Logic parts of our business that are spending more. We're also exposed to parts of WFE that maybe some of our peers aren't in the wafer houses, in the reticle houses. When you add that all together, I think that given our relative exposure to logic foundry, which is higher than memory, it's positioning the company to, I think, do a little bit better than the broader industry this year. Maybe, if you look at consensus estimates where they are today, maybe mid-high single digit down versus industry, to your point, at down 15% or 20%, mid-40s WFE levels. I don't think it's changed much. There's certainly a lot of noise in the system, on the trade front and so on. I think the impact to our business has been minimal.
There's been some tariff exposure we've had for a while, but most of our sourcing of commodities or commodity parts in China usually go into broader subsystems and other places. The impact has been as country of origin changes and value gets added, the impact has been fairly minimal. Very little impact there. Certainly, customers get cautious and try to figure things out, you've got some of those dynamics. I think procurement organizations try to leverage the noise that's out there, we have to be disciplined. I'm spending a lot of time talking to my sales team about what's actually subject to tariff, what de minimis tests are, all these kinds of things, which I never thought I'd be having a lot of these conversations.
In general, it's holding together, I think, pretty well as much as we had said, and I think, in particular to the quarter, the quarter is shaping up as we thought and the second half dynamics as well. I think foundry in logic business will be relatively balanced across the year. It seems like it's shaping up that way. Have a little bit of improvement into the second half. In memory, more related to technology migration than capacity investments, and we tend to participate in that, so that's good for us.
Certainly not, I would not call it a recovery. I think the second half is stronger. I think the way we characterize the quarter-to-quarter apps in Orbotech, the quarter-to-quarter sequential path is the June quarter be up, then we'd see sequential growth through the second half of the year. We still see it that way.
Got it. If you look at within memory, the DRAM versus NAND dynamics, what you thought at the start of the year versus what you're seeing right now, has there been any noticeable change?
No, not really. If you go back to the middle of 2018, that's when we saw things change in a very dramatic way, particularly around the flash investments. You saw a number of new fabs effectively push most of the investment for those facilities from the second half of 2018, early 2019, in most cases, completely out of 2019. Despite that, I think for those customers, because they're doing technology work, we still have a level of business there that's been good for us, and it's been, I think, pretty consistent. Customers have been, despite all the noise in the system, I think, pretty disciplined about how they've managed capacity. They have to get or satisfy big growth today with more capacity than just shrinks. Those are bigger bets, and I think they're more sensitive and more disciplined to the timing of those bets.
Both turning on or turning off in this case, also, when we get to turning them back on as well. It's played out, I think, as much as there's noise in the system, it's I think pretty consistent with the way we saw it, and we continue to see it.
In terms of China demand, both sort of from indigenous Chinese customers versus the multinationals, give us some relative sense of how big your business is in China and if there is a way to kind of break it down in terms of indigenous versus multinational and how they're trending this year.
Yeah. The multinational investment has been mostly around one project this year, the rest has been indigenous. I would say that our China exposure overall, including multinationals, mid-20s. I've been pretty public that our expectations for native China business is probably about 10%-15% lower than what we saw last year. The last couple of years has been relatively strong. It's been very good years. In the ballpark of around $500 million.
Right
For the company on the systems side. That has held together, I think, pretty well, too. I think we have seen some puts and takes there. Initially in the March quarter, we talked about a project, the second phase of a project in memory that we thought would shift out of the year.
Right.
It looks like it will actually come into the year. I'd say maybe at the tier 2, I've seen some movement around some of the project timing. Overall, underneath the forecast, it's moved a little bit, but we still feel pretty comfortable about the business levels we expect.
Got it. Looking at longer-term trends, I think you have given a growth target longer term of 7%-9%. What are the underlying drivers, and how do you come up with that? If I look at the broader semiconductor industry, if I ask management teams there, they will say, "We think this industry is going to grow somewhat faster than GDP because there is content growth, there is data, and so forth." How do you come up with your growth targets? Do you kind of tie it to semiconductor industry growth, or is there something else that goes into that forecasting?
