Hi there. I guess we'll kick things off. I'm Brian Chin, the Semiconductor Capital Equipment Memory Analyst here at Stifel. Thank you, everyone, for joining us today for the Stifel CSI Conference. Very excited to have one of the key semiconductor equipment companies, especially in the ion implantation market, Axcelis, with us today. Long commute in from where they're located in the area, but we've got CEO Russell Low, CFO Jim Coogan, and Doug Lawson, who EVP of Corporate Marketing. I believe a couple of different hats over there. But yeah, I think they have a couple slides maybe to kick things off, and then we can hop into maybe a fireside discussion. I'll turn it over to Russell or whoever.
Thanks, Brian. So really just a one-minute intro. So, we are a global leader in the design and manufacture of ion implantation equipment. We basically play in every single device segment there is. So you use an ion implanter for memory, whether it be DRAM, HBM, NAND, you use it for image sensors, mature technology, advanced logic, power devices. So we are used in every single device. So as you can imagine, all of our customers are the big names you know of in device manufacturing. So whether it be STTI, Intel, Samsung, those are our customers. So, you know, the ion implantation market is about $2.7 billion, and that's been growing relatively quickly in the last few years, which I'll talk about in a moment. But it is, it is a very much a global market.
So we have customers in Taiwan, Japan, Europe, North America, Korea, China. We have, you know, a global customer base, so then we also have a global customer support infrastructure. We are located in Beverly, so it's about 30 miles north of here. So if you're driving, just, you know, like 2 hours. We have about 1,700 employees worldwide. We have an installed base of over 3,000 ion implanters, and we have been in business for 45 years. We are one of two companies that have an entire portfolio of products that serves the entire implant TAM, and we are one of the innovation leaders. So, you know, we have a very strong IP portfolio. So basically, I mentioned that the implant TAM has doubled in the last few years.
So I joined the implant industry in about the year 2000, and it was about $1 billion. Today, that market is about $2.7 billion, like I mentioned, and it's going to be growing towards $3 billion in the next couple of years. A large part of that growth has come in the mature segments. So when I think about mature segments, I'm thinking about power, both silicon and silicon carbide, image sensors, and kind of general mature. General mature would be kind of analog devices, but logic devices that are 28 nanometers and above, basically anything that's planar. So those technologies have grown. We are incredibly well positioned to benefit from these mature technologies.
So we have been working very closely with our customers for a number of years to put together basically, you know, specific portfolios of products that really do address these emerging mature markets. And, you know, these markets not only are utilizing highly specialized implant equipment, but also these markets are highly implant intense. So this has been a really great opportunity for us. So that's kind of one of the things we've been- we'll be talking about is, you know, the mature markets and how we have a really great product portfolio to address the market needs. And just finally, this will be the last slide. So we are really proud of our growth over the last few years. So we started off with the Purion platform in around 2012, 2014 timeframe.
We created the entire portfolio of products off this platform, and now as time has progressed, we've added more and more products that are basically derivatives off this platform, and they've gone after these specialist markets. You know, so our Purion Power Series, for example, really does focus on the needs of silicon and silicon carbide power. We also have a number of products focused solely upon image sensors, and you know, we really have done very well in those markets. So, you know, we've had strong opportunities. We've had a highly differentiated product portfolio. We are one of the two companies that are driven by innovation. We have a strong balance sheet, and I'd say we've got very strong customer relationships after all the years we've been working with these customers, and we've got great employees as well.
With that, I'll hand it back to you, Brian.
Yeah, no, I appreciate that. Thank you, Russell. It's a good quick overview of the company. Maybe to fan this out for a moment and then kind of get closer, then kind of fan it out again for a moment. You talked about how the implant market really has gone from $1 billion when you joined the company to close to $3 billion. We'll talk about the market in a moment. And a lot of the incremental growth is driven by more mature technologies.
