So thanks to all of you for taking time to spend a little bit of time with us today. Just a few housekeeping items before we get started. Obviously, there's some refreshments in the back. Help yourself. The restrooms, if you didn't notice, are out the door and to your left. The presentation we're going through today is on our website, and there are some instructions on all the tables to get the Wi-Fi connection there. So hopefully, that takes care of all the needs you have for today. Our agenda: we've got a pretty packed schedule. So after I complete my opening remarks here, Peter Kirlin, our CEO, will come and give an overview of our strategy. Then that will be followed by Chris Progler, our CTO, to talk about our technology and how we're using that to continue to innovate and grow.
We'll provide some outlooks for the high-end markets that we have. FPD will be provided by H.K. Park, who heads our FPD business in China and Taiwan. Frank Lee, we'll talk about the IC Logic business. And then Pete Broadbent will discuss the memory business. After that, John Jordan will come and give us a financial overview. And then we'll have some time, hopefully, to answer your questions. We should be done with all that around noon. Lunch will be provided. If you would like to stay and have lunch with us, you are welcome to do so. Obviously, we'll be making some forward-looking statements today. Those statements do have some risks and uncertainties. Please refer to our SEC documents for more disclosure about those risks. And at this time, I will turn the floor over to Peter Kirlin.
Okay, Troy, thank you very much. And thank you, everyone, for coming and your interest in Photronics. Let's see. So I will begin with a brief reprise of our earnings release yesterday. It really, I think, will largely be a description of what we've done to reposition the company today, be sitting more or less at our record revenue run rates, which provides, I think, a very good platform to move into the initiatives that we have underway to drive our company forward and upward over the next three to five years. And finally, I'll conclude with why we think Photronics represents a compelling investment opportunity. So as far as Q2 goes, revenues stepped up to $131 million, driven by China and driven by high-end IC. We saw what is the typical 50% or better drop-through at the operating margin and gross margin lines.
Net income was up 60% to $0.15 a share. And of course, our cash balance is quite strong, which is quite important because we are now, as we approach the second half of the year, accelerating our investments in China to build out the factories there. So why, over the last year, have our revenues been dropping when the industry has been increasing? And what have we done to remedy the situation? Well, if you look back, we hit a record revenue run rate of $531 million in the first quarter of 2016. 2015 was our record revenue year at $524 million. First thing that happened in the second quarter of 2016 was the Micron JV ended. And they pulled a lot of their reticle business inside. Second half of 2016, Samsung opened a captive, not an IC, but FPD.
They ramped that through the first half of 2017 to volume. Finally, in the first few months of calendar year 2017, our largest customer in Taiwan lost significant 28-nanometer business. Those were three big hits representing about 20% of our top line. That's not an easy hole to dig out of. What we did first was to chase the silicon. We chased the 28-nanometer logic business. We found the home of that silicon at another large Taiwanese foundry. We started to make a significant number of the reticles that we had lost. We also found a 3D NAND flash business with a large multinational company in China based on the Micron bit cell. We now have recovered that business. Finally, one of our existing foundry logic customers has a lot of 14-nanometer business. Now we're benefiting quite nicely from that.
So that generated about half, or filled up about half, of the hole. The other half of the hole was filled up by aggressively developing the China market. We're building factories there. Developing the business has both a tactical as well as a strategic importance to us. So we really doubled our efforts in China. And between the two, we have now reached more or less run rates that are close to record levels. So lots of hard work, lots of places. And I'm very proud of what the team has done and what amounts to be a relatively speaking short period of time. So why invest in Photronics? First of all, we're the merchant market leader. And we haven't gotten there by accident. We've been in business 50 years. And as we've grown, we've developed a set of core competencies that we believe separate us from the competition.
It's really this that, over time, is what's created a company that continues to win. Given where we now sit, we're well poised to take advantage of the dislocations that are in front of us, not just in China, but also in the display market. Finally, based on the investments we're making, we expect to see revenues of $630 million. That's 40% up relative to 2017, with a quadrupling of EPS by 2020. That's a year, well, it's two years from now. It'll be the first year where both China facilities should be fully ramped. At least the first phase of the ramp should be complete, stated more succinctly, I think, also more accurately. Because I don't think the first wave or we don't think the first wave will be the final wave in China.
So for those of you in the room who are new, I see some new faces. Tried to capture Photronics on a single slide. That's not so easy. But here we are. As I said, we're number one, the number one merchant. We have a global footprint that's as large as our next two largest competitors combined. And as we build out China, again, this will separate us from them even further. We've got three market dislocations we're investing in simultaneously. We've done that in the past, but typically, it's been one at a time. And we've never had customer commitments in advance of the investment. So standing here, we have contractual commitments for approximately $300 million, which should allow both of those facilities to ramp the profitability quickly. And finally, growth is great, but it's not great if you need to dilute your shareholders in order to achieve it.
And with the cash we have, funding the growth is not problematic. So there'll be no dilution required to grow the business. As a reminder, where we sit in the overall manufacturing flow, our customers, people like UMC or BOE or CSOT or Samsung, doesn't matter whether it's a chip or a display, they sit at workstations, do their designs. Designs come to us. What we do is we make masks. And what they are, they're the layer-by-layer negatives that go into the ASML Canon or Nikon scanners that project the circuit pattern on either the silicon wafer or the FPD panel. That's what masks are. If it's an FPD panel, that's, believe it or not, a really big IC. There's maybe six layers, the simplest display, now the most complicated one approaching 15 layers in a mask set to make one complete display. IC, simplest, maybe 20 layers.
Today, most complicated, bucking up close to 100 levels, so IC masks generally, and Chris will discuss this, more complicated, and what that does is create the chips and the displays that sit in almost everything we carry around or drive. Your microwave oven, when you punch it in the morning, it's got semiconductors in it, so the chips are basically the oil of the information age. They're not going anywhere. They're ubiquitous. Now, we've been at this for a very long time, 50 years almost. It's kind of remarkable. There aren't too many companies that have been in the semiconductor business for 50 years, and what's enabled us to stay in business for 50 years is, over that entire period, we found ways to grow. The first 20 years of our life, we were a small private company, capital-constrained, largely servicing U.S.-based customers.
But nevertheless, it's kind of up and to the right. In 1986, we did an IPO. And we went through what I would call a golden age of growth for Photronics. We did more than a dozen deals in this period. And we went from 50 to, this is actually $387 million in revenue. And at the end of that period, we were the number three. Oh, I have to go back to the microphone. We were the number three mask maker globally. So rapid acceleration growth. Phase III last 15 years, still growing. And I might just point out that every other merchant during that period has been shrinking. So we've been growing when all our competitors have been contracting. Now, middle is much more volatile. But these three peaks are actually meaningful because they correspond to technology dislocations that we either created or exploited.
I'll come back to that in a subsequent slide because I think it's very important that everyone in the room understand why we keep going up and to the right because I think it's poorly understood. So again, still up and to the right. And finally, what do we expect? Well, what we expect by simultaneously investing in three technology dislocations is a re-acceleration of our growth. So it's effectively back to the future and back to the golden age of growth for the company moving forward. So what's the secret sauce? How have we been able to do that all those years? And the answer is we have built core competencies into our business successfully over time. We've got better doing things that allow us to win. And we've institutionalized them. So phase one, small private company. What did we do?
We found ways to make high-quality masks inexpensively. That's what we did. We did it largely by buying used tools, using minimum adequate materials. And we had small to mid-sized customers. And we gave them inexpensive masks. That was the first core competency we developed to separate ourselves. Next, phase two, go from small supplier to the number three global mask maker, acquire big customers, people like TI, ST, at the time, Motorola. They expect a lot more from us. What did we do? We listened to them. We really listened to them. So someone like National Semi would say, "Oh, we have the factory in Maine. It's running ahead of the factory in California. We might have, oh, 100 reticles in the line at the time." They say, "Reticle number 50 and 73 need to be one and two." We'd say, "Sure.
Figure out a way to do that." In fact, those reticles might even be running in different factories in our global network. We found a way to do that. So during phase two, we became the photomask industry's leading service provider. And when you talk to customers today, still 30 years later, that's what the brand represents. We represent to the industry the leading service provider. That's our hallmark. But that's clearly not all of what we do. So move on to 2006. As I said, we peaked in 2002, the industry's leading service provider, low-cost producer. But we were a technology follower. So in 2006, we did a JV with Micron to enter the high end of the business. We acquired lots of know-how. But to build those high-end masks, the tools are very expensive, as everyone who invests in us know. The materials are expensive.
And we took this customer intimacy core competency and actually flipped it back to our suppliers. And by doing that, we have found ways to make those high-end masks less expensively by managing how we service those expensive tools and what we do and what materials we choose to use to build our reticles. In addition, we took this process know-how and pushed it back into our mainstream factories. And although our yields today are not 100% in the mainstream fabs, they are approaching 100%. And we have efforts underway continuously across the globe to drive our mainstream yields to perfection. So stepping up to today, we've taken customer intimacy again, bundled it with our technology. And we've developed PORs that actually allow our customers' yield on the silicon to go up. And that's not done by cranking down the reticle specs.
