Good morning, everyone. Welcome to Day two of Citi Tech Conference. It's my pleasure to welcome Bren Higgins, CFO of KLA Tencor. We also have Ed Lockwood from Investor Relations. Bren is going to make a few remarks opening remarks, and then we'll do the Q and A.
Bren?
Thanks, Atif. Thank you for having us here today. So before I get started, I'll just do the normal safe harbor statement. I'll make some forward looking statements, and those statements are subject to risk. There's a whole list of the risk factors outlined in our SEC filings.
You can also find those on our website, klatencor.com. So 2018 continues to be a very strong year for KLA Tencor and for the industry. For the industry, it's certainly driven by a lot of memory investment, probably 65% to 70% of the WFE, the wafer fab equipment market investment this year is memory. Greater than $50,000,000,000 2017 was greater than 50,000,000,000 2018 looks like it's greater than 50,000,000,000 as well, probably up somewhere low mid single digits. There's some adjustments that are happening here in the second half around some of the capacity planning that's been discussed and I think pretty public.
So we'll see where it ends up. But generally, I would expect it to be in that ballpark. And early indications on 2019 as you see very similar trends into next year, segment shift, more foundry logic investment next year, continuation of strong momentum in flash, NAND flash, three d flash and then DRAM, probably some adjustments on that front. But overall, a very healthy environment. On the second half of this year, I know there's been some questions about that.
We had guided six weeks ago, we thought that the September would be the trough of this current digestion phase that we're seeing now, mostly second in the second half and driven by adjustments to capacity, mostly in the DRAM space, but also some logic pushouts into next year. We thought the December would snap back pretty strongly. When we look at it today, we see September still is the trough. We think the December is up sequentially. Feels like it will be up a little less than what we thought.
And we had indicated the second half of the company versus the first half from a shipment perspective, now this is shipments, but from a shipment perspective, would be flat to up low single digits. And I think right now, given what I see today, it's probably flat to down a few single digits or so. So, some adjustments still mostly around DRAM, and we think that business shifts into next year. So on the margin, modestly weaker, I think, trajectory wise in the way we had laid it out in terms of September, but also growth into December, still pretty intact. So for the company, we expect revenue growth in low the low double digits into for calendar twenty eighteen against the industry backdrop I talked about, so a nice outperform relative to the industry for the calendar twenty eighteen.
China has been good for the company. EUV investment has been good for the company. Inflection and Wafer business and reticle business has also been good for the company this year. So we think it's good momentum as we head into next year, more foundrylogic investment next year, which is a sweet spot for KLA, EUV development, all of that that's happening next year and a continuation of these dynamics that I mentioned around wafer and reticle. In terms of Orbotech, the Orbotech process is ongoing.
We got the Korean approval a few weeks ago. So the only outstanding approval that we're waiting for is from China, and we're working with regulatory authorities in China, through the process. The process is going according to normal course. I don't think there's anything particularly unusual about the process or unique in terms of what we're the questions we're asking, the dialogue we have. We would expect to close it in the fourth quarter, which is what we've said.
So no real changes on that front. We're feeling a lot of momentum around integration, and integration planning is picking up as we move closer to close. And we look forward to integrating that business and broadening the served markets of KLA Tencor as we move into calendar 2019. So with that, I will happy to take any questions.
Thank you, Brent, for those prepared remarks. Just on the update that you're providing on the December, curious if the DRAM weakness you're seeing, is it a continuation of the weakness you saw in the September? Or are you seeing more kind of a broad participation from the DRAM makers adjusting to the correction in the market?
Yes. No, it's a good question. And there is some churn, but I would say it's more puts and takes. We saw one large project push out into next year, and so that's what drove the change. If it weren't for that or it was just minor, I would have probably not said anything.
We aren't guiding the December. But since I gave second quarter or second half commentary in the earnings call, I thought it was prudent to update it. But, I still expect September to be the trough. It's down about 10% at the midpoint in terms of the guidance we gave. And I do expect shipments to grow into the December.
And based on what we see today, it looks like with the shift into the first half, that the first half looks pretty good of calendar twenty nineteen.
Sure. Thank you. And then on the foundry side, we've seen a global foundry make an announcement that they're going to disengage on the leading edge and focus more on the lagging edge. Has that had any impact to your outlook?
No, it's very minor. I mean most of those adjustments we've adjusted out over the last six months. It was becoming pretty clear that some of the commitment to seven nanometer and beyond was waning. So from an outlook and expectation perspective and how we see that flow through the P and L and the ultimate guidance that we're given we've been giving very minor impact, if any at all. Okay.
