Applied Materials, Inc. (AMAT) Earnings Call Transcript & Summary

Hello, and welcome, everyone, to the Applied Materials 2021 Investor Meeting. I'm Mike Sullivan, Head of Investor Relations. I'm going to describe today's agenda and explain how you can join us in the Q&A session later. Today's event will conclude by 2:00 p.m. Eastern time, including the Q&A session. The first section of our meeting will be hosted by Gary Dickerson, our President and CEO. He'll talk about our markets, present our strategy and take you deeper into our semiconductor technology. He'll invite several of our general managers and customers to join them along the way. The second section of our meeting will be hosted by Dan Durn, our Senior Vice President and CFO. He will talk about our business model and invite Ali Salehpour to discuss our opportunities in service and display also with the insights of our customers. Then Dan will share his perspective on the semiconductor industry outlook and deliver our new financial model. After Gary summarizes the presentations, I'll join Gary and Dan to moderate the Q&A session. When that time comes, you'll be able to submit a question using the Q&A button on your screen. If you've requested an audio link, you'll be able to ask your question directly. But there's one more thing. Applied has a strong pipeline of new products and technologies that we've been developing over the past several years but can't fully describe in today's meeting. So we're also hosting a series of events called Master Classes, where we'll preview technology inflections and show you how the road map is evolving together with industry experts, including our customers. Our first Master Class will cover memory technology and take place on May 5. The second class will be on the 16th of June and focus on logic technology. In the second half of the calendar year, we'll hold Master Classes covering specialty semiconductors, heterogeneous design and advanced packaging and inspection and process control. We hope you'll join us. And now with that introduction, I'm pleased to welcome Gary Dickerson.

Thanks, Mike, and thanks to everyone joining our event today. While our investor meeting in 2021 looks different from the past, and we're missing everyone in person, we're excited to connect with you virtually and share our outlook for the industry and Applied Materials. As many of you know, I spent almost 4 decades working in semiconductors, and I strongly believe this is the most exciting time in the industry's history. Semiconductors are more strategically important to the global economy that at any time in history, and this is driving new waves of silicon consumption. As an industry, we're in a privileged position where our combined technologies can drive a huge positive impact on a global scale. Before we dive into the details, I'll highlight the 5 key messages that I would like everyone to remember by the end of today's event. First, Applied's markets are more attractive with more opportunities than ever. The AI inflection is driving a new era of innovation and secular market growth. It's clear that the future is not going to be like the past. AI computing workloads require new semiconductor solutions at a time when traditional Moore's Law scaling is slowing down. The new PPACt playbook that will shape the industry's future will be enabled by advances in materials engineering and materials-to-systems co-optimization. Second, we've aligned our strategy and investment to position Applied as the PPACt enablement company. We have the broadest and most enabling portfolio of unit process technology, spanning materials creation, modification, removal and analysis. This allows us to combine these technologies in unique and highly enabling ways that no one else can. And we've developed a proprietary methodology and suite of solutions to provide actionable insight acceleration and significantly improve the time to market of new PPACt innovation. Third, we're shifting more of our business to subscription-style revenues. We have a highly valuable installed base business. And over the past 5 years, we converted a large portion of this from transactional revenue streams like traditional on-demand spares and support to stickier long-term service agreements. In addition, we're starting to monetize new products and services using subscription-style approaches. Fourth, we continue to optimize our portfolio of businesses to drive profitable growth and higher free flows. Finally, We have a high ROI operating model that generates attractive shareholder returns. We have the highest growth potential within our peer group, and we have strong operating leverage as we grow. In our model period, we expect to grow our earnings per share at 1.7 to 2x our top line growth. Here's the framework for our event today. We'll begin with our market outlook and you'll hear from me as well as some of our customers and partners about major inflections, driving secular industry growth and new opportunities for Applied. Then you will hear from leaders in our semiconductor business, Prabu Raja and Raman Achutharaman. They will describe our strategy, investments and unique capabilities for PPACt enablement. We believe our portfolio is highly enabling, and we'll talk about proprietary new capabilities we've been deploying to accelerate actionable insight and time to market. Then Dan Durn and Ali Salehpour will explain how we're shifting more of our business to subscription-style revenues and integrated solutions. By doing this, we're delivering more value to our customers and capturing more value for Applied. Dan and Ali will also describe how we're generating attractive free cash flow in our display business. Finally, Dan will bring all of this together by translating our strategy into economic and financial impact, summarizing the progress we've made over the past few years and then walking you through our new financial model and capital allocation strategy. So let's get started. Today, our markets are being reshaped by 5 major inflections that are happening in parallel. First, at a macro level, we see accelerated digital transformation of companies and the economy at large. Second, we see a major inflection in computing toward AI. These new approaches are needed to make sense of the massive volumes of data being generated and will work best with workload-specific software and hardware. That hardware will be built from customized and entirely new types of silicon, which creates great opportunities for Applied. Third, The semiconductor industry is inflecting as the benefits of traditional Moore's Law 2D scaling slowed down and the industry transitions to a new playbook to drive power, performance, area cost and time to market. And there is no company better positioned to enable this new playbook than Applied. Fourth, as the industry grows, there's an increased focus on ensuring that growth is sustainable and responsible. And finally, there is a business model inflection. Consumers and companies are migrating away from products and transactions to outcomes and deeper collaboration and partnerships. These 5 factors combined to create tremendous new opportunity is for Applied Materials. Before I provide more insights into these inflections, I would like you to hear from an industry leader, I admire deeply for his creative and thoughtful vision of the future. Until the end of 2020, Young Sohn was Corporate President and Chief Strategy Officer at Samsung. He is currently Chairman of the Board at Harman International and a very successful venture investor, including a founding investor in Zoom Communications. Young, over to you.