That's done by understanding the mask signature and how it affects what prints on the silicon or on the panel and tuning it. And for anyone who is in the business and knows customers, wafer yields are the crown jewels. That's not information that is readily disclosed. But it's disclosed to us in many cases because we can help our customers make it better. So that's kind of a long-winded story. But it's quite important because these competencies have stepped in over time. And even after they've stepped in, we found ways to refine them and to make Photronics better. And this is what separates us from the competition. And this is what has enabled us to grow when the others have shrunk. So this slide is a quite important one because now I'm going to go back to a point I made earlier.
That is, I think, how we grow, what we do to grow, is not understood, generally not understood by our investors. It's not really understood, well, I think, because we have always discussed what we've done as a point in time. But in fact, there's been a method to the madness and how we have systematically stepped our revenues up. I'm going to take you back to 2002. Remember, this is the end of phase two of our growth history, the M&A growth. Revenues peaked at $387 million. Five years later or four years later, we hit a new revenue peak, $455 million, $60 million up approximately. That was done by having the right photomask technology on the ground in Korea for Samsung and LG to exploit the G6, the G7.5 substrate-sized transition in that market. That's what drove our revenues up.
And I'm going to discuss several more bars on this chart. They're all different. But what makes them the same is what enables us to be successful is having the right technology in the right place at the right time for the right customers to take advantage of it. This was an industry transition. We didn't create it, but we certainly took advantage of it. The next transition we created for ourselves, and that is in 2006, as I said earlier, JV with Micron. 2008, we opened the nanoFab in time to take advantage of the DRAM to drive our revenues up $60 million again. That was no accident. That was not an industry cycle. That was a deliberate set of actions that we took to drive our revenues higher.
The next peak, again, JV with DNP, not high-end memory, 28-nanometer logic ramp in Taiwan drove our revenues yet higher. So three different dislocations or disruptions, different partners, revenue step up, step up. What we're doing now is instead of trying to manage one dislocation, the company is actively managing three simultaneously. We're trying to take advantage of the G8.5 to G10.5 display substrate transition. This is primarily for TVs. Trying to take advantage of the LCD to AMOLED transition in mobile displays and taking advantage of the fact that from a geographic base, the semiconductor industry is shifting from the rest of the world to China. So those are three dislocations that we are aggressively chasing simultaneously. If you look at these, again, 60, 60, 12, why only 12? Well, industry peak, industry peak, industry trough. So we are definitely affected by the cycles in our industry, no doubt.
But we don't reach higher growth plateaus based on industry cycles. We reach higher growth plateaus by putting the right technology in the right place at the right time. And we've been doing it consistently for the last 15 years. It can't be done in a year or two. It takes four or five years. But we've demonstrated it repeatedly. We can do it successfully. So that's only $106 million why because there's three. And the answer is these are three dislocations superimposed on top of an industry downturn because we don't know when the industry turns down. But we think it would be foolhardy to think that somewhere between now and 2020, it does not. So since we're looking out far beyond our regular guidance horizon, we're guiding for a number which we believe is quite conservative.
If this happens to happen at 2020's industry peak, this is a very conservative number, at least 10% conservative. Shifting gears to China or expanding on China. Again, we've been hard at this since 2015. We're targeting 2020. We've been hard at developing the business. 40% CAGRs are not easily achieved anywhere in the semiconductor industry yet. We're demonstrating it. Right? And if you look at this, black is IC, red is FPD. So you can see that our efforts to develop the FPD business in China are really now accelerating. And H.K. will talk more about that in a few minutes. But there's not a single customer in China that makes FPD panels in volume where we are not presently doing business. We have them all. Likewise, on the AMOLED side, the China business, there's nine fabs. We're qualified at every one. And we're POR in most cases.
So there's been a lot of work already done to develop the China business. And this is quite important, as I said earlier, tactically, but strategically. It's not the bookings. It's the customer relationships. That's the coal that we need to fuel the boilers in Xiamen and Hefei that are going to propel our company forward over the next three to five years. So strategically as well as tactically, extremely important work being done today, a year ago, a year before that. So this is an old slide. I think many people in the room have seen it now many times. But it's quite useful, I think, in describing where we're investing in China and the similarities and differences between the two sites. So our IC facility is being built in Xiamen, investing about $160 million total.
But of that $160 million, $40 million of it is tools that we're moving from other places into China because some of the business that's going to be built there is going to shift from either Taiwan, maybe a little from the U.S., a little from Korea. But anyways, some of the business is relocated. So likewise, the tools are being relocated. This is a JV with DNP. And it addresses every market segment in IC: logic, memory, high-end, low-end, all basically a full market basket of products. FPD, different, located central China, same investment, all new tools, wholly owned, really focused on G10.5 with secondary focus on AMOLED. And by the way, this G10.5 business, we have no G10.5 market share today. This is all incremental. So the tools are all incremental. That's what's old. What's new is for us, again, it's not build it and they will come.
They have come to the tune of a $300 million contractual commitments, and we are building it, and anyone who is in this industry that makes high capital investments knows that the return on the investment is highly impacted by how quickly you ramp that facility to both first break even and then profitability. We have the business in hand. We need to do that as quickly as we can make it work, so I started out with a discussion of how we repositioned Photronics to be now basically running at record levels, moved on to discuss how we expect to drive our revenues up over the next three-to-five years. Beyond that, what do we think we're going to do to grow, and the answer is we think the right way to grow beyond that is business development.
What gives us confidence we can execute on that? Well, if you look at the photomask industry, it's a history that's rife with consolidation. We were born way back here, as was Micron Mask. The first wave of consolidation was done by the DuPont company. It was done in the mid-1980s through the early 1990s, more than a dozen deals. Those of us that have been in the business a long time, it was quite a joke back at the time because when DuPont got into the semiconductor industry, they owned Conoco Oil. They were $50 billion in revenue, which was larger than the entire semiconductor industry, one company. They had deep pockets, obviously, and they did the first wave of the roll-up.
We then, as a public company, took over the leadership mantle and did, as I said, more than a dozen deals in the 1990s second phase. If you look at 2000-2010 next decade, all the remaining players were active in cleaning up the business during that period. Since 2010, there's only been two deals done. They've both been done between Photronics and DNP. And in both cases, they're JVs where we have both management as well as board control, so given that and given that the other three guys are all Japanese, we believe we're in a very good position to drive the next phase of consolidation in our business, so the past is the best indicator of the future in our view, so if it's a photomask-driven deal, what are we looking for? The answer is capacity, technology, and perhaps market share.
Any two of the three are enough. Of course, three is better. If we step out of mask making, it will be into adjacencies where our core competencies will add value. And in any event, whatever we do, we'll need to demonstrate deep double-digit ROIC and will need to be accretive. So we've walked away from several bad deals over the last few years. But we're pretty confident the right one is going to materialize in the timeframe that we're thinking about. So finally, who's going to deliver on that plan? And the answer is the senior leadership team is sitting in the room. You're going to hear from all of them today. There's really two categories of people here. One is a group that has more than 10 years with Photronics and more than 30 years in the industry. So these are all gray hair, experienced individuals.
We have two newbies to the organization, but they also have gray hair, so they have lots of experience. And they actually give and bring, I think, a very healthy perspective to those of us that are old and jaded in the mass market. But one thing that's the same with all these people is separately, they're winners. And collectively, they're all of a single mind and a single purpose to make the company win. Every day when I walk into the office, I feel lucky to be working with them. And the reason I feel lucky is I know I can count on them. And you can count on them too. I assure you of that. So finally, why invest in us? Because you're investing in the global leader. It's the team that can make lemonade out of lemons on one hand or drive aggressive growth on the other.
We are targeting 40% growth in revenue and a quadrupling of EPS to $0.80 by 2020. So the next person to the podium is our CTO, Chris Progler. Chris.
Thank you, Peter. I'm going to go a little bit through Photronics' technology portfolio, how we guide what we develop, kind of markets we serve, and why we think we have some advantages over our competitors for sure, and why we're positioning ourselves to meet some of these growth targets. But first thing, to just talk generally about trends in the industry, I'd say next three to five years that guide how we think about technology development. First, Peter already talked about China's IC and display aspirations, investments. You'll hear a lot about that.
In this case, the technology team looks at what we want to develop locally in China, what we want to ship into China, what kind of technology should we transfer there, what are the ramp rates of customers in China and some of these new fabs for both IC and FPD side. So strategically, keeping an eye on that is very important for how we do our tech trajectory in China. The second one, if you follow our industries, you know there's a lot of excitement on the next wave of advancements, the kinds of things that are going to shape society maybe next decade. Some people call it the fourth industrial revolution. Whatever you want to call it, there's always been this healthy tension between chip technology and chip speed and algorithms. And now those are merging together with artificial intelligence, deep learning, that sort of thing.
