Good morning, and welcome to Lam Research Analyst Day 2016. Thank you for joining us live here in N.Y., and those of you joining us via webcast. We have until about 12:30 P.M. today, and we have a lot of material to cover, so let me go through the logistics for today. We will have four presenters today. We plan to take a short 10-minute break after the first two speakers. I request that you hold all your Q&A until all the four speakers have completed their presentations. We will have plenty of time in the end to get to all of your questions. Lastly, please do not forget to collect your headset, and you will notice that they will feature in our presentation. Do not forget to turn off your cell phones as well. Before we begin, I need to read you our safe harbor.
Today's presentation and discussion includes statements that are not historical in fact or forward-looking statements covered by the Safe Harbor Act. Such statements are based on current expectations and subject to risks and uncertainties reflected in the risk factor disclosures in our SEC reports. Please be aware that actual results may differ from these forward-looking statements, and the company undertakes no obligation to update them. In addition, today's presentation and discussion includes non-GAAP adjusted financial measures. A reconciliation between GAAP and non-GAAP results can be found in the appendix to today's presentation. With that, let me turn it over to Martin Anstice, our President and CEO.
Good morning. It is a pleasure to be here this morning, Special thanks for those of you that are joining us in person, and thanks to those that are listening to our webcast. As always, we appreciate your interest and your time, your interest and support for our company. Before I begin this morning, I would like to offer a few words of context. We can sometimes forget that leadership is a privilege, and leadership at Lam Research is a tremendous honor. When I think about our company. I think about our increased strategic relevance as we occupy a seat at the very center of the most fundamental industry inflections. I think about the technical depth of our company, and the mutual trust and respect that we enjoy with our customers. I think about our financial and operating performance, meeting or exceeding expectations over multiple years.
I think about the quality and the competency and capability of our employees worldwide. I have come to appreciate the value of employee pride in the culture of our company, and in our values, and in our performance. More than any other commitments that we can make to you today as an assembled leadership team, the commitment to sustain and build and architect employee pride is the most compelling value proposition we can commit to. As you might imagine, we are very pleased with the performance and our accomplishments so far along this journey, but we are not yet satisfied. Perhaps like no other in our industry, we believe we have a world of opportunity and the wherewithal to deliver on that promise. That is the chapter of the Lam Research story that we would like to share with you today.
By way of introduction, joining me in speaking today and for Q&A, will be Rick Gottscho, our Executive Vice President of Technology and Products, Tim Archer, our Chief Operating Officer, and Doug Bettinger, the Lam CFO. Combined, we will characterize our opportunity and our confidence in the value creation of Lam Research. We will attempt to make the case for compelling stakeholder value for our customers, our employees, our suppliers, for our communities, and of course, for our shareholders. Our thought process is humble, yet optimistic. Likely, we will see more change in the semiconductor ecosystem in the next five years than perhaps any other, except for the formative years. Innovation in the silicon world is the very foundation of creativity for the global tech economy. Without mass storage, advanced computation, and connectivity, the most visionary entrepreneurial appetites risk being unfulfilled. Lam is an enabler. We make things possible.
We deliver extremely complex technology and productivity to our customers. It is true to say we deliver equipment. It is much more accurate to say that we deliver a result on the wafer through the integration of hardware, software, process control, and materials. That totality will be the scope of our presentations today. More simply stated, we believe we have the right products at the right time in markets that are growing. We have multiple years of demonstrated outperformance in our recent past, and we are committed as a leadership team going forward to continue to deliver high-quality earnings growth long-term. My focus today will be on the industry drivers of our business. I will attempt to increase your depth of understanding of the strength and the opportunity of Lam Research.
Rick will next focus on the critical technology headlines characterizing why the market growth claim we're making has substance. Characterizing our leadership broadly and providing you perspectives on our strategic relevance to the customer and our business momentum. After a short break, Tim will provide you a window on our execution, without which vision and opportunity is worth nothing. Tim is responsible for making it possible for us to scale and grow Lam at an 18% revenue CAGR for the last five years. Scaling and operational execution generally is an important element of competitive differentiation. Hopefully, there's incremental perspective for you today on the value of that in our presentation. Bringing up the rear is Doug, and Doug will talk money. He will update our long-term financial model for you and summarize our plans for capital redistribution.
He'll provide a reference for our continued growth into the March 2017 quarter. Then we'll host Q&A, and I'll attempt to facilitate that to accommodate, as Satya just referred, your questions. Again, before we begin the presentation, thank you very much for your attendance here today. We look forward to this morning, and we look forward to sharing our opinion on why Lam Research is a must-have in any investment portfolio. Simply stated, at Lam, we believe our opportunity is greater than our risk. Let's make a start on telling you why. This is one of the very few slides today that actually looks backwards. It's important because it provides context as well as a commentary on the competency and capability. It provides the foundation that increases our confidence in achieving our long-term vision.
Last year, we were recognized as the second-largest equipment company in the global semiconductor supply chain. This year, we celebrate our fifth consecutive growth year, averaging 18% over that period. We now actively compete for approximately $12 billion of wafer fabrication equipment spending globally, and that proportion is increasing steadily. We picked a two-year reference on this slide because it is approximately 18 months since we hosted our last investor and analyst event. Since we were last with you, annual WFE increased from approximately $32 billion in 2014 to slightly more than $34 billion this year. During that timeframe, we increased our revenues by 30%, a three times growth rate compared to WFE. A differentiated level of performance in the community of capital equipment companies, we believe. Important is not just revenue growth, but profit growth and the quality of our earnings, both areas of intense focus in our company.
As an operating philosophy, there is a clear relationship between the profitability and growth of our company. We hold ourselves accountable to EPS accretion, that is the test that we apply to everything we do. Our focus is less on discrete P&L line item percentages, more on the fundamental and sustainable expansion of profits and cash. We typically target operating income growth at a rate greater than one and a half times the growth in our revenues, you will see that in the model presented by Doug a little later today. Again, in the timeframe presented here, we meaningfully exceeded that profit multiple, while at the same time increasing our investments in the future. Worthy of note, in the last three years, we increased our investments in R&D by 30% in a period where SG&A expenditures only increased by 10%.
Doug will also comment about the importance not just of how much money we spend, but where and how well we spend it. Focusing on the middle of this slide, I would like to introduce to you the balanced growth drivers chart. The so-called Lam wheel of fortune. You will see this a number of times presented today, the message of balance is really important for us to convey to you. One of the most frequently asked questions from the investment community of Lam is: What is the most misunderstood or underappreciated value of your company? I would probably say the most misunderstood reality of Lam is our balance. It is true to say that for many, many years, we have focused on investing in creating competitive differentiation and delivering enabling capabilities to our customers at key technology inflections, patterning and 3D NAND as primary elements of that story.
In the last two years, we have strengthened our position in these inflections, there is so much more to the Lam Research story. As Tim and Doug both will share later today, our installed base business continues to grow faster than the pace of Lam process chambers in fabs around the world, that number now exceeds 40,000 process chambers. The shipments of our installed base business in this two-year period, you can see, represented approximately 25% of the company, that is the nearest thing that we have to an annuity in our industry. The so-called other segments here includes specific business, such as image sensor and IoT investments by our customers. In this period, it also includes non-patterning logic and DRAM investments by our customers. Again, a level of importance to our company that exceeds 25% of total shipments and revenues for the company.
3D NAND is now the clear industry technology of choice, succeeding the NAND planar scaling shrink roadmap. As we will characterize in the next several slides, we are in the early stages of participating, by most independent opinion, in the single largest growth driver of the entire semiconductor industry. We are obviously very pleased about our leadership here, particularly in the most critical and technically challenging applications, but there is so much more to come. Patterning is the story of two integration schemes. Actually, the litho etch scheme is by far the smallest expenditure of the industry. The most dominant scheme in production is self-aligned spacer-based implementation, which Rick will speak to a little later. Bottom line, again, by most independent analysis, the implementation of EUV will be focused on the immersion-based litho etch multi-patterning, the holes and the cuts applications.
Worthy of note for our business, that litho-based multi-patterning segment represents only 6% of our total shipments in the last couple of years. Balance is a commentary on how we optimize growth and profit percentages for enhanced earnings and cash from operations growth. It is a commentary on our comprehensive engagement today with all memory, logic, and foundry players. It is a statement of balance. It is a statement of commitment to support every customer at every technology node, and we are certainly seeing a broadening of active investments today across multiple nodes. Last but not least, the balance message is a commentary on how we realize and seek to strengthen competitive differentiation. We are more balanced than ever relative to our unit process and integrated products focus. We are more balanced in our thinking about delivering value from technology leadership and operational excellence both.
We are more balanced in the creation of enabling results on wafer through our holistic focus on the integration of hardware, software, process control, and materials. As a closing headline of this slide, we are very pleased with the demonstrated multi-year company outperformance. We do not see signs that suggest our outperformance opportunity is being diminished. Quite the reverse. On the demand side of our industry, there is no killer app. The world of tech is no longer defined around one dominant electronics product like the PC, but a multitude of consumer and enterprise application drivers that have been enabled by the foundation of the cloud for mass storage and advanced computation. The smartphone as the primary vehicle available globally for connectivity and mobility today. The drivers of the last five years are clearly still relevant.
Smartphones will still drive leading-edge silicon, and the cloud data center will provide the infrastructure that makes it possible to convert enormous quantities of data into valuable information. The combination of ubiquitous connectivity, mass cost-effective storage, and high-powered advanced computation make possible the sustained innovation of the entire tech economy. The next five years are different. The semiconductor industry today is like no other. Now, we collectively touch, in some way, shape, or form, every human being on our planet. The definition of success for our customers is no longer make a faster chip at lower cost. While that is still important, it is no longer sufficient. Our customers today literally aspire to change the world. The next wave of application drivers, including AR and VR, artificial intelligence, machine learning, and connected automobiles. That world is just emerging.
Connectivity is only beginning to revolutionize the world, and that revolution is our collective opportunity. That opportunity, in our opinion, is endless. The new tech economy and the expectations defined by their investors is also just emerging. The foundation of silicon and the integrated circuit roadmap make that possible and sustainable. Restating a message of earlier, probably we will see more change in the next 5 years than ever before, except perhaps the formative years of our industry. For Lam, that is a world where we are committed to demonstrate that our opportunity exceeds our risk. Moving a little closer to home, to the markets that directly influence the economics of our company. We see accelerating demand and a world of opportunity. The data center and cloud provides the economic infrastructure for mass storage and advanced computation.
The platform of mobile, including the cell phone, tablet, PC, and other derivatives with 4G and 5G technology are driving content, creation of connectivity to transport our data efficiently and effectively. The world of connected devices, the world of now more than 1 trillion IoT devices, who are expected to be shipped by 2035, creates the capacity for the measurements in a distributed sensor environment previously unimaginable. The vast majority of the investments necessary to support these content increases is leading-edge technology. That very same expansion of the need at the leading edge creates the sustainability of investments in WFE. In some respects, the world of technology hardware is exactly as forecasted, with PCs, smartphones, and server units all trending at best to single-digit growth.
Here we are with an opportunity for NAND demand beyond supply, with the biggest month-over-month increase for several years in DRAM prices last month and compelling performance value being assigned to 10 nanometer and seven nanometer logic. Why is this happening? If it's not units, it's content. The new drivers of machine learning, AR, VR, and IoT are driving silicon content demand in electronics. 3D NAND is a major headline here, but not the only one. The 3D NAND inflection is transformational. In 2016, the amount of storage in a single SSD is crossing the amount of storage available in a hard disk drive. Combined with faster IO and the economics of scale, this is completely changing how people design storage, compute, and networking in the data center. We expect more than 60% and more than 25% bit demand CAGR through 2020 in NAND and DRAM, respectively, here.
Similarly, the smartphone or perhaps the super phone is seeing some of the largest content drivers in years with AR, VR, imaging, and other applications creating more than 40% and more than 25% in NAND and DRAM bit growth, respectively, from roadmaps of our customers focused on graphics, integrated circuits, sensors, and networking. Heading into our third straight year, we see increased demands for equipment and installed base support across multiple technology nodes, with investment levels being material at the greater than 28 nanometer technology node. In fact, we have an expectation that calendar 2017 investments above 28 nanometer in logic exceeds the investment levels that we have seen in calendar 2016. At the same time, the semiconductor industry is stable, it's disciplined, and consolidated. Much more evidence speaks to supply and demand balance today.
