Good morning. Welcome to the 28th Annual Needham Growth Conference . We're going to continue our morning schedule with a discussion with the management of Advanced Energy. Joining me today from Advanced Energy are Steve Kelley, President and CEO, Paul Oldham, Executive Vice President and CFO. Front row, we have Edwin Mok, whom many of you know, Senior Vice President of Strategic Analytics and Investor Relations. My name is Jim Ricchiuti, Senior Analyst in the Equity Research Department at Needham, covering advanced industrial technologies companies. So welcome, Steve, Paul.
Glad to be here.
So before I start the question, I'm going to lob it over to Paul for some great FD disclosure .
Great. Thanks, Jim. Just a reminder that our comments today are subject to a number of risk factors. You can find those risk factors in our SEC filings. Also, we provided our last earnings guidance at our earnings release on November 4th, and we will not be providing any updates to guidance as part of this meeting.
Okay. Great. So I think most of our audience is pretty familiar with AE. But I guess as we're all here this morning, a year after the '23 Needham Conference, probably dare to say that I think from folks I talked to, there was a much better appreciation, I think, of the company's operations in the data center computing business. So before we start, we're going to spend a lot of time on that. But before we do, how about just outlining to give us a sense of the revenue breakdown, and then we'll jump into the deeper Q&A.
Sure, and maybe I'll just use Q3 as a reference point. In Q3, our semiconductor equipment business was roughly 42% of revenue. The AI data center, data center in general, was around 37%, then the rest was industrial, medical, telecom, and networking, and so basically what we do is we sell into the high-end power delivery market. The roots of the company are in semi equipment. Our two biggest customers are Applied and Lam, but over time, we diversified into other high-end applications in data center, industrial, and medical.
Good. Okay. Let's hit a topic that's on a lot of folks' minds. On the data center side, you actually gave some indication of what your expectations are for 2026, 25%-30% growth. And that's on the heels of what is expected to be a doubling of the business in 2025. So help us get a better sense of what the breadth of the design wins that you've gotten in 2025, and just your overall confidence in providing this outlook already for 2026.
Yeah. What distinguishes the AI data center market versus the conventional data center market is the pace of change, and so what we're seeing is every year there are new GPUs and other silicon, and that's driving higher power requirements every year. It's more or less a continuous process. What that means is we need to come out with something new every year for these customers. We need to be engaged early, and so the revenue forecast we have for 2026 is based on designs we secured in 2025, as well as some continuing business. We have a high confidence in the forecast. The biggest variable is the mix. Because as we look into 2026 and our customers look into 2026, there are other constraints in the system, supply chain constraints, where we may see sudden changes in the forecast.
So it's our ability to keep up with those changes, which is really important this year.
But going back to last year, Steve, we had expectations to start the year, and we ended up seeing something that looked significantly different and much faster. So what were some of the usual upside drivers that really drove that outperformance versus what we were thinking?
I think our customers were successful, and they were able to build their data center capabilities faster than they originally forecasted. The good news is we're able to react quickly and build additional capacity. We have two primary sites today where we build these data center products. One's in the Philippines and the other one's in Mexicali, in Mexico. We invested quite a bit of money in both capacity and capability in those two sites. We have the capacity necessary to meet the 2026 forecast. In addition, we've also built a new factory in Thailand. That factory is ready to go later this year when we get the signal. I'm happy that we have the brick and mortar necessary to meet any upsides we see this year.
With the capacity you added, do you think you gained share with this kind of performance?
It's hard to say because I haven't seen the numbers out there, but it feels like we're doing pretty well. The objective we have is not share gain, though. The objective is to keep margins very healthy in this business. So the share gain is a byproduct of that strategy.
Yeah. And we're going to touch on the margins in a moment. But if we think about the customer base that you're dealing with, you have diversified it, but you're obviously selling to some of the leading hyperscalers and across various platforms. The conversations you're having with these customers, what kind of forecasts are you getting as you went last year, as you went through the year? Are they significantly different per customer? And just trying to get a sense as to how good those forecasts have been.
So at the top level, the forecasts have been pretty consistent, I would say. But what's changed is the mix within the forecast. And so again, there's a lot of pushing and pulling going on depending on what's happening with their end customers, as well as what's happening with their supply chain. And so the capacity we're building is pretty fungible. We could shift it as necessary to different models. So most of the discussion is not the absolute number because it's still healthy, but it's what's going to make up the shipments over the next 12 months.
Yeah. And that mix also does have some impact on margins, right?