Where that's really been a pivotal driver. Now, that market has taken a bit of a respite this year after a very strong and differentiated growth profile last year in 2023. I'm curious, if we just look at your outlook for the year, and the half-on-half sort of comparison, you had some optimism that after a little bit of a dip in the first half-
The revenues could improve in the second half. Have you seen indications, firstly, to confirm that perspective since then?
Sure. Sure, so, you know, I think we kind of, you know, Q1, Q2, we're on a run rate for approximately $1 billion. We definitely see additional strength in the second half. We expect the second half to improve. I think there's a couple of things. So we have seen, you know, requests for quotes picking up. We have, you know, these are kind of like anecdotal things. We've seen the backlog of $1.1 billion has started to firm up in many areas, an now we're starting to see, you know, tools being pulled back in again.
Yep.
What I would say is that silicon carbide power has grown 25% year-over-year. Really, it was the other elements of the business, image sensors and memory and, you know, mature, that have been a little bit on the slow side, but we are hearing from our customers that they are looking to see, you know, 2024 back end and 2025 get better for them as well. So I do think we have a lot more, what's the word? Confidence.
Yeah, yeah
In the second half, recovery.
Yeah, it comes down to, again, those, there was a little bit of lack of visibility in the beginning of the year, just as our customers were trying to firm up their CapEx budgets, expectations, fab readiness, and others, as they entered the year, looking at the markets. And, you know, with that, we saw, you know, a little bit of what I'll call squishiness relative to some of the delivery date expectations. That has actually now kind of shifted for us, where we are now seeing those pull-ins occur. We did not see cancellations over that period of time, and really, when we think about our backlog, our backlog is tied to our customers' needs relative to their fab readiness and bringing that capacity online, and so we feel really good about the orders we have.
They're firm, firm purchase orders that we put in there. It does represent our systems sales only, so it, our backlog does not include, the revenue associated with our, our aftermarket business and upgrade opportunities. And so we still feel really comfortable and confident about what, what we're seeing in the back half of the year. In addition to that, we're seeing a lot of really positive momentum on the memory side, and so, as we think about the capacity, that's being sort of chewed up by HBM today, and, and as the ASPs increase and as the profit realization of the memory manufacturers continues to increase, you know, you know, there's, there's likely going to be a need for them to step up their investments for incremental capacity sooner than their, than their initial plans. We're preparing the organization for that.
As Russell noted, we've got a very strong balance sheet, and so we wanna utilize that to the best of our ability to take advantage and be opportunistic where for when our memory customers kind of turn back on. They have a history of turning on sort of fairly quickly in this space, and so they'll place orders with us, and so we wanna make sure we've got the long-lead materials necessary to build the high-turn tools and components that they will typically request from us.
We're saving some incremental capacity and manufacturing slots to be able to satisfy those requirements, but we don't want to over-index relative to that memory recovery either, and so in our first quarter call, as we talked about the back half of the year, we pushed that relative expectation for recovery to later in the fourth quarter. However, we are prepared to move sooner in the event that they turn on orders sooner in the year than what our current expectation is.
Okay, great. Just even, Russell, just to clarify your commentary around sort of the firming up of the backlog and maybe even some tools pulling in, is that largely since silicon carbide already, you had better visibility and a growth profile in this year, is that more on the general market that you're talking about, general market mature?
Yeah.
Got it.
Yeah.
The memory is more of a, like an indication of, of-
Yeah, there's
Sort of
I would say green shoots. So again, some positive indicators that, again, 2025 and 2026 should be relatively strong years for memories, starting with DRAM, with NAND probably recovering maybe in 2025. We'll see how that picks up, but we do expect DRAM to pick up pace and recovery as we go through the year. But your point is what we're seeing now is, on the general mature technologies, which was showing some incremental signs of weakness, is actually now firming up in a much more meaningful way for us in the back half of the year.
Got it. And just since we're on that topic, the... I mean, I don't know that you have this level of visibility beyond, but if I think of, like, RF, which is, like, a super depressed market, image sensor, you know, probably similar.
Yep.
You think of, like, handsets, you think of phones. Do you think that is the connection point there in terms of either, you know, the phone market improving? Is this a, is this sort of a referendum on the improvement in the phone market, on the China phone market, on.
