in addition to all of the battery backup. So these, these systems have the kind of the UPS, uninterrupted power supplies, that all uses lots and lots of power devices.
And if you specifically use a silicon carbide device, you get a 10% bang straight there and then for power usage. So you know, I think as I kind of like said that, you know, it's a, You know, AI is a data hog, it's a compute hog, and a power hog. AI is gonna impact every single type of device. It's not just NVIDIA. It really isn't, or the HBM. It's every single type of device can get impacted by AI. And you know, although there can be a little bit of hype around AI, it's not, it's not going away. There are real business applications driving value for these, for this technology. Renewables. So I'll talk a little bit about renewables. So globally, we continue to add more renewable energy.
So you can kinda see here that it's going up and up and up. I'm not sure if I've ever mentioned this before, but Beverly, our headquarters, actually is 100% renewables. We use hydropower from Canada, I think it is. Korean manufacturing site will be 100% renewables by the end of this year, I think. So, you know, we are definitely all on board with renewables. But the reason I mention that is look, we're not alone. So if you look at the hyperscalers, you know, Microsoft announced a $10 billion investment recently for renewables to support their business. Amazon have been quite vocal, saying they're gonna be 100% renewables in 2025. This is really happening. We need more energy, and we need to reduce the carbon footprint.
So, you know, this really is happening. Interestingly, all these advances made in storage, and I'm kinda like talking about battery, whether it be a data center, battery backup, or car batteries, we're making lots of advances in batteries. So now those systems, you know, solar or wind, coupled with a battery, are now becoming an awful lot more valuable. So when people talk about, "Well, the sun goes out, what are you gonna do?" Well, you basically fill these batteries up, and it's becoming actually a very good standalone technology. Our very own Doug Lawson has on his island installed a very large solar array with all of its batteries. So when Armageddon comes, he's good. He's got his own little system.
You know, we all know that there are always gonna be power interruptions, whether it be on the islands, where you have hurricanes blowing through or, you know, there's too much load, like in Texas. You know, these systems are great because not only do they have the storage to support the cyclicality of the wind or the solar, but they also mean you have a backup. They're not glitching. So that's a certain value to those things. You know, once again, all of these things, solar, wind, battery backup, infrastructure, all going to use power devices. And if you use silicon carbide, those power devices are going to be 10% more efficiency. Silicon carbide is an incredible material.
I mean, the density of current that it can take is superior than pretty much any other material. So I left automotive to the end. You know, this is kind of like a slide, just on kind of like automotive. So, this, so automotive is the killer app for silicon carbide. It's driving really large volumes of silicon carbide, which, you know, as we know from other technologies, that drives the price down and down and down. So that's why I'm excited about seeing silicon carbide turning up in all sorts of other applications. So absolutely, electric cars and quite honestly, hybrid cars, electric cars, we are somewhat agnostic because they use silicon IGBTs, or they use silicon carbide, MOS transistors. You know, they still use a large amount of ion implantation steps.
But I do think that, you know, silicon carbide specifically, it's an amazing material, and it's going to find new applications. I mean, kind of just anecdotally, I was talking to a customer yesterday, and they mentioned, "Well, you know, what if you could, you know, digitize the electricity net?" So, you know, you've all seen those really large transistors sitting on poles by the power cables. They could be replaced with silicon carbide, and you get a 10% improvement. He also mentioned about, like, large motors. You know, so take an HVAC motor. There's lots of those around, really high current motors. Okay, in time, as the price comes down, you know, you'll start to see lots of applications.
And the reason I mention that is 'cause I think, you know, that's very exciting for Axcelis, but we've seen this before. I remember, probably wasn't that long ago, maybe 15 years ago, I replaced all my bulbs in my house with LED bulbs. I went down to Home Depot and had a heart attack about the price. And now they're in kids' toys. They're, you know, they're in McDonald's Happy Meals. They're just everywhere. I mean, we've almost given LEDs as kind of like they're just ubiquitous now, but in the old days, like, "Holy moly, look at this!" So I do, I do think, you know, these things will come down dramatically in price. Every single time the price drops, there's a new parity point for a new application to be transferred out.
I do believe that silicon carbide uses a lot less energy, and it's going to be efficient for us. So, you know, so, you know, and then just to kind of talk about cars. Cars, electric cars, people kind of saying, "Well, you know, they're slowing down." Yeah, yeah, they are, but, like, that's relative to 60% growth versus now they're 25. 25% growth in electric cars is still a really big number, and that's globally. So, you know, look outside of the US. China is about 30% penetrated, it's getting to 50% really soon. Electric cars are taking off, really dramatically. And, you know, hybrid cars as well, like I said, we're relatively agnostic, 'cause they still use a lot of power devices, and I think hybrids will be a great solution for the US.
I think it takes away the range anxiety, and I think it takes away concerns regarding, say, charging network. So I think, you know, we've heard that the Camry, the best-selling car in the world, is going to only be available as a hybrid in 2025. So these things are coming. And so Jamie bought a hybrid, I bought an electric, and I think Jamie is now like, "Oh, I'm not worried about the end anxiety. I want electric now." And I got to tell you, I love my electric car. I love the acceleration; it's so much fun. And I love the fact that living in Massachusetts, I don't have to stand out in the cold, filling it up anymore. So, you know, I do think.
Yeah, just, like, plug it in and walk away. It's great. But, you know, I really do think that these trends are continuing to happen. They're not slowing down. Sorry, slowing down. They're still growing at 25%, though. So as I mentioned, artificial intelligence is going to impact every type of device. So we talked about it's a compute hog, so it wants the fastest processors, it wants the most amount of memory, like HBM memory. It's going to want to store all that wonderful stuff that it creates. So that's going to touch advanced logic, DRAM, NAND.
It also wants all of your data, so it wants the information from your fridge or your doorbell, but it also wants it all from your phone and from your laptop and wherever this data is coming from, and it's got to transfer all of that data. So it's going to store it locally, so there's even more memory and storage, and then it's going to transfer it. So it's going to be communication, it's going to be Wi-Fi, it's going to be mobile networks, it's going to be Bluetooth, all of those, mature nodes. And then, as I mentioned, it's a power hog. It's going to need more and more energy, and we need to reduce the amount of energy that it needs. Electrification, so we talked about that. So that's why I kind of say they're, they're kind of interrelated. Clean energy.
Clean energy, that stuff's, you know, solar, it's DC, windmills. Wherever there is high current, any step up, step down, any conversion, you're going to see power devices. So, you know, solar, wind turbines, the battery storage, really good for silicon carbide and silicon IGBTs. Cars, you're going to see, you know, hybrids may have more silicon IGBTs initially, and the silicon carbide might be in more of the battery ones. That mix is going to change over time, but it still uses a lot of power devices. And then finally, you know, cars. They really are basically a computer on wheels. So you're gonna still see the info systems, you're gonna see the driver, the autonomous driving features. All of those are gonna be in every car, and in addition, you're gonna add in these drivetrains.
The nice thing about the drivetrains, whether it be some form of hybrid, some form of battery, they're all taking away market share from combustion engines, right? They're not competing with each other, which is why we say, you know, it, it's great to see this growth. So this is my last slide. So to summarize, I really do believe we're well-placed with our Purion Platform to take advantage of these two long-range secular drivers, you know, advances in artificial intelligence and power. And when I say power, I'm trying to get across electrification in general. Whereas cars, which people think is power, yes, they are. They are the big application. There's so much more out there as well. So, you know, so I think we're very well placed. There are two large drivers.
I think that's gonna drive growth in every type of device out there. So it's gonna be power, it's gonna be memory, it's gonna be mature, it's gonna be advanced logic, and that's also gonna support geographic expansion as well. So with all of that, you know, we believe we can achieve a $1.6 billion revenue target in the 2027 timeframe. So, you know, with that, I'm gonna hand you over to Greg, who's gonna tell you a little bit more about how do we get there. Thank you.
Morning, everybody. I'm Greg Redenbaugh. Good to see you all. Thank you for joining us today. Really appreciate it. So Russell talked about our position in the market and the direction we're going as a company. I'm gonna spend some time talking about a little bit more detail on the markets, and then the path that we're gonna take to achieve that target that he just talked about. So we're gonna talk about market drivers and positioning for future growth. So a good place to start off is just a reminder about the way we think about markets, and this is an example of our revenue by device and market segment. So our 2023 systems revenue last year was $884 million, and the way we break that down is really into two big categories.
We break it down into mature nodes, which include power devices, like Russell just talked about. So that includes silicon IGBT, as well as silicon carbide. It includes general mature and also image sensor. That mature nodes portion of the revenue made up about 88% of our revenue last year. The other 12% was comprised of memory and image sensor. So this is the way that we think about our market and the way that we break things down, and this is how we're gonna focus on the market opportunity going forward. So if you take a look at the overall ion implant TAM, you know, it's really more than doubled in the last several years, and it continues to grow.
So for those of us who've been involved in ion implant for a long time, you know, it used to be that the overall market would cycle around $1 billion. But starting about 5 years ago, the market really started to take off. And this is really driven by several things. It's driven by increased wafer starts across all segments, but it's also especially driven by the rapid growth of implant-intensive mature nodes, so power and image sensor devices as well. So that's been one of the main drivers. So if you look at the TAM, we see it being around $2.5 billion this year and last year, with the opportunity to grow in 2027 up to $3.3 billion.