So, with that,
and you guys have talked about exceeding wafer fab equipment growth, this year and next year. And even with those adjustments, I think you'll still outgrow this year and into next year. Can you just talk about the key drivers that are contributing to your outsized growth this year and your expectations into next year?
Sure. Yes. No, and it's important to reiterate that for calendar twenty eighteen, we had said low double digit revenue growth. And even with what we talked about, I don't think that, that changes those expectations. So you have consensus estimates $4200000000.04300000000.0 dollars in revenue, and I would expect us to be in that ballpark for the year.
Yes, it's interesting. I mean we've seen certainly an improvement in process control intensity in the three d NAND area, and that's been very good for the company. The metrology requirements in three d NAND are higher. So between thin film and optical CD, we've seen some improvement there. Customers are continuing to one of the ways to control defectivity is to run monitor wafers through unpatterned inspection tools before they start to process.
And so we've seen part of the growth we've seen in the wafer segment has been driven by the three d NAND sort of cleanliness requirements. There's also been flatness requirements that we've seen not only adoption of those types of products in the fab, but also in the wafer houses. And it's an area of our business that we don't talk about a lot that when we talk about memory intensity, there's actually a pretty big driver for us that happens in wafer houses because wafer houses drive wafers. Memory is what drives wafers. And so there's a lot of investment that's happening there to add capacity to support the really robust memory environment we've seen and can expect to continue to see.
But also in three d NAND, wafers have to be very flat. They're increasing specs around flatness. And so there's wafer metrology requirements that customers are buying in the wafer houses, but also in the IC fabs. So that's certainly been a good driver for us. China has been great for KLA TENCORE.
We had this year was memory centric from a shipment perspective. Big DRAM projects shipped right at end the year in the first part of the year, three d NAND project middle of the year, we're shipping, this quarter, and a little bit last quarter into another DRAM project. So it's been more memory centric, but a very good driver for the company. Small fabs, greenfield fabs ramping these up, a lot of driving to try to progress process technology. So market share has been good with those customers and also adoption to try to get that insight into their process development.
So that's been very good for the company. Customers are preparing, certainly, on the foundry side for seven nanometer ten and seven nanometer designs to come through the fabs. And so you're seeing a step up in tape outs for seven nanometer designs in the foundry, and so that's driving radical activity. EUV development is also driving radical activity. So that's been an inflection for the company as well.
And I think on the wafer side, some of the dynamics I talked to continue. And for the next couple of years, I think on the reticle side, we'll continue to see strong performance from that business because you'll start to transition to EUV and the EUV requirements are going to be pretty significant there as well. So, those are probably the biggest drivers for us. I mean, I'm pretty we're pleased to see the performance of the company through this year with so much memory investment. And then as we move into calendar 2019 and you see foundry intensity, which is about twice that of memory, that it should be good for the company as we move into 2019 with more foundry investment, which we expect to snap back pretty strongly as we move into the first part of the year.
Great. And just to dive a little bit deeper into those drivers. On the memory side, we are hearing about yield issues at 1x, 1y nanometer on the DRAM. Three d NAND continues to struggle every time you make a jump to the next layers of the sandwich. How are you, kind of benefiting from some of these yield challenges that we're seeing on the device side on memory?
Well, there's certainly lots of problems out there, and sometimes there aren't always solutions for them, but certainly customers, there's no shortage of problems. We talked about some of the issues in three d NAND, and we expect those to continue. And as you start to add layers, you go to double deck. Overlay is going become increasingly challenging, which we think is another driver for us in three d NAND. On the DRAM side, more multipatterning schemes, ultimately, EUV at a layer or two will drive, will impact yields and certainly drives opportunities for process control.
One of the things about DRAM is they're not getting as much big growth from scaling as they used to get. So they're having to add wafers to meet big growth demand. And so you're actually starting to see capacity adds. And this year, probably 100,000 or so capacity adds in the DRAM space. So that's good overall.
In general, so you've got technology challenges, you've got capacity requirements. So no shortage of opportunity for us. We've got a tool in development internally on the three d NAND side, bouncing between the two a bit, to do channel hole measurement. The channel hole is, today is managed or measured, destructively with a cross sectioning wafers to be able to make those measurements. So we have a technology that we shipped a couple of tools.