Digital transformation is powering data explosion. The pandemic, in particular, accelerated data transformation of the economy everywhere. Data is expanding nearly double every 2 years. Yet, only 2% of data are currently processed. 80% data are unstructured. Turning data into insight through AI is very compute-intensive. Silicon and software are critical enablers. New compute architecture innovation is needed for higher bandwidth and energy efficiency, and hence, we're thinking about infrastructure for how data are created, stored and analyzed are critical. Generational trends like 5G, autonomous driving, industrial IoT, hyperscale computing are powering the use of AI. The future of economy and the semiconductor industry is extremely positive and strategic. Applied Material has exciting opportunities for enabling the future of computing with materials engineering.

Thanks, Young. It's clear that digital transformation will impact almost every area of the economy in profound ways, some of which we may not fully appreciate today. For most companies, embracing digital transformation is nondiscretionary because it changes the basis of competition. Those who are first to embrace these new ways of working will emerge as winners. And those who don't adapt to the changes won't be able to keep up. This is equally true inside the semiconductor industry. In our presentations today, you'll hear about how we're using new digital tools and AI to accelerate the time to market of new technologies and change the way we work, both inside the company and with customers. This is an approach we call AI to the power X. or actionable insight accelerator. As the world navigates COVID-19 and prepares for a post-pandemic era, the digital transformation of the economy is accelerating. Companies are rethinking and reengineering the way they operate, and consumers are making different choices about the way they spend their time as well as the products and services they buy. I strongly believe that many of the changes we're seeing today are irreversible since these new ways of working offer compelling advantages in terms of time and productivity. Digital transformation is built on semiconductor innovation and has significant implications for the electronics ecosystem. We first showed this data generation model in early 2018, and we've updated it regularly. Internally, we call this the data explosion chart. If you compare 2015 to 2021, data generation has increased more than 150 times in just 6 years. In 2018, machines surpassed humans in creating data, and we moved from an application-centric to a data-first world. We predict that by 2025, even though human data generation will continue to grow, it will only account for about 1% of total data generated. In other words, in the future, the industry's growth and economics will no longer be limited by the ability of humans to create or consume data. As everything gets smarter from our phones, to our cars, to our homes, we're seeing increasing silicon content. Here, I'm showing some illustrative examples, but the trend is clear. What this data shows is significant growth in silicon content per unit over the past 5 years and then continued acceleration of growth over the coming years. We're still in the early innings of adoption of these trends as illustrated by the smart home numbers. I'm sure that many of us already have voice assistance, security cameras, smart thermostats and appliances in our home today. But when you divide global sales of smart home products by the 1.6 billion households around the world, you see that on a global basis, penetration is very low today and there are decades of growth ahead. All of this is great news for Applied and our customers. But when you look deeper inside these trends, they are actually even more favorable for us. Let's take the first line on this chart, smartphone and double-click to see a more detailed breakdown of the silicon content of a high-end phone. We've split this by leading-edge logic, the application processor, modem and RF; specialty logic, the image sensors, WiFi and so on; and memory, the DRAM and NAND. 5G is driving a richer mix of high-end phones that feature more cameras and sensors combined with specialized AI computing. All this adds up to more leading-edge and specialty silicon content per handset. This is a great example of the robust growth we see in the specialty market. And the reason that 18 months ago, we formed a new group inside the company focused on these customers. We call this group ICAPS, which stands for IoT, communications, automotive, power and sensors, and they will generate over $3 billion of revenue this year. We're still in the early innings of adoption of these trends. And another way to look at this is by geography. This model normalizes silicon consumption per capita in the U.S. in 2015. Let's call that baseline 1x. Our model shows that silicon consumption per person will grow faster in the next 5 years than it has over the past 5. By 2025, China will have only reached the same spending per person as the U.S. experienced in 2015, and India trails China by another 8 to 10 years. As the global build-out of IoT and digital transformation expands, we can look forward to multiple