They're merging with sensors, more interactive displays. And what's that causing as well is Moore's Law slowing down a little bit. And we're starting to see a resurgence in the ASIC model for chips. And these are kind of purpose-built chips to do specific applications, whether it's Bitcoin mining or to handle a data function on a system. We're starting to see a return back to more ASIC-driven things to serve the way those technologies are merging. And our business is design-driven, really. I mean, we depend on wafer starts or panel starts, but mostly we depend on designs. So this return back a little bit to customization and diversification, very valuable as far as Photronics goes. The third item, lithography-intensive display form factors and enhancements. Display, we know now Gen 10.5, a lot of fabs going up in China. This is mostly driven by the desire for larger-sized TVs.
H.K. is going to talk about that, but also now we're seeing particularly non-Chinese incumbents in display really trying to differentiate through technology, so these are mobile displays, much more lithography-intensive, finer dimensions, more complex flows, start to look more like IC-like things. That's very good for masks because we bring a lot of tech value into those applications. The fourth one, process of record integrated solutions. Process of record really means when you're kind of hand in hand working with a customer to develop a system that can deliver content to a wafer or a panel. The interaction between masks now and lithography is completely tied, so the best way to approach this is in a collaborative way and develop a process that works from top to bottom and gets the right content onto the wafer or the panel.
If you do this, and this has been a big emphasis for us, you also have a very strong position with that customer. Very hard to be displaced. They rely on you. There's a trust built. So we put a lot more emphasis on building our POR relationships. You'll hear that throughout. The complexity, cost, and scale. We have done joint ventures. The most recent ones, as Peter mentioned, are with DNP in China and Taiwan. Many of our customers are consolidating. So we used to talk to multiple customers. Now we talk to one, and they control multiple fabs. The reason for that is our industry delivering these solutions getting extremely complex, very expensive. It's hard to get ROI, and you really need scale. These are global businesses now. Even though we service regionally, we have to think globally.
So it's pushed us to consolidation, and we actually see more opportunities in this area. And finally, new innovations in manufacturing. Photomask kind of trails IC and display for things like automation, the way process control is done, that sort of thing. Now we're seeing opportunity to pull more of that into our fabs, the way we automate industrial machine learning, process management, automated feedback. So that's also an important area for our technology team to improve our manufacturing and yield. So what are our served markets, at least from a technology perspective? There are really three. Of course, we talk a lot about IC and FPD, but I want to break it down at least how the technology team looks at it. And what this chart shows here are the three main things that kind of drive more intensive mask-based lithography. The first, obviously, is IC.
This line shows kind of the adoption curve of technology in these different applications. On the IC side, the industry has been evolving patterning in lithography for more than 30 years. It doesn't mean there's not a lot of innovation still to come on the IC side. The reality is the technology implementations at a very, very refined level. We're talking about small incremental improvements, tuning things to get a little more yield. Of course, EUV could be disruptive. Nevertheless, the amount of new technology going into this space is different than, for example, display. Display is at a growth trajectory now on mask-based lithography that maybe we saw 15 years ago on the IC side. This is when the masks started to interlock with software tools and exposure systems to create much more value on the panel.
We're seeing this kind of growth phase in technology adoption for our sorts of products now on display. And the third we track, we're starting to see what we consider a much more interesting lithography roadmap for masks on the advanced packaging side. And that we put kind of at the accelerating phase. And what you see on these three charts, very simple graphic, we really talk about technology adoption. In these cases, the Photomask is used to improve the quality of the patterns. For advanced packaging, the patterns are still relatively simple, although we'll talk more about them in a second. So you see, generally, the masks look a lot like the patterns we're trying to print. On the other hand, in display, now we're starting to see users decorate these patterns to improve the lithography, improve the function in the display.
And then on the IC side, very, very sophisticated things done to the layouts to get them to print properly. The applications driving these, if you follow the industry, you know them well. I already talked about data management, memory, storage, sensors, and Internet of Things, more advanced algorithms connected to processors. On the FPD side, it's AMOLED, G10.5. And then for very high-resolution displays, virtual reality is really pushing the envelope on that. And that's actually driving at least a lot of this sort of technology adoption to enable those sorts of uses. And then lastly, on the advanced packaging, fan-out approaches, wafer-level packaging, and then migrating packaging to panel size is a trend we're watching. We actually have a few projects in those areas I'll talk about. So first, let's go to the IC side. As I mentioned, it's industry-wide, which is in Photronics.
I would classify it kind of in a leveling phase. I've been working in IC lithography 30 years, 35 years. I started with IBM. So I've seen this progress from G-Line all the way up through today's EUV work. The way Photronics looks at it now is we really invest in specific opportunities. We don't as much just deliver new IC mask technologies and bring them to the industry. We're very specific. We're directed. We work with customers that commit to us with POR, and we're opportunity-oriented. And that's, I think, working out very well for us. We have a great history of getting technology into products in the company. We were the first merchant to enter 14 logic, and actually it's all the way back in 2014. First merchant to bring 20-nanometer DRAM, also in 2014.
We were the first merchant in production on 3D NAND in our standalone facility. We've always been aggressive on new writing tools for masks. The writing tool is kind of the engine that makes it all work. That's where the content is created. So it's an enabler on a mask process. We've always been early and aggressive on writing tools. We were the first to bring EBM-9000 writing system to market. And that's the same tool still we're using to scale our current optical mask process all the way down through to 5-nanometer. So today, we're in this regime. We're starting to prototype and pilot 7-nanometer masks this year. We're sending them to customers. We're already certified for so-called 1y DRAM and Gen 3 NAND. Gen 3 is usually classified as 96-layer kind of NAND flash.
And we're doing that on equipment that was state-of-the-art at this time and still delivering very advanced masks today. And then where are we headed in the future? We have an EUV program now. We're shipping multiple EUV masks a month. We have a joint development agreement with IBM to prototype EUV for a particular application. And that's to replace multi-patterning for optical lithography with a single-patterned 193 level. So this is just an example of some of the EUV masks we're building. And we are in development on the so-called 1z DRAM node and Gen 4 NAND flash. On the FPD side, also opportunistic, but really our strategy and technology is to widen the lead we believe we already have. We believe we are industry leaders in this for sure on IC as well. But here, our goal is to widen our lead.
And we also have a long history of being successful in that regard. We entered what's called phase shift mask technology, which was something IC people used, I would say, 15 years ago, started to get pulled into FPD. And it's a much more complicated mask to build. This just shows a picture of its cross-section on the side. It has thickness and depth to it. We delivered first Quad High-Def masks for OLED in 2014. We qualified two types of new technologies that allowed panel makers to reduce the number of masking layers by combining two into one. You think, "Why would you want to do that?" It reduces the total mask count. But those individual masks had much more value to the customer, higher ASPs. And it allowed the process flow, particularly for LCDs, to be made much more efficient.
Without going into a lot of detail how they work, you're basically combining two exposures on one mask. This is different than if you hear multi-patterning for IC lithography. There you're splitting one exposure into two masks. In this case, you're effectively combining two into one. So it's opposite. And it's really an efficiency sort of solution. We were, as I mentioned, writing tools very critical for us. We were first to market also in FPD for the P-800 tool, which is the most advanced FPD mask writer you can buy today. And we have the first and we believe the only one. And we're also first to market with the P-80 writing systems. And this just shows a simple example of why that's important. Contact hole patterns are very common in all these sorts of devices.
This just shows the difference in the resolution we're able to achieve between the P80 and the P800 without really doing a lot of process work. This is a new tool for us. So what you're seeing here is just writing tool-based scaling. And there is no way you're getting to this kind of resolution with this sort of tool. So if you look at why that resolution is important, particularly I mentioned mobile display driving resolution. Today, we're at 6xx PPI in displays, working on 8xx. Where virtual reality and other things start to push us is PPIs over 1,000, 1,200. And you really can't even do the development work without this sort of capability. The other trend I mentioned on the TV side, we're going to larger frames, larger TVs as a standard. And that's pushing to Gen 10.5 for efficiency.
But also, resolution continues to go up on the TVs, 4K, and then we're moving up to 8K. And then the last application we want to profile, which we don't talk much about. It's a little bit of a new focus for us, but we actually already have a lot of traction in this, is advanced packaging. You follow the industry. Some of the larger chip makers are distinguishing themselves through how they're doing packaging now, putting chips together. The things that drive that are wafer-level packaging, panel size, fan-out methods. And then on the system side, what's called heterogeneous integration.
Basically, what this means is instead of designing single, very, very dense, large chips, getting all the SoC onto a single chip, going back to my first slide, people are designing point chips that do various things and then integrating them together on a package to deliver some kind of function, so heterogeneous integration is multiple functions built into one, and moving that up to being able to print on larger substrates, nine-inch and 14-inch gives the efficiency. The lithography is very, very different for this, and as I said, we're starting to see what we call an accelerating value roadmap now in mask-based lithography. We have a few development projects going on with customers, and we also ship masks in this space. To give you just an example of why this kind of lithography is different, the features are large. Don't even need to show the features.