It's a good question. I think there's a baseline expectation for growth of the industry that's given that capital intensity has flattened out. It's no longer declining. There's lots of efforts in the industry to keep it from going up. We think that baseline industry growth for semi equipment through cycle ought to grow generally in line with where semiconductor revenue ought to grow. That puts you at a baseline growth rate of 4%-5%, let's say, worldwide GDP plus.
Right.
We have an element that we believe is related to share improvement opportunities. It could come in the form of process control intensity improvement opportunities as well. As we look at those opportunities that are out there from opportunities around EUV, opportunities around increasing intensity and memory, and then overall share in existing products. If you look at the market position we have, we're in the low 50% of process control. We don't participate in all of it. Segments we participate in, we've got a pretty strong share, but there is money that is being spent in process control in certain areas that we believe that product offerings that we have today and will have moving forward will enable us to gain additional share. I think that customers like second sources, but at the same time, I do believe that there is additional share opportunity for us.
Our service business is growing. It's growing faster than the core business. Historically, it's been around 9%-11%. It looks, moving forward, to grow in the same range. That incrementally will contribute about one to two points to the compound annual growth rate. You take that, you take the share opportunities, and you couple that with baseline industry growth, and that gets us in that range. If you add Orbotech into that, certainly in the 1%-2% to the top line, I think their businesses have the same GDP base growth rate expectation.
Right.
The share gain, I think piece sort of fits in that 1%-2%. There is some service opportunity. The PCB business has an imaging product line that has a very strong, which is the majority of the revenue, very strong contract component to its service stream. The other businesses don't have a lot of service. In particular, the specialty semiconductor part of Orbotech sells to a lot of the same customers. They have the same that we do, and we have the same customer list. We have infrastructure and capability around the world that a larger supplier can help enable the smaller supplier to try to get more out of that business.
I think, if you add Orbotech into it sort of contributes to those pieces, the one to two in the service, one, two in share, and that's how we get to the seven to nine.
Got it.
Against that backdrop, I believe given the operating leverage targets of the company, we ought to be able to deliver bottom-line growth of about one and a half times that top-line growth rate. One thing about the Orbotech business, it has lower margins overall as you aggregate them, and they're different across the three businesses. But I believe the operating leverage opportunity that exists as we grow revenue of driving 40%-50% drop through on the revenue growth in terms of incremental operating margin growth is attainable in that business. Just like we've been managing our semiconductor process control business.
Got it. Hypothetically, if next year WFE is flat, what would you be doing differently?
Well, I think the way we've managed the company through 2019, we've seen this as generally a digestion more than a downturn overall for us. We're talking about mid to high single-digit declines, that's not significant. Including Orbotech, we're talking about 30% plus operating margins or low 30 percentile operating margins. Against that backdrop, we believe that we're investing in some unique opportunities moving forward. I think the opportunities that we talked about a little bit around EUV and some of the memory challenges are multiyear challenges and multiyear investment cycles, that we are investing into those. We've always made sure that we invest in the right products, almost independent of what happens in the top line, and continue our product development because it's so important to the differentiation model of KLA.
Our margin profile is dictated by our ability to do things that our competitors can't, to introduce products at a cadence that our competitors can't, to tie the data and output out of the tools together so we can try to leverage the portfolio of the company. We will continue to do those things. To answer your question, I'm not so sure we'd do anything differently. As we look at next year, we say, okay, if it is to, I'm not so sure I have deeper insight into 2020, if you're talking about a flattish environment, we'd see the service business grow. I think, against that backdrop, I think you'd see our operating expense profiles probably level out a bit. I don't think we'd do things that were meaningfully different than what we're doing right now.
All right. What does the introduction of EUV mean for process control intensity? Does it increase your opportunity set in the industry? Does it kind of flatten it? What is the impact for KLA?