Perhaps the world of cyclicality has been replaced in our industry by the competing tensions of consolidated supply, with WFE spending concentrated to the major players, and absent the killer app, more fragmented demand. This is more complex to model and analyze, no doubt, and you have our sympathy. But it is the environment that defines our opportunity. Our focus on operational flexibility and operating model leverage is a key element of our leadership. These content drivers are not a one or two-year phenomena. We see sustained multi-year roadmaps for increased silicon content across a variety of devices. That means there's potential for sustained growing WFE over multiple years across the entirety of the device spectrum. We have assumed here customary bit growth numbers. NAND with sustained investment to capture approximately 50% of the storage SAM. Foundry logic spending strong and diversified across multiple technology nodes.
DRAM with value and performance improvements from continued shrink through the 1Y node in this time horizon. Despite the sometimes popular doomsday commentary on the limitation of physics or the description of unsustainable economics in our industry, at Lam Research, we are optimistic about the health and the future of our industry. There are simply too many smart people in this industry with an enormous capacity to innovate, with demonstrated capabilities to overcome these challenges. The problems are more complex for sure, but we are equipped today with increased learning at our customers, in our companies, with the academic community, and other partners in consortia. In the semiconductor industry, we now live in a consolidated world, one that is more disciplined in its sensitivity and approach to balancing supply and demand of silicon capacity. We are less cyclical. Some important headlines here.
Measured by the concentration of spending in our top customers, consolidation stopped almost five years ago. What is perhaps more striking than stable customer concentration is the healthy profitability level of our top customers. Both these messages we consider important positive elements relative to our future. Honestly speaking, I find the reference to cycles or mid-cycle earnings, or questions about where we are in a cycle, almost impossible to process, let alone answer today. The world that we live in at Lam is a world where key customers individually make strategic commitments to technologies and businesses in time frames that reflect market opportunity for them and their unique strategy implementation choices. That is a world I would characterize is healthy, sustainable, less volatile, and more rational.
Certainly, the entire ecosystem, including Lam, is more efficient and effective in this environment than the wild swings of our past. We have a vested interest to keep it that way. Our ability to create value for our shareholders all improved. Value is created by the enablers and the technology leaders. The definition of success today has moved away beyond having a stock price at the peak of a cycle higher than it was in the cycle before, even if it lasted for only one day. Sustainable leadership and profitable growth is our primary focus today. The next three slides describe our shared challenges, often characterized as the limitation of physics or the pressure of unsustainable economics in our industry. We have a very different view.
These slides offer a perspective on how we might overcome the challenges and exploit the opportunities through partnerships in the ecosystem that are more collaborative. They certainly make a case for an exciting future. Last but not least, they describe the commitments of Lam to the semiconductor industry at large, a commitment to support the success of our customers. No doubt, we live in a world that is more than more. No longer is innovation limited to traditional shrink. Today's dialogue is about power, density, cost, bandwidth, next-generation memory, and increased productivity across many simultaneous technology nodes. As if the challenges of technology were not enough, the economics get more challenging also. The cost of R&D is increasing for the entire ecosystem. Being successful requires we work together to overcome common challenges.
This is the primary reason why our most important vision objective in Lam Research is to be number one in customer trust. With trust comes opportunity. In many respects, this is a slide of context. It describes our specific contributions to the definition of success of our customers. That is our purpose. Because the traditional economics of Moore's Law are potentially less effective than before, certainly they're more challenging. Our customers and their customers are talking about novel innovations at the system level to open up the interface between compute and memory, to take advantage of the rising attach rate of accelerators and greater quantities of memory, often through systems integration and advanced packaging. This is a fantastic opportunity to drive a market transition for semi. We are obviously very focused on contributing to those new inflections.
I've heard often some variant of the following described as the reality of our industry, described by some as the new law of Moore's Law. That new law states that every year, the number of people that predict with confidence and conviction the end of Moore's Law actually doubles. I was hoping I was going to get a laugh. Our window on the world could not be more different. Our future challenges, some of which Rick will discuss in more detail, include process complexity, where process controls and the interdependency across processes, including pretreatments, are driving opportunities for our company. Where uniformity now measured at the five-nanometer technology node at 0.3 nanometers is a critical focus for our company. Cost and productivity always. Killer defects. We now are held to a standard of less than one defect per wafer pass.
Atomic level scaling is critical to the success of our company, and we are a leader in that area. Aspect ratio scaling, the non-volatile 3D NAND roadmap is rapidly approaching 80:1 aspect ratios with bottoms-up fill-in deposition requirements. As an industry, the demands on our leadership for innovation have never been greater. This trend is set to accelerate. With more layers, we will mean higher aspect ratios and the alternative integration schemes that are associated with that implementation. RC reduction needs will drive interconnect metallization changes through the industry. Process variability reduction is clearly driving the need for process and process control integration. As part of the semi ecosystem, our challenges have context. Our challenge is to get comfortable as an industry and as a company to follow the world of new tech, but also to lead it.
Our challenge is to create the opportunity for differentiation and also for growth. That's an important message, not just for our investors and our employees today, but for the talent and the youth of our industry necessary for tomorrow. The good news is we actually do have a roadmap, although clearly significant questions remain, and the technical and business paths of our customers will vary more than ever. We have visibility for the next five to 10 years. There is more than $5 billion invested annually by the global equipment industry to support this roadmap. We're focused on enabling this roadmap. Over the last several years, as I've noted already, not only have we significantly increased our investment in our future, our commitments measured by increased investment in R&D is greater in Lam Research than any top equipment company.
What are the key messages on this slide? Scaling continues, but it's not just about traditional shrink. It's about new materials and vertical architecture. Vertical scaling roadmap is more comprehensive and aligned today than ever before. The path to more than 100 layers in non-volatile memory roadmap exists. The post-FinFET era, defined by gate-all-around transistors exists, likely horizontal implementation first, before vertical. Multi-patterning continues with or without EUV, and the trend towards spacer-based integration schemes continues at pace. DRAM is likely the next technology to undergo structural innovation. Hence, the dialogue focused on next-generation memory throughout the industry. Even here, we have visibility to the 1Z technology node. We estimate that by the end of calendar 2017, the DRAM install base will be comprised 25% 1X capacity, 50% 20 nanometer, and approximately 25% at the 25 nanometer and above technology node.
Beyond traditional scaling, as I've just described it, our focus is also on the agenda of eco and green. Our focus is on reuse strategies of our customers and the delivery of advanced services for productivity solutions to the entire install base. The roadmap and opportunity has increased our diversity, no doubt. Customization is more likely as the semiconductor industry seeks ways to create and sustain competitive differentiation. In most areas, we have a 10-year horizon that underscores our ability to innovate and differentiate for growth and for value. I made the statement earlier that we have the right products at the right time. I might have said we are in the right industry segments at the right time, or we're not cluttered with a portfolio of products that is less relevant to the enablement of our customers' success.
Through good strategy, alignment, and execution, Lam's a pretty special place to be at currently. All four significant technology inflections that we have discussed for some time have tremendous relevance and prominence in the industry, at least through the end of this decade. For those that have gray hair, you will remember, just like me, that the story of Lam 10 years ago was the etch unit process company with an opportunity to compete for 11 percentage points of wafer fabrication equipment spending. Subsequent to the integration of Lam and Novellus, we announced that we now competed for 25% of WFE. As a result of two things, first, the integration choices of our customers, and second, the new product pipeline and new product release activity of our company, we compete this year for 34% of wafer fabrication equipment spending.
Next year, we expect that number to increase to approximately 37% by 2020. You might remember this slide. As a frame of reference, when we first established this SAM expansion metric in 2013, we competed for $8 billion of wafer fabrication equipment spending. We systematically had our projections exceeded by our actions and the actions of our customers. You might remember that we first sized the SAM expansion opportunity from technology inflections at $1 billion, then $2 billion, and then $3 billion at SEMICON West in 2015. To the best of our abilities to model today, we now estimate a SAM expansion from 2013 through 2019 of approximately $5 billion. The single largest change since our presentation 18 months ago is in 3D NANDs, where we estimate now more than a doubling of our SAM.
Consistent with prior presentations, 3D NANDs represents approximately 40% of the SAM expansion presented here, and patterning in logic and DRAM combined is of the same order of magnitude. As is customary, we've provided you some details on the slide of key assumptions we make which quantify our SAM growth in future technology nodes for your review at a later time. More specifically, I want to take a look at patterning and in a few minutes, 3D NAND. Patterning is a significant and growing segment of our business. It is extremely critical to the success of our customers. It is illustrative of one of the most significant increases in strategic relevance for our company in the industry. The patterning market opportunity is likely to grow by 40% from now through 2020.
This is inclusive of the 1Y DRAM and 5 nanometer logic roadmaps, and integrates the consensus assumptions available today for EUV adoption. As I noted with our wheel of fortune reference earlier, the overriding patterning integration scheme is spacer-based, not litho etch, in a ratio of almost 3 to 1 for Lam shipments. This fact is particularly important to Lam because the process flow of spacer-based patterning is dominated by etch and deposition steps, and focused on lines and spaces, and also cuts applications for pitch scaling. Importantly, as previously mentioned, at least through the 5 nanometer logic node, our patterning opportunity continues to grow. If you believe in the substance of the opportunity for NANDs in storage that I presented a little earlier, particularly the data center penetration, then our customers will need to spend $50 billion on wafer fabrication equipment investment through 2020.
A commentary on opportunity and responsibility both, we recognize at Lam the strategic relevance and critical importance of our company's execution to the success of our customers with this inflection. 3D NAND is without doubt the single largest growth opportunity in semis, and the 3D NAND construction has created a new paradigm shift in cloud architecture. We are in the early phase of this opportunity with a strong roadmap for non-volatile memory to capture 50% of the storage TAM in the data center. In this context, there are four critical messages on this slide for Lam. The first, the Lam SAM as a percentage of WFE in NAND grows from 27% in 2013 to approximately 45% next year. We have more than 90% market share in critical etch and deposition applications, and we have gained 7 to 10 percentage points of share in the transition from planar to 3D architecture.
Conversion of the installed base to 3D capable outputs will take until 2020 at least, and increased vertical scaling in next generation memory extends SAM expansion beyond that timeline. As shown in the second chart on this slide, we gain 50% more revenue opportunity. We project 50% more revenues per incremental bit from 2D to 3D. We are actually indifferent to whether the incremental bit growth investment comes from greenfield or conversion. Our revenues will be the same. Perhaps it looks easy, I am not sure. On behalf of the Lam management team, I can tell you it is extremely difficult to architect a transition for our company and to grow at an 18% five-year CAGR. Again, I would like to extend my sincere thanks to our customers for giving us that opportunity, and also to our employees who create the substance in our vision.
We have two primary legs of growth. The first is our market growth, which comes from process flow and integration choices of our customers, combined with our product pipeline actions. The second is the increased competitiveness that is ultimately measured by our market share. As I stated many times, we do not defend or win new business with 100% success, but we have defended and won new applications at the 90% level again this year. Extremely good performance from our perspective. Our target going forward is consistent with the trends of the past, to gain one to two percentage points per technology node inflection. No doubt, gaining market share is harder as your share increases. As context, we are approximately a mid 40% market share company, all markets of Lam. Said differently, we believe in the substance of our share growth ambition. Gaining share requires a lot.
It requires technology leadership and productivity leadership. It requires customer trust and speed of execution. In the later presentations from Rick and from Tim, both will specifically call out elements of our performance that we consider key factors of the Lam history and of the Lam future. The definition of success for companies is most typically described as an aspiration to increase our value, and that is indeed ours. In addition at Lam, we care passionately about technology enablement, and we define our primary objectives around the success of our customers. We aspire to be number 1 in customer trust. We aspire to be number 1 market share in every single market at every single inflection of our company. We have an objective to deliver best-in-class products and services to our customers and to make Lam a place where successful people want to work.
We do have an objective to make some money and also to invest appropriately in the future of our company, returning cash excess to that strategic and operational need to our shareholders. If we simply define the vision of Lam, what would we say? We would articulate a vision including our focus on continued patterning, multi-patterning, and 3D NAND scaling enablement, delivering growth outperformance. We would say increased strategic relevance through the delivery of differentiated solutions that focus on our customers' grand challenges and the next-generation technology inflections of the industry. From disruptive technology investments, creating value beyond unit process, harnessing the full value of our multi-product company. Growth and value balance with a priority to fund profitable growth before capital redistribution. Finally, to have accelerated innovation and increased value from process and process control, now through collaboration, and continued evaluation of adjacent market opportunities.