So the opportunity for Axcelis is that we can bring those high-value, Purion product extensions that Russell talked about, that are required by specialty markets like power devices and image sensors. So a very strong market growth. If we take a little bit more of a detailed look again at the ion implant TAM segment breakdown, again, the mature process technology segment is really driving this, market growth. And it's growing, that mature process technology is growing based on increases in wafer starts, high implant capital intensity, which I'm gonna talk about, as well as higher value implanters for specialty applications and those product extensions. So that part of the mature process technology makes up about 70% of the overall implant TAM.
Memory, which makes up about 16% of the market, is expected to be driven by new wafer start activity to support bit growth demand. And then advanced logic, which is about 13% of the market, is really being increased based on accelerating wafer start growth. So if you look within the mature process technology, the power device portion, it says it's about 36%. We'll talk more about how that breaks down between silicon and silicon carbide. But it's a good reminder, as Russell focused on the opportunity and the growth possibility with it, continuing growth possibility in different applications for silicon carbide. Those devices are expected to grow within the power device segment, driven by that end market growth that he talked about, device performance enhancements, as well as fab optimization opportunities. So this is the market segment breakdown.
So just taking a step back, maybe in a really basic way, we think about in terms—we think about things in terms of where is ion implant used and how do our customers use it to make their products? An ion implant is used in all of these different market and device segments. So starting on the left-hand side, Advanced Logic. The way to think about that is that's, those are the devices that do the math, that do the computations. So the types of devices that fit in that category would be microprocessors, also known as CPUs, GPUs, which you've probably heard a lot about, and they typically contain billions of transistors, so very complicated. The device technology is fairly advanced, so it's less than five nanometers in terms of technology node.
These advanced logic devices are increasingly being used in data servers, as shown in the picture below. So a lot of growth potentially around artificial intelligence, which we've heard a lot about. So that's Advanced Logic. Memory, as the name suggests, you know, saves the data. And so typical devices here would be DRAM, NAND, which I think we've all heard about, and these are also devices that contain billions of transistors. Fairly sophisticated technology involved there, and these are used any place you want to save information. So desktop computers, laptops, our phones, everywhere we need to save things. So memory saves the data, obviously. So power devices, as we've talked about, are a little bit different. As the name suggests, power devices manage the power.
What we show here, but the difference here is that power devices are a single transistor, and unlike memory and advanced logic, which are typically done in silicon, there's a lot of growth in silicon carbide, which is a pretty amazing material. And so single transistor controls power and is used in EVs, industrial applications, and growing areas, as Russell talked about. It's an amazing material. I had a customer visit us in Massachusetts, and in this case, he brought an example of one of these transistors out of the package, so you could actually see the small chip.
It was about 2 millimeters on the side, and when I asked him, "You know, how much current can this little 2 millimeter on the side device handle?" And he said, "Oh, about 150 amps." So that's about the same amount of current that your entire household would take in a very small device. So the use of these things in inverters and EVs is really, obviously empowering. So that's power devices. Moving on to image sensors, or also known as CIS, CMOS image sensors. These are the things that allow us to take the picture. An ion implant is used in making the pixels, which are shown here on the slide. And we're all very familiar, I think, with CIS devices because we have them in our phones today, so very prevalent.
This is where all the data gets pulled in to what is the next segment, mature logic. So mature logic are the devices that control, collect, and convert information. So everything from RF signals to analog, processing to digital processing, and while these are sophisticated, they're not typically used or built with the most advanced, process technology. So we look at this as more like a 28 nanometer node. Very implant intensive, as I'll talk about, but these things are everywhere. So they're everything from washing machines to automobiles to, you know, lawnmowers that are self-driving, and they also connect us wirelessly to many things. So this is how we think about how our ion implant enables our customers' products.
So when we talk about growth, our growth roadmap is really aligned with the market segment opportunities in, broken down in the way that I just talked about. So in addition to wafer start growth, different parts of the market actually are more intensive in terms of ion implant versus others. So what we show here is the market segment breakdown, the estimated percentage of implant CapEx, and some examples of leading chip makers that exist in this space. We've also calculated the capital intensity, in other words, the number of implanters that would be required in a typical facility or fab that are building these devices, relatively normalized. And what you can see here is that, especially for power devices, but also for mature process technology nodes, it's very implant intensive. So these facilities would need between 50 and 70 implanters, which is great, power devices.
Similarly, the mature process technology needs a large number of implanters as well. Memory needs a fair number, advanced logic, a little bit less. All of these device segments require implant to help, define the device, but especially in power devices and mature process technology, that's the case. So just a few slides on the power market and the opportunity. So this is a breakdown of the overall, power market, the end-use market, which we see growing over the next several years to $33 billion, with a compound annual growth rate around 8%. This market is broken down into silicon, which is the predominant part of the market today, silicon carbide and gallium nitride.
If you look at the end use of the devices that are power devices, the top one, of course, is automotive and mobility, followed by industrial applications, and then it goes down from there. The other interesting thing to note is if you look at the silicon carbide portion of the overall power market, that's growing faster. That's growing more like 25% over the same relative timeframe. And the automotive and mobility will account for more than 80% of that silicon carbide market. And as Russell talked about, battery electric vehicles will remain the main market driver for silicon carbide. If we zero in a little bit more on silicon carbide as the market and talk about that 25% growth, it's expected to grow to be more than $10 billion by 2029, again, driven by automotive, mobility, and industrial applications.
But the other interesting thing is that silicon carbide, being implant intensive, you know, requires a significant investment, and the growth of implant over the timeframe is expected to be almost 10%. So it's not only implant intensive, but there's also a very healthy growth rate as we see it going forward. Okay. I think as Russell mentioned, we are the power leader in ion implant for power devices. We started working on developing solutions for our customers more than 10 years ago in the medium current space and also in the silicon carbide space. So we've been involved in this since 2014. We believe we've created a significant incumbency advantage, and we've developed a Purion Power Series product portfolio that covers all elements of ion implant.
So high energy, medium current, high current, and we've developed tools that really serve that entire range of what our customers need. These systems are production proven. They've been used for quite a while now, and we designed them to be flexible. So one of the benefits of working with customers for so long is we're able to identify what their specific needs are, and incorporate it into our product portfolio. The other key thing is that the Purion Power Series is extendable. So today, power devices are largely built on what's called a planar technology. But customers want to move to new device types like trench or super junction, which allow them to get both better performance and also scalability out of power devices.
So to show that extendibility, if you look at the graph in the upper right, this shows an example, kind of a representative example, of the kinds of implant recipes that our customers are using, broken out by the implant dose, how many ions customers want to put in, versus the implant energy, how far they want to put the ions into the substrate. So each one of those dots is an individual representative recipe, and we've overlaid the Purion Power Series product elements on top of that. And what you can see is two things: there's a very good overlap of what customers are looking for today, but as we see them wanting to go to higher dose and also higher energy to achieve better and scalable device performance, the Purion Power Series has product extensions that allow us to help them cover that as well.
So, the products are also extendable from a wafer size standpoint. So not only are customers moving from planar to trench, they're moving from 150 to 200 millimeter. You know, we've had good evidence of success in this market. We've worked very hard with customers in it, and you can see the graph in the lower right shows the revenue growth across all products that we've developed in this segment. So in short, we view the power market as really the basis of our business on which to grow. So Russell talked about some growth opportunities. The first one I'm going to talk about is Japan. So we have a full product offering that can address all of the available markets in Japan. It's a pretty big market.
It's $450 million a year, roughly, and we have a relatively modest penetration in Japan today, less than 10%. However, we have a good number of resources in-country working with customers, and we've been working with leading device manufacturers to evaluate technologies and products in a variety of different market segments. Recently, we've been winning applications over local ion implant competitors in the power market, given the strength of our Purion Power Series, like I just talked about. So we've had some good success there, but we also plan in power, but we also are looking to extend into image sensor, memory, and advanced logic. And recently, you may have seen the press release where we've expanded our footprint in Japan by opening two new offices, one up north in Hokkaido and one down south in Kumamoto.
So Japan is an area of opportunity of growth with a healthy, fairly big market. We don't have that strong of a presence, but we've got some good upside opportunity to grow into it. The other focus area for growth is Advanced Logic. So this is a market that grows with direct correlation to wafer starts and technology shrinks. It's also a pretty healthy size market. It's about $400 million a year, and we also don't have a really strong presence in that, honestly, we're less than 5% in this today. However, we have a product in Purion Dragon that we developed specifically to meet the needs of Advanced Logic customers.
If you take a look at some of the results of our current efforts on the right-hand side, we have what we call a learner system, which we have placed at a leading European Advanced Logic research institute, where we're working with that group to identify what things on customers' Advanced Logic roadmaps would be needed to be incorporated into our product most effectively. We also have what we call a learner, which is an evaluation system with a customer's Advanced Logic R&D center, where we've been working with them for more than a year now, identifying applications in the Advanced Logic space. And then as a reminder, we've had production systems in Advanced Logic facilities for a number of years already. So the Purion Dragon is being positioned for Advanced Logic.
We are developing extensions to the product, including upgrades, refinements, and new capabilities to serve this advanced logic space. So the timing for this, although we're getting some traction in it, in terms of meaningful revenue, is probably in the 3- to 5-year timeframe. Okay, so now I want to shift to talk a little bit more about the products and some of the customer challenges that we're addressing through new products and product extensions. So as we've said, I think before, Purion is really the core of our growth strategy, and we started developing this product platform again about 10 years ago, and we designed it so that we could address all elements of the high, of the, ion implant TAM, including high current, high energy, and medium current.