But it has some unique challenges. For example, the substrates on these packages are very uneven. They're warped. So you have to be able to print the patterns over a very deep Z structure or deep depth. And that causes in imaging what's called depth of focus. You have to hold the quality of the pattern to a very deep range to accommodate for the warping of these substrates. So we work with a customer now. This was their baseline, increasing 40% the depth they could print and still maintain the fidelity of the pattern by pulling in some technologies we know from the IC and display side to work on this. So we are going to continue to evolve the capability as that lithography importance grows. So that shows kind of the roadmaps and at least how we think about these three driver applications.
The technology as standalone is not really valuable to the company as an asset. How we get that worked into products is really through partnering. There's a lot of sensitivity now between masks, patterning tools, and yield. And so customers are really encouraging us to partner more, and we're pushing for that too. It allows us to be designed in for them. I think since I've been with the company, we have a record high number of process of record relationships, now seven across IC and FPD. This really means we're co-developing things kind of hand in hand. And the customer is designing around our mask process and vice versa. We're designing around their wafer process, particularly for the high-end tools, very critical for return on investments. You spend the money on R&D, and you also want to get the benefit of the early business.
The POR is the way to do that. This is a metric we track, and we're doing very well in that regard. The other thing Peter mentioned is we have a lot of know-how in the mask printing onto the wafer. We've developed a suite of custom design tools that start to do wafer yield prediction and how mask sensitivities and mask spec change and how that'll impact wafer yield. We've deployed those in a few projects already. We've showed measurable yield improvements at wafer level. We're starting to codify that more and make it more routine. Between the POR and these design tools, it gives us a real stickiness in terms of how we work with customers. Finally, we want to be viewed whether the customer has a captive or doesn't as having a virtual captive model.
We want to be their partner, customize solutions for them, flexible, interlocked, very fast, very responsive. That is kind of how we define the Photronics process of record. Collaboration, teamwork, we have a great roadmap, and we have scale and transferability for them. The other thing that's very important for the technology team is that we have early access, early joint development. This is kind of the seed to tree sort of argument. We have to have things that get us engaged early. We do have quite a few of these, although most customers don't let us publish. This one we can publish. We signed an arrangement with IBM to co-develop some EUV mask technology. It's mainly focused on metal patterning for EUV. We have a series of those. That's what kind of starts our trajectory going and how we get technology into products.
So the question, why Photronics can everybody do this POR stuff and what's the big deal? It is quite a unique strength, and it's built up over the many, many years that we've been working on mask-based lithography, as Peter mentioned. Just to give kind of a visual example, the blue are our chips in this particular project that are functioning. The red ones are ones that are not functioning. And this just shows an example of two masks. One came from us. Another came from someplace else where the masks all met spec, if you will, customer defined the spec. They were delivered. But the set we delivered showed right off the bat much, much higher yield at the wafer. So the question is, why? Both of them met requirements. What made those masks function a lot better? And there's a lot of reasons why.
The interaction between the mask and the lithography now is quite sensitive. So small changes and a lack of understanding on what you're doing at the mask level has a very, very quick impact on panel or wafer yield. So the successful masks for high yield, of course, you have to have a capable process. You do need the right specs. You have to be qualified. You need good pattern quality. You can't have any defects. Photomask print on every chip, every panel. So you have one defect on the mask. It repeats itself at a very, very high rate as these things go into production. So it's effectively a zero defect environment. But defects are generally defined in various ways. And we have a very good understanding on defects that really matter and those that don't. The second thing is we have a lot of know-how on mask-to-substrate transfer.
I talked about that a little bit in the previous slide. We have the ability to simulate now, characterize, and improve the masks. We tie it directly to the wafer or panel performance and not just our little piece of this. We are starting to capture that knowledge into software tools, application projects, and that sort of thing. Third, reproducibility. Photomasks generally are custom. They're built to order. You have to be able to build them over and over again. They have to work functionally identical. They have to be exactly the same because customers do not have time to requal. You build the same mask twice. They can't requal the second one. Functionally, they have to work identically. We've developed over years of doing this a suite of tools that allow us to make sure our masks are matched.
If a customer orders a second one, it works exactly the same as the first one. Then the last thing that makes us, I think, a very strong POR partner and why customers are coming to us is we have great systems for protecting IP, especially as we go into China. We have methods to monitor mask lifetime. These masks get a lot of wear and tear in fabs. You could see 40,000, 50,000 wafers on a single mask. They interact with the exposure. They pick up contamination. So you need to understand lifetime issues with masks. We work hand in hand with customers on maintenance, monitoring, and these sorts of things. So all this together creates a package of why I think we're getting a lot of traction in this POR.
Another strength we have, I think, helps us very much is we view technology in our company as an asset, not to be necessarily protected and hidden in one site, but as a fungible asset. We want to be able to move around. We want to be able to put it where it can best be used, can make the most money for the company, can get us the most business, being aware also protecting IP.
We've had very many years of working on matching processes and moving things around our network have given us a lot of capability in this area. So we call intermask matching. There's a lot of reasons why you'd want to do this. The first is very simply, we have a process we developed in the U.S. Maybe we want to run it in Taiwan because we want it to be closer to a Taiwan customer.
Here's some examples of actual cases. We've moved to Taiwan from the U.S. 2.5D RAM, 3D NAND, 14-nm logic. We move very efficiently from our U.S. site there. And we have a good development team in Taiwan. They bring it up, and they make improvements. The second is for captives. Captives, they want to know you're good in technology, so they don't want to waste their time with you. But mostly, they want to know you can just do what they did. They want you to match perfectly. They don't want any headaches. Here's what the mask needs to look like. Make it look like this because we know it works. So there, this matching technology goes into making our mask look like a captive mask. And it's much more easier said than done. These patterns are very complex. There's often not standards in how you evaluate them.
So the know-how we have developed has allowed us to match into some very, very advanced captive processes. For example, all the way through 7-nanometer logic now, we have a module match for some of the most advanced captives in the world. And lastly, the ability to move within a site and across site. For example, when we ramp our facilities in China, we don't want to start over again. We want to be able to efficiently move processes that make sense to put there. And all these technologies come to bear as we transfer things from U.S. to Taiwan and on the FPD side from Korea to Taiwan and China. So it's a very powerful skill of ours. It's cost-effective, and it allows our technology to be deployed right place, right time for our global customers.
Of course, what that leads to, we believe, is technology is really a tool to support our business growth. This is kind of a standard cycle of how this works. We do invest in technology development in these three main applications. We develop new capabilities. We're very tuned in with our customers, so we're doing things that we know we're going to be able to deploy. We use that partially through this process of record model and partially through matching and good support to capture our share of high-end business, and of course, if we're successful in that, this is the earlier higher ASP work, increases profitability and cash flow, which we allow ourselves to reinvest. We have multiple of these circles going on in FPD, IC, even on the packaging. We really track each phase of this.
And crossing the hurdle on each one of these is very critical for our technology program, and our results in technology and high-end percentages, I think, show this result has been a positive one. So for technology summary, Moore's Law is slowing down. I think everybody kind of recognizes that. That follows the industry, but tech and innovation on the IC side is still very, very important. It's how you extend things, how you be cost-efficient, but still get to the next node. All very, very important things, so we still have, while it is, I think, saturating a little bit or in the leveling phase, IC, we still have a lot of strong technology programs there. We believe we have proven leadership across all the main lithography applications, extending IC, FPD, rising complexity, getting up this growth curve, and really leading into that and not following.
And then for advanced packaging, it's more of a stage setting for us and building our network and building our infrastructure. This is very nice, but we have also partnerships and process of record that takes this technology and puts it into products. Results, trust, know-how, consistency. These are all things very, very important for mask making in particular and delivering integrated solutions that actually can show a customer, if you use our products, you're going to get better final results on a wafer or a panel. That's very, very important, particularly in emerging regions like China where there's not as much institutional knowledge from history. So the ability to partner there actually is very high. And we have a lot of good relationships and good projects. And lastly, I mentioned technology portability, our ability to move things around, not keep it stuck in one place.
Don't develop technology just for its own sake. Get it to where it can be deployed rapidly. We have years of experience matching processes between our sites with captives and transferring this effectively to where it needs to be. And we verify things, and these allow us to have cost-effective multi-site solutions, so thank you for your attention. With that, I'd like to turn it over to Mr. H.K. Park. He's our General Manager, China and Taiwan FPD, and he's going to talk about high-end FPD.
Okay. Thanks. Good morning to everyone. Today, my presentation is mostly to talk about the high-end FPD business, starting with the FPD business performance. As you can see, we had reached a peak in revenue in 2016, but it fell in 2017.
This decline in revenue was mostly because our biggest FPD customer, Samsung Display, sharply shifted to the AMOLED while reducing its LCD business from 2016. Besides, Samsung also established their own captive shop to produce their high-end AMOLED mask by themselves. However, our revenue, as you can see from the last 12 months, has been recovering. How could we recover from the revenue plunge in 2017? And moreover, what would make our sales more likely from now on? It is because we are well-aligning with the market movements. Mobile market is shifting to AMOLED, and the more and more display require higher resolution. Photronics is extending the technology leadership in the market by leading portfolio of the advanced mask writers. Also, we have IC photomask know-how, which can be leveraged to developing high-end FPD process. The other point is the TV market is moving to larger screen.