In closing, before I hand to Rick, Lam's commitments to the semiconductor industry is most simply characterized by our commitment to deliver innovative technology, trusted productivity, and speed. As I stated many times already, our customers are literally aspiring to change the world, to make the lives of the global population better. This indeed is an industry like no other. For Lam, we believe we have an opportunity like no other. Thank you all for giving me an opportunity this morning to frame our vision. I hope I provided context for the presentations that follow, and at a minimum, on a Friday morning, I hope I kept you awake. With that, thank you. I hand off to Rick.
Good morning. Really appreciate Martin setting the stage for me. He's a tough act to follow. I'm going to focus on the origins of our outperformance as Martin described them, and most importantly, what are we doing with that $1.8 billion that we spent over the last two years, and we'll be spending a comparable amount of money over the next two years. We're spending that money to extend our leadership position and continue our outperformance going forward, in the markets that Martin described. In particular, you've seen a version of this slide already. I'm going to focus on the patterning story, what we see going on in the industry with the advent of extreme ultraviolet lithography in particular, and its impact on multiple patterning and its impact on our business.
I'll then turn to 3D NAND, and what we see happening there, as the number of layers and the density continues to increase by exploiting the third dimension, and again, why we expect to outperform the industry as that segment of the industry continues to grow. Lastly, I'll talk about new devices in logic and memory, new interconnect technology, and what we're doing in process control that basically constitute the other category that Martin referred to, how our investments in those areas will enable us to outpace the industry. First, we're going to talk about patterning, and to introduce the subject, we're going to show a video to remind you how multiple patterning takes place and the critical role that deposition and etch processes play in multiple patterning. Let's roll the video.
Today's semiconductor demand is increasingly driven by new applications that continue to push the limits of possibility. Capabilities such as artificial intelligence, automated cars, and advanced virtual reality require greater adoption of cloud computing architectures and rapidly increasing system-level optimization. Within the server, this translates to increasing demand for ever more powerful processors, accelerators, and memory, which in turn is driving smaller and denser chip features. Yet where we used to rely on conventional photolithography to manufacture chips with relative ease, today that process is complemented by a series of ingenious techniques that have been developed and combined to pattern the most advanced devices on the market. These devices contain various patterns that require different techniques to create. Let's take a quick look at two of these right now: the litho etch technique used for irregular hole patterns, and self-aligned multiple patterning, commonly used for unidirectional lines.
In litho etch, litho etch approaches, patterns are formed by lithography and then etched. The process is then repeated. This is required because at advanced nodes, the desired hole patterns are so densely packed that they cannot be created using a single litho etch pass due to the current resolution limits of lithography. Instead, the dense pattern is decomposed or split into less dense patterns, which are then formed using multiple cycles of lithography and etch. At the sub-20 nanometer nodes, self-aligned patterning is now widely used for line pitch scaling. This innovative approach relies on an increased number of deposition and etch steps to achieve dimensions beyond the lithography resolution limit. For example, the self-aligned quadruple patterning process is capable of producing a pattern four times as dense as the original. The self-aligned technique produces excellent pattern fidelity and delivers remarkable scalability for pitch shrink.
The semiconductor industry is moving towards unidirectional designs for continuous scaling. No longer limited by lithography resolution limits, we can now manufacture these advanced patterns using a combination of self-aligned patterning to form the lines and litho etch patterning to cut those lines. Use of multiple patterning techniques will continue to grow with current and future lithography solutions. Lam Research will continue to play a critical role in enabling chip makers to deliver the increasingly powerful devices we have come to expect from this rapidly changing industry.
One of the things you may have noticed in that video towards the end was we took an 80 nanometer pitch as printed with immersion 193 lithography. Using quadruple patterning techniques, intensive use of deposition and etch, we reduced that pitch to 20 nanometers. You might think, 20 nanometers, what's 20 nanometers about when 10 nanometers is in production in logic and seven nanometers is close behind, and people are working on five and three nanometers? 20 nanometers is the minimum pitch required to interconnect those five-nanometer transistors. The 80 nanometers, as I said, is limited by the resolution of immersion 193 lithography.
To put that in further context, with the advent of extreme ultraviolet, the resolution limit there is about 26 nanometers, which means by the time we get to five nanometers, we're going to have to employ multiple patterning with extreme ultraviolet to get around the resolution limits of extreme ultraviolet. That's reflected in this patterning roadmap. What we see going forward is immersion 193 multiple patterning continuing indefinitely over the next decade. It won't be the only patterning technique. We'll also see the advent of extreme ultraviolet. We think there'll be some layers introduced at seven nanometers and more at five. Initially, the application of extreme ultraviolet will be for the patterning of holes for contacts and vias. The next application would be for the line cuts that are necessary to isolate the lines formed by spacer-based double and quadruple patterning.
As I just said, subsequent to that, we'll see EUV multiple patterning, litho etch, litho etch for holes and vias, but ultimately, EUV will be used to set the template for spacer-based double patterning with EUV, and ultimately, quadruple patterning. You can see here the number of wafers that have been shipped and how many of those wafers went through Lam machines to date to extend the patterning roadmap. Looking at the makeup of multiple patterning going forward in a little bit more detail, this is a foundry logic picture, and I'll come back to memory in a minute. You can see the number of wafer passes here by technology node and the makeup of those wafer passes.
As Martin said, the litho etch, litho etch, and in some cases, it's litho etch, litho etch, litho etch, litho etch, is actually a relatively small fraction of the total multiple patterning market. The bulk of the market is made up of self-aligned spacer-based double and quadruple patterning. The assumptions that we've made here with the introduction of EUV are consistent with the assumptions others in the industry are making. That's customers, that's other suppliers. You should also note in this slide that the numbers we're showing here are higher than what we showed about a year ago. The 1.6x, for example, at 10 nanometers, I think a year ago, was about 1.4x. In each case, we actually underestimated the number of wafer passes that are going through a multiple patterning process.
The other thing to note here that's very important, even with the advent of EUV, the patterning market for Lam Research, dep, etch and clea n, continues to grow. EUV does not shrink our market. EUV is part of our growth trajectory, and I'll talk a little bit more about why that is. The first thing to consider is why self-align patterning, double, and quadruple patterning is really going to continue even with EUV. The conventional wisdom might be that EUV simplifies lithography, which it does, and would enable the return to two-dimensional design rules, as shown here. We've migrated with the advent of multiple patterning and self-aligned patterning from a 2D design layout to a 1D design layout, which is tailor-made for the self-aligned patterning, followed by lines and cuts.
You might think, well, there's a real penalty associated with that, and there is to the extent that if you want to connect one device to another, instead of laying out an interconnect within the plane, that is 2D design, you actually have to connect one line to another by going out of the plane in the third dimension, over and back down again through the holes that are fabricated. There's a certain cost penalty and complexity penalty associated with that. Why, with the advent of EUV, wouldn't we just return to 2D design, and why wouldn't spacer-based multiple patterning just go away? First of all, there is a problem with EUV that has yet to be resolved, and we don't believe this will be resolved even as EUV is introduced into volume production as early as seven nanometers.
That is, there isn't a good means to protect the mask. There isn't a good means to detect mask defects and repair mask defects. If you're starting to print lots of features and the mask is defective, you're going to print lots of defects. You don't want to use EUV lithography under those conditions for printing dense patterns. You want to use EUV lithography for printing sparse patterns like holes and vias first. Subsequently, the lines and cuts. It'll be a very long time, we believe, before EUV will be used for dense lines and spaces for defectivity reasons. Further, I'll show you a picture shortly, line edge roughness, which is basically a fundamental limitation of resist technology, causes transistor variation and degrades performance effectively, gives you a broader distribution of transistor performances, and therefore limits yield.
To minimize line edge roughness, you need a lot more photons, a lot more power coming from the EUV source. As I think you all know, one of the gates to the introduction of EUV for fabrication is the fact that the power hasn't been adequate. That's going to continue to be a problem because more power means more cost. In fact, right now, the power levels aren't there to generate the kind of line edge roughness that is needed. Lastly, EUV requires or utilizes reflective optics as opposed to transmissive optics. This increases mask design, litho process complexity, cost, and cycle time. Those are all reasons why we don't see a major shift back to 2D design, even with the advent of extreme ultraviolet lithography.
By contrast, unidirectional design with self-aligned multiple patterning inherently has lower defectivity because masks can be protected and repaired if you're using immersion 193. Line edge roughness is naturally reduced as you go through the sequence of processes that you saw in the video with self-aligned spacer-based multiple patterning. Essentially, each one of the dep and etch processes that you saw provides a smoothing function for the line edge roughness. I'll show you an example of how we're working with ASML to achieve that result with EUV printed resist. The mask design litho processes are simpler in 1D. You can just see that in the picture. Of course, the self-aligned spacer-based multiple patterning has been in production now for many years, is a production proven, reliable solution.
As Martin said earlier, we see 40% growth over the next four or five years in multiple patterning in our served available market. That's really great that the market's growing by 40% over that time period. The question is, how do we outperform in that market? To answer that question, I first want to remind you of something I said about 18 months ago in this meeting about what the challenges are in multiple patterning. Overcoming these challenges is key to winning. There are two overarching challenges. The first one is minimizing variability. The more wafer passes you have, the more unit processes of deposition, etch, and clean that you have, the more variability you introduce into the result on the wafer. That variation stacks up and eventually affects the yield.
We have to reduce the variability in every single step, far beyond what is required for single-pass lithography. The other overarching concern is cost. The more passes you have, the higher the cost per device for our customers. We need to boost productivity. These are the two challenges that we have been focused on for years, and are continuing to focus on as we focus on extending our leadership position in multiple patterning. Looking more to the past, then we'll look a little bit further into the future, how have we established the leadership position in patterning, and how have we outperformed the industry as a result? These are three primary examples. First of all, we introduced some years ago, a technology that we call Hydra, which enables unprecedented within-wafer uniformity, and gives us control on a die-by-die basis.
It's also useful for wafer-to-wafer repeatability, and in fact, helps us with chamber-to-chamber matching. We've seen a 10-fold increase in the shipments of this module that goes on our conductor etch tool, over the last couple of years. In the area of deposition, the spacer deposition itself, we provide a tool called Striker ALD Oxide, which has best in class within-wafer uniformity and within-die uniformity. Using a quad architecture, that is 4 wafer processing stations per process chamber, we have the highest productivity, the highest throughput density, addressing the cost challenge of anybody in the industry. As a result, we've seen a 3-fold increase in our shipments over this time period of the Striker ALD spacer tool. The third example is our Kiyo F Series product. That's the latest released conductor etch product in which the Hydra module is inserted.
There are other features in the Kiyo module, which is also the basis for one of our atomic layer etching products. We have focused on reducing wafer-to-wafer and chamber-to-chamber variability, and we've seen a 4-fold increase in the shipment of these products over the last couple of years. That's where we came from and how we've generated the outperformance over the last several years. Looking forward, we want to extend our leadership to continue this outperformance, and we're focused in three areas. Two are shown here. The first one is, because there's so many steps in multiple patterning, the processes are complex and there are interactions between upstream processes and downstream processes, we provide a capability to take an incoming wafer with a certain amount of variability.
That variability could be in the litho, in the lines, in terms of the critical dimension of those lines or the line edge roughness across the wafer, or from wafer to wafer. It could come from variability in underlying films deposited upstream. In both cases, we can measure what's coming into the etch system and compensate that variability using our Hydra technology. You can see here a reduction of 4-fold in the critical dimension variation, from what came into the etcher versus what went out. It's not just about very tight control in the etch process or the deposition process itself. We have the ability to correct or to compensate for incoming variation. The second thing that we've developed, this is newly released, it's in beta at the moment, is an ability to extend the uniformity all the way out to the extreme edge of the wafer.
In the last eight millimeters of the wafer, typically 10% of the die reside in that last eight millimeters of the wafer. There's a natural variation that's going to happen as you approach the edge of the wafer. It's like falling off a cliff. If the wafers were infinitely large, you wouldn't have this problem, but unfortunately, nobody's figured out how to make an infinitely large wafer. You must compensate for the discontinuity that occurs in electric fields, in chemical reactivity as you fall off the edge of that wafer. We have new capability that's being introduced that allows us to tune that edge, and this is an illustration on an etch tool.
We're doing very similar things on our deposition products across the whole product family, that allows us to tune the edge for any process condition and flatten out, whether it's etch rates or critical dimension control. This will enable us to extend our leadership position and to continue our outperformance in the patterning market. Turning to EUV and how Lam can add value to EUV patterning. We've been partnering with ASML. We're doing that not because we view EUV as a threat and we want to somehow undermine their ability to be successful. Quite the contrary. We see EUV as critically important for the industry. We think it's good for everybody.