It does. But most of the products we ship to the customers today have healthy margins. So there are very few laggards left in our portfolio.
So you talked about design wins from 2025 that you expect to be ramping in 2026. Talk to us a little bit about the design win activity now as it relates to what you would expect would have some impact in 2027?
Yeah. I would say the design-in activity r emains very intense. So we have a targeted set of customers. It's a relatively small number of customers in data center. But the amount of projects within each customer is quite significant. So what we try to do to maintain the pace is to reuse as much technology as possible from generation to generation. And where we need new technology, we look across the company first and see if we have any technology blocks we could use to help speed the transition. But the biggest challenges, I think, in data center are in power density because the space in the rack available to us does not change appreciably from generation to generation, but the power requirements go up. So the question is, how can you generate that power, same amount of space, reliably? And it's an engineering challenge.
There have been questions as well on design wins that you guys have talked about, 800 volt. What can you say about designs in that area and the activity you're up to?
Yeah. We have a number of development projects for 800 volt type applications. But I would say that's not going to generate any revenue this year. I think assuming 800 volt develops as expected, we'll see initial revenues in 2027, but they will probably get more serious in 2028 and 2029. But I think we're well positioned there in a number of different areas. So no matter which way it goes, I think we're there.
Yeah. Somewhat similar question as to market share, but do you have some sense of your win rates that you've been seeing with some of these key customers?
Yeah. I think the win rates for us. It's close to 100%. So we're very selective about where we engage and there are customers that we deal with today for the most part. And so we choose to engage, we generally win.
Yeah. You alluded to the fact that you feel like you have adequate capacity in place. What about any other issues as it relates to components and tightening? What are any things that we need to be mindful of as we think about what's happening in the market right now?
Yeah. So we went through an extended training course during COVID on component shortages. So we learned a lot. And so we designed out as many of the poor-performing suppliers as we could. And where we couldn't design them out, we put in place strategic inventory. So if you look at our balance sheet, you'll see a fairly good amount of raw materials inventory. Much of that is strategic because we don't have confidence that the suppliers can ramp up quickly. And so I think we're pretty well positioned going into any upsides this year, whether it's data center or semiconductor or industrial medical. And certainly we have the capacity, as you mentioned, in the company to handle upsides. But we're taking, I think, an aggressive approach on inventory so that we don't get caught short.
Paul, this may be one for you. I'm wondering how investors should think about the margin profile in the data center area of the business. In the past, we've had some questions about how much lower is it than corporate average. But given the mix and given the volume, we've seen things move up. So how far are we from corporate average? Can you talk about that?
Yeah. I think historically in this market, when we entered it, margins were quite low. In fact, I think you can read as we acquired Artesyn in the early years, margins in these products were probably in the teens. We brought those up, as we've said recently, approaching corporate average. We haven't broken it out exactly. But that's a function of the strategy that Steve described, where we have a much more targeted approach that plays to our strengths rather than trying to just play with everybody else. As we look forward, our goal is to hold those margins. I mean, ideally, we could improve them a little bit. Certainly, as we bring new products to market, we'll try to capture more value as the products move to higher value, higher power ranges. Maybe there's some opportunity to do that.
But we want to keep them kind of in this close to corporate average range or better. I think the good news to that is as this grows and becomes a larger percent, then it doesn't have as much effect on the company from a mixed perspective. And in fact, if you look at, as Steve described, our business in data centers up to 37%-38% of the total, and yet we've been able to meet our margin goals. To put that in context, if you went back a year or so ago, that was in the low 20s. So we've significantly increased the volume here, and it hasn't had much of an effect on overall company margins because we've been able to absorb that through manufacturing efficiency and other things. That's our goal, is to sort of keep it close to corporate average, capture value wherever we can.
But we recognize that this isn't the semi market where it's a very enabling technology. It's important technology, but one where we think we can get very good returns on our investment and continue to see meaningful growth.
Let's turn to the semi market. You guys sounded a bit more optimistic coming out of Q3, which is consistent with what we're hearing from others, some customers of yours, other suppliers to the market. Just from where you sit, what's driving that optimism?
Yeah. I think the key factors driving the optimism are in leading-edge logic, obviously TSMC leading the way there, and in memory with SK Hynix and Samsung making some pretty strong statements. So I think that is leading to a wave of WFE forecasts that have gone up based on what's happening, especially in DRAM and leading-edge logic. So as I look at our forecast for 2026, I would reiterate that we think it's a growth year. And our confidence level has certainly gone up in the past few months based on statements from our customers and their customers. We're also pretty enthusiastic about the success of our new products. So we're going to see a greater contribution this year from Everest and eVoS, our new technologies, but it's really going to become significant in 2027 and 2028.