So we have a common Purion Platform on which we have core products that we can then build extensions to. We developed those extensions, like the Purion Power Series, to address specific market opportunities. We've also developed specific products for image sensors and, as I said, advanced memory and logic. So we use this product roadmap to really address several customer challenges. So in terms of those customer challenges that we're focused on, I'm going to go through four of them. Two of them are focused on power devices, and two of them are focused on advanced logic. And these are all customer challenges that we're working directly with customers today. The first two involve power devices. One is a silicon IGBT implant, a proton implant, and the second one is addressing that evolving device design that I talked about before.
In terms of advanced logic, we're looking for targeted new applications in advanced logic that we can use Purion Dragon with. And then the last one is a unique capability we've developed that I'll go through, that addresses both power and advanced logic, customers' challenges. So the first one is a power device solution, regarding a silicon IGBT backside implant. So we talked about the power market being broken into silicon carbide and silicon. One of the prominent silicon devices is IGBT, Insulated Gate Bipolar Junction Transistor. And this device, in order to work efficiently and to work with high frequency, benefits from having a backside implant with proton. So the customer challenges is how to increase the speed and reduce the power during the device operation. We've developed an optimized solution, which is based on a Purion high-energy platform.
It's based on a Purion VXE. It's a proton implanter, and that product is based on that Purion high-energy platform, as I said, of which we have more than 300 installed around the world today. So that's the product basis on which we're building this new solution. Our high-energy product is LINAC-based, so it's designed for higher reliability and throughput, and we've also developed optimized cooling. So these devices are made on thin wafers because of the way that they operate. So you have to be able to handle much thinner substrates, which we've also worked on carefully. So customers are very excited about this. We have systems in the field that we're working with customers on now, and the opportunity, we believe, for the served addressable market on an annual basis is more than $50 million.
So we're very excited about this, and customers are as well. So the second customer challenge area is that evolving power device design and the transition to new product requirements. So what we're showing on the right-hand side is a cross-section of the device, and this is showing a silicon carbide trench MOSFET power device, and it's showing different implant examples of how customers would use our product solutions to make the device. As customers are moving from planar to trench, they need to be able to better refine the device, meet, be able to reach deeper depths in the device to actually form it. So we've developed extensions to that Purion Power Series, which enable the customers to make these trench devices and eventually super junction devices to improve the device capability.
An example of that is the high-energy Purion EXE silicon carbide, part of the portfolio, and we've developed it to have higher throughput across the platform, new wafer heating technology, which customers need to make the devices, an updated ion source technology, so the source of the ions, going into the device and optimized cooling. And like the other Purion products, it is upgradable from 150-200 millimeters, customers make that transition. So, the goal here is to capture that device transition and to get to, as Russell said, the production fan-out of power devices using this, solution for the customer challenge. Okay. So transitioning from power devices, kind of the basis of our business, to advanced logic, where we're focusing on growth. The customer challenge here is that advanced logic devices require more materials modification, implants, and new applications.
Russell talked about electrical doping applications and material modification. This is one where we see an opportunity for potentially new applications for implant. And what we show on the right-hand side are examples of the number of implant steps per technology node, going from 7 nanometer down to 3 nanometer. If you look at it, there's a fair number in the front end of line, which is where implant has been used for a number of years for electrical doping. But there are growing and new applications in what is called middle of line and back end of line.
Again, our proposal and our plan is that we're using Purion Dragon with a scanned spot beam technology, which is a little bit different than what else is available on the market today, to be able to address these new applications in advanced logic. And Dragon brings productivity advantages, angle control advantages, and the ability for customers to control the uniformity and the angles of the implanted ions. So we see incremental opportunities for high current growth in advanced logic using this approach and focusing on these, this different part of the advanced logic process flow. Okay, the last example of customer challenges that we're focusing on to enable our growth helps both power and advanced logic customers. So the customer challenge here, it's a little bit technical, but not too bad.
Customers need to create what's called a box profile with uniform dopant concentration versus depth, with multiple implants. So the way to think about this is if you look at the graph on the lower right-hand side, this is showing the concentration of implanted ions as a function of depth, basically. What customers do today is they implant a series of different energy recipes into the wafer, which they can then add up to form that box-like profile that you can see on the graph. So it's called the chained implant. Customers like this 'cause it allows them to kind of refine and optimize their device performance, and chained implants have been around for quite a while, but they take a while to do, because for each one of those implants, you need to tune up the beam and stabilize it before you can actually execute the recipe.
So it's not always that cost-effective to do it this way. We've developed something novel, which we call MUSIC, which stands for Multiple Steps Implant Chain. It's a unique process capability to implant multiple energy recipes in a single recipe step. So this provides a significant throughput gain to customers, and you can see that in the graph in the upper right-hand side, which shows the comparison of throughput between a conventional chained implant and the approach using MUSIC. It can be tens of % faster. So customers are very excited about this possibility. Purion has an architectural entitlement, which allows us to deliver this to customers with this throughput advantage, and it's also IP protected. So the high-value opportunity is to be able to capture applications in both power and advanced logic, which use chained implants. Okay.
So, we've gone through a number of the customer challenges that we are tackling right now and have stepped through a couple of the power examples as well as advanced logic examples. So hopefully, in the time that we've gone through here, I've given you an idea of how we think about the market, where the market is growing, the specific targeted areas, both geographically with Japan and market segment-wise with advanced logic, that we're using to build on our power basis for growth. And with that, I will turn it over to Jamie Coogan, who's going to talk about finance. I should have said value creation.
Yeah, value creation. All right. So, you know, this morning, again, we talked through with Russell, you know, the, some of the macro events and drivers that we see that are going to drive some long-term growth. And, you know, Greg, I think, did a nice job of breaking down the sort of implant TAM, the technologies, how we fit into that space, and then ultimately, what we're doing to drive, you know, new opportunities through new products, extensions, and to continue to drive our technological moat. Fortunately, all that costs money, so we've got to figure out now how to pay for that.
Over time here, what we've done, you know, so before I go into some more details on the long-range model, we did allow Russell to kind of give you the top-line revenue number, but I saved some of the margins for myself. I want a little bit of the glory here today, too. But ultimately, what I want to do is show you where we've been. So, we have had a lot of new interest in the company. We've done a lot of onboarding meetings, and so we felt it was important through Russell's section, through Greg's sections, to be able to educate people on sort of what is an implanter, right? Why is it important, and ultimately, how do we fit into the space?
I think then it's also important to show you, you know, on a historical basis, where our financial performance has been, and, you know, before we get to where we think it can ultimately go. If you look at the chart here, you can see in 2019, we were approximately $343 million, right? At that time, the implant TAM was about $1 billion.
You know, we had begun the journey at this point already to work and partner with our customers on the challenges that they were seeing, specifically on the power side and specifically with silicon carbide, and really developed a novel solution here to be able to launch into the market and take that, you know, sort of lead in the space on silicon carbide, as both Russell and Greg talked about. As the implant TAM grew, right, it grew in part due to the addition of silicon carbide into the TAM, but it also grew given the sort of, I'll call it, explosion of general mature technologies. The implant TAM has more than doubled, right? It's grown to almost, you know, $3 billion over the course of this period of growth.
And with that, given the implant intensity in those general mature spaces, the need for our technology has really proliferated to help support that growth. And so you see here us growing to over $1 billion, which is record performance for the company in 2023. You know, as we look at the operating margin performance, you know, there's a couple of really nice stories here. You know, the team had done a lot of work over the last 10 years to drive gross margins significantly higher. But from 2019 to 2023, during supply chain challenges, right, during issues with COVID, rising costs, labor pressures, right, and the risk of inefficiency, we were still able to drive gross margins another 150 basis points improvement.
Given the growth in the business, we were able to leverage, right, our SG&A base more significantly, and we saw almost an 800 basis point improvement of SG&A as a percentage of sales over that period of time. We anticipate to continue to work through those improvements, add technology to our stack here to continue to drive efficiencies, and the team is working very closely with our suppliers to drive costs down on our systems and units as we go forward. Looking at how that dropped through to EPS, a couple of benefits here. One, just the improved profitability really did drive EPS. So operational improvement overall, saw us go from about $0.50 a share in 2019 to almost $7.50, right? So we're just shy of $7.50 here as we exited 2023.
That is, given the operational profit improvement and margin improvement that we saw over that period of time, helped drive that growth. In addition, we had share repurchase program, and we executed against that share repurchase program in a really kind of thoughtful and methodical manner. We initially designed to be able to offset dilution from management grants, but as you can see, we reduced the share count over that period by approximately 3.8 million shares. So we were opportunistic and continued to buy, continued to be steady in the market and buy on a systematic and rational basis over that period of time, which reduced the share count outstanding, which had then helped to drive the incremental earnings per share. Conversion to cash flow, though, is most important.
And you can see we've done a great job of driving incremental cash flow into the business. This has positioned us to be able to make incremental investments in the operations. You know, we opened up our Asian operations center over this time. We made the investments that were necessary to expand capacity. We opened up a new logistics center as well over that period of time. Those investments are gonna drive incremental efficiencies, which aren't fully realized in the P&L today. We expect that, you know, that reduced our warehousing footprint very significantly in the greater Beverly area, and we're now one facility with automated tools and systems, that are really driving the operational performance, making us more efficient to get product to the shop floor and more efficient to get that back out to the customers.