decades of growth. Many in the industry have described the evolution of the market as having 4 eras of computing, each with increasing pervasiveness, mainframe computing with tens of thousands of units; PC and Internet with hundreds of millions of units; mobility, a supercomputer and camera in every pocket selling billions of units; and now IoT and AI, hundreds of billions of connected devices in the coming years. Here, we've mapped those eras to semiconductor industry revenues all the way back to 1980. We've roughly defined the start of each era. 1990 for the PC era, that was the year when PC sales passed 25 million units for the first time; the year 2002 for the mobility era with the first BlackBerry smartphone; and 2018 for the AI era as machines generated more data than humans. What you see is each era adds more than twice as much semiconductor demand as the previous one. I believe AI is going to be, by far, the biggest wave yet. As an industry, we have line of sight to $1 trillion market in the next decade. So what are the key elements of the AI era that we need to pay attention to? At the edge, we're seeing more and more connected devices in many different flavors. Cisco estimated $0.5 trillion by 2030, creating a yottabyte of data per year. That's 10^25 bytes of data. Next, we have huge advances in communications, including 5G, which means it's cheaper and faster to move data around. Next-generation communications do more than just enable new applications. They also determine how training and inference workloads will be distributed between the cloud and the edge. Then AI is the way to make sense of all the data that's available. Simply put, data isn't very valuable without context, pattern detection, determination and prediction, and that all requires AI. So I like to think the value of AI is providing actionable insight. I believe this provides incredible value across every industry. And today, you will hear about how actionable insight is a key strategy to enable Applied to accelerate value for our customers and outperform our markets. Traditional computing approaches are not very well suited for the extremely large data sets used in AI, and that's driving significant innovation in software and hardware, which is fantastic for the industry. As we move from general purpose computing to application-specific customized computing, it's important to think differently about the market's needs. Here, we show a way of segmenting their computing landscape. Along the x-axis is power consumption and the y-axis is computing performance or how many operations per second. What you see is that various zones of innovations emerge, each serving a different set of applications; cloud data centers, where most of the training workloads will be carried out; edge data centers like autonomous vehicles; tethered edge devices like cameras and voice assistants; and battery-powered edge devices like smartphones and sensors. Each of these zones will become a battleground for innovation with customized silica needed to differentiate and win each type of application. The major focus for all our customers is to deliver the lowest power, higher-performance chips at the best cost to serve each application. A company that has great visibility into this evolving landscape is TSMC. So it's my pleasure to ask Dr. Mark Liu, Executive Chairman of TSMC, to share some of his insights.

Thank you, Gary, for inviting TSMC to this event. It has been a great journey to work along with Applied Materials over the past 30-some years. I remember vividly working with Applied Materials to introduce the first ion implantation tool, when I was working, building TSMC's first 8-inch fab. Years have passed. I feel our partnership grew stronger than ever before. We advanced semiconductor technology frontiers together. Our innovations continued. Today, the role of semiconductor industry as a foundation technology is more important than ever. More than half of the world's population are online. There are more than 3.6 billion social media users and 2.63 billion people stream videos with it. Clearly, semiconductor innovation is at the heart of modern technology advancement. All kinds of innovation ideas are being realized one by one, from human genome deciphering, drug discovery, disease treatment with AI, autonomous driving to all kinds of digital transformations. We continuously rewrite our vision for the future and enrich people's lives. There may be doubters suggesting semiconductor technology advancement is slowing down. Our product data show otherwise. The power reduction at the same speed or speed gain at the same power and logic circuit density are still on track to sustain the historical rate for the past 5 generations. That is, power efficiency doubles every 2 years. Our coming 3-nanometer will be another full node stride to fulfill such trend. Beyond 3-nanometer to sustain this rate of improvement, and I believe we will, we need to work together closer than ever before. We need to innovate in new transistor structures, in new materials, in new system architecture and a new 3D integration. It is an exciting time. We look forward to working with Applied Materials to discover the future semiconductor innovations. Thank you.