I've already shown that our markets grow even with EUV. We want to try and accelerate its introduction because we know that overall it will lead to a lower cost structure, and therefore stimulate the demand that Martin referred to earlier for all the new applications, be it virtual reality, augmented reality, cloud computing, what have you. As I said earlier, one of the biggest problems in extreme ultraviolet is the line edge roughness. What you're looking at here, as printed by ASML, is a pattern at fairly low dose. That is, we can get away with relatively low EUV power, but there's a lot of line edge roughness. After an atomic layer etch treatment of that EUV exposed resist, you can see that the line edge roughness has been reduced by more than 50%.
That translates directly into lower dose for the same line edge roughness, if you will. That's going to lower cost, and that should accelerate the implementation of EUV. In this case, the dose was lowered by as much as 30%. For us, this represents an increase in our SAM cumulatively through 2021 of more than $300 million. We believe by collaborating with ASML, and with our leadership in atomic layer etching, and our investments to extend that leadership position, most of that $300 million is going to flow into Lam Research. In summary, we have a number one position today in patterning. Patterning, it's as defined as deposition etch and clean segments of the patterning market, with more than 50% market share overall.
The multiple patterning market is going to continue to grow, with the advent of EUV, and self-aligned multiple patterning is complementary to extreme ultraviolet lithography. We have a continuous focus on reducing process complexity and variability and improving productivity. We're driving new technologies to further extend the patterning roadmap and put more distance between us and our competition. We see a 40% increase in the SAM, and we see an increase in the SAM in all segments, foundry, logic, and DRAM. Next, I'd like to talk about 3D NAND and what we're doing to extend our leadership position in this market segment. Once again, I'll kick this topic off with a video to remind you how 3D NAND is fabricated, and again, how critically important deposition and etch are in the fabrication of the 3D NAND device. Let's roll the video again.
To address the demand for more storage capacity, 3D NAND structures build vertical strings of memory cells. Let's take a look at some critical steps in how 3D NAND is made. The manufacturing process starts by depositing thin alternating layers of materials, such as oxide and nitride films. Each layer must be highly uniform and extremely smooth. To define the vertical channels, a hard mask is deposited, and openings are formed, followed by high aspect ratio etch through all the layers, which define the size and shape of the memory channel holes. Contact pads to the word lines are created using controlled etches that produce the distinctive staircase shape structure. To form the slits, a hard mask layer is deposited, and openings are patterned. Slits are then etched to separate the columns of channel holes from each other, creating an array of memory cells.
In some 3D NAND schemes, the nitride layers are removed, and tungsten word lines are then created with an inside out atomic layer deposition process. By stacking cells vertically, 3D NAND offers higher bit density for increased memory demand. Lam's advanced deposition and etch systems are making high volume production of 3D NAND devices a reality.
As you saw, the fabrication of a 3D NAND device is very deposition and etch intensive. As this slide shows, our products are used throughout the fabrication sequence. The stack of oxide nitride is deposited with our Vector Strata system. The slit is opened up with our Flex Dielectric Etch system. The channel, which is the most critical process step because that's defining transistor performance, and it's the etch process that defines all the critical dimensions, is also done with our Flex Dielectric Etch product. The word lines are filled with our tungsten metal CVD process in the first couple generations of 3D NAND, and now with the latest generation of 3D NAND, that's migrated to atomic layer deposition. We're the first ones to introduce atomic layer deposition tungsten for this application. The stair etches are done with our Kiyo Conductor Etch. Multi-level contacts, again, Flex Dielectric Etch.
The fill of the contacts with our tungsten CVD ALD deposition system. On the back end of the line, connecting all the arrays of devices together is our SABRE electroplating product. How are we going to extend the leadership position again? We are focused today primarily on 96 pairs. 48 pairs is in production, 64 pairs. We have leadership positions all the way through 64 pairs. 96 is where we're focused. There's even development going on up to 128 and beyond as we try and test the limits with our customers on our process technology. The challenges in the stack deposition are wafer bow due to stress. This industry, is a very stressful one, and it shows up in the films and it shows up in me sometimes. That stress builds up, and each layer has some stress in it.
As you add more and more layers, the stress builds up, and the next thing you know, the wafers look like a potato chip. You can imagine that might be a problem in terms of ongoing, continuing the process. Also, the uniformity of the layers is critically important because it's the uniformity of these layers is going to, again, determine the properties of those devices. If the layers are non-uniform, it means that one device in one part of the wafer or one part of the chip is going to perform differently from another part of the chip or the wafer.
Productivity, as you add more layers, if you just allow the process to run at the same rate, the cost is just going to scale proportionately with the number of layers, and that's not going to give us the kind of cost scaling that has always driven Moore's Law. We have to get more productive as we go up. Again, our quad station architecture, which is a big chamber, which enables us to run lots of wafers before the tool is taken offline, usually for an in-situ clean, before you actually take it hard down for a real Wet Clean. We can run longer than our competition because of our chamber architecture. We have the highest throughput density because we have four wafer stations per chamber.
We have best-in-class uniformity. We have developed techniques within the same chamber to compensate for the inherent stress in the films as we deposit them. We've seen over the last couple of years a six-fold increase in the Vector Strata shipments. That's looking backwards. Looking forward, because of the improvements I've just described, we expect to hold on to our strong position in Vector Strata as the SAM is going to grow because you got a lot more layers that you're putting down. Even though we're improving productivity, the market's going to grow faster than the productivity is improved. Turning to that most critical of etches, it's not only the most critical in 3D NAND, it's the most difficult etch, I've seen in my almost 40 years in this business.
It's a very high aspect ratio etch, the aspect ratios, as you can see here, are increasing as the number of layers goes up. The dimensions of the hole aren't changing significantly, so you're just changing the depth. That makes it more and more difficult to get reactive species, radicals, or ions down to the bottom. That means you have problems with profile control. You want a nice vertical hole all the way down through 96 pairs, and we're going to go beyond that. It also has an adverse impact on throughput and cost. It's a fundamental limitation. If you can't get the reactive species to the bottom, the etch rate's going to fall off. How do we deal with that? We have developed proprietary ion energy control technologies, and we continue to evolve those technologies. There's an enormous room for innovation in this space.
We have many different options available to us, they're all based on our use of a small volume confined plasma, which has been at the heart of Lam's dielectric etch technology, since the early 1990s. It's highly differentiated. Nobody else in the industry takes exactly the same approach, it's the reason we have 100% share in this space today, we expect to maintain that share position as we go to 96 layers and beyond. We've seen a six-fold increase in shipments over the last couple of years with this product. The next critical technology is that wordline fill. This is a tungsten fill today. Again, stress is a big problem in the deposition of this film. The tungsten is actually very highly stressed and can cause that potato chipping effect on the wafer. There's another problem.
The way the tungsten is deposited is we use a fluorine precursor today. That fluorine tends to attack the gate dielectrics and can cause reliability problems or leakage problems. We have, to combat that problem, released a new product, which is ALTUS ALD tungsten. Using an ALD process as opposed to a CVD process, we're able to reduce the fluorine concentration by at least two orders of magnitude, which enables our customers to continue scaling these devices. It also reduces the inherent stress in the tungsten by fivefold. We have in development a new technology which eliminates the fluorine entirely, although it's not really clear that that's going to be required anytime soon. The tungsten market, as a result, has grown dramatically from 2015, or will grow dramatically from 2015 through 2020, threefold.
We have here also 100% market share position with the customers that use tungsten wordline fill, expect to hold that position at 96 layers and beyond. In short, we're extending our 3D NAND leadership with enabling differentiated solutions. We're number one in deposition, etch in 3D NAND served markets today. We're addressing 45% of the 3D NAND wafer fab equipment market, which is up from 20% that we addressed with planar NAND. We have more than 50% market share overall, as Martin said, we have more than 90% share in the most critical applications in 3D NAND, which makes us most relevant to our customers, which naturally opens up more opportunities as they develop new technology and expand this technology. We see 3D NAND being an important market segment, a growth driver, for at least 10 more years.
Going from planar NAND to 3D NAND, we picked up 7 to 10 points overall. Lastly, I'm going to turn to what's happening with new devices, what's happening in interconnects, and what's happening in process control, and the opportunities that they provide Lam. In the time I have, I cannot be exhaustive in this area. This is, as Martin showed, roughly 25% of our revenues or shipments is coming from these segments. They're all growth opportunities. I'll just touch briefly on each one of these in turn. First, new devices. In the logic world, the FinFET was introduced some years ago. The fins will evolve with the introduction of new materials. We see silicon fins migrating to silicon germanium fins next. We see the fins getting taller and taller, which means high aspect ratio etching on the front end will become more and more important.
We believe that the fins will evolve to gate-all-around structures, as Martin said, initially horizontally, and then maybe vertically. It's not the only option out there. It's still a pretty uncertain space. In any case, for both the FinFETs or gate-all-around structures or similar structures, atomic scale precision is absolutely required. The critical dimensions, six nanometers, for example, corresponds to 30 atoms. What's really more important than the 30 atoms, which is the critical dimension, is the tolerance that is required in making that device, such that every device looks like every other device, and you can maximize yield. A rule of thumb is that tolerance is going to be on the order of 5%-10% of the critical dimension. We're talking about three silicon atoms or less, 1-3 silicon atoms kind of control. Clearly, we're processing on an atomic scale.
That's good news for us. We have been the first to introduce atomic layer etching into volume production. I'll show an example of that on the next foil. It's also good news for the industry, because it turns out atomic layer etching, somewhat counterintuitively, actually simplifies the process space. It makes the development of these processes simpler than what we used to do, and some examples are shown here. In a conventional etch, there's something called microloading, which has been a problem for decades in the industry, and you can see it. What you get is different profiles and different etch rates for dense features and isolated features. You have to try and minimize that effect.
At the same time, you're trying to create a uniform result across the wafer to wafer, lot to lot, while you're trying to maximize the etch rate, while you're trying to optimize the profile. Everything is done simultaneously. You also have to minimize damage. Atomic layer etching, because it's a self-limiting process, theoretically completely eliminates the microloading effect. In fact, it's not just theoretical. You can see the evidence here that we can demonstrate the virtual elimination of microloading. It simplifies the process by eliminating one of the most challenging problems. The other thing is atomic layer etching is a sequential process, you can break the process up into sequential steps and optimize each step independently, rather than trying to optimize everything at once. It's a simplification, reduces process complexity, that ultimately leads to faster solutions, shorter cycle times, and lower cost.
The other thing that's really remarkable is that when you use atomic layer etching, you get atomically smooth interfaces. That's like the ultimate selectivity, and you can see that here as well. Conventional etching tends to give you a very rough interface, lots of damage. Can adversely impact device performance. Atomic layer etching inherently gives you a very sharp, smooth interface with much less damage, much less leakage. Net result is we have been number one in transistor etch for some time, and we expect to extend that leadership position and outperform in this market segment. Part of the transistor is the contact, where you make a contact to the source and the drain.
One of the most challenging aspects of the contact fabrication, it has to do with overlay errors when you are printing the hole that you're going to then open up, then you have to etch down to source-drain area. The overlay error, a lot of people think that just comes from litho. It's actually much more complicated, and litho is probably not the biggest contributor. There are deposition and etch processes upstream that also contribute. The net result, as you can see in this picture in your lower left, is that what's printed tends to be bigger than the actual contact area of the source and drain. The reason for that is to allow for this misalignment. The net result is when you open this up with an etch, you're going to expose the spacer on the sidewall of the gate.
If you erode that spacer, which is sticking out and exposed to significantly energetic ion bombardment, what will happen is you'll create a leakage path or worse to the gate. You'll short the gate. You either have leakage between source and drain and gate or an actual short. It's one of the biggest yield limiters in making a logic device today. Our solution to this problem, and you can see in the graph, we're plotting corner loss versus landing area, which is that bottom area of the contact. Conventionally, is quite significant. As you make that dimension smaller and smaller, as you shrink the devices, the corner loss becomes so severe that you just don't yield. We resorted to atomic layer etching techniques, which gave us more than a twofold increase in selectivity over the range of landing areas.
This has enabled the introduction of 10 nanometer technology with self-aligned contacts to overcome these overlay limitations. Now, on the right, you see a graph projecting the increase in atomic layer etching applications. The first one introduced in volume production is the contact etch that I just mentioned. We see atomic layer etching as being crucial for fabricating higher aspect ratio fins, partly because of the reduction or elimination of the microloading I talked about, but also because of the sensitivity of materials such as germanium and silicon germanium, much more fragile materials than silicon. The precision of atomic layer etching and that atomically sharp interface, that high selectivity that's inherent to atomic layer etching, will be critically important in those applications going forward. I already talked about the application of atomic layer etching to smoothing low-dose printed EUV resist.