So really looking ahead, what do we see for the long-range model for each one of these components? As Russell noted, in the 2027 timeframe, we anticipate revenues to be approximately $1.6 billion. With that, you know, we do see, given the systems mix, given our CS&I revenue mix and those expectations of growth in the various end markets, we see margins at greater than 45%. That's about 150 basis point improvement from where we are today. We believe that we can get there. It's fairly consistent with our prior models, but, as many of you are aware, and for those that are new, system mix really does drive, you know, how we see margin performance. So where we're gonna see the growth is ultimately going to drive the margin performance overall.
You know, if we see higher high-energy tools, if we see, you know, outsized growth inside of our power device segment, if our image sensor business returns more dramatically than we are anticipating in the model, you could see performance in excess of the 45% that we're putting forward today. On the operating margins, we see that at 27%. So again, a nice basis point improvement there. That's continued leverage of the business. There's some offsets, though, that come with that, so we are gonna make investments in technology to increase our efficiency. With that's gonna come incremental depreciation and amortization. Our CapEx from 2019 to 2023 has more than doubled as well.
So those historical investments we made, we're sort of now gonna live with higher depreciation, so that non-cash expense is gonna be flowing through that operating margin in those future periods, and we expect that to continue as we make investments in the business. In addition, we've more than doubled our R&D spend from the 2019 to the 2023 timeframe. We're gonna be opportunistic, so it's roughly, kinda, call it 9%-10% of sales is, you know, what we're targeting on a more long-range basis. However, we wanna make sure Jamie's got as much money as he needs to be able to drive those incremental opportunities. And so as we identify those opportunities for the business, we're gonna allocate capital to those that have the highest and best return on the R&D side.
We can't get to this organic growth rate, right, without those investments, and so you will see us putting a little bit more money towards R&D, where necessary, in future periods at those high-value opportunities that Greg identified and talked about. Conservative estimates on tax of 15%. Really looking at that sort of global minimum tax here and trying to make sure we just stay aligned with that overall. And ultimately, that leads to greater than $11.50 estimate on earnings per share. Very conservative estimates on share repurchase authorization. We want to really let the operational performance of the business shine through as part of this long-range model.
But, you know, there probably is opportunity for us to go above and beyond the share repurchase estimates we have here, but we didn't want that to necessarily skew, you know, the performance of the business, relative to that long-range model. On the long range, we are looking at free cash flows of greater than 75%. We still think the business can do a wonderful job of generating cash, throwing off free cash flow over that period of time. We are gonna measure that off of EBITDA, on a go-forward basis, just to see how efficient our conversion ratios are. And so you're likely gonna start to see us, you know, talk through that level of performance, going forward.
We're still gonna be really focused, though, on making sure we keep a clean operating profit performance for you guys. That's a point of pride for us as we look ahead. You know, we wanna make sure that in the operating profit side, that we continue to be sort of true to who we are here relative to that. Growth in revenue really is gonna come from all the markets that Russell talked about, as well as Greg talked about. You know, we see.
You know, the approach we took to building this up was a bit of a bottoms-up based on where we see the TAM growing, what markets within that TAM are growing, where our penetration is with our customers very broadly, and ultimately, you know, what we think some of these new products, technologies, geographic expansions, and the push into advanced logic can do for us. We are anticipating, you know, certain recovery in memory spending, you know, kinda modestly above our prior historical highs within this space. So for those that aren't aware, approximately $130 million was our peak memory, you know, prior to the downturn in NAND.
We are anticipating, you know, memory to be a little bit higher than that, but not significantly higher than what we saw on the prior peak. On the image sensor recovery side, we expect this to come back to what we'll call more normalized levels. And then, honestly, with general mature technologies continuing to grow, we anticipate power to be very, durable for us with some modest growth expectations relative to how power is going to drive the business. There's CS&I revenue load, right? That's really driven by the number of Purion products that we have out in the field, as well as our GSD multi-wafer tools that are out there today. That we anticipate the continued addition of Purion power systems into the field will drive incremental CS&I revenue opportunities.
And we're also looking at ways to contract with our customers in a more efficient manner that allows us to capture a little bit more share of wallet associated, you know, with those products that are in the field today and improve the customer, you know, up times and yields and throughputs, you know, for those customers through these incremental contracting mechanisms. And so, you know, we see that as a contributory factor driving incremental revenue growth as well . . . . Going into CS&I a little bit, so for those that may not be aware of what that is, this is our aftermarket business. We think of this as our spares, our consumables, our upgrades, and from time to time, you're gonna see things, you know, the used tools can fly in there as well as some service revenues.
You know, given the growth that we've seen, so this is made up of the top chart here shows all of our installed tools. So this includes our Purion, our multi-wafer tool, which we call GSD, as well as some other legacy systems that Axcelis has had over time. The beauty of the systems that we put out there is that the lives of these are very long. You know, on average, you could expect them to be 10-20 years. We typically see 15-20-year lives on these tools. And once we're through installation and warranty, there's a you know, relative annuity stream that comes along with that.
The team's goal and efforts here has been to increase, right, the share of wallet associated with that aftermarket to the best of our ability, and we're gonna continue to drive that. The chart on the bottom shows that the additions. This is looking just at the multi-wafer, right? So this is just the implant systems that we have out there. And you can see the contribution as Purion continues to push out. You can see the relative contribution that we continue to receive on an annualized basis from the incremental units that we push out into the field. So, you know, this is a really nice, solid foundation of our business.
It's very margin accretive for us, so this is above, you know, consolidated margins in the aftermarket business, with certain parts of the, CS&I business, being, you know, very margin accretive. That's primarily associated with the upgrades, just given the, the efficiency and production, capabilities that we provide to our customers through that. The Purion product family was designed in a way that allows those to be field upgradable. And so, you know, as we identify upgrade opportunities, we do have the, the chance to go back to customers who bought those prior Purions, push those back through the Purion, and then bundle those upgrades together into a new product system and a new product family.
Allows us to drive incremental ASPs, launch new products into the field, and really satisfy those emerging, and future customer needs and requirements with the Purion. Operating margin expansion, I talked through this a little bit already, but, ultimately, what we do see is the mix, system mix between high energy, high current, and medium current are gonna drive the operating margin expansion on the gross margin line, as well as the growth in the CS&I business. We expect the incremental leverage that's gonna come from the sales opportunities, to flow through on the G&A line item, you know, offset in part by some incremental R&D spend over that period of time, as well as some incremental depreciation as we, you know, institute technologies to drive cash efficiencies, in the business on a go-forward basis. And I got to have cost controls.
If I don't have cost controls, am I not doing my job? I guess probably is not. Continuous investments in ion implants. So you can see where we've been with research and development over time. As sales have grown, we've continued to put more dollars on a per dollar basis. We've significantly increased the amount of research and development dollars that we have today. And as I said, we're gonna continue to try to find opportunities to put incremental dollars towards research and development to continue to fund the organic growth opportunities we see in the business.
You know, the team has done a wonderful job of segmenting our products, creating, you know, very specific solutions that meet the customer's needs and requirements, driving up ASPs, driving up gross margins, and extending the technological moat that we have around many of our product families today. So with that, you know, we'll continue to be very thoughtful as we go through and make those investments and make sure that we have, you know, return characteristics associated with those types of investments so that they meet our long-term capital allocation goals. Interesting enough, that leads into our capital allocation strategy. As you can see from the model, we do have a significant amount of opportunity on the organic side.
This remains our highest and best return, and we're gonna remain focused on investing in the business to drive towards not only the $1.6 billion model in the 2027 timeframe, but beyond that, you know, into future periods. You know, as we think through, that's going to be continued, you know, views on capacity, capacity expansion, you know, incremental systems for our applications team to test out new products, you know, development of our test center here for our customers to come in and try out recipe sets, understand how the new products are going to work, and really increasing, you know, research and development opportunities for those new product developments.
On top of that, we're gonna now weigh, you know, the other two parts of the component here between M&A, so these are our inorganic opportunities, and our share repurchase authorizations. You know, we're gonna execute continuum. Today, we have a $200 million authorization in place. At the end of the first quarter, we had approximately $175 million left on that authorization. We continue to execute against that, in a really thoughtful and rational, methodical way, moving forward. But we gonna look to be opportunistic if we can, and so at a minimum, we're gonna try to offset dilution. But as you can see from our historical practice, we've done more than that, and we'll continue to find ways to do that in future periods.
So with that, I'm gonna turn the call back or the presentation, I should say, back over to Russell and allow him to wrap it up and then invite the management team on before we begin Q&A.
Great. Thank you, Jamie. Really appreciate all the work from the team putting this together. So hopefully, through this presentation, we've talked you through why ion implantation is so critical for making devices. We've talked about some really large secular growth trends, you know, power, electrification, and artificial intelligence, and obviously, our insatiable need for chips in general, I mean, if you talk to the guys from SEMI, they say basically, we're gonna be $1 trillion of sales by 2030 in devices, right? So that's a doubling in the next few years. So lots and lots of devices out there. So hopefully, you've seen the growth, and you've also seen how we're positioned to take advantage of that growth. And then Jamie took us through the numbers. So, hopefully, you've kind of seen where we're going.
With that, I would like to invite the management team on stage and turn ourselves over to Q&A.
All right.
Okay, go, go easy.