Especially, China has emerged as the biggest panel producer in the world, as well as being the largest consumer market. Photronics has extended China business in the last two years significantly. So 33% of our total FPD revenue comes from China. This is the very reason why we should accelerate with our new Hefei production. Here are strategic priorities. First, our global footprint will be perfectly aligned with the dominant geographic panel technologies. As you know well, Korea is the place where the most advanced AMOLED technology comes from. So our Photronics Korea has the best lineup with the most high-end capability and capacity to support AMOLED there, even by recent investment. In the larger format, LCD TV, especially represented by the generation 10.5 +, will be all created in China in the future.
Photronics Hefei new facility will be created next year, which will be the first G10.5 photomask producer in China to support the larger TV market there. Second, we'll be successfully established China operating presence. At the time Hefei facility, Hefei manufacturing facility is complete and start production and start business, our Hefei facility will be the first G10.5 FPD photomask maker to support the larger screen TVs. And also, this fab should be integrated into our global network to help us optimize customer service. We'll also extend the technology leadership, and we'll be able to provide the higher resolution AMOLED displays and masks in terms of capability and capacity, and also even by IC technology, which can be crossed over to FPD. All of this, I believe, can make us grow to be global and merchant market share leader.
Here, let me dig a little bit deeper into our technology leadership for mobile display. As I mentioned earlier, mobile devices is shifting from the LCD to AMOLED owing to its many benefits. AMOLED delivers superior visual characteristics like higher resolution, but much lower power consumption. Especially, AMOLED can enable flexible format. We forecast AMOLED will be able to have one more big momentum to grow once flexible format such as bendable, foldable, or rollable devices start. On the top of that, the virtual reality and the augmented reality devices get popular. They will definitely lead to much higher resolution over 1,000 PPI. This trend will, accordingly, increase mask complexity, having more mask layers per set and much tighter specs on each mask layer. Photronics is moving very fast in this regard, being ahead of the market, being ahead of the competitors.
We invested the most advanced mask writer into our Korea site. This should be the first P800 to be placed in the market. We are the first having P800. This machine is currently being qualified in Korea at the leading AMOLED company. We'll produce the highest resolution AMOLED mask from this year. And also, our capability by adding the P800 can deliver us to have a much competitive advantage, even to compete with the AMOLED captive. It will also be able to make a bigger gap between us and our competitors in terms of the AMOLED capability. Also, it will attract many Chinese companies who are aggressively investing in AMOLED to come to us. Okay. Let's move to the larger screen. That's a G10.5 +. TV size is getting bigger and bigger. Currently, the popular typical TV size is already over 40 in, up to 55 in.
But we forget the future TV size will be much bigger. That's probably 65 in and 75 in. The why? China panel makers are so aggressively investing in going into the biggest path. That's a G10.5. In fact, G10.5 can improve glass efficiency significantly for ultra-large screen TVs, especially for 65 in, 75 in. For 65 in, as an example, the current G6 and G8.5 fab can make two cuts and three cuts per glass with 88% and 64% of the glass efficiency. However, G10.5 glass can be cut into a total of eight pieces with 95%. A decade ago, this industry forecasted future TV size should be over 40 in. So G7.5 came out because G7.5 could make a big difference in glass efficiency at that time in comparison to G6 and below. Later on, some others took a little bit bigger glass.
That's a G8.5, which can reach the glass efficiency up to 55 in. In the same way, now it is going up to G10.5 +. So for 65 in and 75 in as the future biggest home TV size. So anyway, the better glass efficiency and the China government's subsidy policy to the panel makers will cause China to produce larger TV panels at a cheaper price, which will accordingly lead to a decline in the TV set. That's the reason why China makers are aggressively investing into the G10.5. As I mentioned, China is definitely the fastest growing display region. Over 60 fabs are in production, and under construction or being planned. From the panel shipment perspective, China already came up to the number one position, followed by Korea as the second spot.
Especially BOE, the biggest Chinese panel maker, already came up to the first place who shipped the most panels last quarter. China will definitely be the largest market, even for FPD photomask. So as you can see the graph here, China G10 mask will be significantly growing in coming years. It will be increased by more than 2.5 times between 2017 and 2023. The main driver to hike China G10 mask is G10.5, which is the forecast expected to have over 60% CAGR owing to its much higher ASP structure. At this moment, there is no one merchant, no merchant photomask producer who has high-end technology or experience in China. Of course, no G10.5 photomask inside China yet. This is the map to show the current China display fabs. As you can see, China has already various fabs in operation.
Many of them are conventional LCD widely from the smaller G4 and G5, G6, up to G8 and G8.5 and G8.6. However, in recent years, there has been also several LTPS AMOLED high-end fabs open, although they still have been fighting with the low yield. Photronics, we have two sites. One is in Korea, which has a five-writer capacity to support the most high-end business. The other one is we have a Taiwan facility to support mainstream business. Also, we support the China business. Important thing is we are already qualified at all of the China existing fabs. This is the map to show the future China fab, which will start the mass production from this year till 2022. As you can see, most of the fabs are AMOLED and G10.5+ fab. Especially, we forecast, we foresee the total eight G10.5 fabs will be operated till 2022.
This is the market that Photronics will be able to dominate in the future. Our location, again, is here, central China, close proximity to some of the very important customers that includes G10.5 and a couple of the AMOLED fabs as well. This is the picture to show our new Hefei China facility 2019. Our construction will be completed by the end of this year, and the tools will be moving during the first quarter of the next year. And then we expect to start the mass production in three to four months after that. And our facility, our China facility, should be one of the biggest FPD mask facilities in the world. And our production line should be fully automated with the super class one of the cleanroom. For your further information, the total building size is about 250,000 sq ft.
At the cleanroom floor, size is about 46,000 sq ft. This is the picture to show our current construction status. As of last week, our construction progress is about 60%-70% into each area, office, and fab, and utility. Now the loop is being covered. How will we win the market? We will become the G10.5 mask share leader. How? As mentioned, our China G10.5 fab will have a geographical presence benefit. And also, the more important thing is, as Peter mentioned, we got the two China main customers' business commitments. That can, we believe, enable us to have a dominant market share in the future, especially G10.5. And also, we'll extend the AMOLED technology leadership. Our Korea site will be a high-end AMOLED center to build most critical AMOLED layers.
Our China and Taiwan side will support most of the other mainstream business plus some others' AMOLED business as well. In the same sense, we will be a market leader in China as well. We are the only photomask maker who has three facilities in three main markets: China, Korea, and Taiwan. By integrating three global manufacturing facilities, we will be able to provide outstanding customer service like immediate response, quickest response, immediate after-service locally. Also, I think we will be able to maximize our technical technology performance by co-developing with the local customer in each local area. This is our business outlook. As you can see, we estimate about 30% of the CAGR from 2017 to 2020. Our revenue would be expected to double up in the next three years. How? Key success factors should be first, Hefei fab, of course.
Once G10.5+ is launched, our Hefei fab should be the largest FPD facility from 2020 in terms of our revenue. Second, AMOLED mobile display technology inflection. Then we'll diversify our customer base. This is the last FPD summary. We are repositioned, especially we are repositioned for future high-end FPD business. China investment has been already made. Everything is on track. All of our investment will be well aligned to industry trend. We would expect to have a double-up business over the next three years. Okay. That's it. Thank you. Next is about the high-end IC logic to be presented by Frank Lee, who is the General Manager of Photronics Asia IC.
Good morning, ladies and gentlemen. This is Frank. I'm going to report on high-end logic IC. After repositioning our business, Photronics' overall IC revenue is growing off the 2017 chart. There are several major achievements.
First one, the strong demand from Asia logic foundries in Korea and Taiwan, especially in the high-end IC, as you can see in the chart, is the major growth of our logic IC. We expand our penetrations at captive. Also, in the past couple of years, we have established a very strong technology and business partnership with our key customers. That is because of our high-end capacity and capabilities. Another very important achievement, we have a major M&A. We formed a JV with DNP, Dai Nippon Printing Company in Asia. First one, 2014 in Taiwan, Photronics DNP Mask Corporation. And then 2018 with DNP in China. It's a Photronics DNP Mask Corporation in Xiamen, China. These two JVs actually combine the technology know-how and customer relations of the two parent companies.
Also, the China JV will reduce the risk of our China investment by ramping up quickly and also by competing more effectively. In both JVs, Photronics owns 50.01%, and both JVs are operated by Photronics management team, including myself. JV with DNP is working very well and has been viewed by the industries, by our customers, as a success story. The JV is working and, most importantly, is growing. Early this year, we held a technology forum in Shanghai, joined by Photronics and DNP. Here is our CTO, and this is Hayashi-san, DNP CTO, both making technical presentations. Big groups of customers, suppliers, our partners, and government officials attend the event. The China JV has demonstrated our strong commitment to provide a local supply of high-end photomasks to our China customers. Focusing on the strategic priorities is key to grow our market share.