I think there's kind of like two really big drivers. You know, clearly, you know, AI is gonna touch everything, and, you know, we're waiting for the refresh of phones and PCs, where everybody gets the updated AI assistant. I mean, I've heard that they're gonna make Siri.
Siri
smarter
Yeah
which I think people have said is a pretty low bar, but people will be
Us less smart too.
Yeah.
So I think, so we're seeing, you know, that, so, so just kinda like backing out a little bit, so, so AI, so clearly you read about NVIDIA, GPUs going crazy, high-bandwidth memory. You know, from us, we mostly process wafers, so the great thing about high-bandwidth memory, as Jamie said, is, one, they're bigger. They're basically DDR5s, but bigger, and they're using up capacity. They're giving our customers profit. Once they start to make a couple of quarters worth of profit, they start investing. But that's what people have been thinking about in terms of AI, but I mean, clearly, it's just got a massive thirst for power, and we're seeing that come through. We're seeing, you know, a massive thirst for data and then communicating all that data.
So all of the chips are gonna get driven through that, whether it's Internet of Things and it's gonna be dragging everything off your phone, off your computer, off sensors, off your fridge, your doorbell, wherever it is you get this information, it's then gonna be transmitting it somewhere. It's gonna be crunching that stuff, and it's gonna be storing that stuff as well. There's a huge amount of storage. So we're seeing that that is kind of a secular trend. The other thing we're seeing is electrification. So when people think about silicon carbide, they think EVs. It's actually kind of one step above that, is electrification in general. So we're relatively agnostic as to whether it's a silicon IGBT or a silicon carbide going into an EV.
So obviously, hybrids are taking off, and you can kind of see that in the fact you can only buy the 2025 Camry as a hybrid which is the world's most popular car. But we're also seeing, you know, the generation of power the transport of power. So data centers, I mean, I was a little upset to hear that we're gonna be turning on coal-powered, coal-powered power stations, coal-fired power stations to, to give the energy required to to data centers.
But, you know, it was quite nice to see that Microsoft had put $10 billion into renewables. So, you know, you'll start to see more renewables going into this, this area as well, which, quite honestly, it'll be... If it's solar or if it's wind turbines, they all use power, which, you know, power devices, which is fantastic. And, even last week, you saw Infineon release a new chip, a 400 V silicon carbide chip, that's focused on the data center. So you're seeing, you know, electrification in general and the desire to use more power, but also make it more efficient. So we're kind of seeing those two drivers being the things that are kind of pushing us on.
Yeah, and just, just to add to that, sort of bringing it right back down to Axcelis. Whether it's silicon carbide power devices or silicon IGBT power devices, we're pretty agnostic. Both are very capital intensive for ion implant. And so whether it's a hybrid car, you know, that's taking market share away from ICE vehicles, that's all new inverters, that's all new, parts that have to be produced. And if it's EVs or data centers or other things that are using additional chips, all of that is, is good. Where it really doesn't matter whether it's silicon carbide or silicon.
Yeah, maybe this might be, you know, review for some people, but would, would you mind maybe talking a little bit about just, what it is about, you know, 28 nm and more mature nodes and specialty technologies like power devices? What makes it so much more implant-intensive even than when we hear about, you know, people hear about Gate-all-around, they think of the most sophisticated technology and the smallest features. But what is it about these older nodes where there's been a lot of, you know, renaissance in spending and now, you know, a lot of potential wafers needed in the future? Why has that, that been such a vibrant market for, for implant?
All right, so, it starts off sort of like a fairy tale. So a long time ago, in the 1970s, ion implant came into semiconductors, to produce the transistor, and that, that was its role. It was invented by, or productized by several folks coming out of MIT. They moved 30 mi north of here and started a company. The initial implants for those nodes back in the early mid-1970s were 2. It was a source implant and a drain implant. Over the course of the next 30 years, as the devices got smaller and smaller, it became more and more difficult to control the electron flow, in those transistors. And to do that, they kept adding the ion implant steps. So they were different angles, different doses, different energy levels in order to control that.