Just like going to a punk rock concert, you always play faster live than you did in rehearsal. So we'll all you New Yorkers trying to catch the 1:00 P.M. flight, we'll probably-
Make sure you get there.
Probably in good shape now.
Yeah.
Yeah. You're welcome.
It's at 2:00 P.M.? Well, whatever. All right, so we'll open up for questions. You know, I think there's a couple of things that are interesting. I just want to mention, from years of hearing your questions, there's some interesting new slides in this deck. CS&I has always been a question in terms of how it grows and so forth. I think the slide Jamie showed there provides some new interesting detail on that. The other one is the sort of resurrection of the slide showing the number of implants per 100,000 wafer starts. So, you know, we do listen. And so even when we say, "Can't talk about that," eventually it gets into the presentation. So with that, I'll open it up for questions.
Charles, by the way, Kara has got a mic, and Maureen, and so she will, she will bring them out. So Charles, right here in the middle. That's why he picked the middle seat, I think.
Well, that was not the intention, but thanks. Charles Shi here from Needham. So I wonder if you guys can provide an update about the original target of $1.3 billion in 2025, because you laid out the journey for the next three years to get to, like, 1.6. If you're still upholding that $1.3 billion goal for next year, you're gonna be halfway there in the first year of that three-year journey. So wonder if you want to provide some update, that obviously, that was a long-term target, now it's getting a little bit not quite the long term.
Yes.
It's next year, but any thoughts on that? Thanks.
Okay. So, so I'm going to just, so everyone knows, I'm just gonna rephrase the question a little bit for the folks on the webcast in case it's not clear. So, Charles is asking about the, the former 1.3 model-
Yes
That we had in the presentation until this event, and ask Jamie to make some comments on that.
Yeah, maybe have Russell start, and then I'll, I'll jump in and-
Sure. So, I think we kind of, so we've said that there's multiple paths to achieve the $1.3 billion in 2025. So, you know, obviously, silicon carbide this year has been very strong. It's continued to be strong, but every other market has been taking a pause. So silicon IGBTs, memory until recently, general mature image sensors. So you know, we have said previously, this year, we're on a $1 billion run rate based on Q1 and Q2. The second half is gonna be slightly better than the first half with a kind of a run-up in Q4. So we are starting to see things improving. You know, Q4, we're expecting to see some kind of a little bit of DRAM kicking in, a little bit of image sensor kicking in.
Mm-hmm.
So some of those markets are coming back. But so we do see multiple paths to get to $1.3 billion next year, but we still need to see an improvement in-
Right
Consumer spending, industrial, and also, automotive.
Yeah, and just to, you know, as we think about the way the process works, Charles, I think you kind of nailed it on the head. The 2025, 1.3, was a long-term model assumption developed under a prior set of assumptions, right, you know, a few years ago. When we think about providing color relative to our expected performance for 2025, we're gonna actually kind of go back to our process of, you know, sort of annual view, so our annual budgeting and forecasting process, which occurs in the third and fourth quarter of this year, you know, which will ultimately inform, you know, where we see the remaining year, the year going next.
You know, as we look ahead to, you know, how does that then stage out, right, to the 1.6 going forward, as Russell noted, it's the underlying recovery and the timing of that recovery is going to be difficult to predict and how it layers into each of the years there in between. But, you know, we really did try to take a thoughtful approach, you know, in creating that 2027, you know, $1.6 billion, you know, target, to to make sure that, you know, it it was realistic, achievable, and the underlying assumptions there, you know, were within the the reasonable growth expectations for each one of those markets.
Great.
Yep.
Um, Dave?
So, um-
Wait. Let Maureen get a mic to you. I know, Dave, everybody can hear you.
We can hear you, yes.
I was just wondering if you could address, whatever detail that you can, the adoption of SiC in the data center, and not on the back end, on renewables and anything like that, but on the rack-
Yeah
Or on the board itself. 'Cause we're hearing all sorts of things recently that that's starting to happen, and obviously, that would be kind of an exciting development for the SiC market and TAM.
Mm-hmm. Okay, so, Dave's question is talking about silicon carbide in the data center and how it can expand, you know, both in the facilities that Greg talked a little bit about, as well as-
Mm
Potentially into the rack. So do you want to start?
Yes. So, thanks for the question. So, I remember a long time ago doing a project at school talking about, you know, data centers and what technology they're gonna use, and it was like, "It's gonna be GaN." I think GaN still has a place inside the boxes, but I think as the price comes down, that might get discussed more. But thinking about the actual infrastructure of how you deliver all of this huge amount of current to those boxes, I think that's where the focus is at the moment. So there's kind of battery backups and power management in these data centers. And, I was thinking about, like two or three weeks ago, Infineon came out with a 400-volt trench device focused solely on the data center market.
The interesting thing that they mentioned was that the trench architecture was what allowed them to be cost-effective against the incumbent. And naturally, you know, silicon carbide is coming down in price, and people are thinking at the system level, and that's where the real value comes from, in addition to the cost reduction. About a week later, onsemi came out with a whole slew of silicon carbide devices, also focused on the data center. I think at the moment, and we are working with analysts, you know, like the usual kind of market analysts, to work out what does this mean? We think it's relatively early at this stage to kind of, like, try and work out what that means in terms of future business.
But I think it's a really good indication of a market that's gonna be enabled by the sheer volume of the car business, you know, that running down of the prices.
Yeah, and just maybe add that, you know, as Russell noted, this is still, You know, I think we're all kind of learning, right, the emerging, you know, use cases potentially for AI, AI data center usages. And so, you know, as we know, we see that line of sight to the sort of facilitation, the power usage, you know, as price points and parities increase and, you know, as technologies advance, you know, David, it's certainly possible, right, that you could see these end up in the, in the data center systems. I think the one thing that we do know is that relative to, you know, the current market expectations, I, you know, we don't necessarily believe that the current market expectations take into consideration some of these other use cases, today.
Mm.
So as we start to get more information and more data on that, you know, we could see that being additive to the TAM, and then also being, you know, incremental opportunities, you know, that are not, you know, necessarily fully realized in the long-term projections.
Yeah, just to kind of add onto that, so like I say, I was talking to a customer yesterday, a silicon carbide customer, and I guess, these are new things that come to me out of the woodwork, right? So when they mentioned about the digitization of the grid and saying, you know, you could get rid of transformers on poles, it'd be a lot cheaper, you're gonna get 10% improvement in energy, it's like, well, nobody's actually mentioned that. And he said, "You know, the high voltage components have been demonstrated. They just need to get into manufacturing." And then the person talked, when they talked about silicon carbide or just power devices in general, she said, "It's amps per dollar.
Right.
I was like: Okay, that's an interesting way of looking at it. So that's when they were talking about, you know, any motor, any big motor, and which then they mentioned about HVAC motors. So, you know, think about the number of HVAC systems there are around the world. I mean, whenever I travel to Asia, every single room has a box bolted on the window. And, you know, you think if you could reduce the power usage by 10% for that motor, that's significant, and I hadn't really thought about those items. But then, when you think about what happened with LED and the fact it ended up in kids' sneakers, it ended up in McDonald's toys, you realize that, you know, the price just comes tumbling down.
You know, I was at SEMICON China this year, and you know, I was pleased to see a whole bunch of 6-inch silicon carbide wafers. There was an awful lot of people showing 8-inch silicon carbide wafers, and the quality is improving. And when you think about what the most expensive part of a silicon carbide device is, it's first the substrate, then it's the epitaxy, then it's the implant. But you know, these substrates are coming down 20%-30% a year, and at every single step as that price reduces, it turns on a new application, which we're just beginning to learn about, really.
Yep.
So, I can't just moderate. I gotta.
I'd expect nothing else, though, right? Yeah.
No, I want to bring it home to Axcelis, right? So, you know, all of this discussion hinges on the cost reduction of silicon carbide or silicon IGBT continuing to open these markets. All of that is dependent on implant. So the trench device that was discussed, that requires the high energy implants. Greg introduced, you know, the Exe silicon carbide tool as a tool that's designed to enable that. We talked about the proton tool for the backside implant for IGBT. Again, that improves the performance, lowers the cost of an IGBT chip. So the implant piece of this cost reduction and this enabling of the markets-
Yeah
That Russell and Jamie are talking about is critical, which, you know, really positions Axcelis well to participate. Back to moderator.
We're gonna miss you.
Uh, Craig?
Yeah, Craig Ellis, B. Riley Securities. Team, thanks so much for providing so much very useful information. Appreciate you all being here to do so. The question that I have is for Jamie. Jamie, can you help us understand some of the underlying mechanics in the $1.6 billion target revenue number. For example, it looks like when I do the math from some of the slides, it's a flat to slightly up share assumption that you have as a percent of the market. And as you think about the $1.6 billion, how much of a contribution are you baking in from things like either new products-
Yep.
or success with initiatives like Advanced Logic in Japan? Thank you.
Yeah. That's a great question, and I have my handwritten notes here, Craig, because I knew that was coming. So the reality is-
Do you want to repeat the question?
Hold on. Yeah. Oh, yeah, sorry. I'll let Doug repeat the question.
Yeah.
I can't help myself.
Just quickly for those out in the webcast land. Greg asked for Jamie to give more detail on the $1.6 billion model.