And also, it's very important to generate return on our China investment. So we leverage our JV with DNP in Taiwan and China to win the market shares. We increase our business globally with large foundries, IDMs, including captive. We try all the efforts working with our customers to prevent any additional captive mask shop in China. Between Photronics and DNP together, we strongly believe we can enable our customers to advance their photomask technology. So basically, there's no need for them to build their own captive. Lastly, we will bring up our China facility successfully. With the extension of our Taiwan JV into China, we believe we can compete very effectively, and we can utilize our Taiwan JV to become the leading and the largest mask provider in China. So looking at this chart, also, as Peter mentioned, China represents the fastest growing semiconductor region.
I think everybody knows about the ZTE case. I think after that case, semiconductor is becoming and will be the focus of China in the next 10 years, and if we look at the chart, the focus for China local-made IC is about $33.6 billion in 2022, and compared to the China consumption of $200 billion, this is only 17%, which is much, much lower than the target set by the Chinese government, so photomask opportunities in China are very attractive. Typical photomask term is about 1% of IC revenue, so we look at $33 billion IC revenue. 1% of this number is $300 million. The same as H.K. showed, in China every city is trying to build an IC wafer fab. There are 23 new China IC projects, and most of these fabs are scheduled to begin operation from 2018 to 2020.
Of course, photomask demand will follow the ICs. If we look at a map, when our China Xiamen facility is ready, Photronics globally will have four high-end manufacturing sites to support China customers: a Taiwan site, Korea site, and a Boise, Idaho nanoFab. Here is the building model photos of our Xiamen facility. From the map, we are only four kilometers away from our largest customer, UMC Xiamen. Four kilometers, and we are going there because we get a strong endorsement from UMC. We have extended our business agreement with UMC from Taiwan into China. This will greatly reduce our financial risk at the beginning stage of the fab. The facility is scheduled to be complete by the end of Photronics fiscal year this year, and tools will move in Q1 next year. This picture is taken four days ago.
The cleanroom, the building will be complete by the end of this year. How do we win? We have developed a winning plan, which is aligned with Photronics corporate strategy and, of course, with our customers' growth plans. Again, technology leadership to meet our customers' roadmap. Close relation with our key customers to have a partnership. Grow market share. The last 12 months' results do indicate that we have been winning market share in both Merchant and Captive. We expect IC logic revenue will grow 7% CAGR. Made in China, as I just reported, the known migration, new design, and new application will all drive the market growth. Photronics and our JV together, we must take advantage of our position as a technology and market leader to win the business and to grow our market share.
In summary, we have repositioned our business, and we grow our high-end market share in the past year. Our China investment is progressing very well. We grow our business and partnership with our major customers. We have JV in Taiwan and China, which allow us to compete very effectively with our competitor, and because of our scale, our size, our technology, we have been viewed as a trusted technology partner by several key customers. Thank you. With that, I turn to Peter Broadbent.
Thanks, Frank. We're now going to focus on our high-end memory. First, a brief look at our recent memory performance. We've essentially repositioned our customer base in memory. I'll speak to that on the next slide. Of late, we've seen extremely strong demand from memory, and that's reflected in the data point there, which is showing 78% year-to-date growth versus 2017.
We've established strong high-end market share with specialty and emerging memory foundries, and we're benefiting from the Micron technology agreements. So that technology agreement, we qualified for the current nodes and the next two nodes. And that is driving our growth with key customers that have adopted the Micron bit cell. And we expect to benefit from that for the foreseeable future. So today, our reputation is as a trusted technology partner. And that's built on both our high-end capacity and our capability, but importantly, our technology and the global engineering support that we provide our customers in the marketplace. And that's twofold. That's our R&D support led by Chris's team, and it's the engineering strength in each of our factories worldwide. More specific look at Boise. Through aggressive sales efforts, we have replaced nearly all of the revenue lost from the Micron outsourcing.
So today, we are approaching peak revenue levels in Boise, and we expect to exit the year on that trajectory. Through that hard work, we now have a strong diversity of high-end customers, both memory and logic. And that's reflected in the chart you see on the left. 2015 represented our last full year in the Micron partnership, and we were reliant on one large customer. Today, we have multiple customers filling back that volume. And that diversity is across development and production masks. During the dip, we focused, as we always do. It's the DNA of our company, and it's what Peter talked about as a core competency. We focused on cost out. And we were able to significantly reduce our break-even in Boise. So as we reach these peak revenues again, we're driving meaningful contributions to the bottom line.
And strategically, a robust qualification pipeline with POR development with key memory customers is providing a strong foundation for growth in the future. Our strategic priorities for Boise are straightforward. We intend to maintain and grow our market leadership that we've established in Asia memory. We will provide, through technology, capability, and customer service, a mirror-like captive performance for our customers. And that virtual captive strategy that Chris spoke to will provide us with the foundation for future growth. We will increase our high-end business with key foundries in both China and the U.S. And we'll integrate our worldwide network to better serve our customers. We will lead with Boise at node plus one development and at node production. And then for node minus one, we will transfer process and technology to achieve scale and efficiencies for our customers.
In general, the memory market today is strong and diversifying across technologies and applications. If you think about exiting FY 2017, calendar 17, three large DRAM companies and five large NAND companies dominated the market. But today, we're seeing a host of companies investing significant capital in the memory market, and that's largely in China. Without a merchant in the market and with our position as the leading merchant provider in memory, we see it as our market to take with our technology and our production. Most of this emerging memory production is occurring in Asia, and we have key customer engagements in Taiwan and in China. Frank showed the 23 new IC fabs going into the market over the next several years. Six of those are memory, and we're laser-focused with our virtual captive strategy on this market.
We will serve them from Boise and our locations in Asia with an integrated pipeline and process technology transfer. Chris spoke about the productive synergy between chip design, mask design, and wafer production. That synergy is something that we've demonstrated we can perform for our customers. By bringing to bear Dai Nippon's and Photronics technology on this market, we believe that we can convince these customers in this market to continue investing in fab and chip design and allow us to carry the load on photomask design, preserving this market as a merchant opportunity for years to come. How do we win? We win by one, leveraging our technology leadership in both logic and memory. By establishing POR for new nodes and developing joint development programs with customers for next-generation nodes, we intend to extend this.
We will win by building customer relationships that lead to business partnerships. We'll act as a virtual captive provider and, through our network, provide collaboration integration for scale, and by doing these first two, we will win by growing market share. We will win the lion's share of the merchant opportunities this now represents, and we will grow the merchant market by serving current captive providers and serving the new fabs as a leading virtual captive shop for them in the future. Before I sum up on memory, a brief mention and look at the mainstream sector, which is near and dear to Peter and my heart, representing about two-thirds of our IC revenues. Mainstream has been a stable and steady contributor to both revenues and earnings for our business. It's characterized by low use of capital, but it is highly competitive and more like a commodity market.
We win by being the low-cost provider, another core competency of Photronics. If you look at where this is headed, take a look at the chart on the left, you'll see from IC Insights, they say that over the next several years, five years, that the total IC market will grow at about 5.1%. Analog at the top is projected to grow at 6.6%. This is mainstream. On the right, another chart that shows the speed of growth on the bottom versus the share of total IC sales on the top, and the bubbles represent the size of the IC market. You see that cell phones, which you all know, have driven the market. Most of the discussion on that is around high-end logic and memory. There are mainstream components in those devices, and that's driven the market.
You have optoelectronics, sensors, gyroscopes, power devices, etc. But on the far right, you see the emerging high-growth areas of automotive, which have driven our customers of late, provided stability in the market. And what we have yet to see is the ramp in the Internet of Things. So as we look forward, we believe that this market is not only going to be stable, but potentially growing. In terms of our memory outlook, Frank mentioned that we expect in our business to grow at a CAGR of 7% in ICs. I should mention and note that this is about 2x or a little more than 2x the growth that SEMI forecasts for mask demand in the marketplace. And as Peter noted, by the 2020 timeframe, we're anticipating somewhat of a muted cycle in ICs by that time.
But we see this growth being aided by memory, and the key success factors are our China expansion plans, the node migration opportunities before us in memory, and some strong market drivers for our customers in terms of memory with edge and cloud, etc. We will maintain our leading market share that we've established with specialty foundry producers, and we will achieve success with a strong integrated network to serve these global customers. And finally, we see mainstream stability and accelerating growth of auto and IoT as we move forward. So in sum, we've repositioned the business in terms of memory and in terms of Boise with a strong diversity of customers and revenue streams. Boise performance is nearing back to JV levels, and it's also with the lower-cost structure contributing meaningful numbers to the bottom line as we approach these levels and exceed through them.
We're growing the business with Asia specialty memory foundries, and we're extending our technology leadership and developing partnerships for continued growth in the future. With that, on memory, I'd like to now turn it over to John Jordan, our CFO, for a financial outlook.