Thanks, Mark. I strongly agree the future is going to look much different than the past. Classic Moore's Law and 2D scaling is not going to be enough to enable the future AI infrastructure at the edge and in the cloud. As Mark described, the industry is transitioning to a new playbook to drive performance, power, area cost and time to market of new devices. This PPACt playbook includes new architectures, new structures, new materials, new ways to shrink geometries and new packaging technology. Applied is uniquely positioned to accelerate this playbook and to solve our customers' most critical bottlenecks. Based on our collaborations across the ecosystem, it's clear that PPACt is the battleground for the future. But the t in PPACt, being first to market with winning products, will determine which companies are most valuable in the global ecosystem. There is tremendous pull and urgency from our customers to accelerate R&D and quickly ramp to volume production. To illustrate the value of being fast, let me use an example from another part of the ecosystem, specifically cloud computing. Company A started with a lead over company B and company B has been trying to close the gap. The value of t is the area between the curves, which in this case, we estimate to be $78 billion of revenue and counting. In this period of time, the basis of competition has also changed. Cloud companies started out by using off-the-shelf silicon and, in more recent years, have formed their own chip design teams to create custom silicon to support specific workloads and differentiate themselves in hardware. Later in the presentation, you'll hear how we're working with our customers to address the area between the curves and capture the value of t. As you can tell, I have never been more excited about our future growth opportunities. At the same time, I believe that as an industry leader, Applied Materials has a responsibility to ensure this growth is delivered in a responsible and sustainable way. Last summer, we announced our new 10-year sustainability road map. We've taken a holistic approach using our 1x, 100x, 10,000x framework, which describes the relative magnitude of our impact. 1x is our operations and how we run our business. 100x is the industry's impact, including our customers and suppliers. And 10,000x is how our technology can be used to advance sustainability on a global scale. Since launching this initiative, the response from our customers and partners has been tremendous. Let's hear from some of them who share our vision and passion to make possible a better future. Sanjay and Shaheen, over to you.

Micron is highly committed to reducing our environmental impact. We appreciate how Applied Materials has embraced similar goals and made strong commitments to increase the eco-efficiency of its manufacturing systems.

At Intel, one of our key initiatives is improving diversity and inclusion throughout the supply chain. And I'm very happy to say that as one of Intel's strategic suppliers, Applied Materials has really embraced this commitment. Applied is an industry leader driving change through investments in unique supply diversity programs. Supply diversity and inclusion programs provide new perspectives, expand access to competitive innovation and create a connection with a critical portion of the customer base. So on behalf of Intel, I would like to thank the entire Applied Materials team and look forward to continuing our efforts together to create a more inclusive supply chain.

Thanks, Sanjay and Shaheen. We're making strong progress against the challenging new goals and commitments we've set for ourselves. With respect to climate change, we're on track to reduce our Scope 1 and Scope 2 carbon emissions by 50% by 2030 and to transition to 100% renewable power in the U.S. by 2022 and globally by 2030. We're reinforcing our culture of inclusion by committing to even greater transparency, clearer targets and comprehensive training to improve our diversity and inclusion within our company and in the electronics industry at large. We're expanding our customer and supplier engagements through our SUCCESS2030 and 3 by 30 programs. And we're working broadly across the ecosystem to drive the advances in technology needed to accelerate AI and the digital economy in an environmentally sustainable way. We've aligned our strategy and investments around this vision of the future and look forward to bringing more technology innovations to market while continuing to make a meaningful difference in the communities where we operate and to society at large. Let's move on to the next section of the presentation and talk about the core of our strategy at Applied Materials, which is to be the PPACt enablement company. Over the past few years, we focused our company around our vision of the future and our customers' evolving technology needs. Our PPACt enablement strategy has 3 differentiated pillars. First, we have the broadest and most enabling portfolio of technologies, spanning materials creation, modification, removal and analysis. No other company has anything close to this breadth of capabilities under one roof. Second, we can combine these technologies in unique and highly enabling ways that no one else can. One way we're doing this is with a new class of highly differentiated products, we call Integrated Materials Solutions or IMS, which combines multiple processes and customized metrology and sensors often in a single platform, almost like a fab within a fab. And third, we are more focused than ever on velocity and time-to-market acceleration. We've developed a proprietary methodology and suite of solutions to provide actionable insight acceleration and significantly improve the time to market of new PPACt innovation. Underpinning all 3 pillars of the strategy is our highly differentiated silicon and packaging lab capabilities. We are committed to significantly expanding infrastructure in the United States over the coming years and plan to make major investments to establish new research and development labs that would supplement the innovation infrastructure we currently have, which includes the Maydan Technology Center in Silicon Valley and the META Center in New York. With these next-generation facilities, world-class talent and strong partnerships throughout the ecosystem, Applied Materials will establish a sustainable new platform for technology leadership to advance the state of chip-making for years to come. Now to provide more details on our strategy and progress in these areas, I'll hand it over to Prabu. Prabu, take it away.