Yeah. So not gonna give specifics because it's very difficult for us to predict the exact timing, you know, and then the, the full load. So I'll give sort of, you know, vague generalizations that you guys can use, to, to triangulate to the best of your ability. We haven't really overweighted any one specific, you know, area relative to growth expectations. So, you know, as it relates to advanced logic, we do see some, I'll call it incremental, gains that are there, but we are not expecting this to become, you know, a very significant, share gain for us, in the advanced logic space. So this is, you know, gonna be incrementally, you know, over the opportunities we have.
That's really predicated on the success that we see with our Dragon tools inside of those research and development centers at the European Center and at our customers' R&D center, and the uptake there. We've always framed that as a bit of a 3- to 5-year play, and so depending on when, you know, we start to see adoption of that, we expect to see some, you know, incremental adoption of that over the forecast period, but we're not assuming that to be, you know, kind of go gangbusters here and grow dramatically, you know, within that forecast period. On the image sensor side, you know, it's likely gonna kind of trend back to what we'll call on a percentage of a system, system sales, I should say, normalized, you know, sort of revenue levels there for CIS, very broadly.
Talked a little bit about DRAM already and let you know there that, you know, there is some incremental gains. It's again difficult to size the meaningful impact that, you know, HBM is going to have on the DRAM market, as well as the evolving needs of the devices and the DRAM requirements inside of those devices. So we do expect to see memory sales up relative to that prior peak, but again, we're kind of, you know, trying to be modest in terms of, you know, what that ultimate impact could be. On the power side, power SiC is going to, you know, continue to be really durable for us over the period of time. So, you know, we anticipate some continued growth in the silicon carbide space for sure.
The question will be as the adoption of, you know, hybrid vehicles, and how quickly we see that level of adoption, what happens with silicon IGBT? So again, we tried to make sure that, you know, we were—we didn't overweight ourselves, right, to the, the potential optimism associated with that. So I think that leads opportunity sets there, you know, kind of for the approximately, you know, you know, $1.6 billion that we put in place, for performance above that, you know, if possible. And then, you know, finally, I'll, I'll end with, you know, on advanced logic. You know, actually, I talked about already general mature technologies.
You know, general mature technologies, we do see recovering, and so, you know, as that recovers back into sort of, you know, call it the more historical ranges for us on the other general mature technologies, that is a contributory driver for us over the forecast period.
Yep. All right. Tom?
Oh. Thank you. I was hoping to ge t a little bit more on the Dragon. It's been around for several years. We haven't seen a ton of momentum.
Mm-hmm.
What is the technology or the key differentiator that you think is gonna drive, you know, success in certain markets going forward?
Okay. So Tom's question is about the Purion Dragon, which we announced a few years ago, and a little bit more about the technology and how it drives into the market. I'll start with Greg, and maybe Jamie can chime in.
Yeah, absolutely. So yeah, thanks for the question. So we have had Dragon around for a little while. As I said during my presentation, we've identified new and growing applications in advanced logic, where we think we can bring some of the capabilities of Dragon to address those, primarily in materials modification applications. So we've learned a little bit more. We've made some modifications to the product to make it accommodate what customers are looking for. We've also developed more direct experience-
Yeah
Through some of these engagements with these sites. So that's really helped inform where the problems are and how we develop those solutions a little bit faster. So I'd say, in short, we're taking a really focused approach to where are the new applications that are growing in different parts of the process flow that we can differentiate around.
So, Tom, I just want to add, I mean, so this tool is designed for really low energy, high current, which is great for doing material modification, and it also differentiates itself in the sense that it's a scanning beam, so you can separate angle control from, you know, uniformity. So we've been kind of like looking at things that our architecture enables, you know, our customers to solve and looking to differentiate ourselves that way. But, you know, the technologies are quite differentiated, and we have had a couple of opportunities in really high aspect ratio, kind of doping because of the angle control.
Some of the stuff we're talking about now in advanced logic is more middle of line and back end of line, where there isn't actually an incumbent, which then gives us a greater opportunity because now we're not competing against somebody else. So you work with a customer, you understand their valuable problems, you come up with solutions, and then obviously the best solution wins. So you're not guaranteed of finding opportunities when you start working with customers, but we find that once we're in R&D, we're engaging with the customers, you'd be amazed the number of things they try and do.
Yeah. You know, just to add a couple of things, some of these new applications, you know, are critical to, you know, sub-2 nanometer type of technologies, like backside power, for example. Implants being looked at for, you know, applications in that space. That's technology that didn't even exist and is just going through R&D right now. Second thing I just wanted to remind folks is we do have Purion Dragons in production in advanced memory applications. So, you know, the tool is out there and being-
Mm-hmm
being used. The focus of the tool and its design, you know, was to help penetrate the advanced logic, so.
Okay. We have a question coming from online that David will,
Oh
-read.
Thank you. Thank you so much, Doug. Thank you, everybody. So first question comes from, Jed Dorsheimer at William Blair. "Silicon carbide in industrial-
Mm.
Where do you come out on planar versus trench? It seems clear that planar has been the path for auto, but in data center and other industrial apps, it would seem that trench has some unique features. This would obviously have implications for implant.
Okay. So
No, go ahead.
So the question, the question was, you know, what's our view on the, basically the roadmap for planar transistors for silicon carbide versus trench? So, when I look at planar transistors, they all use medium current, but planars have a higher dopant concentration, so they tend to use the H200. We love planar. When it comes to trench, they tend to have much higher energy implants, which use the high-energy implanter, and those devices are deeper and smaller. So think about, you know, a planar device maybe being three times bigger than a trench device. I personally believe, and I guess Jed would probably know more about this, I think a lot of the times people are moving to trench is it gives you a higher yield.
I mean, a smaller device on a substrate that has multiple crystal defects, you're going to miss the defects more often.
Yeah.
I think that makes a lot of sense. So, I kind of think when we look into the marketplace, definitely the advanced manufacturers are very much talking about eight-inch trench. That's where they're going. The more rudimentary manufacturers are quite happy churning out lots of six-inch planar. There is one very, very large player in the market who's adamant that they want to stay with planar. Although, you know, everybody is looking at-
Yeah
At Trench at some level. It really is the next iteration. You can get more devices, higher yields, and obviously, it goes hand in hand with the transition at the right time from 6-inch to 8-inch.
Yeah, and the beauty of that is, as Russell noted, with the higher energy, those are—you know, that's our bread and butter, right? Tool set. That's where we are, you know, the leader, given the architecture of our tools and the ability for us to deliver the highest energy, specifically in the power space. Those also represent good margin opportunities for us as well. And so, you know, from a transitional perspective, the ability for, you know, trench becomes the predominant, you know, power device inside of the data center, whether that's in the facilities, whether it's in the, you know, racks themselves or, you know, any combination thereof, you know, that will provide an advantage to Axcelis.
And just kind of an interesting tidbit is that as we talk to more and more customers, and, you know, we have multiple customers in every area, and we see the entire market-
Yeah
People are moving actually to higher and higher energy. So it's almost like silicon all over again, where, you know, people started off with a medium current implanter, and then this did it, added more. But now we're finding that what started off with people requesting our Purion XE, they're now looking to take the Purion EXE because they want that little bit more extra energy to take them to the next step, and that gives us even more differentiation. So I quite appreciate that.
Yeah, and the last thing I'd like to add to that is the competitive environment. So, you know, when we look at these higher energy, high-dose applications, this is a place where Axcelis, you know, has a tremendous competitive advantage. It's also the higher margin tools, higher ASPs. It's very good business for Axcelis.
David has one more question.
I have another one. This comes from Mark Miller at Benchmark. "What does Axcelis see as 2027 revenue mix between power, image sensors, mature, advanced logic, and memory?
Yeah. So again, I'll probably keep that in vague term, Mark. Appreciate the question, and glad you're on the webcast here for us. But, you know, ultimately, as you can see the breakdown. And by the way, we will post these slides online following this meeting. We just didn't want you reading ahead, you know, while we were here. So we will post these online for everybody after the event. You know, as we think about it, right, as we look at the growth of the markets and we look at the breakdown, we do expect the general mature technologies and power technologies to continue to be, you know, a large portion of the revenue for Axcelis as we move forward. That includes the image sensors, right?
That includes the other general mature, and, and as well as silicon IGBT and, and silicon carbide. You know, we, we don't have, You know, memory, like I said, is gonna be, you know, back on a dollars perspective, will be closer, you know, modest growth over those prior, revenue peaks that we had of around $130 million in the historical period. So some modest growth above and beyond that. And on the advanced logic side, you know, that's going to be, you know, incremental, you know, to what we've seen historically, but it's not gonna make up a, a large portion of the revenue load for the $1.6 billion model.
You know, Mark, thanks again, but it's probably the best I can do right now, and I'm gonna burn my notes after reading, so.
Brian, in the back?
Thank you. I, Brian Chin at Stifel. Thanks for the presentations and the discussion. This might kind of pull in a couple different panelists here, but maybe just start with Jamie. In the 1.6, what's the assumption roughly for the CS&I? You know, it was $250 million last year, maybe it's $300 million at the next, or conceptually, it probably grows a little bit lower than the equipment business over that horizon? That's the first part of-
Yeah.
I'll just repeat it. So Brian's asking for a little more detail on the breakdown of CS&I in the $1.6 billion model.