Then when we go to gate-all-around, there are even more steps that will require the precision of atomic layer etching and the selectivity and the low damage that it affords. We see this as a significant growth opportunity for us. We're already leaders, first ones to introduce atomic layer etching into production, and you can expect us to outperform the industry going forward with the investments we're making in this area. Turning next to new memory devices. I'm just going to focus on two. There are many under development, but these are the two closest to volume production. First, you have Magnetic RAM, which is comprised of a very complicated stack of materials that are traditionally considered non-volatile in using etch technology. The result is, if you use conventional etch technology, these materials effectively reactively sputter and then redeposit on the sidewall of the structure.
That redeposition shorts out the device. It's a killer. In the case of Phase-Change Memory or PCRAM, you have a different kind of damage mechanism. These materials are also very, very sensitive, but in this case, they're volatile when you etch them. Once you etch and then you expose to ambient conditions, like air, before you go to the next process step, the air reacts with those materials and causes device damage and can kill the device. We have developed a new etch technology that can pattern the Magnetic RAM without any sidewall deposition, clean process. This is being evaluated by leading memory manufacturer right now, and we expect this to go into volume production in the not-too-distant future.
In the case of the PCRAM, we have developed a new solution to manage that sidewall exposure, protect those sidewalls, and prevent the damage created by exposure to the ambient. Because of our very strong position in memory, which we've enjoyed for a long time, we are the preferred strategic partner of all the memory manufacturers, and we have the inside track to the applications emerging in new memory like MRAM and PCRAM. Next, just a few comments on interconnect processes and materials. Interconnect, this is now connecting one device to another throughout the circuit, the performance of that circuit now is largely limited by the resistance and capacitance of those interconnections. One of the problems that can lead to very high resistance or actually yield loss, occurs when you electroplate copper in very high aspect ratio structures.
The plating process relies on the physical vapor deposition of a seed layer of copper, because you have to complete an electric circuit. Because of the high aspect ratio nature of the structures these days, the PVD process that puts down the seed isn't very effective, and quite often you have a very thin seed. When you go to plate on that thin seed, you may actually dissolve the seed, and that leads to voids. Voids lead to, at best, high resistance, and at worst, zero yield. We've developed a new product that overcomes that limitation. We can take a pretty bad seed and prevent the dissolution of that seed and get a very good copper fill, as you can see in the after picture on this slide.
What's also happening is there's a transition going from copper to cobalt. That transition is taking place for several reasons. One is, with the cobalt technology, you don't need as thick a barrier liner seed, which is a high resistance set of layers and limits the resistance of the overall lines and vias. The other thing that cobalt provides relative to copper is it's less subject to electromigration, so it's inherently more reliable to more reliable contact. We have worked in collaboration with our customers to transform our SABRE copper electroplating system into a SABRE cobalt electroplating system. We've been able to leverage the number 1 position in copper electroplating to a number 1 position in cobalt electroplating. Electroplating cobalt is the most cost-effective way of putting down cobalt for lines and vias as compared to, say, for example, ALD and CVD cobalt.
Electroplating, we expect to continue to be prevalent in interconnection, and we expect to maintain our leadership position in this market segment. Lastly, I want to touch on the opportunities in process control. What's driving the need for advanced process control is, again, the requirement to reduce variability as dimensions get smaller, as you have more and more steps needed to make a chip, not just patterning steps, but the number of steps is increasing in general. The need to detect faults so that you don't process a lot of wafers, all of which are bad. You want to detect those faults as they occur. We have three different approaches to process control. The first one is real-time process control, so sensors and algorithms running in real time on the deposition and etch and clean systems that enable us to have precision control.
The example shown here is a control system that allows us to etch to depth. The traditional endpoint detector is basically your watch. You just time the etch. We've put a lot of effort for decades into making our etch very repeatable, such that you can rely on just timing the etch. The problem is the upstream processes, as I said earlier, are not necessarily repeatable. The wafers coming in may have different thickness films. Even though your etch may be very precise, you're going to get a different result because what came in wasn't so precise. You need a way to detect what's coming in and to correct for that and to endpoint to the same depth. Particularly important if you're doing a blind etch with no etch stop. We've had a capability like this for some years. We're continuing to improve this.
Of course, this market is tied intimately to our systems business. It is something that is an option or is included in a system. The stronger our base business, the more opportunity we have in this kind of process control. We also have integrated metrology efforts. In particular, we have an ability to measure the mass of the wafer. We can detect, for example, in the 3D NAND stack, if one of the layers was misprocessed, either completely or partially, and alert the fab host control system that something's wrong and the chamber should go offline. You can minimize the number of misprocessed wafers effectively to one in that chamber.
We have a significant investment, increasing our investments in advanced algorithms, such as machine learning, big data mining, to compensate for incoming litho and material variation, as I talked about before, and also to provide the capability for faster chamber matching. Chamber matching, when you're talking about matching the performance of one chamber to another at an atomic level, is an increasingly difficult problem. The net result here is we see an opportunity of about $400 million annually by 2021 in this space. In conclusion, we believe that we're investing to extend the technical leadership, which is key to outperformance in this industry. I've shown you differentiated and production-proven products. We've got a strong product pipeline. As Martin said, we're spending a lot of money. That money, I believe, we've demonstrated as being well spent in creating that pipeline and addressing these unprecedented technical challenges and opportunities.
We also have broad and deep investments in fundamental research within the company, we're collaborating with people outside the company. We have strategic collaborations with customers and across the ecosystem for faster cycles and speed to solution. Thank you. I think with that, it's break time.
Thank you. All right. With that, let's take a short break for 10 minutes, let's plan to resume at about five minutes to 11 o'clock for those on the webcast. Thanks. All right. We should get started with the second half of our presentations today. It's now my pleasure to welcome Tim Archer, Executive Vice President and Chief Operating Officer.
Okay, good morning. As Satya said, I'm Tim Archer. I'm the Chief Operating Officer of Lam Research, I think that's really just a fancy way of saying that it's my responsibility to make sure that the company is not only prepared but also executing on the tremendous opportunity that you just heard about from Rick and Martin. If you look at the title that we determined for my presentation today, it's about executing to drive value for customer and company. I just want to point out the inclusion of the customer in this statement is extremely important because we firmly believe, as Martin said, that customer trust and consistent execution in the long term leads to growing market share and a successful business. When we think about successfully executing on our opportunity, we like to think about it as a platform that's comprised of four distinct elements.
The first of these elements is investing in the infrastructure that's required for the growth of the company. The second is the creation of an efficient manufacturing operation that's capable of responding to quickly changing customer demand. The third is the development of a field operation that can regularly deliver differentiated support at the customer interface. The fourth, we'll spend some time talking about today, is our installed base business, and it's about having an installed base business that is really focused on customer collaboration and mutual value creation. In the area of infrastructure, we have been investing, and just recently, we've completed the construction of a new $100 million state-of-the-art laboratory at our facility in Fremont, California. This new laboratory is somewhat unique in its design.
We ensured that it was not only capable of supporting the products that are currently on a roadmap, many of which Rick just talked about, but it also contains an area that's specially facilitized for concept and feasibility. In this area, we can rapidly prototype and evaluate entirely new hardware concepts and advanced materials in a way that can drive much greater innovation in our technology, much faster. With it, I'd like to have you take a short look at a video of our new technology center in Fremont. Please roll the video. Okay, you don't build a $100 million video building without making one of those time-lapse videos. We fully expect that this new facility will help us meet our development needs in the Bay Area for at least the next 10 years. From an infrastructure perspective, I think we're very well-suited.
Perhaps one of the other things we want to emphasize for you is, as a company, we remain intensely focused on efficiency in the company and also focused on asset utilization. One of, perhaps the best parts of this story about this new lab is that we're able to take advantage right now of the, let's say, the strong real estate market in the Bay Area. The majority of the cost of this new facility actually was offset by the recent sale of our San Jose facility, which had actually become somewhat underutilized as we've been consolidating our headcount up to the Fremont campus. A pretty nice trade for us for the long term. The second foundation of successful execution, as I mentioned, is the factory. Rick and Martin Anstice both talked about the tremendous growth of the company.
I'll kind of put it in perspective for you. Over the last four years, the output from our factories has more than doubled. It's been pretty challenging for our manufacturing and supply chain organizations to keep up with that pace of growth. I'm happy to say that during that period, despite the rapid growth, we've been able to consistently meet our customers' expectations for best-in-class on-time delivery and improving quality. At the same time, we've been able to drive our operation spending as a % of output down every single year. We think that's a pretty nice balance. We've been able to accomplish that by focusing on building flexibility throughout our manufacturing and supply chain network.
We focused on flexibility in creating a high-quality mix of full-time employees and flex workforce so that we can respond at each of our global manufacturing locations to changing customer demand. We focused on flexibility in our supply chain. We have over 100 strategic partnerships with suppliers, we utilize them regularly to outsource non-core activities to create additional capacity and capability to respond, again, to changing customer demand. Those customers share our vision for growth and are willing to invest alongside us to create capacity, carry inventory, and invest in new technologies that we need for our products. Finally, we've invested in our global manufacturing network. We've added, over the last several years, 2 new sites to that network in Asia. Just last week, we actually celebrated the fifth anniversary of our Korean manufacturing operation.
The Korean manufacturing operation, while originally envisioned as a site that would primarily service our Korean customers, has become an important part of our global manufacturing network. Over the last 4 years, the output from that factory has increased more than 4 times with equivalent quality and efficiency to the rest of our worldwide sites. Just this year, we announced an activity in Taiwan focused on the refurbishment of older Lam equipment to put that equipment back into service to meet the growing needs of the IoT and non-leading-edge markets. We expect that operation in Taiwan to similarly ramp over the next several years, to meet the growing needs of the market. I'd like to now move on and talk about the field, kind of the third element of our platform for execution. Now, actually, you probably hear a lot of talk about customer consolidation.
From a sales perspective, customer consolidation truly has made our lives a little easier. From a field perspective, actually, the number of sites ramping really hasn't changed that much. In fact, one of the biggest challenges for field operations is that the customer's expectation and requirements for the size of those ramps and how steep the ramp is from time of first equipment delivery to full factory output, has actually been increasing tremendously. Our customers see a fantastic opportunity for their business, their time to market is critical. The burden that they're putting on us is an expectation for fast, flawless executions on all of our products. How do we address this need? We've created dedicated ramp teams that receive specialized training in how to safely and quickly start up new fabs.
We've partnered with regional service providers who are able to provide us with skilled engineering labor to supplement our own teams during periods of peak activity, so that we're never the bottleneck for the startup of a fab. We've also worked internally at improving our own install and qualification processes. Most recently, a significant accomplishment is that we've been able to reduce the install and qualification time of one of those most critical 3D NAND etch tools that Rick talked about, by 50%. The little graphic at the bottom here shows, as you've heard many times, the addition of 3D NAND wafer starts per month, last year was somewhere on the order of 270,000 wafer starts. Next year, maybe similar, maybe bigger. That's a lot of tools to install.
Cutting that install and qual time by 50% is a significant productivity gain both for our customers and for Lam Research. Now, the ultimate payoff of doing well in field operations, since it's at the customer interface, is really increasing customer trust. It's always with some pride that we report that our efforts in this area continue to receive greater customer recognition. In fact, in 2016, we've already received more than twice as many awards from customers recognizing our partnership and support as we did in all of 2015. We think that customers are standing up and taking notice of the investments we're making in the field to better support them. There's no better time to demonstrate customer support than when they're at their peak of busyness.
Now, one area that we are focused on, the development of our field operations in particular, is in China. I know there's lots of questions about what is the activity currently and in the future in China. I think as most of you know, China, through its 2025 initiative, has stated plans to significantly grow semiconductor revenue in the region. We're closely watching this activity, obviously, and the impact that it might have on WFE in the region, whether that WFE is spent by international or domestic China companies. We can watch this development ideally and happily from a position of strength. Lam has been in China for a long time. We have already seven regional offices operating in the country. In the last two years, we've roughly doubled our field engineering staff.
One of the things we're doing is we're making investments to ensure that when China does ramp in a very significant way, we actually have the field operations and local capability required to again support that ramp in the same way we've come to support in every other region of the world. We've partnered in the last couple of years with five of the top leading universities in China. The focus of these partnerships really has been twofold. One is to support research targeted at the semiconductor industry. The second is to support education programs in semiconductor-related skills to ensure that Lam has the ability to attract and hire the resources we need over the next several years to build a strong field operation.