Thanks, Gary. It's great to connect with you and share how Applied is well positioned to enable PPACt for our customers. There are 3 elements to our strategy: first, leveraging our unit process leadership and broad portfolio of products; second, leadership in developing enabling solutions by uniquely combining our technologies; and third, leadership in accelerating the time to market of PPACt with innovative solutions. Let's start with a quick look at where the industry is today. First, process complexity is increasing. For example, materials interfaces are more important than ever, and every atomic layer is critical. Variability is now a show stopper, and we see more 3D device inflections. Second, cost is increasing in both R&D and manufacturing. Leading-edge fabs cost nearly $20 billion today. Third, classical 2D scaling has been slowing down, and at the same time, complexity is increasing. The market opportunities are becoming larger than ever. And fourth, there is an ecological challenge. We expect massive growth of digital devices in the data center and at the edge. And this is leading to huge potential increase in energy demand. The need for high-speed data connections using today's interconnect technology will consume too much energy. To address these big challenges, a new playbook is needed to simultaneously optimize power, performance, area and cost and to accelerate time to market. This new PPACt playbook is enabled by new architectures, new structures for 3D devices, new materials, new ways to shrink and advanced packaging. And the letter t, arguably, most important to capture economic value is about accelerating time to market. The foundation of the playbook is materials engineering. The new PPACt playbook is driving number of device inflexion. I would like to highlight a few. Everything is becoming 3D. NAND was the first, now 3D gate-all-around transistors, 3D DRAM and 3D interconnects. Wiring resistance is one of the biggest challenge for both DRAM and logic. Also, we see major inflections focused on high-performance computing, like packaging and metal gate transistors in DRAM. These inflections provide big growth opportunities for Applied. Here, I will provide you a high-level summary of growth drivers that you will hear about today. One, mature nodes and specialty semiconductors driven by IoT is benefiting from our unique broad portfolio; two, the future 3D inflections in DRAM and logic play to our strengths; three, we are just in the early innings of foundry interconnect and high-k/metal gate for DDR5; and finally, our differentiated patterning and packaging solutions are seeing strong customer adoption. Together, between 2020 and 2024, these provide a total incremental revenue opportunity of $7 billion. And we expect our revenue to grow at a 13% CAGR compared to overall market growing at a CAGR of 8% to 9%. I'm frequently connecting with all of our customers. And based on our deep R&D collaborations, I feel very positive about growth outlook for us with waves of growth opportunities, following one another, especially over the next 5 years. We are the PPACt enablement company. We are in the right place at right time. I will talk about our unit process products first and then cover our unique solutions. Starting with products. Our portfolio is unique in its breadth and depth. Our products span the spectrum from creating the materials that enable power and performance, shaping and modifying the structures and unique analyzing capabilities, including our leadership in e-beam. Applied is the only company with both process and metrology capabilities, the only company with full flow labs for device and packaging integrations and their technology accelerator, META, to speed up innovation from lab to fab. No other company has this combination of technologies to accelerate the PPACt road map. So our portfolio is unique and built on over 50-plus years of materials innovations. Additionally, with our breadth, we have connected these leadership products in new ways, allowing us to deliver completely new PPACt solutions. The first category that I want to introduce you to is ICAPS. ICAPS stand for IoT, communications, automotive, power and sensors. The ICAPS business group was formed 2 years ago and is focused on delivering compelling PPACt solutions for specialty devices and mature nodes. By developing new ICAPS-specific products and solutions for both 200-millimeter and 300-millimeter, we have built deeper collaboration with our customers, resulting in a strong revenue growth for Applied. We now generate more than $3 billion per year in our ICAPS business. Now let us hear from your leading ICAPS customer, Tom Caulfield, CEO of GLOBALFOUNDRIES.