Okay. And has the way you're thinking about 2026 in terms of the stronger second half, which a lot of folks are talking about, and I assume there'll be some benefit from the new products as well, will we begin to see that, Paul, in the guidance for June quarter? How early will we be seeing it? It may even be in the marketplace, correct?
Yeah. I think we'll just have to see how it plays out. I mean, coming out of SEMICON West this year, which was in October, I think we were already seeing more bullish comments in the market. But I think coming out of that, the view was, that's great, but it hadn't translated really to orders yet. And it does take some time from sort of the bullish outlook to translate back down to the OEMs and ultimately to the sub-suppliers. And I think if you look at our commentary in November, we talked about Q4 and Q1 still being kind of sideways with really seeing more upside as we moved into the second half. So since that time, the commentary externally has continued to be positive.
I've told people, it seems like this time when people are projecting a recovery, there seems to be a little more support foundationally about what could drive that, which is great. But there's a number of factors that affect the timing. So we continue, as we said back then, we talked about really a second half recovery. The pace of that we'll see. As things become more positive, we'll see if that can pull in a little bit from second half.
Okay. Maybe you might want to just remind the audience. You guys tend to be a little bit more underindexed in that, where there's been some, I guess, decent upgrade business. But talk to us about just on you touched on a little bit. We obviously have been seeing the reports on memory, but just your sense, is that the biggest driver in terms of what has changed in the near term?
I think the biggest driver is really on the DRAM side because of the consumption of DRAM and High Bandwidth Memory modules, right? And so that's where most of the action is today, is in DRAM expansion capacity and in leading-edge logic expansion capacity. I think what we're seeing in NAND is more of an upgrade proposition. And you're right, we're less indexed to that because in NAND production, there's more of a bias towards dielectric etch, which is not our strong point today, but it will become a bigger factor over the next few years.
You brought up the new products, but let's talk a little bit about that. Remind us again what kind of contribution you were anticipating this year from eVoS, Everest, and NavX.
Yeah. So eVoS, Everest, and NavX, we launched those technologies in 2023, and they've been well accepted by our customers. But it's a long process, right? So you qualify our direct customers, the OEMs, then they qualify these technologies with their customers, the fab operators. What we said was we were going to see $10-$20 million in revenue from those technologies in 2025 and that we expected to see an acceleration in 2026. And I think based on what I've seen, I think it will become very significant in 2027 and 2028 because these technologies are really enablers of increased throughput, increased yield at the leading-edge nodes, both in logic and memory. And you'll see increased incorporation of those technologies as the fabs build out the capacity.
You're more excited about one of these technologies or is this the case you love all your children?
I love all the children. They're both really interesting technologies. They give the customer a lot more flexibility, a lot more capability, and so each of our customers is making their own choice. Some are choosing Everest, some are choosing eVoS, depending on a variety of factors, but it's nice to have both red and black covered on the roulette table. That's our position right now with eVoS and Everest.
WFE, what are your thoughts as we look at that market over the next few years? You guys have demonstrated an ability to grow in excess of that, and maybe you might want to remind us of your targets in that respect.
Yeah. I think our target has always been to grow faster in WFE. We've used the number 1.3x WFE growth. And that's over time because obviously any particular year, you may have other factors that either amplify that or reduce it somewhat. But I think if you look in the past three to five years, we've been able to keep up that pace. There are a couple of things working in our favor. One is the acceptance of the new technologies. That helps, of course. The second is as these geometries get smaller, the number of etch and dep steps increases. So that's called etch and dep intensity. And that's where we operate largely with our eVoS and Everest technologies. So that means more of the WFE is spent on etchers and dep equipment. So that's good for our customers, and it's good for us.
Yeah. I think for the first time in a couple of years, we're at this conference and talking about your industrial and medical market, which we've been waiting for a turn. And it's about, what, about a third of 15%-16% of revenues. It's a decent-sized market, but we're finally starting to see some recovery. So what's your sense in terms of the inventory correction and just the overall demand environment there?
Yeah. So we'll just take a step back first and talk about industrial medical. We think looking at our strategy, we have three market areas that we compete well in. One's semi equipment, one is data center, and the third is industrial medical. And industrial medical is important to us because over time, you're going to see ebbs and flows in semiconductor and data center. They're inherently cyclical markets, right? So we see industrial medical as our most stable pillar. So looking at the past four years, however, you would say, "Hey, not so stable." But they went through a huge inventory correction after finally getting their parts after the COVID supply chain issues. And so it changed the market for a couple of years. So today, what we see in industrial medical is, I think, most of our customers, the inventory issues seem to have abated.