Yep, we do continue to, you know, show our, our prior examples on CS&I, sort of twenty-five-ish percent or so, right, of sales, plus or minus, right? The difference here is we're gonna be continue to add, you know, incremental period power units. We do expect there to be, you know, some potential significant opportunities for upgrades, you know, over this forecast period as well, which is going to, you know, continue to drive probably CS&I levels. Because you're not just adding the sort of annual annuity that comes with, you know, the delivered systems into the field, you're also now going back into the, you know, the entirety of that install base to push those upgrades, you know, through the entire install base, you know, over that forecast period.
You know, so and then the final piece there is the contracting mechanisms that we're currently using with our customers. You know, we're gonna continue to evolve those in a way that increases their productivity throughput, output, while increasing, you know, our share of wallet of the aftermarket. You know, so there are certain components within that aftermarket stream that, you know, the customers have no choice but to come back for us too. But there are other pieces where there is competition, right, in that space.
And so as if we can continue to provide, you know, higher quality, higher throughput, higher productivity through a more fulsome level of service to that customer, we anticipate being able to capture the incremental share, and sort of to deflect some of the second sourcing that does occur from time to time, for some of the more common parts and components that are out there.
Okay. 'Cause if it was a quarter of the revenue of 16, theoretically, it would actually grow faster than the equipment business.
Yeah, and that goes back to, as we think about the upgrade opportunities, Brian, as well as some of these incremental contracting. You know, precisely where that falls, right, is, you know, and how quickly we get up there is, you know, part of the reason why we have the build day there. And so, you know, whether it kind of the incremental growth comes from systems, right? You know, as we've talked about some of the opportunities to potentially go above and beyond, you know, in some of the other areas of the systems business, that could, you know. Those opportunities could more than offset, you know, a little bit of, you know, overestimation on the CS&I side.
Okay. So the other part of the question is, if, you know, if I apply that math and you'd still to 27, you would still outgrow the implant TAM by about 1%, like, in terms of the CAGR over that horizon. So in terms of that 100 basis points of maybe premium, how would you break that down? Is it sort of, is the base case having a similar market share than you have today? Do you see a little bit of a tilt towards the power market? 'Cause you, And also, the kind of the last part of that is, you know, you shared some math about implant intensity for power.
How many, you know, what sort of wafer capacity do you think could be added from now through 2027 for the power device market?
Maybe I'll start with the first-
Let me start to summarize. We'll take that mic away from Brian, because-
Yeah.
Brian, you know, I'm getting old. I can't remember all of that to repeat. So, I'll just summarize. Brian's looking for a little bit more breakdown, sort of on share gain and, you know, the breakdown in that 1.6 relative to TAM growth, and so forth. So Jamie, you wanna-
Yeah, I'll start with the. So, you know, Greg talked a lot about, you know, trying to break into advanced logic in a more meaningful way, as well as the geographic expansion into Japan. So as, you know, part of that performance above the TAM, right, is our ability to, you know, break into those, what we'll call underrepresented sections of the Axcelis business today, is going to provide that opportunity. You know, we do see through the introduction of some new tool sets, you know, an opportunity to potentially gain incremental share in other parts of the business as we move forward. But we have not overweighted share gain, you know, in this model, as a means to be able to achieve that. And Greg, if you want to add anything-
The thing that I would add to that, in addition to Japan and advanced logic, is the backside proton implant, and that market opportunity, which we're very well poised for and executing on today, could also contribute to that $1.6 billion model.
Yep.
Okay. Next, Tom? I'm sorry, yep.
Jack Egan, Charter Equity Research. Yeah, thank you, guys, for hosting this event today. It's been very helpful. I had a question about GaN on the power side. So, you know, a lot of the big opportunity in the power market is in, you know, vertical device structures, so whether it's trench or, you know, those power FETs. But GaN devices are a bit different in that they are they're power devices, but they're planar. And so I was curious if that kind of reduces the importance of them for implant just in that long-term TAM for power.
Does that still kind of remain primarily silicon carbide and silicon, even if we see GaN get to, you know, a larger portion of the market, or is that kind of an opportunity for further growth in implant?
So just summarizing, so Jack's question is around the potential impact on the implant TAM relative to GaN's growth, compared to silicon and silicon IGBT, silicon carbide. So, Russell?
Yeah, so, definitely, so, I've always been interested in gallium nitride, and if you go back 10 years, they're kind of both silicon carbide and gallium nitride on the starting grid, and there's always a discussion on which one's gonna go first. I was actually betting on the GaN at that stage. So, yes, you're absolutely right. It's a planar device. You put a sapphire wafer or a silicon carbide wafer into a furnace, you leave it for eight hours, you know, it's into an MOCVD tool for eight hours, it comes out as a multi-layered stack. You can carve them up, and you've got your devices. I think they're always gonna have a place. So if you look at the high voltage versus frequency graph, you can see this really big IGBT blob.
Then you kind of see silicon carbide going to higher voltage, but slightly less frequency. Then a slightly lower voltages, but really high frequency of GaN. GaN, obviously, you're gonna see it appearing all over the place. I mean, it's already in your phones anyway, but it's gonna turn up in those wall warts. It may turn up in data centers, and its price was driven down by LEDs. It already worked out how to do GaN on something during the LED days. So I think it's gonna always have a place. And, you know, we still do actually implant gallium nitride devices. So we've shipped tools to customers to implant gallium nitride wafers and modify them.
What I would say is it's not as intense, as, you know, an IGBT or a silicon carbide, but power is still very good for us. And I think personally, gallium nitride will be more of a niche player than a hardcore player. I think silicon carbide can do most of what gallium nitride can do, so it's gonna be in those niche applications where it's gonna win. Did that—Jack, did that kind of help?
Yeah, no, that was awesome. Thank you. Christian?
Christian Schwab from Craig-Hallum. The trench having high energy, you know, versus medium current planar, you know, on the side, you did a nice job of telling us how many implanters were up there. But, I'm wondering, you know, math being equal, you know, how much dollar content more would be in, say, an 8-inch trench fab versus, say, a 6-inch planar fab, dollar content to you?
Okay, so Christian's question is sort of to translate the, you know, number of implanters per 100,000 wafer starts chart into into dollar value. Yeah, 'cause it depends on the mix and the recipe set. So, Greg, do you want to?
Yeah. So we did normalize that based on an existing flow, and although I can't really comment on exactly how that dollar figure figures in. As we talked about before, having high energy in the trench device is beneficial for Axcelis, and so we're working closely with customers to make that happen. That intensity does come from the fact that implant, as I said, helps to define the device, and so that, that's what makes it overall so intensive for, for implant.
Yeah, and, you know, I'll just add—difficult question, sort of a, the standard answer that kind of give a lot. You know, all these process flows for these customers, you know, do differ.
Yeah.
Their manufacturing sort of operational, you know, mentality or approach is different too. As you can see from the chart Greg had and the chart that's been in the presentation, there's a lot of overlap for all the different implanters. So if a customer is thinking, "I've locked in, I'm doing this," you know, a medium current solution might be just fine. Many times, the customers are looking and saying: Well, this is what I have today, this is where I'm going. I should probably buy the XE or buy the H200 to satisfy the fact that, you know, I want to be ready for that. We see that a lot. You know, we've had several H200 silicon carbide tools out for evaluation. Part of the purpose of doing that is the customers are running with medium current.
They know they need the high current, and so, you know, we work with them, use the evaluation process as a way for them to get things qualified, so then when they start to ramp, they get the efficiencies. And so, the dollar value will move around relative to what makes up the total number. So Dave?
Hi, it's a bit of a two-part question on your memory business. You threw up a slide up there talking about memory being 16% of the implant TAM, but for the first half of this year, for you guys, it's 1%.
Mm-hmm.
So let's talk, maybe help us understand what the difference is in that percentage. Also, you mentioned that you expect the memory business to kind of exceed its previous peak a little bit by of $130 million. If you're gonna do $1.6 billion in revenue in 2027 and memory is 10%, why-- right, and it's 16% of the TAM, why wouldn't that number be like $200 million instead of slightly more than $130 million? I'm, I'm kind of curious. It seems like the memory numbers are way understated.
Yeah. So, so let me just repeat it, and summarize a little bit. So, so, first of all, you know, thanks, Brian, for opening up the, multi-question. The twofer.
Easy one to remember.
Yeah. Yeah. So I'm gonna summarize Dave's why not more into one question. Dave wants more detail on the memory market relative to this year and going forward. So maybe it was Jimmy, you want to-
Yeah, I'll start, and, you know, maybe Greg can add in where we—you know, one of the things that we wanted to put out in front of the group here, and, you know, for everybody, was a model that we had line of sight to. So as we think about how we achieve that, we think about where the relative opportunity sets are, we wanted to ensure that we were balanced relative to the expectations on relevant growth. I think, Dave, your question, it's not a loss of share necessarily. It's not necessarily a, you know, any reflection on the inadequacy of the technologies.
What it really represents is the difficulty in exactly predicting the timing of the ups and downs relative to the markets within the time frame in which we built the model, relative to the contribution of each portion of that. And so, you know, as we think through, you know, building out that model, could memory be higher? The answer to that is yes, right? Absolutely. Especially if you look at some of the trends that are going on and the need for, you know, DRAM going forward and the storage needs that are gonna go forward. But we wanted to make sure that we were balanced in our approach, and didn't overweight one specific section to another, you know, given the sort of broad nature of the opportunities that we have in front of us.
The implications of that are either the memory TAM is gonna go way down as a percentage of the market, or you're going to share, so it just has-
Well, so you've been talking about, you know, the last upturn we had-
Yeah
It was about, 1, what was it?