With that, I'd like to move to the fourth foundation of execution I talked about, and that is an installed base business that's really focused on customer collaboration and value creation. The customer support business group, as Martin showed, makes up about 25% of Lam's overall business. It's a pretty significant chunk of business for Lam. It has in and of itself four distinct segments, spares, services, systems targeted at the non-leading edge markets, and upgrades. I want to state, this business really has two charters. The first charter is to improve the performance of the installed base to make it a competitive advantage for Lam in the sale of new equipment. There's a saying in this industry that the best salesperson is a well-running tool in the customer's fab, and we firmly believe that.
The first priority for customer support is to make sure that every tool we've sold meets the commitments upon which that tool was sold to the customer. We have roadmaps to continuously improve that installed base performance. The second charter of this business turns a little bit more inward, which is we want to create new products and services that serve now, as Martin said, the more than 40,000 module installed base to create value for the customer, improve productivity in the installed base, but also generate incremental business and value for Lam. We've been quite successful with that so far, and we have plans to accelerate growth in this market. In order to grow this market, you need to understand a little bit about the four different segments. As I said, spares, services, systems for the non-leading edge markets, and upgrades.
I'd really like to talk just about one of these, which is the services business, because it's a part of the market that's undergoing a pretty fundamental change. Traditionally, the services market was really all about break-fix contracts, man-on-site contracts, essentially where a company like Lam was selling labor resources to the customer. It's not the most efficient way for us to use resources. It's also not the most profitable way for a customer to spend their money. Over the last couple of years, we've transitioned our efforts in this area away from those traditional people-related service contracts, and instead towards what we call advanced services or productivity services. I'll give you an example of one of these that we've recently introduced to customers with some significant success.
It sounds pretty simple, but it's an example of the type of knowledge-based product and knowledge-based service we're trying to move towards. We developed a tablet-based product that helps our customers actually do their own service. What they do is they can use this tablet to improve their capability and success in performing a wipe clean on our edge tools, which is a pretty common routine type of maintenance. The end result is, once customers started using this advanced productivity tool from us, the success rate of doing their own maintenance and bringing that tool back into production more quickly went way up. The net result is the customer saves money, they become more self-sufficient, but at the same time, Lam gets the opportunity to sell a high-value knowledge-based tool. We're looking to continue to expand investment in the development of these productivity services.
In fact, based on the success of the first couple we've introduced to the market, we actually believe we can grow productivity services by 10X over the next 10 years from its current levels. It's a significant opportunity for us. We talked about the 40,000 module installed base. The better we do on the systems business, the faster that installed base grows. This is a business that definitely benefits from the success of the technology inflections Martin and Rick talked about. The breadth of products, the breadth of opportunity across those four segments, really gives us a chance to actually grow the installed base business faster than the overall market. The overall installed base is growing, and that's what we've been able to do.
The installed base business has grown at a double-digit CAGR over this entire timeframe from 2013, and expected to continue that pace of growth into the future. Just a couple of examples to support that. If you look since 2012, we've actually engaged now about 50% of all of our customers in some type of service contract, whether it be traditional or productivity. That's 125% increase. As I just described, the productivity service offerings is an area where we think we can actually accelerate growth at a rate much faster than the installed base because of the value being delivered to customers. Maybe the final point on the installed base business is, we didn't talk a lot about the non-leading edge tool market, but clearly you can't read anything today without hearing about the IoT market or some of the 40 nanometer and above legacy business. It's growing.
Even 200 millimeter today, we see increasing demand from customers who want to extend the capability and also deal with perhaps the obsolescence that's occurring on tools that were sold more than 20 years ago. That creates opportunities for us. If you just look at that non-leading edge business, we sold tools to more than 80 different customers in 175 different fabs in the last four years into the non-leading edge market. There is demand, and maybe the one takeaway is every one of those tools that we sold comes with it the opportunity to sell spares, services, and upgrades. Whether it be leading edge or non-leading edge, both are creating this opportunity for an annuity-like revenue for Lam Research. Maybe the bottom line there, the growing installed base business is an important part of Lam and enhances our profitability and cash generation.
With that, I'd like to summarize. Hopefully, I've been able to show you that Lam is investing in the necessary elements of execution. We've put in place a flexible global manufacturing capacity that we think will meet the needs to support the growth that you just heard about. We think we have a world-class, perhaps the best field support and service capability, as recognized by our customers. We also are working hard to develop and grow a meaningful, profitable installed base business. With that, I'd like to turn it over to Doug Bettinger, who will talk about the financials. Thank you.
All right. Thanks, Tim. Are you guys freezing? It's absolutely frigid in here. We're trying to get the heat turned up, but we wanted to make sure you were going to be awake. We knew it was early. When I left sunny California, I was worried coming to New York. This time of the year can be completely unpredictable. Packed a big, warm jacket. I didn't know I was going to need it in this room. Anyway, I've got three objectives as I go through this presentation. Only a handful of slides. I'm going to do my best to give you some context for what Martin talked about, what Rick talked about, and what Tim talked about in our recent financial performance, one. Two, all of you guys have been asking me about a new financial model. Guess what?
We do have one, and we're going to share it with you. Finally, I can't tell you how many times I've been asked this morning, "Hey, give me some idea of what you're going to do on capital allocation." I'll tell you in a minute. Sadly, I don't have a video. Every year, I do my best to figure out. Satya, write this down. Next year, video in the CFO's presentation. I don't have a video. Quick, last couple of years. I look at this, and it's been what I would characterize as amazing performance from Lam Research. We've grown at three times the rate of the growth rate of the industry. WFE over this time frame has grown at a CAGR of about 4%. Lam Research's shipment and revenue has grown in the mid-teens.
More importantly, our operating income has grown at a CAGR of north of 20%. We are absolutely delivering leverage to the bottom line, which at the end of the day, that's what this is all about. Critically important, in terms of the value of the company, is the cash generation. Over this time period, $2.3 billion of cash from operations generated. Extremely profitable business. The new equipment part is really important for cash, but what Tim just showed you about our installed base, really important about the sustainable cash generation of the company. Nice performance. Whenever you show a financial chart, you want it to go up and to the right, unless it's Rick's spending. Rick's spending is up and to the right also. This is actually pretty amazing performance.
As I was looking at this chart a couple of days ago, I thought, "What story do I have to tell you guys about what's going on in these numbers?" I went back. I joined the company in 2013. March of 2013 was my first quarter at the company. I remember meeting with Martin and Rick and Tim about what's going on at Lam Research. This was the end of 2012, the beginning of 2013. What is unique about the business? I heard about these inflections, did some homework, and I said, "Okay, I actually believe what these guys are telling me." In 2013, first quarter I joined the company, our shipments were $894 million. Revenue that quarter was $844 million. It was before the inflections began to roll through our numbers. Two quarters later, the company celebrated its first billion-dollar quarter.
We were really excited about that. That was a milestone for the company. Fast-forward to the quarter that we just guided you to, four years later, a little over four years later. Shipments of a billion, $850 million. We have doubled the size of the company. It's come from the technology inflections we've been talking with you about for a while. These are going to continue. Just kind of a data point. In that quarter, and again, in any one quarter, you can get permutations about who's spending in the quarter. The memory investments from that March quarter to the most recent quarter that we announced, up by 5X. It's all about 3D NAND. Our customers are investing. We are enabling the roadmap in 3D NAND, and it has grown our business very nicely.
If you look at shipment and revenue from 2013 to 2016, it's up 60%, a little over 60%. Gross margin, similarly up about 60%. You always want that earnings slope greater than the top line, and the slope is double. We've grown earnings per share in this time frame by 120%. Let's step back and look at this and feel really good about what we have achieved as a company. This is phenomenal growth in my opinion. A little bit about how did we do it. Martin described for you an expanding market. Talked to you about our objectives around market share. We've already had an expanding market. We've already had expanding market share. We've executed extraordinarily well in a growing market. Tim shared with you a little bit about the customer acknowledgments. Right place, right time, stepping into the arena, executing well.
It's nice to have a growing market. Our SAM as a percent of WFE, Martin showed you, has moved from the mid to high 20s into the high 30s, we believe, by 2019. Control. Tim is all about this at the company. Operating expense as a percent of revenue. This is one you don't want up and to the right. You want down and to the right, and you're seeing that in terms of OpEx as a percent of revenue. We're delivering leverage and scale to the bottom line. Importantly, though, is during this time frame, we've actually grown R&D investment by over 30%. We've done that by reconfiguring a little bit about how the company spends money, and we've moved dollars from the SG&A area into R&D as much as we can.
We've shifted the composition of the spending so that over this timeframe, that's actually $3.3 billion of investment in R&D. This is how we stay ahead of the competition. The lab is part of that. The investments in the things that Rick talked about in terms of enabling new technical capability. It's all about getting the right dollars in R&D in the right place. As a result of doing that, our return on invested capital in this business is actually very strong. Three times the cost of capital. We have a pretty rigorous process that we go through every year where we identify where are we going to make those R&D investments, what's got the highest return, rack it, stack it, make sure the stuff with the highest opportunity gets funded first. As a result of that, the return metrics in the business are quite strong.
That's a little bit about how we've delivered on what we've delivered. Quick, you guys probably will remember, July of 2015, we put a target model out there up on the page just to remind you what it was. We pinned this around a WFE of $35 billion. Whether that was right or wrong, that was our going in modeling. We expected in that timeframe, with an assumption of industry spending at $35 billion, that the company's revenue should be between $5.8 billion and $6.3 billion. In that, 22% operating income was the model between $6 and $6.75 in terms of earnings per share. How'd we do? We did pretty well. I think we expect this year we'll come out a little above $34 billion. The company's revenue, if you go to the midpoint of our December guidance, we're going to be at $6.3 billion.
High end of this target model. SAM's been a little bigger, share has been a little stronger, the top line's been at the high end of the range. Operating income is 23%. Actually, I think it's like 22.6 or 22.5. It's rounding up to 23. Again, delivered on that operating model. Spent a lot of time making sure that this is front and center of us when we're thinking about investments and things we're going to do. Earnings, again, if you go to the midpoint of the December guidance of $2.18, we'll deliver this year $6.97. This is a company that delivers on aggressive commitments. We've got a management system at the company that Tim has custody for that ensures we deliver on the numbers that we set out to deliver on. What's the next model? Here's how we're thinking about the future.
First I'll go through 2017, 2018, then I'll do a 2019 and 2020 view. First column there shows you, here's where 2016 is coming out. Again, I'm pinning this around a $35 billion spending environment. In 2017, 2018, if the industry is spending around $35 billion, which I think is a reasonable assumption, we expect the company's revenue to be between $7.25 billion and $8 billion. We expect to continue to deliver some leverage in the model, 24% operating income, and earnings in that environment should be between $8 and $9. Getting a lot of questions, "Hey, what's going on relative to your numbers right now and the future, and what does 2017 look like?" All that kind of stuff. It drives me a little bit crazy sometimes, all these tactical questions.
I'll give you just a little color in terms of how things are looking to us right now. We peer into the March quarter, I'll remind you, the December quarter shipment number of $1.85 billion is a record for the company. As we look into March, it's going to be stronger. Our expectation today is that shipments in March probably are going to be around $2.3 billion. Okay. We continue to see absolute strength in the business. As we think about what 2017 might look like, admittedly, early days, often when you're at this point in the year, looking into the back half of the year, you don't have perfect visibility into. It does feel like 2017 is going to be a little bit of a year weighted to the first half, a little bit. Again, early days.
Wanted to give a little color because I can't tell you how many questions I got this morning about what's going on near term. That's as much as we have for you right now relative to near term. We really don't think about the business near term. We think about a multi-year, stable WFE spending environment that Martin showed you the last several years have been demonstrated by. Quarter by quarter, things are going to ebb and flow. They always do. We have a sustainable multi-year business in an industry that has a modicum of stability that it didn't use to have. The 2017, 2018 model. Let's look at 2019 and 2020. Here I'm giving you two scenarios, a $35 billion and a $38 billion, consistent with prior practice. This is how the world looks to us in 2019 and 2020. The SAM expansion.
Martin showed you a $5 billion SAM expansion. Showed you 48 percentage point share gain objectives in both etch and deposition. That is the underpinning of what we're seeing here in these models. In the $35 billion scenario, you can see in the middle of the slide. We think the top line for the company is between $8 and $8.75 billion. We're going to continue to deliver leverage in the model. 24% operating income is going to go to 25. Earnings, between $9 and $10. That's if it's $35 billion. If it's $38, it's going to be a little bit better. In that scenario, we expect top line between eight and a half and nine and a quarter billion dollars. Again, leverage in the model. We're going to keep driving efficiency. We expect we'll be able to move that operating income to 26%.