It's taken a couple of years. If I take a look at our distributors, after six quarters in a row of reduced inventories, we think they're at an equilibrium level, and we're seeing order rates increase. We've seen revenues increase every quarter in 2026 after Q1, so we think the market's recovering. It's recovering in a measured way, but I think it's reaching equilibrium this year, and I think we're well positioned to gain share based on the design activity we've had over the past two years, so one byproduct of our strategy is even when industrial medical was down, semiconductor was very steady, data center was strong, and so we kept investing heavily in R&D across the board, including I&M, and that's put some space between us and our competition when it comes to new technology, new product offerings. That's going to help us moving forward.
And fair to say I&M is a diverse vertical in verticals. I probably characterize it. But it was interesting in Q3. I thought you said that you saw some activity coming out of the aerospace and defense market. And just because we're hearing a lot about that at this conference, what's your exposure to that market?
Right now, the exposure to aerospace and defense is relatively low. But that's going to change. So where we participate is in the COTS space, the commercial space for military. So with the increased focus on speed and agility in the defense sector, I think that's playing to our strengths. We have great technology for many applications, and we're able to get it to our customers quickly. And so you're going to see our aerospace and defense revenue go up, I think, significantly in 2026.
Yeah. Any sense as to what that represents if we think about that I&M market? It doesn't have to be a dollar number, but just in terms of the pecking order.
I don't really know at this point. I just know it's all an upside for us right now.
Okay. Good. One of the things that you guys have talked about, even going back to the analyst day, is that the level of your revenues from sole source products, which 70% or so, but you have a goal of pushing that up, getting it to 80%. I'm curious, does it vary much across some of these markets?
Yeah. There is some variation. So if I look at the semiconductor equipment area, almost 100% is sole source. So that's a case where there's very little second sourcing going on because of the time it takes to tune our solution to the customer's needs. If I take a look at industrial medical, the vast majority of that is also sole source because we engage in the high end of the market, and most of those customers want something special. And so we customize our standard platforms for those applications. And then the market where we've had the most exposure to multiple source has been in data center. That's where we started a few years ago. Too many multiple source opportunities that drove prices down and margins lower.
So today, we've made quite a bit of progress in converting the data center from a multi-source business to a single source or one of two sources type business. So I think we're making good progress on the march to 80% of our revenue coming from sole source products.
Okay. And that also plays to some of the margin improvement. You touched on capacity. As we think about where you're adding capacity, bring us up to date. You talked about Thailand, Mexicali. Thailand is probably a good place to start. Where do we stand with that?
Yeah. So we broke ground in 2023 on a new factory in Thailand. Actually, our first purpose-built factory for the company in Asia. So it's 500,000 sq ft on one floor. So we're going to have the ability to build all products in Thailand. We expect to start with data center products there. That'll happen likely sometime in the second half this year. So as we reach a point in the Philippines and Mexicali where we start to bump up against capacity limitations, we'll activate Thailand. But yeah, I think we're in good shape from a capacity standpoint.
Any headwinds that we could anticipate from the ramp-up of Thailand in technology in the early days?
Yeah. We're always concerned when you start a new factory, but one of the benefits of starting with a high-volume product like data center products is you're able to absorb a lot of that fixed cost more quickly than if we started with a high-mix, low-volume type product like semiconductor.
I'd also add that the cost of Thailand was always contemplated in our model. We put out a model a year ago that said by 2030, we'd get to $3 billion in revenue and $15 per share with margin improvement that went from the mid-30s% to around 43%. So what you're seeing with Thailand is we're actually getting to the volume levels we anticipated where we would need that capacity. It's just happening a little sooner. So it's contemplated in the model. Now, you could see quarter to quarter some variation as you fold that in, but we will fold it in as we have capacity needs. We don't have to bring the whole thing online at once. So I'd say broadly speaking, that incremental investment is already contemplated in our overall model and shouldn't be too dilutive to our results.
While we're on the subject of margins, gross margins, Paul, remind us a little bit again about the impact of tariffs and where we stand with that and how you're navigating that.