Well, there's short term and longer term, I think as it moves into higher volume production, there's a different opportunity. In the short run, certainly from a development perspective, any EUV development in and around the reticle ecosystem is being done with KLA systems.
Right.
We've enhanced some of the offerings that are in our existing 193nm reticle inspection offering, which customers are using, also using our high-end wafer inspection to do reticle qualification. This is how it's being done today across any customer that's doing EUV development. As you start to transition to more and more layers, you start to need production solutions. There's a High- NA and other dynamics that are probably playing out as you get to crossover in terms of EUV layers. When I say crossover, I mean more than 50% out in the 2021, 2022 timeframe. Against that backdrop, I think that the ecosystem around the reticle will be a challenge for the industry. There's issues with pellicles and what that does to production and also defectivity management. We're investing in a couple of products to support that volume environment.
It's our belief that you'll see an inflection around the reticle in terms of process control intensity. Certainly historically, anytime we've had significant changes in process technology, you see increases in process control just because the immaturity of the process and certainly at higher volume levels, that poses a challenge for customers. We think it's a great opportunity for us. Smaller defects is also enabled now that you have linear scaling. Linear scaling drives smaller defects, which is something that's always been part of the core capability of the company. Enabling that to happen and that driving value to solving those problems to our customers, we believe that is an opportunity for us.
Got it.
We're excited about it.
What is the trade-off between, let's say, if there is WFE growth next year, but it comes more from memory, where process control intensity is lower than foundry logic, but there is more use of EUV, where process control, you're saying you have more opportunities. Should we be worried about that trade-off right now, or does it kind of balance itself out, or how do you look at it?
It's a good question, I think there are fundamental dynamics about chip architecture design, how customers behave in memory and foundry that won't change. At the same time, foundry logic customers will spend more on process control. High-mix foundries will spend more on process control.
Right.
The chip complexity is higher. They have to deliver yielded product to tighter market windows and so on. In memory, you have less complication, you have more redundancy, you have repair. You have those dynamics. At the same time, there are unique challenges that are out there, we're seeing process control intensity in memory improving. You look at last year, 70% of the industry spend was memory.
Right.
We actually had a performance that was in line with the industry. We are seeing some improvements there. It will never look like foundry logic, at the same time, there are significant challenges, if we can solve these problems, customers will spend with us. We have products under development to improve, I've talked about it more publicly recently, is to improve metrology and in-stack metrology and using new technology that's available in the market that would replace a destructive process that customers use. As you're increasing layer counts and going to double stacking, that creates challenges. It creates challenges in the overlay area of registering layers on top of each other. There's opportunity out there if we can solve these problems. I'd like to think that we'd see continued improvement in process control intensity and in the memory space.
I think last year is pretty good evidence of it improving for KLA. I think historically, we would've been in a very different place, and I think these investments we've made, these changes that have driven more complexity, have been good for the company.
Got it. When you made the decision to acquire Orbotech, first give us a sense for why you made that acquisition. How has the integration process been so far? I think initially it came as somewhat of a surprise to the investment community, right? We are so used to seeing your business kind of focused in a certain direction with very high margins, and Orbotech has a slightly different structure, although it's very interesting when you actually reported Orbotech results, their business is very resilient, right? It's holding up among the best in the industry. Just help us understand why you made that acquisition.
Sure.
What has been the feedback so far?
No, it has held up, and I think the way we looked at the business is it really is three distinct businesses in one. There's a specialty semiconductor business that they have that is exposed to a lot of the, what I'll call the More than Moore markets of WFE. It's WFE that is spending more today.
I think going forward, has higher semiconductor content and probably less cyclicality given that it's serving automotive markets, it's serving industrial automation markets, IoT markets, and so on. It's a business that has designed for particular niches and has seen those niches start to inflect, and they've been able to have very differentiated positions and I think a margin profile that reflects that. We certainly give them more credibility. There's perhaps an opportunity for us to enable additional process control insight into helping them deliver better products to market. So we're excited about that opportunity, both from a strategic perspective, what it exposes to, but also the ability to leverage their position better as part of a bigger company with KLA and synergy with channel and service. There's a printed circuit board business.