130 .
130 .
Yeah.
It was 50/50 NAND versus DRAM.
DRAM.
Last year, we were about 10% DRAM, and it was only DRAM, right?
Only DRAM. Yep.
It was about $80-$90 million. So as NAND comes back, it, we're expecting it to be about the same as DRAM. That would be about $180 million.
Yeah.
Yeah. So I'll just add to it. So, you know, the content per 100,000 wafer starts for memory for implant is pretty similar to what it was during the last couple of cycles. It hasn't really changed a lot. You know, implants, the recipes and so forth move around a little bit, but in general, it's still basically the same. So Dave, you know, the question that we wrestle with is, you know, how many new wafer starts get added and when?
Right.
That's what will determine what the upside to memory is within that model. And so—and it's still too early. We're, you know, just starting to see memory recover.
But the great thing is, though, Dave, so normally in a slowdown, there's a shrink in the devices. This time, the devices got bigger, right? So to take on this kind of additional packaging for stacking up these basically DDR5s into an HBM, they've had to get bigger. So that was kind of a change. The other thing is that they're cannibalizing DDR5 to build these HBMs, so there's going to be less and less DDR5, and a lot of these, so that's what's going on in the transition. Christian's made it very clear that until these guys make money, they're not gonna be buying anything more.
That's right.
They are making-
They're making money.
Very good money right now, and I think they're kind of like hanging back, going, "God, this is really good." But there's gonna be a breakout. Obviously, Hynix has already broken out to some extent. They did have a Yongin cluster capability that's like four mega fabs. Like, there's gonna be 800,000 wafer starts in four fabs. That's like 2027. Because they're doing well in HBM, and they found themselves at the number one position for in DRAM, first time ever, I think, you're seeing them pulling in this extension fab to M15. So it's the M15 extension. That looks like it's coming online in 2025. But we are all about wafer starts. So HBM has been really good for us in the sense that it's used up all this excess capacity, right?
The utilization was very low. That's the first thing. Now, these companies are going to become forward-
Right
Constraints. They're gonna be looking for productivity upgrades. They're gonna bring online these new fabs, and that's gonna give more wafer starts, which will need more ion implantation. But we are wafer starts, and as we kind of alluded to before, DDR3 to DDR4 to DDR5, the number of implant steps hasn't really changed. Maybe the recipes get a little bit more dose, a little bit more energy, whatever, but ultimately, we haven't been like the DRAMs of this world, where you get this definite blow up. You know, we are number of wafers.
Yes. I'd like to, you know, just broaden that a little bit to make sure people don't lose, lose sight. So when we're talking about memory, when we're talking about wafer start, so that's what drives our business in memory. When we look at some of the other markets, it's implant intensity. So the power market has, you know, multiplier wafer starts, plus the increased power, increased implant intensity based on the device designs and so forth. In advanced logic, we're going after essentially greenfield new applications, that are all upside to the total TAM, that we're on equal footing, you know, with the competition to go, go and win by being in the R&D sites. And then, of course, in the mature markets, it again comes down to wafer starts.
A lot is, you know, that's—those are mature, they're well known, but they're often upgrading to Purion products to improve productivity, reduce costs, get better process control over the new devices. And so, so anyway, I just wanted to clarify to make sure people understand there's different growth factors relative to what drives, the revenue and the different implant choices. Craig?
Thanks for taking the follow-up question, Craig Ellis, B. Riley Securities. I was hoping to get more color on where the company is with one of the capital allocation points, M&A. Can you just talk to us about where you are in terms of developing a funnel of candidates, moving potential candidates to the more actionable part of the funnel?
And things that would be either, A, interesting and, and likely to be acted on, whether it's software, services, or technical capabilities, or things that are just off the table? And what are the implications for target model attainment if we are actionable on M&A? Thank you.
Okay. Craig's question is around capital allocation focused in on M&A, so.
Yeah, so not much we're going to say on the specifics there, Craig, but what we can say is the aperture is wide open for us, right? So we are, you know, we're building out a team of folks. You know, in the room, we've got Atul, who's been helping us from a technological side, you know, comes with many years of experience. We've just recently added David to the team as well, to help us think through sort of on the financial modeling strategy perspective in terms of the execution against that. You know, our goal here is to, you know, find those opportunities that meet the return characteristics that we would look for relative to the alternatives. Our priority is going to continue to be organic growth, right?
And then we'll evaluate the alternatives between share repurchase at this point and M&A opportunities as they come forward. We have, you know, really particular strengths, right, in the semiconductor capital equipment space, right? We manufacture and assemble very complex semiconductor tools. We've got a broad base of field service representatives, application engineers, and, you know, inventory staged throughout the globe at the largest fabs, manufacturers and producers, you know, today. And so, you know, with those strengths, you know, you could see us add a technology in. You could see us find ways to try to bolster our aftermarket revenue, you know, given the fact that we can leverage that global field service support team that we have today. So, you know, that. I think that's probably the most, you know, specific-
Mm
I can get right now, because, again, it's very difficult to predict timing, you know, and ultimately how that will impact. The model that we presented today is an implant-only model, right? It doesn't, it does not take into consideration, you know, any contribution associated with that, and at the time, you know, if we were to do an acquisition at that time, we would then speak to, you know, the potential ramifications or implications, you know, to the long-range model if any. Yeah.
Uh, Hendi?
Thank you. Hendi Susanto from Gabelli Funds. I would like to ask about China. How do you see future growth in China for your long-term $1.6 billion goal by 2027? Do you see yourselves in China to grow faster than overall sales? And then second, considering that semiconductor market can be cyclical, what can cause investment in ion implants in China to take a pause? Some areas are in the early innings, some areas may be in the mid-innings. At some point, China may deem it has enough capacity.
All right. Thank you, Hendi. So, that's a record half hour before a China question got asked.
Yeah. That was actually the first sheet that I had on my handwritten notes here, Hendi, so.
Just to summarize the question, Hendi's question is more around the market relative to China, you know, how big it can be, what, you know, where it can grow, and so forth. So,
Yeah. Maybe I'll start with just the expectations of the model. We, you know, we expect China to continue to be durable, right? From, you know, and to grow, you know, accordingly. We anticipate continued investment in the space. What we see, though, as the other, you know, geographies, you know, within our portfolio come back, we do anticipate the percentage of revenue to come back into that sort of, you know, more normalized ranges. You know, so we're not going to be at the, you know, elevated, you know, rate that we see today in that sort of 40%-60% by quarter. We would see that moderating down, you know, fairly meaningfully, you know, as we move ahead, into that model.
But we continue to see growth, strength, and investment dollars, you know, as part of that. You know, relative to the specifics, you know, for us, that's going to be in the general mature space very broadly. And so if there were, you know, a pause, right, it would, you know, there could be, you know, some other, you know, economic factor, global economic macro factor that, you know, could hinder that. But as of right now, we don't, we don't see that coming forward, and putting any temperance on the opportunities there.
I agree with that.
Okay, yeah.
Gus Richard, Northland. I'm just wondering, in advanced logic, is your window, your opportunity like, gate-all-around, a second-generation gate-all-around, or is your opportunity, you know, getting into the advanced logic guys through IMEC or maybe some of the new, you know, government-funded entities like Rapidus?
Okay, so the question is around Advanced Logic and where the opportunities are, both from a technology perspective as well as kind of the market customer opportunity.
Yep, so I can, I can take that one. So as I talked about, our area of focus is really around more of the Middle of Line and Back End of Line. So the Gate-All-Around opportunity you talk about, it's fairly well subscribed right now. The implant intensity there is a little bit lower, actually, but the opportunities we're focusing on would be more materials modification and some doping in the Middle and Back End of Line. In terms of, you know, how we're doing our learning, and, yeah, you may have guessed correctly, the place where we are working, absolutely, and there are opportunities, potentially in other Advanced Logic sites, definitely.
Other questions? I think Cara ran all the way to the front.
I know.
Hey, thanks, guys, for clarifying that the HBM opportunity is more tied to wafer starts. But, is it fair to say that the HBM may favor higher energy implantation tools, and does that favor something like a Purion XE?
The question is around HBM and the impact on ion implant, and so.
So, okay, I'll kick it off. So, HBM is basically, as Jamie says, is DDR and DDR5 in disguise, right? You're taking these DDR chips, sticking them on top of a logic chip, and stacking them up. So essentially, it's the same implant flow as a DDR5 chip. It is DRAM. So in DRAM, what you see typically, versus NAND, for example, is you see more high current, less high energy. NAND takes more high energy, less high current. And from generation to generation, it doesn't change significantly, which is why we say it's essentially the same number of implants as needed for the same number of wafer starts, each and every generation. Did that answer the question?
Yeah, and so and I think, you know, it kind of comes back a little bit to, to Dave's question, you know, that HBM has been great for Axcelis in the sense that it's using up the capacity, underused, utilized, capacity-
Yeah
Is being absorbed, and that speeds up the opportunity for when they need to expand their wafer starts.
Yeah, absolutely.
That's when we'll benefit.
It's definitely used up the, you know, the fab utilization, and now they're making money as well, and that's really critical to needing to expand further. So it has been a real positive.
Other questions? All right, if not, I want to thank everybody for coming today. I want to thank the Axcelis management team for putting the time in to be able to have this great dialogue. And I personally want to thank everybody here for the last many years of you know, working with me and putting up with me. So-
Yeah.
Thank you very much.