Thanks, Peter. Good morning. I've been at Photronics for just under nine months. So this is probably my last opportunity to give an observation as a new guy from outside. Photronics is a lean, no-frills, well-organized business with transparent boundaries and little to no unnecessary redundancy. All the key players are well-respected in the industry and have been with Photronics and in the industry for a long time. There is a single-minded insistence on operational excellence.
And you've seen in the presentations that all my leadership teammates have a keen strategic insight into the industry and a comprehensive grasp on the industry and the confidence to invest where we think the industry is headed. There's a deep respect for customers and relationships, and there's a passion to win and a passion to improve the value equation for shareholders. From a strictly financial standpoint, it's a very strong controls environment and a tone at the top that only wants to do things right. There is no earnings management. Always the accounting has to be as it should be. That said, in many organizations, as in many organizations, there is plenty of opportunity for process improvement to achieve best practice.
We set up these charts to show first quarter and second quarter of fiscal 2017 and fiscal 2018 to illustrate that second quarter of 2017 was the nadir in our recent trough. Fiscal 2016 and I can use that. Since then, we've had four quarters of improving operations so that the revenue in second quarter at $130.8 million was 6% over the $123 million of first quarter and 21%, as Peter mentioned, over the revenue reported in second quarter of last year. We've often talked about the importance of the value of operating leverage in our model, and this quarter and last quarter illustrate how that manifests. In second quarter of 2018, we were 270 basis points above the gross margin in first quarter of 2018 and 650 basis points over the margin of second quarter of 2017.
At 11.7%, the operating margin was 210 basis points over the first quarter operating margin and 660 basis points over the operating margin in second quarter last year. Income before minority interest is a good indicator of the results of operations before the joint venture partners share, and we were 60% over the first quarter income before minority interest and 10 x the income before minority interest in second quarter of 2017. Net income and earnings per share were similarly strong compared to first quarter and second quarter of last year. Net income at 80% over the first quarter and six times the net income reported in second quarter last year. Earnings per share at 15%, 67% over the $0.09 of first quarter this year and five times the EPS reported in second quarter last year. Year-to-date was similarly strong with 14%, 17%.
With 17% year-over-year improvement and EPS at four times the amount reported at the same period last year. So we've heard my teammates discuss these achievements and success factors, and the continually improving results of the past four quarters are testament to the success that that repositioning is delivering. And this chart sets forth our outlook of 40% revenue improvement by 2020, about a 12% CAGR. Our strategic priority is to support the organization to achieve the targeted revenue and earnings. We have a three-level support that will keep the organization focused on investments that will meet our hurdle rate for return on invested capital.
All investments must clear the hurdle rate to be considered to ensure that the balance sheet stays strong enough to support the operation in a downturn and to provide the flexibility for investments in organic and other M&A and will continue to drive cost out of the business. The company consistently generates strong operating cash flow. The capital investment required for the business sets a high fixed cost threshold, but as I mentioned before, the positive upshot from that is that once that threshold is met, the contribution margin is significant. We target 50% contribution margin, but we can achieve 70%-80%, and that's not unusual. Every operating location, mainstream and IC, FPD, and high-end, mainstream and high-end, FPD and IC, are profitable and generate positive cash. So since 2009, the recession in 2009, we've generated positive free cash flow. Our working capital management is tight.
We've got a blue-chip customer base, and our investments are well controlled and based to a significant extent on customer commitments. So the strong operating cash flow has enabled debt repayment and continued strengthening of the balance sheet with an adequate cash position and borrowing capacity to support the current operations, to provide insulation during a downturn, and to provide resources for organic and external M&A. This chart has the operating cash flow, the CapEx, and free cash flow to show the change from the prior years to what we anticipate after the China investments are complete. So we've got 2016 there because that was a normal year, if you will, before we started the China investments in 2017 and before the CapEx in 2018 that has my share of the China investments in it.
So, where we previously had $72 million of free cash flow in 2016 with $50 million of CapEx, after the China investments are complete...
We need to use the right button here, John.
After the China investments are complete, we expect our maintenance CapEx to be on the order of $50 million, which will generate a free cash flow well in excess of $100 million. Just to recap the China investment, we've got a total investment of $320 million, and the redeployed tools from our existing locations, the JV partners' contribution, and the investment incentives that are provided by the high-tech zones where we're building aggregate $170 million, and that will reduce our total out-of-pocket cash, if you will, to $150 million for the China investments. So our use of capital has been well-defined. Our organic growth now is focused on China, where we see the growth in the industry.
We continually work on developing our collaborations with customers, vendors, and competitors to expand the relationships and expand into JVs and other M&A where those relationships can lead. We're also continually searching for other opportunistic M&A, but we won't engage in any projects that don't meet our hurdle rate. The convertible securities will be repaid in the beginning of 2019, in April 2019, and we're confident, when we are confident, that the China projects will be completed at the planned costs, and we've minimized the risk. We'll then consider revisiting other uses of capital, including share repurchase. So in summary, we've defined a clear path to profitable growth while maintaining a strong balance sheet that will support the investments and insulate against down cycle threats.
We have a financial model that reflects faster growth of earnings than revenue and achieves our financial targets of $630 million in revenue and 15% operating margin, and we will adhere to the discipline of accepting only projects that clear our hurdle rate and improve the return on invested capital, and now I'm going to return the podium to Peter to reinforce the attractiveness of our investment thesis.
Okay. Thank you, John, so in conclusion, we are the photomask industry leader. We have momentum. We see plenty of opportunities in front of us that we have been and are actively developing. Targets for 2020 were minimum revenues of $630 million and EPS of $0.80 a share. So with that, I'd like to thank everyone in the audience for your attention.
I'm going to ask the management team to come up here to the podium because I think we need to use this microphone, and I think Amy will walk through the room with the other microphone to take your questions. Again, thank you all, and the floor is now open. Edwin. This is being webcast, so we need to be disciplined here.
Thanks for the presentation. A lot of detail and a lot of useful information. I'm Edwin Mok. First question I have is I want to understand kind of your 2020 target, the $630 million number, right? I think in one of your presentations, you guys start talking of FPD to grow to $200 million. That implied that your semi or IC is $430 million, which is basically like 5% above what you just reported, right, on a run-rate basis.
Is that the right math that we should think about with the model, or is that just some margin risk that you factor in? That's why it's only 5%, and I have two follow-ups.
Yeah, so again, the way we are looking at 2020 is indeed, I think you did that math correctly. The FPD industry, we really think regardless of where the state of the business is, given the market drivers in China, that we can power through whatever headwinds there are. The demand is that present, and the contractual commitments we have from the customers are that solid. As far as the IC business goes, we don't know where 2020 is going to be, so in our guidance for 2020, we've layered on top of the number a downturn.
So if the industry is healthier at that point in time, as I said in my remarks, that 630 is a pretty conservative number. So the way you did the math is exactly right, and it incorporates growth in China, but it also incorporates an industry downturn layered on top of the overall semi business.
That's helpful. It's always useful to factor in the risk so that you can drive upside. Chris, I have a question for you all on the virtual fab or virtual captive strategy that you guys have, right? It seems like a right strategy to kind of go after the captive customers, right? But obviously, the last 10, 20 years, we've seen more captives, right, done first in the IC and also later in FPD, right? Just how do you kind of think that strategy would change the tide?
Or as you discuss with a customer, my concern is basically always you gain share in the captives, and then they decide to add a new factory and also you kind of lost that business, or you just go on and off, on and off. How do you kind of combat that trend?
Yeah. I think the total number of captives in the industry has definitely shrunk, but those that remain are larger and invest more, so that's clear. When we think about virtual captive, there's two pieces to it. One is for customers that don't have captives, we want to discourage them from building one, and we love our captive customers, but we would prefer no more of them be built, so for those that don't currently have captives or haven't expanded to a second or third site, we want to discourage that.
So we'll do that by trying to make ourselves give them as captive-like experience as possible. So this is a little bit new for us as far as the strategy, and I think it's working. We have broken into one or two captives that we had no positioning with previously. So I would say that part of it's been effective. Whether we can convince them not to continue to invest or expand, I think that remains to be seen, but that's our strategy. So we share roadmaps. We share future investment. We share our situation and how we might help them meet their capacity needs without having to invest internally. For those that have never taken that step to jump into captives, I think that barrier is higher than ever.
There is a lot of investment money available, particularly in China, but the technology, I would say on the IC side in particular, to jump into captive is not straightforward. Most of the captives that are remaining now have been evolved the last 20 or 30 years. So to just jump into that and be an impactful player, I think not such a simple step. So for those, we push our virtual captive model and say, "Hey, we will invest in an annex for you. We'll customize processes. This will be your capacity, your capability. We'll link our systems together as though we were a captive, and here are all the reasons why this is going to be a better solution for you." So we recognize the challenges you make. Captives have taken a lot of market share from merchants. We can't deny it, and that's well known.
But we think this is the right strategy to level it off and then to discourage additional ones. And last point I would make, the pendulum definitely has swung in the last four or five years to captives, but as Moore's Law evolves over the next five, seven years, we actually do expect a few more captives to get less commitment from their parent companies than we see today. It wouldn't be appropriate to elaborate on which ones, but we do see the possibility also of pendulum swinging back a little bit and opening up opportunities maybe to JV, a few captives, and that sort of thing in the future, so.