Right
Which is mostly focused on imaging, which is direct writing of lines and spaces in high-end printed circuit boards. Imaging is technology synergies related to optics and related to lasers, so we think that there's some crossover opportunity for us to help enable not only their position, but you're also seeing this transition in the industry to more from high-end PCB boards to changing substrates to advanced packaging. So this exposes us to more of that. There are efforts in the company already, then efforts that Orbotech was doing to try to address this market. This is an area where customers are doing a lot. If you look at our customer base and what they do to enable cost or capability, they do front-end process node shrinks, which we all know real well.
They're doing a lot of things with packaging, with PCB board technology to enable, I think, new innovation and cost and so on. So this exposes us to those markets, which we think there's strategic and sort of technical synergy as well. They also have a flat panel business that, after 6 years of growth, is down this year. They're more exposed to the transitions and less to capacity, similar to the KLA business, that it's more technology transition-centric than it is capacity. It's a business that I think that we've got to do some work structurally with the cost structure to get it in the right place and ensure that we can deliver leverage over time through the cycles. I don't think those cycles matter that much to KLA, frankly.
In a business that's $200 million-$250 million level, it cycles up to $350 million or $400 million in upturns. I think in a $5 billion company, that doesn't matter. It's much more about through cycle growth, can we enable.
Right
them to solve bigger problems and get a structure in a place that we can deliver through cycle average profitability that makes sense. That's how we see it. Obviously, there's service opportunities. There's the ability to take the cash flow and deploy it through our model, do things with it that perhaps they couldn't do on their own, given the small company, three cycles, different cycles, and things like that.
Right.
The ability to finance some of the transaction with some, I think, pretty attractively priced debt. When we looked at it financially, I think it was a pretty easy decision for us to make against the alternatives. It made a lot of sense, and I think it exposed us to more of where we think innovation is also going. It's going in our front-end business, but we also see it going this direction, too, and we're excited about those opportunities. In terms of integration, there are things we're doing that are longer term in terms of leveraging our internal supply.
Right.
Doing some things with cost structure around manufacturing and so on. There's common supply chain. We're doing that work. From an integration perspective, in some ways, it's two separate companies, so it's a little bit more complicated in terms of how we're executing that. We're working through that planning now and would expect to achieve the $50 million synergy target as we start to see the bulk of it, at least from a run rate perspective, as we move into the first part of 2020.
Got it. The gross margin, there has been a short-term hit because of Orbotech's gross margins, I think for June are 58%, 59% or so. I think generally your gross margins have tended to start with the six. How soon do we get back to model? Do you need to do anything special, or is this a course of mix that helps gross margins get back to trend?
I think we had some mix issues in our process control business that drove our margins a little lower than the trend we've seen. At these revenue levels in our process control business, our model's been sort of 62%-63%, which we've been very public with. I would expect to see us get back into the target model range as we move into the second half of the year. Orbotech obviously comes in with a different margin profile, and there's a couple of points of dilution kind of related to that. I think that we're in that sort of 60% ± kind of range as we think about the year. I provided some guidance to that effect in the last call.
I think in some ways in our process control business, what we're shipping is what we revenue, and we have a lot of different products, and some products carry different mix dynamics than others in terms of the profitability model. Structurally it's, I think, pretty intact and we'll see it in the model range, if you will, as we move into the second half of the year.
Longer term, what is the right way to think of KLA's gross margins?
A lot of it depends on these mix dynamics and what grows, I think, as we look at the core business and the growth rate expectations we talked about earlier, I feel comfortable that you'll see a couple of points of dilution related to the new businesses. If we were 62%-63% before, it'd be a couple of points off of that. Our objective to drive the incrementals on the incremental growth of 40%-50% on the operating margin line.