So by 28 nm, depending on the company, there were 25-30 different implant steps to make that same transistor that 30 years ago required two. So when they moved to FinFETs, FinFET happened, you know, after the 28 nm node. So let's say 14, 16 nanometer is the one that became very popular. FinFETs are three-dimensional; they're grown above the silicon. Planar transistors, which were the 28 nm type, you're actually creating the transistor into the silicon, so you would implant in to make that happen. When they started building the three-dimensional transistors, they would deposit that transistor. And so technologies like ALD and epitaxy were used to create the transistor. So implants fell off at that point.
We're starting to see implants grow again in advanced logic using with a need for material modification. So opposed to making the transistor, now the technique that they're using ion implants for modifies the physical characteristics of the materials, changes the edge properties, the stress properties of it, maybe the resistivity, so films connect better. And so there's growth in that market as well, so.
Okay.
A little history.
Yeah, that's good. Maybe just to double back on the size of the implant market, I think in 2022, third parties had talked about it, maybe a $2.5 billion-$2.7 billion kind of market size. Now, when I'd heard about what the 2023 numbers, which are kind of more recent, they actually put the the TAM in the, I think, $3.5 billion or so, mid-$3 billion. And, you know, it's a very strong growth rate, you know, 30%-40%, something like that, year-over-year. And, you know, just curious, what's your impression or take on the data, on sort of the way the market share was situated, kind of the growth in the various implant markets? There's kind of three main markets within implant.
I know it's the high energy segment, maybe kind of maybe the weakest growth of those markets. Maybe that has to do with flash being, you know, very weak last year. But anyways, any thoughts just in terms of how that market data came out for us?
So Brian's referring to, there are several research firms that put out market size and TAM estimates. So he's referring to one in particular that estimated the TAM to have grown from $2.5 billion-$3.5 billion over a downturn, which is quite unusual. Our belief, from our ground-up work, is that the TAM for 2023 was $2.7 billion, so about $800 million less than what that firm. There are other firms out there that are aligned with Axcelis on this. We're not sure the methodology that this particular firm used to get that, or if they started counting some other technologies that they haven't in the past.
But we do believe that it's around $2.7 billion growing over the next few years to $3 billion. Relative to the positioning, when memory is slower, you know, that can affect high energy. The areas that are most impacted by high energy are image sensors, NAND, and power devices. Those are probably the three biggest areas. Power was very strong in 2023, but flash NAND was very, very soft. Image sensors was extremely soft. So they probably balanced out, leading to relatively small growth. Medium current is a general-purpose tool. And so, you know, that tends to be one of the smaller shares. And then high current is the largest market share, usually representing around 50% of the TAM.
You know, those percentages really didn't shift too much in that data, so.
Okay, got it. So throw all that data in the bin.
Yeah.
I mean, usually when there's a, like a discrepancy, right? Like even WFE broadly, right? Like, it usually leads to discussion around China, which, you know, we should probably talk about here, 'cause China is there's less transparency and, and kind of different patterns maybe that for China OEMs versus. So but, but just even to cut to the chase, you do not have a significant competitor in China for implant, is that correct?
No. So I mean, the primary competitors in the markets, you know, are Axcelis and Applied Materials. We together make up, you know, roughly 80% of the TAM. There's two Japanese competitors that pick up most of the 15%-20% that are left. There's some small competitors in China that have been around for years, but really are producing very low-end implanters. So they're not impacting the TAM, and they're not really impacting the business of, you know, us or our primary competitors. So, you know, relative to that, you know, $3.5 billion dollar number, and it coming from, from China or, we have visibility to... You know, implant is our business, and so we're in all the fabs, you know, with all our competitors. We see all the opportunities.
If we were off by $100 million or, you know, even $200 million, maybe that's possible, but $800 million seem
Yeah
pretty, you know, it, it doesn't make sense.
I'd be a little upset with the sales team at that stage.