We'll start to hopefully see a little bit of revenue next quarter or next year in 2019, more so as we move beyond, that we can actually use X-ray technology to manage or to measure the channel hole in die and in process. So, we think that's an opportunity for us to get to serve a market that's really being served by nonprocess control, through cross sectioning and some of the other lab activities that are happening today.
And then that product, is it for a certain number of layers? Or can you, kind of use it in reverse for prior generations as well for three d NAND?
Yes. It's primarily for it's a three d NAND product, and it's basically the technology enable us to measure through the stacks. And so you can actually go through the materials demand to measure the hole that goes down the center. So yes, in some ways, independent, right, in terms of its usefulness. Okay.
And moving on
to foundries, we're seeing almost a bifurcation in leading edge versus lagging edge foundry models where on the leading edge, have TSMC and Samsung. On the lagging edge, have now global foundries and UMC and even the Chinese foundries. In terms of your market share, are you more exposed to one or the others? And how do you benefit from these two kind of separating trends?
So I would say our business today is probably 60% leading edge, 40% trailing, and I define trailing as 28 and above. I think where you drive that line is kind of important for that metric. Now it used to be 75%, 80% leading edge. So it has been a significant shift. The market shares tend to be very consistent, whether we're at 28 nanometer node versus what we're selling at the leading edge.
And in a lot of cases, they're adding incremental capacity to existing facilities. So we will and a lot of times, it's this is what we had before, and we want to continue to add those tools. We have invested in adding some new capability into older tools. We restarted older tools to meet those markets. So, it's been a great opportunity for us as a company.
And I think the margin profile, even on the older tools, tends to be pretty consistent. It's part of the strategy of the company and discipline around maintaining pricing platform to platform sort of enables you to once you start to have a lower price in one platform, it carries forward into the next one. So you have to be pretty consistent in terms of how you think about that. So, it's been a good opportunity for us certainly in new fabs and in China. The comment I made around memory also fits in foundry, where they're actually ramping new fabs.
And so support from a big supplier like KLA that can ramp to those needs, can provide, application resources in the field that we can support, a full fab is also a competitive advantage for us there as well. Last year, we booked about $170,000,000 in trailing edge automotive fabs. So interesting about the demand and where it's at that we'd see that level of activity from fabs that are basically supporting some of these. We talk about end market diversification a lot. This is an area where pretty significant level of investment supporting that.
And Orbotech probably exposes more to those kinds of opportunities, frankly.
Yes. So EUV is expected to go into production next year for seven nanometer fuse steps. I feel like your narrative on EUV has gone from being agnostic to EUV to actually positive as you guys are benefiting from a lot of these stochastic nuisance defects that are coming up on EUV. Can you just talk about kind of your latest assessment on how you're benefiting from EUV and what kind of upside you're seeing from introduction of EUV?
I'm not sure we were agnostic. And we were concerned about timing and at what node you'd actually see EUV. And we were, I think, a little more cautious around the timing over time. And frankly, I think we've probably proven right about how that would play out. But linear scaling has been great for KLA 10 core.
Smaller defects are great for KLA. And that's the thing that we haven't had over the last few years is that Moore's Law has moved in the transistor. It's moved vertically. It's been less about scaling linearly. So EUV, we believe, will be has it's been good for the company, but we also believe that process control intensity should be able should stay flat ish as we move into EUV over time.
Anytime there's a meaningful technology transition in the industry, there tends to be an inflection around process control because they're dealing with new materials, new architectures, new defect mechanisms and so on. So if you go back historically, and I've been in the industry for eighteen years, you go back to whether it was copper introduced or wafer transitions or different gate structures and all these different things that have happened in the industry, we've always seen an inflection around process control in the near term as that happens or early on. Now over time, customers optimize and the processes mature, and so we see it settle back. But we do believe that going forward, we should be able to maintain the intensity of process control in an environment where EUV and the ecosystem around EUV will be pretty expensive for our customers. So that's a positive for us.
Now to your point about the opportunities, I mean, stochastic defectivity, which is random defects where you have inconsistent patterning, is driving demand for products like our Gen five, where you need full wafer coverage because you have a randomness to the defects you need to see across the full wafer. You can't use really evening technology when you're looking for one defect in 1,000,000,000 contacts on a wafer. So and it could be spread out in a random way. So it's been a good driver for us, and we think that it's one of those things that as you we improve optical inspection, the sensitivity of the signals and noise, the computational methods within it that will be a driver for that product as you see EUV start to get adopted into production. So it's a good thing for us.