Great. One last question I'll let go. So John, thanks for providing a little more detail to the 2020 numbers beyond the 630 revenue number. I noticed that you put 15% operating margin as your target for that, right?
If I go back and look at last two peaks as just where you guys were, you were almost already there, but this is a much higher revenue level. Why not more operating leverage? I would expect more operating leverage from your model because historically you're always going to be going up and gross margin expand to almost 30% relative to revenue. Just can you walk through that math there for us? Maybe kind of help us think about the operating margin and potentially upside if there is no downturn like where you guys are factoring.
Yeah. Edwin, I think the time you're referring to, there was a much different mix of the sources of revenue. So things have changed. ASPs are different. And this is where we think the mix of revenue is going to bring us up.
Mix is the biggest driver.
Sorry?
Mix is the biggest driver of that operating margin part...
Let's face it, we're making significant progress toward it, so at 11%, 15% is within striking distance, so depending how that mix works out as we get to those revenue levels, it may turn out to be conservative. Yeah.
Thanks.
Wait one second. Thank you.
All right. Thank you. Brian Chin, Stifel. Thanks for the great presentation today. A couple of questions. Maybe first, just to frame up the flat panel market in China. If you had it aggregated today, what is the TAM of FPD photomask production right now, taking a snapshot of the 34, maybe 30 or 34 flat panel fabs, mostly LCD focused today? But what's that TAM today? What is it either in 2020 or 2022, which is that second math that you put up there? The second part of it.
As far as China goes, the China TAM at this moment is about $160 million. Okay. But as I reported and we also mentioned, due to G10.5, which has a significant ASP, so the $400 million TAM will be expected in 2020. That's what we showed in our presentation. Yeah.
Is it $400 million in 2020?
Yeah.
Okay. And let me think.
Excuse me. I think that was 2022.
Yeah.
Gotcha. Yeah.
Yeah.
Great. Thank you. And then I think you said about a third of your flat panel business today is derived from China. I'm curious, who is the predominant merchant supplier you see in the mainstream LCD as well as the AMOLED and LTPS part of the market?
You mean currently?
Yeah. The other merchant supplier that you tend to run into.
We have several competitors that compete in the market for mainstream business. For example, Hoya is one of the strongest in the market. SKE follows. SKE has currently the monopoly for G10.5, although there is no G10.5 market mature yet. Those two guys are big players. In Korea, there is LG Innotek, who is a subsidiary of the LG Display. They are mostly focused on LG Display. Of course, they also send their mask to the market. I think those are the big players in the market at this moment.
Okay. Maybe one more flat panel. I've got you over at the mic. I think you gave targets out to the 2020 horizon, but there was, I think, a talk of tripling the flat panel revenue over a three-year horizon.
Thinking about what your capacity is in Hefei, beyond what you expect to install in early fiscal 2019, what is the additional headroom you have there to expand revenue capacity?
You mean the total capacity for Hefei? Yeah.
Yeah.
Well, theoretically, if we have a lot of the G10.5 orders filled up, our total capacity is almost $200 million. But in fact, given the LCD market demands and the customer's plan, and then also based on the business commitment from the customer, we expect to have $100 million business expected in 2020.
Yeah. I just want to correct. We didn't say triple. We said double.
Double.
Oh, double. Yeah. Interesting. Maybe one more one for Chris. Can you contrast again the 10.5G mask set ASP and also versus the Gen 6 AMOLED and then the LTPS just to kind of give that contrast over whatever capacity?
ASPs.
ASPs.
Yeah. I don't know if we would. You mean probably. I think we can certainly.
Even if it's like an expression of 100% baseline versus.
Yeah. So let me answer that. So right now, G10.5 mask sets, as H.K. mentioned, there's only a single merchant with the capability, the tools to build them. Mask set prices are about $10 million per set. So that's a significant step up relative to anything else that we make. By comparison, high-end AMOLED mask sets are about $1 million a set. Mid-range AMOLED masks. So Chris had the 600 PPI on the board. That's a million-dollar mask set. A 300 or 400 PPI mask set is about half of that. So that's the kind of step-ups that you see or we see in the market right now with the different technologies.
Maybe one last thing for John on the financial side. Just going back to Edwin's question, but also reconciling with your, I think, 50% incremental operating margin target, it would suggest op margins could be kind of 18%, that kind of territory if you want to sustain that kind of incremental margin at 50% through the horizon period, and so I'm curious if that makes any sense. That would kind of suggest maybe gross margins could be sort of creep up to the 30% level.
Yeah. We're not going to subscribe to that, Brian. We're pretty happy with the metrics that we set out. The ASPs will change as we increase the IC business in China and in our other businesses as they get more competitive. So we want to stay with the 15%, and like I said, it could be conservative.
Fair enough. Just in terms of the walk-up to that, based on the depreciation increase as you install the tools in the two separate China fabs, where do you think gross margins likely buy them out? Is it fiscal 1Q, or is it?
There's a mix. There are several tools that are coming off depreciation as we put the new ones into use. So I don't want to speculate where those margins are not speculate, but we just want to stay with the 15% operating margin and have to kind of fill in for gross margin and depreciation.
Okay. Fair enough. Thanks.
All right. Congratulations on refilling the Idaho factory. Could you give us any more color on the types of customers that kind of filled in for what Micron left behind?
Sure. Chris, thanks for the question. In the initial phases as we worked to convert new memory opportunities, it was some mainstream business, actually. We contributed with additional customers from across the U.S. and Europe. We then started to get traction with memory customers, predominantly a key one in Asia, and then we started to diversify into new emerging memory customers that are doing some good development work and are starting to put some masks into production in Asia, and then lastly, we filled it with some logic customers with EUV development work, with some qualification work at advanced nodes below 10, and with some other production business from some key logic customers, foundry customers outside the U.S. as well.
Great, and I love the fact that you guys mentioned ROIC probably 10 or 15 x today. Is there any kind of color or proximity you can give us on this hurdle rate that you guys are looking at internally to deploy capital?
Yeah. We use you can estimate the cost of capital for a company this size generally in the 13%-15% range. So we use 15% as our hurdle rate.
Other questions?
Given that hurdle rate, it does seem hard for me, at least, to see how M&A can make sense relative to your own stock price. You mentioned, which was nice to hear for the first time, possible stock buyback somewhere down the road. How can you possibly find anything in the M&A set area that could make sense relative to your own stock? Can you see possibly JVs could make sense, but what could make sense for you guys?
Richard, you're right on. It would have to be a very attractive M&A opportunity. Stock repurchase is a topic of discussion. And when we're at the point where we know that China is going to settle, where we expect it to settle, it will rise to the surface.
And just to follow up, I mean, given that the return on invested capital really has disappointed over the years, I mean, if you look at the return on equity, it's consistently been kind of mid to maybe high single digits, really never double-digit ROE. So when investments are made, including the China investment, which you've really never given any what the return on invested capital is there, how much oversight is there?
To the extent that possibly it would make sense to have some new blood at the board level, some outsider to oversee these major investments, including possible M&A, how much is that being thought about?
Okay. Every project that we put up, the board takes a good solid look at both the pros and cons. I think we elaborated that as far as the China investments are concerned, that we have customer commitments to ramp the facilities rapidly. John also pointed out, and we don't make a big deal, but we also have significant incentives from the high-tech zones where we're investing. The China investments, as they're articulated, meet the ROIC targets that John elaborated. I think, and we'll just leave it there.
Possible new board blood? Is that something you guys would think about?
The company has a nominating committee, and I think we are very rigorous in the governance actions that you'd expect a public company to take. So there's always the topic of directorship is one that's considered every year consistently.
Just a follow-up to the China investments in softening. I would imagine next four to six quarters, that's going to be a predominant topic here, right? With all these tools we allocated, and thanks for the chart, by the way, to kind of talk about your actual cash outlay versus how much actually impact you. I guess two-part question. First is kind of the contribution from a joint venture partner. How does that timing come? Do they just come when you add the tools, or they pay you one time and then you're kind of stuck done? Just kind of want to understand the mechanics behind that.
Then in terms of actual depreciation impact after the factories ramp, if I use what you provide and the fact that you're allocating capacity, is it pretty marginal depreciation that you are actually adding to the, I think, you're at around $90 million kind of depreciation rate right now. Is it just marginal above that, or should we expect it to go to $120 million? Just kind of you kind of create some framework around that.
I think an estimate closer to $100 million is more appropriate. $120 million is way beyond what we're expecting.
Then is it the cash that you expected from the CapEx requirement?
Say it again, Edward.
The cash you expected, yes, from the CapEx requirement.
We haven't broken that out. We've kind of kept it as $170 million aggregate for the incentives. We have some NDAs in place that preclude us from talking about those details, so we just have to keep it at the aggregate amount. Other questions?
Okay. So again, thank you, everyone. Lunch is outside for any of those that wish to stay. The management team will be here. And again, thank you for your attendance.