Yeah. For offline, the other thing would be if a company overstayed their revenue. That would be another way, but we'll, we'll save that for another discussion.
Yes.
Yeah.
So what we'll say is that Axcelis is only implant. The $917 million that are in those numbers and all the others, that's in all our filings. So-
Yeah
... so we are the only company that is sort of held accountable. Our numbers are public.
Okay, fair. Maybe, maybe, let's transition a little bit to the financial model, of the company. Maybe gross margins have been operating above a little bit above sort of that mid 40%, 45% level that you would expect to see when the business is annualizing to $1.3 billion
Yep
in revenue. I think that's the target for next year
Yeah
still.
Yep.
And so maybe can you, James, talk a little bit about sort of maybe the mix situation that maybe is causing gross margins to be a little bit ahead of schedule? And maybe also, I think your medium current platform, which I think is the biggest sub-area within implant, that seems to be the area for the biggest potential to help improve the margins in terms of where they are today
Maybe kind of talk about why medium current is so important to sort of the overall gross margin enhancement of the company.
Yeah. So, you know, as we look at margins, we ended 2023 at around 43.5%, and our expectations for 2024, although we didn't give a discrete number, were that we did expect margins within the period to be higher than what we saw in the 2023 fiscal year. We ended Q1 at around 46%. That was primarily driven by a higher mix of high-energy tools. Those are typically our highest margin tool set opportunities. You know, that was benefited to some extent, you know, by lower memory sales in the period. You know, we had, you know, sort of less than 1% or so of memory sales in the period.
So again, memory is typically going to be some portion of our lowest margin product offering in that space. In addition, we are seeing higher volumes of our CS and I aftermarket business, which is margin accretive overall. So, you know, we do earn margins in excess of our systems margin out of our CS& I business. And as Russell noted, with now sort of 3,000-plus installed tools in the field, as each one of those tools, you know, kind of come online in a more meaningful way and drive aftermarket opportunities, we do expect that to continue to be contributory and be a driver of margin expansion going forward. When we think about medium current, right? The challenge there is the differentiating the products.
The team has done a really nice job of segmenting the opportunities to go after specific, you know, customer use case scenarios that are really high value. And so we think about our silicon carbide tool sets, our power series very broadly, you know, we do get margins in excess of our, call it, standard medium current tool. And so, you know, we've increased the pace of our R&D, you know, investments, and that's specifically designed to create differentiation in those products. You know, increase the technological moat relative to the competitors, but also offer up upgrades and new system opportunities that drive ASP increases.
You know, we're not afraid to go and talk to the customers about capturing our fair share, you know, given the solution set that we're solving on their behalf. We've demonstrated our ability to do that, based on the historical differentiation that we've done. As we think about medium current, you know, that's really where we're putting some time and effort to find incremental differentiations to drive higher ASP there and improve gross margins.
Yeah, I misspoke. High current is the biggest market-
Yeah.
but medium current is
Yep.
Yeah. Sorry, go ahead. What were you going to say?
Yeah, I just. So actually, it's along that line of j ust wanted to add to that. It's not just medium current where there's-
Yeah
-segmentation and the differentiation. We've done-
Correct
significant, you know, margin expansion as a result of growth in our high energy tools for image sensors and, and power devices. And then in the high current space, development of the Purion H200 for both silicon and silicon carbide has, has also expanded margin, you know, in the same manner as what we've seen in the medium current. So, so it's something that the segmentation has proven to be, you know, very successful in this market.
Okay. Actually, one more thing on sort of the ASP since you mentioned how memory is sort of, you know, the indication of the optimism is that it's gonna pick up
Yes
next year for sure. We'll, we'll see about end of this year. But when you think about the mix being so severe, you know, you know, almost practically all non-memory or non-memory advance at the moment, when that normalizes, whether it's, you know, 80/20 or 85/15, whatever it is, could that be like a 100 basis point swing on gross margins or something like that? Just high level.