Got it.
Operating income.
On the OpEx side, I think you got it to about $375 million, I think, for June. I think people were looking for somewhat lower than that. Can you help us level set, is this the baseline level of spending? Do we have synergies on top of this, or how do we look at the progression of OpEx from here?
Yeah, there was some confusion out there. I think there was some modeling that wasn't all that thoughtful about the Orbotech piece, and I think sometimes small companies exclude stock-based compensation.
Right.
It's customary for us to report it, and so that wasn't necessarily reflected in the models. We have seen some increase in our investment. I've talked about some of the things that we're investing in in terms of EUV new products to support the EUV transition, Vertical NAND, and so on. I think where we're at right now, I said that we see this sort of $370-$375 range moving forward. Given the comments we talked about earlier in terms of industry dynamics, I think we'll see spend starting to flatten out in our core business, and we'll start to see the synergies fall off on the Orbotech side.
I feel pretty comfortable that we'll see this level off, and we'll start to see some of the synergy value start to flow through as we move through, maybe not so much in the second half of the year, but certainly as we get into the first part of 2020.
Do you have?
Some of that's just the timing related to actions that you take. It takes some time before it comes out.
Do you have a longer-term operating margin model in mind for the combined company?
Well, again, back to what I said earlier about the impact on the public model from Orbotech, that it was about two points dilution. I would expect that against the backdrop of high 30s% kind of operating margin on our core business, we add in Orbotech, it drops in the mid 30%.
Mid 30.
I think you're talking about revenue levels in excess of $5 billion, we're probably in that mid 30th percentile type operating margin expectation.
Got it. One of the interesting aspect of the business is on the services side. How is your services business different than your peers? They have spares, they have refurb, they have a lot. I think your services business is actually very different than that. I think it'd be useful for people to hear what is that difference.
It is different. There's no consumables, right? That's a big part of process tools, is the consumable parts. Some of it relates to how customers buy process control. Part of the reason is if they buy process control, they try not to spend. It's one of those things that they have to spend, but they only want to spend up to a level, and they hope they don't have to spend any more than that. The trade-off they make is that they spend what they want to spend, and they buy the tools, and they run them pretty hard. They're very complicated systems, and they only buy what they need. They run the utilization at pretty high levels, and they need a contract support structure to be able to keep those tools up.
They also, in contrast to a capacity tool where you have lots of them, and you can offload capacity when tools go down, or just in general, you have enough, you can build capability to provide your own service. The level of complexity is not as high. For process control tools, you don't have that many. You have to match results. It's very hard for a customer to say, "Okay, I'm going to go invest in that capability." A, could I even do it because of the complexity? Secondly, there's just not the volume set to be able to get the economic return. They tend to rely on us for that, too. The supply chain is something that we have control over, so it's very hard to buy parts for KLA systems on the outside from third parties.
You have to buy them from us, then we have to go in and make sure that performance matches. They have to have identical results across different wafers, across different tool sets. Then we can customize our offerings in a way that allows customers to have very custom service offerings in terms of response times, part stocking levels, coverings, components, fab wide versus tool, and so on. As a result of all that, it's a bit of a long-winded answer, as a result of all that, we tend to see about 70%-75% of the revenue stream as contract. They run these tools for a long period of time. We've seen useful lives increase as we've seen more growth in the trailing edge parts of WFE. That's been a nice tailwind to our service business as well.
It's a stream we believe is accretive. Everyone sort of debates about how you do the accounting, but we've been very consistent, and we've seen with consolidation, the ability to drive incremental leverage on the service business. That contract stream has been very predictable. It's had one down year in the last 20 years, and that was 2009. Even in that year, it was only down 10% when the systems business was down considerably more than that. On pace for almost $1 billion now and growing at about 10%. No upgrades in that. It's all the service dynamics that I mentioned.
Got it. How does Orbotech change that? Because I think they also have a very interesting services component to their business.