And what are your Gen five expectations for this year and next year? And are there any additional end market applications for Gen five outside of a print check and EUV based technologies? So we're seeing more Gen five adoption in DRAM, too. So some of
the defect mechanisms you talked about, some of the high aspect ratio measurements that are becoming more challenging for customers in multi patterning is an opportunity for us, too. Gen five, we've end of this year, we'll probably have 17 or 18 tools in the field being used for engineering analysis and defect discovery still. You'll start to see Gen five deployed with sort of combination with Gen four more production use cases as you move to five nanometer. So as we move into 2019, certainly towards the 2019. I would expect Gen five year to year to grow.
I would expect that we'll probably see $100 $150,000,000 of incremental Gen five next year. We're introducing a second iteration of the product that adds some additional capability, adds speed and so on, and that's coming out to support. And it's typical with our product introduction where the new technology comes out, customers use it where speed isn't necessarily a need when they're doing defect discovery or engineering work. But as it becomes production viable or need for more capability, certainly around speed, we introduced that capability. And so we're you'll see that come out as we move into next year, and you'll start to see more production use cases for that product.
Okay. What about back end? Back end packaging, it's seen quite a few challenges, the copper pillar and integrated fan out. Can you just talk about your product portfolio in the back end? Or is there anything exciting happening in the back end where you guys can benefit materially?
Well, so Orbotech exposes us to more of that, and that's a good thing. We have a little bit of exposure in macro inspection in the back end, but a secondary position there. So, one of the things that's intriguing about the Orbotech acquisition is this sort of more than more concept of customers looking for capability and cost in many different ways, whether it's front end process node shrinks, whether it's packaging, whether it's PCB board enhancements, and new processes there. So there's a lot of things happening there that customers are looking at to try to drive capability to improve cost. The Orbotech acquisition exposed us to more of that.
So the packaging opportunity is interesting. I think as we get into, it working with them, I think it opens up some opportunities for us. Today, limited with the present portfolio that KLA has. Going forward, I know there's a lot of investment there, a lot of interest to your point. I mean, TSMC enabled a node transition or an upgrade for their primary customer through a packaging change, not through a process node shrink.
So it shows you the capability and the focus that's there and the opportunity that's there if you can solve those problems.
Sure. Brent, just staying on the Orbotech topic, can you just remind us about the rationale behind that acquisition? You guys have been a pure play on equipment for a while. And with this acquisition, you have a play in the display market now as well as in the back end more materially. So just to kind of talk about three or four big drivers.
Obviously, the deal is accretive, but it takes your gross margins down a little bit. Just walk us through.
Yes. So I think when I look at their business, I mean, their business, their market leading position, we believe we have complementary technology in the road maps of our products that are applicable to their business. You have complex systems. You have a service business that, over time, provides a nice stream of revenue over the life of the tool. We had infrastructure in and around that, so we think there's opportunities there.
And you have rising complexity in those markets. I talked about some of the other drivers earlier about just how customers are behaving, where interest is and how that's driving complexity in these markets. So it's, it makes it more attractive to us. I think the thing that we have become much more aware of over the last number of years, and if you look at the way I've talked about M and A opportunities, Orbotech fits that and the things I just described. And if we believe we can take technology to other markets and solve problems and drive better products, then that will be that's sort of what we will be the primary driver for what we do.
We know we're good operators. You can look at our business model, and I think it's a testament to not just the product positioning and the market leadership, but also to our ability to operate the company and drive improving profits over time. If you look at the ADE business that we bought back in 02/2007, gross margins were in the mid-50s, in the mid-60s today. So it takes time. We have to grind it out.
Sometimes it takes a couple of product cycles, but we believe that we can bring core capability and competency around operating improvement. And ultimately, I'm not married to the ratios. Every business has a different business model. I look very closely at this business in terms of its market leadership position. It's not a cost business.
And while the margin profile is different, if we can drive higher growth rates out of it than our core business, improve the margin profitability margin profile over time, then it makes sense for us to go do it. There's a cash on cash return objective that we have to hit in any deal we do. If you look at the capital structure and some of the decisions we made on that front over the years, we understand that as an alternative. We have to deploy the capital in the company productively. We look at the alternatives, R and D, M and A versus other shareholder return considerations.
In this case, Orbotech fit the bill in terms of the thesis of what we were trying to drive.
Yes. That's a very good point on gross margins because that has been a very positive driver for you this year. Can you just walk us through what's driving the continued strength and incremental gross margin improvement of your core business? And how should we think about the long term gross margins and operating margins?