Yeah. So again, we would have to look, 'cause, you know, the... We'll give some, you know, I guess, anecdotal data on the, the facts of prior sales. So the prior peak of memory was around $150 million of total sales for us, which was roughly split 50/50 between DRAM and NAND, so $75 million apiece. Last year, memory contributed about 10% of our system sales, so approximately $90 million worth of total sales for us, which was all DRAM. So there was no NAND last year, and we don't expect any NAND this year. So as we look ahead, you know, again, it's a matter of, you know, when the recovery occurs versus if.
You know, so when, when the recovery occurs for us, we would expect to see, you know, again, meaningful contribution from both DRAM and NAND going forward. You know, as it relates to, you know, margin expectations, this is where our upgrade opportunities and, and other opportunities contribute on the CS&I. So there's meaningful install base in that area, given Axcelis' history as a memory systems provider. So we've, we've got some, you know, meaningful CS&I and aftermarket opportunities, given the current install base, plus, you know, the opportunity to upgrade our systems to the sort of, you know, newest technology sets that the teams are, are working on and developing to create better efficiency and throughput for those customers. And so, you know, it's a combination of factors, not just system sales that we, we look at to drive margin.
It's taking cost out of the BOM, right? Lowering the cost of the procured components, and also driving these incremental upgrade opportunities to take advantage of the installed base we have in place.
And anybody who's known us for a while knows that, you know, if you went back a number of years, we were basically shipping tools into Korea. That's kind of what we did. And now we've obviously got-
Yeah
multiple products and multiple customers, but we do have a really large installed base in Korea. And what you're kind of seeing now is that, you know, they're trying to take away DRAM capacity for HBM, and they're kind of getting four-wall constrained, so utilization rates are now coming up nicely.
Yeah.
still four-wall constrained, so that's where you start to see this upgrade cycle
Yeah
where anything to get more out of the four walls is valuable. So you'll see that coming through soon, 'cause I think, you know, a new fab's gonna take a little bit longer.
Yeah.
Yeah. I think they've cannibalized all they can. There's-
Yeah
there's no more meat on the bone.
Right.
Maybe, one more question for me here. You've been. You know, you do implant, you've done it well.
Next part, you know, all the regimes that have been at Axcelis over time, but you've also been transparent about... And there's, you know, opportunities to grow. You have a lot of evals in the field. You have to get unlock regions like Japan, where maybe you're under indexed in terms of share
Sure
but you've been open about looking at other adjacencies. Where are you in that process? And, you know, at the very high, loose level, what are the parameters that you're looking for as a complementary or good fit for the company?
So befor, I know you, you're gonna answer the undergone, but before we go there, so we do have an Investor Day in July, and we will-
Yeah
be talking about the next model, which will be implant only. So you're absolutely right. There's opportunity to advance logic and then opportunity in Japan. So there will be another model beyond the 1.3. So I guess I wanna kind of say that we still have lots of opportunities in our implantation. However, that does now give us also a little bit of time-
Yep
to do the right thing. So
Yeah, and with the one second left, I will say, you know, our goal here is to, again, today, Axcelis is a single technology, you know, company, right? Broadly. And so our goal here is to create a little bit of diversification, you know, across technologies, to increase our total addressable market. You know, that, you know, again, there's many sectors that have growing, you know, significant opportunities for long-term growth. We want to participate in those with other technologies. In addition, you know, we see that our global infrastructure to support, right? We're basically in every single fab today with our aftermarket team, our field services team.
So we have the opportunity to leverage that network, you know, pretty meaningfully and continue to expand our aftermarket opportunities, which creates a floor of stability and margin improvement opportunities on a go-forward basis. So that's sort of the rough framework here, but we are, you know, the aperture is pretty wide open, you know, in terms of what we're looking at. You know, we're gonna be thoughtful, deliberative, and we want to be value creative at the end of the day, and so we're not gonna rush to make an acquisition for the sake of one. We wanna make sure that we're doing it for the right reasons.
Got it. Great. Oh, Russell, James-
Yeah.
Doug, thank you so much for joining us.
Thank you.
You got it.
Thanks, Martin.
Thanks.
Thank you.