If you do the math on all of that's earnings between $10 and $11. A continuation of recent history. A continuation of a company that meets its commitments, makes aggressive commitments. I think these are pretty aggressive commitments. We have every expectation of delivering on these numbers. This has been a question that I can't tell you how many people have come and said, "All right, what are you going to do with capital allocation today?" Well, let me set some context first. Obviously we've got a healthy balance sheet. Gross cash of a little under $6 billion. I think it's 5.9. There's about $3 billion of debt on the balance sheet, but this is the gross number. Like I said, this is a cash generative business. The first priority for us is always going to be the investment in the business.
If you look over the timeframe that you see, 2012 to 2016, the cumulative investment in both R&D and CapEx has been $4.7 billion. This is the lifeblood of our business. This is how we generate the outperformance from a top-line perspective is getting dollars in the right place. That's always going to be priority number 1 for how we invest. On top of that, we've had a pretty solid and consistent history of returning cash to shareholders. I'll remind you, in April of 2014, we initiated our first dividend in the history of the company at $0.18 per share. We grew it by 67% in 2019. I'll tell you how we're thinking about that prospectively on the next slide. Over this timeframe, we've actually returned $2.4 billion through our buyback program.
It's been the primary vehicle by which we have returned cash to shareholders, has been the buyback. In hindsight, looking at it's been a pretty successful execution. I think our average share price over this timeframe has been in the 40s in terms of the shares we bought back. It's been well-timed. How are we thinking about it as we go forward? We continue to have a long-term commitment to delivering sustainable cash returns. I'm building this slide to keep the suspense. First, we're here today to announce a billion-dollar board-approved share buyback program that we intend to execute over the next 12 to 18 months. First, buyback. Second, we're here today to announce a 50% increase in the quarterly dividend from $0.30 a share to $0.45 a share. Strong cash generation continuing to fund cash return to shareholders.
Lots of questions right now about probability. Hey, what's going to go on with U.S. tax reform? I have no idea what's going to go on. Clearly, there's a greater likelihood today of getting some type of tax reform in the U.S., and we're going to monitor that. Obviously, what happens there will influence what we do in the future in this area. This is kind of a stay tuned. I'm not sure what's going to happen, but it feels like there's a greater likelihood that we're going to get some type of tax reform here, not too far down the road. We'll all keep our fingers crossed. That's the capital program. That's $1.3 billion over the next 12 to 18 months. That's over 200% of the U.S. cash we generate, and probably 70%-80% of the total cash generation of the company.
That's all I have. Let me just summarize and then we'll invite the guys up, and we'll do some Q&A. Revenue and earnings growth that outpace the industry. I think we've shown a model that we have historically had an ability to deliver upside against, and just gave what I think is a pretty aggressive financial model for the company, and committed to creating value and returning cash to shareholders. That's all I had for you this morning. With that, Martin, I'll invite you back up to facilitate a Q&A session. Rick, Tim. Oops, sorry. Need to match. Me and Matt matched our ties this morning too, by the way. You didn't tell me. I didn't? Well. I think we got mics in the room, right? Martin, you want to pick and choose where we go? Tim, go.
Thank you very much. Actually, I had two things. First of all, there was a slide on the end, and I think you said $50 billion in spending in NAND between 2015 and 2020, so 10 per year roughly. NAND WFE is about 10 this year. I guess I'm wondering, does that sort of imply that you don't think that NAND WFE is going to grow much beyond where it is this year?
Well, I guess $50 billion implies $50 billion, right? It's a proxy over an extended period. There's a lot of unanswered questions. The purpose for putting that reference in place is really to try to address the question that appears to be in front of us around sustainability of investment. My objective this morning was to characterize an incredibly compelling commentary on demand for solid-state drive technology in data centers particularly, and the implications of that in terms of spending in the industry. As Doug said, there's an ebb and a flow, an up and a down from one quarter to another. Frankly, we don't care so much about that. We care about long-term sustainable growth.
Our commentary around demand and our commentary around spending of customers had a purpose, and the purpose was to try and create a better understanding of something that's very compelling as an industry statement, number one, but in the context of what Rick presented, even more so for our company, because Lam sits at the very center of making that technology inflection possible.
I think just as a quick second question. As you look at the company post the failed merger, do you think that inspection is, I know that Rick did talk about the fact that there's a $400 million opportunity for things that you're doing today. As you think sort of out into this model, do you think that inspection is something that you need to acquire, or do you think that partnerships are sufficient to solve some of these very challenging yield problems? Thank you.
You're really harsh in your selection of the word failure.
Yeah
in your question. Thanks. We had a vision, right. Naturally what we've done is we've recommitted to that vision in a different form today. We have articulated now for a number of years, Rick particularly, the importance of controlling process, opening up process windows in manufacturing and open up process windows in device design. That is, has been, will continue to be a critical agenda for the company. The strategy we were trying to pursue was a strategy of acquisition, the strategy of controlling our destiny through the purchase of KLA-Tencor. For reasons that I explained some weeks ago, we weren't allowed to do that. We are still committed in many respects to the vision, and we fall back on the model that we were executing for many years prior to that, which is an organic commitment.
The organic commitment under Rick's leadership in the last several years has been dramatically increased, the collaborative agenda with a portfolio of partners, four or five partners in that space, very integrated with the development roadmap of the company. Do we need to acquire something? Time will tell. We are committed to the vision we've articulated. The $400 million is kind of two-thirds of what we committed to in terms of the revenue synergies of the Lam and K.T. deal. We still think it's a compelling commentary on opportunity, and we're actually quite happy right now in the collaborative space executing with our partners. We continue to evaluate how best to accelerate the delivery on the promise we've described, your question is one of the things we ask ourselves all the time, and we'll continue to do so.
Okay. Thanks, Tim.
Thanks. I guess two questions. First one, for normalized WFE-
Can you get the mic a little closer, C.J.?
Yeah, sorry. I think normalized WFE has grown by about 25% in the last five to 10 years, $28 billion-$35 billion. Curious what your outlook is heading into the next five-plus years. You've outlined a great deal of complexity, and I think many other equipment companies have outlined similar trends for themselves as well. Would love to hear your thoughts there. The second question is your go-to-market strategy for process control, how you think about achieving that $400 million in revenues, and how we should think about anchoring margins for that business. Thanks.
One of you guys want to take WFE, or you want me to do it?
Yeah, CJ, we don't know exactly what WFE is going to do. If you look at the chart Martin showed, it's been a gradually increasing WFE over the last five to six years. I don't think that's an unreasonable expectation, but like I said, there's going to be an ebb and a flow, and when you have consolidated customers, you're going to get big projects that spend in one period and then move in the next. It does feel like when you look at the increasing cost intensity of adding these next-generation technology nodes, that there's an upward bias to WFE is how I would characterize it.
I'll let Rick answer the process control go to market, but one thing I don't want to get lost in that question or our answer to the question is the SAM message for the company. The economics of our company obviously have a relationship to WFE, but it doesn't end there. It's all about the segments of etch and deposition of clean and how they grow in the context of whatever WFE prevails. Our market can grow without the growth of WFE, and in fact, it has done at various points in time.
Yeah.
The story is a WFE story, it's a segment story, it's a share story. We've tried to articulate the picture on all three dimensions today. The headline is a commentary on less cyclicality and more sustainability of investment through compelling demand drivers. Rick, process control.
The process control strategies, as I outlined, there are three basic components. The first is real-time process control. For example, etch to depth or depth to thickness is the counterpart in deposition.
There's integrated metrology, where we see a need for more frequent sampling. You're going to measure wafers going in, you're going to measure wafers going out for feed forward feedback. That's not a new concept. It's been around for a long time. We've partnered with other companies in the industry for a long time in implementing those solutions. I think to some extent, we've been ahead of our time, but in the future, because the requirements are so tight, we need to open up the process window, one of the ways to do that is through the use of more frequent sampling, integrated metrology, feed forward feedback. The third element are advanced algorithms that put all these things together. What are you going to do with the data that you're measuring? How are you going to change the process wafer to wafer?
How are you going to change the process in real time? How are you going to make use of all the data, not just off of our tools, but linking it with other tools in the back? Big data mining, machine learning, advanced algorithms like that. Those are the three elements, there's a multitude of products that we're planning on releasing over the next four to five years that is going to build up to that $400 million number that I cited.
I guess more directly to your question, we go to market directly through a single face to the customer. One here, then down the back there.
Thanks for taking my question. I think you expect this question. I noticed you haven't talked about Wet Clean and you removed Direct and Target. Any comment, can you give us any update on what's going with Wet Clean?
He looks like he doesn't want to answer that. No. Why did we remove it? Because we talked about it like to death 18 months ago. We repositioned our investment. We restructured at some level our ambition in Clean. It still is a strategically relevant part of our business by virtue of the importance of clean process and adjacencies with etch and deposition. It's a business that has the lowest revenue in the mix of the product lines of the company, we focus on the big dollar tickets for the company. It's profitable business and it's strategically relevant to us in portfolio. Other than that, there is no reason why we didn't kind of capture it today. Do we have targets to gain share?
Yes, we do, they're not of the magnitude from a materiality point of view that we thought was worthy in today's presentation.
I think actually, just to add, we probably could have included some elements of the Clean story in the execution phase. Clean's actually been quite a good story relative to, as I said, our focus on efficiency. We looked at a business that really was available for us to go make some changes. Over the last year, we've improved the execution in that business, we've improved the financial performance of that business, and as Martin said, maybe just in light of the other major growth opportunities, that just didn't quite rise to this meeting. We probably neglected leaving it out of the execution phase.
Just a long-term question. A question around DRAM extendibility. Do you see DRAM eventually go through similar what we saw in 3D NAND, where there'll be a major change in architecture as we go down to 1Y, 1Z node, and there's some of this replacement memory technology kind of a solution to that, and people talk about ReRAM, XPoint, et cetera. How are you now thinking about that longer term?
That's a tough question to answer. I think everybody in the industry is struggling to answer that. I think it is possible it goes 3D. I think it's clear that new memory technology is going to come in and perhaps cannibalize some of that DRAM market. The new memory space, I think is quite uncertain and in a state of flux. We have to be nimble and ready to react to emerging technologies and new developments. It's hard to predict right now. What we do see is that the basic DRAM architecture will scale further for another couple nodes, and that's going to extend more than five years out.
Actually, just a little bit of context to your question that I tried to provide in my prepared comments a little earlier. It is the part of the industry that has the most probability of a structural change to it in the way Rick just described, it's not something that's going to happen anytime soon. There are kind of three more generations in the roadmap as currently defined, and it's pretty rare that we'd ever have more than three. Maybe there's a fourth one that shows up in a year's time. Time will tell. If you look at the investment plans that we think play out in DRAM next year, and we expect DRAM investment to be higher in 2017 than it was in 2016. By the end of calendar 2017, we only expect 25% of the installed base to be 1X capable product.
50% of it at the 20 nanometer level and another 25% to 25 and above. If you look at the cadence of conversion, and DRAM is dominated by conversion play, nobody's adding capacity, it's a conversion play. There's a number of years ahead of us before we're anywhere close to having to deal with the question you're asking, even if we do. Please. Then over here.
I had a question on the etch market share gain. You guys said about 48%. Where is that really coming from? Is it on fan expansion or technology inflections, or do you think it will come more from atomic layer etch and newer products? I had a follow-up.
Rick and Tim?
I would say all of the above.
Well, just one clarification. It's not SAM expansion, right? The SAM expansion, when we showed you the slides, share is one category, SAM is another, right? Anyway, go ahead, Rick.
Where we see additional growth opportunities, as I outlined, in patterning, there's room for us to gain share still in the 3D NAND space. Atomic layer etching, as I described, there's a number of emerging applications for atomic layer etching, and we feel confident that we can capture a disproportionate amount of that market.
Got it. A follow-up. You mentioned about working on 96-layer 3D NAND. Is that using the existing technology and chamber, or do you need to redesign the chamber completely? I'm trying to figure out if customers can use the existing install base, or do they need all new tools?
Our roadmap has us making some significant changes. We have a strategy that we'll continue to follow, which enables our customers to upgrade. It's just a question of the extent of that upgrade. Certainly, the challenge of going to 96 pairs and beyond is significant, I expect some of those upgrades to be large upgrades, expensive upgrades.