Yes. So on the blended side, it will probably impact gross margins by, I don't know, three points or something like that as I look at it. It's interesting. Just a few years ago, our gross margins were 59%, now we're 64%. I don't think that there's anything in particular in our business that's driving gross margins to inflect one way or another what I mean by that, there isn't certain products or certain customers.
I mean our product gross margins vary, but the product mix has been fairly consistent. Gross margins across different customers are consistent. There's no real differences there. So we've certainly been able as we've scaled the business, we've been able to drive not just a product introduction cadence and maintain pricing on products in terms of products that have come out and the problems that they've solved and be able to understand the returns to customers and discipline around that pricing. But the operating model underneath and the new product introduction has been very strong for the company.
Service incrementals have been also good. We've seen the service business inflect. We talked about trailing edge. That's driving our service business. And so the incremental margins to service have been pretty positive.
So that's been a factor as well. But I think the product introduction dynamics, that we've seen, new product pricing, product introduction incrementals are probably the biggest drivers. And we're in that 64% today. As I look at next year, it's probably in the same ballpark. And I don't see it changing anytime saying, you have quarter to quarter mix variance, but in general, I think this trajectory is pretty good.
And I'm not sure how much higher it gets in 64%. It's pretty good when you've got incremental gross margins of 60% to 70%. So we're kind of there, average 64%, 65%. But I don't see anything that's going to put change the profile in a meaningful way, to the negative as we move forward.
Great. Let me see if there are any questions in the audience before I ask further questions. Any questions? Over here.
With respect to Orbotech, I'm just curious how concerned you are about it being bogged down in the trade issues between U. S. And China the way Qualcomm and NXP was?
So it's a good question. Very different deal, right? That deal was bogged down for some time. The one thing about this deal is there's no product overlap. It's a pretty straightforward deal from that perspective.
No overlap in top 10 customers. So the conversations we've had, in working and going through it with counsel have been predictable and consistent with what you would expect through this process. And I think the dialogue has been pretty open. So, to my earlier comment, we feel very good about the timing that we at the time, we thought it would close in the fourth quarter. We've been very consistent with that, and that's still my expectation today.
Questions? What is the funding requirement for the
acquisition? And what's the timing of that? So, it's basically a sixty-forty deal, 60% cash. Well, yes, so it's basically so it's a $3,000,000,000 deal. It's about, $2,000,000,000 in cash and $1,000,000,000 in stock.
There is no financing contingency. We can we have the cash on the balance sheet close the transaction. At the time of the announcement, we announced that we would do some financing, to enable us to effectively buy back the shares we're going to issue in the transaction over the twelve to eighteen months or so post transaction. And our Board authorized $1,000,000,000 buyback contingent on the closing of the transaction. So we would expect that when we close a transaction, we'll go out and raise that financing and begin to execute that buyback.
But given that it isn't needed to close, I didn't feel like I needed to raise the money any sooner than, and there's always the questions around timing. The buyback is contingent on close, so I figured I'd wait until we close the transaction before we go out and do that. We would expect we'd go into the public markets and raise the financing that way and go from there.
Questions? Bren, how should we think about your capital allocation priorities post Orbotech deal closure? And Yes.
No, it's a good question, and it's an important part of as we talk about Orbotech, what's changing and what's not changing. So what's changing for KLA is we're serving a broader market where we believe we can bring incremental technology and capability to those markets. What's not changing is our focus on operating profitability and driving profit growth over time and also how we deploy cash that the business generates. It's not a capital intensive business. It doesn't change my and I've been pretty, I think, explicit over time about how we think about how much cash we need in the business, roughly $2,000,000,000 how much leverage we expect to run with roughly 1.5 to 2x growth.
And given the ability to generate high twentieth percentile free cash flow margin, we can return about 70% to 75% of the cash flow we're going to generate over time. In the near term, given the buyback discussions we just have, it will be higher. But over time, that's generally the model that we look at. So what Orbotech also provides is incremental fuel for that plan. Now of the 70% to 75%, 40% is our dividend.
Our dividend target is 40% payout ratio. We just raised it back in February, a pretty significant raise. So it's currently $0.75 a quarter. And it was our ninth consecutive annual raise there. So we will target the dividend.