Yeah. So obviously, tariffs, it feels like it's been around for a while, but it just started in April. Frankly, it's been pretty complex to navigate those waters. I think we've done a pretty good job of it as a company. We're pretty well positioned geographically. We've been very aggressive in trying to look at supply chain flows and other things when we work with our customers to not pay tariffs unnecessarily based on how products are routed. We also have Mexicali, which is fortunate for us because we were able to qualify a lot of our products coming out of there under USMCA. So today, the tariff impact to gross margins for us is about 100 basis points. It varies up and down a little bit by quarter based on mix and timing, but that's about the range that it's been around where we expect it.
Now, having said that, we're working actively both to mitigate that cost as much as possible, as well as we believe over time, there are other levers we have around manufacturing efficiency, material cost, and other things that we can do to offset that increased tariff cost. So we haven't changed our long-term goal as a result of that, but certainly that's been a headwind in the near term, and frankly, we've said we wanted to exit this current year around 40%. Our guidance for Q4 was 39%-40%. If you exclude the impact of tariffs, we would actually be above that 40% level, so over time, we'll continue to work it, but it's about 100 basis points of impact.
Yeah. You gave us a target, organic target model looking out to 2030. I think revenues are about $2.5 billion. Yeah. I wonder if you could, and again, this touches on the margin question, but apart from revenues, when you gave that, we didn't have this issue around tariffs. And so how comfortable are you with some of the measures you just talked about, about getting to that gross margin target that you talked about?
Yeah. If you look at that target, it was really based on moving from the mid-30s up to this 43%, and we made great progress on that. So as I mentioned, Q4 guidance is between 39% and 40%, so we've come up in the neighborhood of 300 basis points over the last five or six quarters. That was based on three things. One, improving our factory footprint. We said that was 200, 250 basis points. We think that's largely accomplished exiting this year. So that's been a big part of improvement. And I think we can continue to wring out more costs from the factory and factory efficiencies. We've talked about we've incurred some ramp costs as data center has grown. So I think there's still more to come there, but we've benefited a lot from that. Clearly, volumes are up. So we've gotten the volume benefit.
At this level of revenue in sort of the mid-to-high 400s, every $50 million of additional quarterly revenue ought to add somewhere around 50-60 basis points of margin. So we expect to continue to see that. The biggest opportunity going forward beyond that is really the change in product mix that comes from the new products. As our new products get into the market and get more traction, become a bigger percentage of the total, and we continue to move up from 70-80% in terms of sole source business, we think there's 2-300 basis points of margin improvement that comes from essentially better mix or better value in the product portfolio. That'll occur over this product cycle. So think over the next one year to three years. So we're still very much feel like despite tariffs, we're on track.
Obviously, we'll adapt where we need to. We'll try to drive upsides to that, but we feel confident in our ability to get to the 43% margins as we move closer to that $2.5 billion organic revenue.
Has the change in the profile of the revenues and the growth you're seeing in some of these markets changed the way you're thinking about OpEx investment? Or is this a case where you're just going to get better leverage with the higher volumes?
Yeah. We're going to get better leverage, and we expect that. So we've talked about growing OpEx as a model at 50% of the revenue growth rate. We actually did much better than that if you look at our guidance for 2025. But from a planning perspective, that's how we think about it. We want to grow OpEx about half the rate of revenue growth. That contemplates the fact that we don't have to invest at this same rate. Obviously, there's variable costs. We will continue to invest in engineering. But even in engineering with our modular architecture, we're getting much more engineering efficiency in terms of the number of products and derivatives that we're getting out of our engineering investment in both semiconductor and data center. So on balance, we feel good about that structural model of being able to see good leverage.
If you put that through with our margin improvement, we think we should see between 35% and 45% operating leverage at the operating income level. So if revenues grow $1, we should see somewhere between $0.35 and $0.45 fall through to operating income.
Balance sheet's in great shape. $750 million in cash. Debt around $565 million at the end of September. There's another piece to that target model for 2030. There's an inorganic piece. Remind us of what that is? How active is that pipeline? Is there any change in the priorities just given when you started talking about that a few years ago?
We're still on the hunt. I think the biggest issue has been valuation. Obviously, there's been a mismatch in expectation of buyers and sellers. I think that valuation gap has narrowed, and so we have hopes that we could do a deal of significance this year, and the focus remains the same. I mean, there's two areas we're trying to beef up. The first is industrial medical. We like to become the number one player there. That's going to require some M&A work. The second is technology tuck-ins, where we identify areas where we have a gap in technology, we'll go off and work with that company and buy them if they perform, similar to how we did with acquisition last year.
Yeah. Good time for a question or two. Anybody in the audience have any? Okay. With that, we'll wind it up.
Thanks very much, Jim.
All right. Thanks.
Thank you.
Thanks very much.