Well, this, restating what we've already said 20 times, the value proposition of the installed base and the extendibility of it, the demonstrated production performance of Lam Research in critical applications is a huge part of the commitments that we're describing this morning. There never is a perfect moat around our business. It takes time and effort to preserve what we have. We have been defending and penetrating at the 90% level for the last couple of years. Once we have it, we tend to keep it more often than not.
I guess one important point I would like to make is we have a lot of confidence in our core technology. Our roadmap is an extension of that technology. We're not looking at a dramatic change in the approach to high aspect ratio etching. We think we understand it well. As I said in my talk, there's a lot of opportunity we see to take the core concepts and expand their applicability and extend their capabilities with respect to the aspect ratios.
It's an evolutionary roadmap for the next decade. If your question is: Is there a disruption at 96 that erodes the value of the position before? The answer to the question is no. Please.
Hi. Thanks. I guess I have two questions as well. To start off with, if I look at the capital allocation policy that you guys have today, about 75% of free cash flow is going back to shareholders. Does that essentially suggest that you established that DOJ is not going to let any deal get done in semi cap space, you guys are returning most of the free cash flow to shareholders? To the extent you have tax reform that happens, could this be a more sustainable model than free cash flow return to shareholders?
You want to do the M&A one first? Inherent in that was, is there any other M&A? I think the way I would answer it, Martin, feel free to embellish, is we're always looking at stuff, we're always thinking about stuff. Don't know if there's anything else. Yeah, when I looked at the cash return, it's, to me, pretty aggressive. It's appropriate, I thought, we thought, our board thought. Felt like we needed to do something meaningful, and I hope you guys think it is meaningful, $1 billion. The question I think at the end was, what about tax reform? I don't know. This is a wait-and-see for us. We'll see how this transpires with the new administration coming in. It feels like there's a greater probability we'll get something, and clearly that would influence our thinking.
Given we don't know what it is yet, it's hard for me to really answer that question.
I think if I could just follow up. You mentioned March shipments, $2.3 billion, better than I think most of us were thinking. Just talk about what's driving this trend from what you see so far and what segment of WFE it's from.
Tim. Do you want to take that?
Sure. Primarily, we're actually seeing ramps across a number of different segments to drive those shipments. Of course, 3D NAND is a big driver in the March quarter. We're also seeing a resurgence of, as we talked about, the DRAM conversion that's coming back as well. We see also continued strength in leading edge foundry. It really is a confluence of all segments taking shipments in the first quarter.
I want to hit home again my message around balance. Right? I respectfully ask that people really process the significance of that message. This is a company that has engagements with every significant technology inflection of substance in every segment of the industry. We gave you a decent amount of color, I hope, to put that in context. The reality of the economics of this company are a byproduct of that holistic engagement in an industry in the right place at the right time, and we have to work really hard to do that, to develop the products that are competitive and sustain, and that's the investment that Rick characterized and Doug characterized in their presentations today. I think there was a question in this area as well. One and then two. Sorry. I messed you up, I think.
Thank you for all the data on EUV and the opportunities it presents for you guys. I guess if I think about that from the perspective of what your customers say, that EUV is an opportunity to maybe spend less in the trajectory that they're already on, and I would think that it's less evident to me that you're able to grow your SAM in the context of flat WFE if lithography is also going up. With all the technology detail you gave us, how do I sort of reconcile that with the idea that it's alleviating capital intensity and if you're taking share of that spend, who are you taking it from?
Okay. I'm not saying No, I'm okay. I think, relative to the last part of your question, who we're taking it from, there isn't an answer to that. We're not taking it from anybody because it's kind of a newly created marketplace. This is a story of the company anticipating a critical inflection for the industry years back, investing in development of a competitive product and technology to enable the two integration schemes that I described. We have a position relative to the EUV roadmap. We believe it's valuable for the industry. We want it to happen. It is going to continue to build the long-term interests of the industry, and as Rick characterized, we have partnerships and collaborations to try and accelerate that and make it more real. The two integration schemes are entirely different.
The spacer-based piece is three to one greater, that is an etch and dep play, that is about filling a void when it existed and building demonstrated production performance so that the integrity of trust with us as the solution provider continues. If we continue to be the trusted patterning partner, then the story of the company will continue. I don't know if you would add anything, Rick.
Thank you.
Thanks, Sean.
Martin, over here. I suspect one of the challenges you have as you try to convince all of us in the audience about the sustainability of your opportunity is the fact that ultimately a big chunk of WFE is still committed towards the PC and the handset market, which are sort of markets that we think of all on next growth. I'm kind of curious, as you think about your business ultimately around end markets, how is consumer today, exposure today versus five years ago? When you think about all these new opportunities around data center, autonomous driving, five years from now, where do you think the ultimate end market mix will look like for Lam? Then I have a follow-up.
It's a great question. I think it's really a hard one to answer, if I'm honest with you. I try to characterize a very different world in front of us. A world of fragmented demand, which means there's a ton of application drivers and a ton of different product sets. For sure, we're not going to be in a world where we're looking at the dominance of one device or one hardware product. I don't think that's part of our future. I think it's done. That could change, but the appearance of the industry is much more diversity of demand. Certainly what we try to do today is also address demand and the impact of increased demand from a content point of view, independent from units.
We can kind of persuade ourselves that there actually isn't much in front of us in terms of those cornerstones of the industry, the PC and the cell phone. The reality is, all of this applications environment is driving a ton of content. That world is just beginning. We're not even seeing 5% of the world we're going to see in the next five to 10 years. It's going to be tough for you. You're modeling the realities of this business are going to get more difficult, not less, which is why Tim focused so much on the versatility and flexibility of our operating model. Less cyclicality, but we still got to move fast. We just got to move fast for different things. You guys see the swan floating gracefully on the surface of the river.
Meanwhile, we're going crazy paddling away underneath, trying to make sure that we preserve the substance of building trust with our customers and having sustainable growth.
Martin, on your presentation, you mentioned that despite all the consolidation that's still going on on the device side, when you look at the guys that are spending WFE, that consolidation is pretty much over.
Yep.
I kind of wanted to ask the opposite question. There's been a lot of chatter over the last several years about domestic China players. There's some chatter that maybe one of your large U.S. IDMs might be getting into the foundry space. How do you view sort of actually your customer potential growing over time instead of just not consolidating?
How do I view the potential of my customers growing?
Well, number.
Number.
Oh, number of customers. You want to take it, Sean?
Sure. Well, that's why I put up the chart about China. There clearly are some unanswered questions about how China evolves. Up to this point, we're extremely busy in China, but most of that investment comes from international players making investments there. Clearly, there are stated ambitions to move some of that activity to domestic China companies. I don't know that it really matters so much for us. That's why I focused on, we're somewhat agnostic to the customer buying as much as what we have to do at the site from a field operations perspective. Even from a sales perspective and a market positioning perspective, I think you kind of alluded to it. The technology generally comes from somewhere else, or it follows one of the market leaders.
We really focused over time on making sure that we are extremely well-positioned with market leaders, and we would expect, whether it be 3D NAND or it's advanced logic or foundry moving into a domestic China supplier, that our strong market positions would translate right to those customers as well. I don't think it changes dramatically for us which way that goes.
Just to add a couple of things. Do I think we visualize that some customers that don't spend so much money that are making a play in IoT spend more of the time? I would say yes. There may be some big players in that space, and clearly, there are folks vying for established leadership in those markets, so that's possible. In China, as best we can tell, it's a pretty consolidated market play. There are some kind of dominant players. There's lots of discussion, and we have 20 projects we're tracking, but if you trace it all the way back to where the money is and where technology is, if we kind of played this thing forward five years from now, do I think we have 20 more customers of substance than we have today? No. Do I think we might have five? Maybe, yes.
It's of that order of magnitude, I would say.
I think that, maybe just adding one thing. The reason I showed the installed base business chart regarding mature markets are not leading edge. We clearly have seen an uptick in business across more nodes. It really is not just about focused at the leading edge, at least for that quarter of our business. We think that that expansion of customers, at least the customers who are active in looking to improve productivity of their fabs, is actually increasing. People are sort of dusting off old fabs and wanting to make them highly utilized for the IoT market. I think our opportunity is growing significantly in that area.
There's one here, I think. No? Please.
Yes, sir. This is a question for Rick. You have two major architectures being pushed by the 3D NAND makers. Just wondering, is the Lam opportunity going to be different between the two fundamental architectures for 3D NAND? Thank you.
Short answer is no. We have strong positions already, with both those architectures. The roadmaps are similar. They're not exactly the same. The products are slightly different. Some of the products are slightly different. The same basic challenges exist. Material systems are a little different, that's what drives different solutions. We see comparable opportunities, with strong positions that we expect to extend going forward in both areas. Any more? Any more.
Thanks for the opportunity. For Rick, I had a question on your patterning opportunity. In one of your slides, you showed your opportunity in 10-nanometer, seven, and five. For the 7-nanometer nodes, if I caught it correctly, you guys are assuming eight EUV passes. I'm just curious, how would that number change if EUV fails to come through for the 7-nanometer node?
You're correct. That's what I had on the slide, and that number is representative, not only of our own perspective, but what we understand our customers and other suppliers in the industry, what their perspective is. How would it change if EUV is not inserted, you're saying, at 7 nanometers?
Exactly, yeah.
Obviously, our opportunities grow somewhat because the first introduction of EUV would take a litho etch, litho etch definition of holes, which is a double patterning, you got twice as many etches, for example. Put it into a single litho etch loop for EUV, it would go back to two. As we said several times, the litho etch, litho etch piece of the multiple patterning market is the smaller fraction. It's 3 to 1 spacer-based multiple patterning versus litho etch, litho etch, it's a relatively small impact.
Thanks. I have a follow-up on your market share in etch. Today, you already have mid to high 50% type market share in etch. You've done an amazing job over the past five, 10 years, we have no reason to doubt the up 48% through 2018 and 2019. You'll be in the 60s by that timeframe. Is there a point in time or a certain level where your customers start to get less comfortable with giving you business? Do you think your technology, your manufacturing capability, your service and support, et cetera, et cetera, is strong enough that potentially in five, 10 years, etch will be kind of like a litho-like situation where you guys are number 1 with 80% share, and the number 2 and number 3 is with 5%, 10%? Thanks.
Well, maybe I'll try to take that one so Rick doesn't have to talk about technology. Actually what will really limit any upper ceiling to our market share has to do with how we perform at the customer interface. What trust do they have in providing us with more market share? I mean, again, a move from the high 50s to the low to mid-60s, I don't think creates significant concern. As Rick said, there are a number of very critical applications where, because of the strength of our technology, we own 100% of that application. It's incumbent, everything I tried to describe, the investments we're making in field capability, in the factory's ability to respond during a ramp. We talked about Q March, ability to deliver on time during that kind of ramp.
Those are the things that build customer trust to give you more and more share. I'd say, at this point, customers are not voicing a concern about giving us more share in etch if we continue to perform.
As much as we would like to, we probably can't realistically target more than 100% share.
Two questions for Doug. One is in terms of the long-term financial model. Is the share purchase of $1 billion, or any future share purchases, are those incorporated in the EPS estimates on the long-term financial model?
The answer is yes, it is.
Okay.
The $1 billion is one program, there's a baked-in assumption about what happens afterwards in terms of how the share count is modeled.
Okay, great. In terms of your excess cash position, in terms of net cash relative to some of your peers, even with a $1 billion buyback and an increase in the dividend
Along those degrees is still going to be relatively significant. Is that something that you feel like you need to have a much larger net cash position than some of your competitors? Or is this just maybe potentially a first step in a direction?
No, not necessarily. Right now the complexity with us is the majority of that cash is parked outside the U.S. Barring any tax reform, bringing it back isn't something we're inclined to do right now. When you look at it, gross cash is $6 billion. We got about $3 billion in debt, so net cash of $3. Just announced a program to return $1.3 billion, pretty meaningful in my mind. We're going to stay tuned and see what happens with tax reform.
Just to add, independent of the financial answer to the question, there's a business that has been growing, and we plan to continue to grow. I want us to stay versatile strategically as well as operationally to invest profitably in the growth of our company to deliver the sustainable valuation expansion that our messaging today implies. That's an important headline as well. We're a different company than most, evidenced by growing by 18% on average for the last five years and the future that we've portrayed this morning.
Thanks.
Any more questions? Counting three, two, one. Before you bail, we have no more slides and no more grand announcements, but I would like to say thank you all for showing up, and investing your time in listening to us describe our visions for our company. Many of you are long-term associates with us, either as investors or analysts, and we genuinely appreciate your partnership telling our story. Thank you again. I wish you safe travels, and as it's Friday, an enjoyable weekend. Thank you. We'll see you soon. Thanks.
Thanks, guys.