We think businesses like ours should have a meaningful payout to the shareholder given the predictability of it, the growing diversification of it. That 40% will be in dividend and then 30% to 35% or so will be, in share repurchases, which we will execute because of the I think the way I described it is more principles based. We will execute it generally of it flows through in a systematic way over time. We have the ability to be opportunistic certainly with the unwinding of some of the cash positions related to tax reform and others is providing some fuel for us to actually go and do more in the near term, but then also the $1,000,000,000 that we'll do as a result of the close of the transaction. So we'll do more in the front end.
And we will do it probably at a more accelerated pace. I'd like to ideally execute all of it as we get through 2019, but market conditions will drive how we look at that. But we expect to maintain that target, sort of 70% to 75% as we move forward. And I think our business has not just the market position, the strength of it, the consistency of flows, the service business, all the things that enables us to, I think, commit to that and maintain it over time.
Sure. China tariffs, China trade wars, a lot of tensions. Can you just talk about the impact to your bill of materials terms of your exposure to stuff that comes from China, if it's significant? And I feel like Kaley is in a great position to be a partner with China as China ramps up these fabs because they have to improve their yields. Any kind of change you're seeing in the domestic Chinese projects' behavior, whether they want to accelerate their spending or slow down?
Anything you can share with us.
Yes. No, the spending has been pretty consistent. And as I look at the last couple of years, we've been averaging, from an order perspective, about $600 or so million of business from native China customers. I think that continues into next year at approximately the same level. The mix has been different between the segments.
This year was a little bit more memory centric. I would think foundry is probably sixty-forty as we move into next year. But the behavior has been consistent despite all the noise generally out there. I mean to your point about being a partner, I think we've been a good partner, and we've been able to support and provide resources. And I think that has played through in share and performance of the company, through this time frame.
And we haven't seen anything really change on that front. In terms of supply chain, we have some components that we source in China. We I would call them more lower value components. So when you go through and do the assessment and look at what's actually coming back to The U. S.
And then staying in The U. S, it's pretty immaterial to any of our numbers. And so we're monitoring. We haven't made any real significant changes yet. Certainly, with some of the suppliers we work with, we can change those sourcing strategies to try to avoid that.
We do do some manufacturing in China, but most of the products we ship in China actually run through Hong Kong. There's incremental value added in Hong Kong, so they're technically not subject to these, to the tariffs there. So, very minor exposure to the business and not anything that I would say, well, we should adjust something financially or in terms of expectations
based on that. Okay. And services business has been good for all equipment suppliers last few years as well as for you guys. Can you just talk about your prospects in services and as well as software upgrades? What portion of your business is dependent on software upgrades and kind of energy model longer term?
If you can just talk about that.
It's always been difficult getting hardware customers to buy software, especially when you have a market position like ours, where we're ForEx our nearest competitor. So typically, where we've added software like capabilities, whether it's algorithms, whether it's simulation software, whether it's correlation classification packages is we've it's been part of the tool and part of the tool purchase. So it shows up that way. But certainly, software content is higher, although it doesn't really work without the hardware. So it's continuing to be a challenge there.
I think that model has worked for us in terms of embedding it or including it with the tools that we sell. The service business has been it's interesting. I mean we thought and we adjusted our longer term growth expectations. We thought the service business would ultimately slow down. We had 8%, 9% growth for a long time.
Very consistent growth, 75% of it based on really service contracts, so 75% of the revenue. So very predictable. There's not as many of these tools and fabs, so customers need them up all the time and given the complexity of these complex systems, hard for them to manage on their own. There's also no consumables, and so they can't source it from third parties. So it's been very sticky from a contract perspective.
We've seen it actually start to inflect. I mean, service is approaching $1,000,000,000 growth rate closer to 10%. So if you would have asked me a couple of years ago, I would said, 6% to 8%, 7% to 9% type through cycle or growth over time. Each year, I keep ratcheting that up. And if you ask me today, I think it's probably 8%, 9% growth off of a much higher base.
Customers running the trailing edge fabs harder and the installed bases, generally running them hotter has been good for our business. And again, I think the growth of the installed base is driving improvement or driving growth there as well. A lot of those fabs are full on the trailing edge. And so customers are looking for incremental capability to the older tools, and I mentioned it earlier. But if you add speed or other upgrades to tools that help solve some of those problems for customers, we're doing some swaps where we'll pull older tools out and put a newer generation tool in to provide incremental capacity.
So all that's been good for the service business, and I think it continues to grow. And I believe it's an accretive stream to the operating income of the company, to the operating margin of the company. So nice profitability on it as well.
Great. With that, I will wrap it up. Thank you, Ben, for coming to the Citi Conference.
Thank you. Thank you for having me.