Hey, thanks, Prabu. Happy to be here. As you all know, GF is the world's pure-play foundry, and we're focused on a broad and important part of the market for semiconductors in the world. We are 1 of 5 foundries of any scale, now 6, if you include Intel's announcement last week. To put GF's partnership with Applied into perspective, I need to first offer some thoughts on the pivot our semiconductor industry made, and it's been in the making for over 1.5 decades. Our industry began a fundamental shift with the emergence of the smartphone, which brought new and exciting features into a device that we carry in our pocket. It had camera. It had power management chips. It had a touchscreen display, had great audio capability. Now these are all chips, by the way, that GF produces for its customers. And so the industry made this shift from a compute-centric focus to pervasive deployment of semiconductors. That's semiconductors everywhere. And in fact, in many ways, the smartphone led not only to this pervasive semiconductor, but it led to the Internet of Things that's now moving from all things connected to all things intelligent. And as a result, our $80 billion foundry semiconductor industry is now dominated by 70% of that industry is for pervasive type devices. So what does this have to do with Applied Materials? Well, Gary Dickerson and team saw this trend early on and created a group dedicated to innovation on adding features to semiconductor products on all nodes. He didn't limit Applied's innovation just to transistor scaling. He created the ICAPS team to deliver on this mission. And today, GF, we leverage much of AMAT's capability in the technologies we innovate and manufacture, things like embedded memory for secure transactions and microcontrollers, world-class RF technology for connectivity, high-voltage devices for electrification of vehicles, low-power consumption devices for untethered Internet of Things devices. The technology list is endless. So thank you for a great partnership, Applied Materials, but I do have a favorite to ask, can you accelerate some of your equipment deliveries to us? We have customers that need our capacity expansion post haste. Thank you.

Thank you, Tom. Moving now to solutions and connected products. We have 3 major offerings where we are seeing tremendous growth in 2021: one, integrated material solutions or IMS; two, advanced packaging solutions; and three, the new applied actionable insight accelerator, which connects our process systems with metrology and inspects and products, along with unique sensors and machine learning algorithms. We are building these solutions from a position of strength. We are already #1 in foundry and logic, #1 in DRAM, #2 in NAND and #1 in packaging. And we are using our unit process leadership and domain expertise to develop integrated solutions for transistors, interconnect, memory and packaging. Now I would like to share our first case study. Applied has enabled the transistor road map over the last 2 decades. We started with selective AP, then enabled high-K/metal gates and now advanced FinFETs. Our leadership products such as epi, implant, metal gate, anneal, CMP and selective removal are highly valuable in fabricating low-power and high-performance transistors. Looking ahead, the new major inflection in transistor structures is gate-all-around. Let us hear from one of our leading customers, Dr. Wong, Executive Vice President at Samsung on his vision for future devices.

At Samsung, we see long path ahead for continued performance and process scaling of our devices. We will take advantage of 3D technologies in the years ahead using 3D technique, new materials and material [indiscernible] technologies. And we look forward to working with Applied Materials as we invent the future at Samsung.

Thank you, Dr. Wong on your 3D device perspective. Gate-all-around provides substantial PPACt improvements, up to 30% lower power and up to 15% higher speed. Building gate-all-around transistor is more complex. It requires multiple new steps in our leadership areas such as AP, selective removal, metal gate, gapfill. Here, interfaces between the materials are critical and Applied's transistor portfolio and decades of transistor leadership is highly valuable. We estimate that Applied's market opportunity for gate-all-around to be $1 billion per 100,000 wafer starts per month. With structures going 3D, e-beam is highly enabling technology and also a significant growth opportunity. Identifying buried defects is a big challenge for the industry and finding them faster will make a huge difference in our customer time to market. Our e-beam inspection enables 5x faster detections of the e-limiting defects. e-beam gives us orders of magnitude more data to see across wafer and across chip signatures for fast and actionable insight. In our process control business, we grew revenue substantially over the last 4 years and are well positioned to double our revenue over next 4 years. More 3D inflections are emerging. DRAM scaling is slowing down and expected to reach physical limits. Going to 3D DRAM opens a new direction in scaling for similar to what happened in 3D NAND. But DRAM is very different than NAND due to its performance requirements. The stack materials need to be conductors in contrast to dielectrics in 3D NAND, also the materials needed to be engineered for fast device performance and lower power. So instead of dielectrics deposition and dielectrics tickets, this will be conducted deposition and conducted etch and selective removal. And these are applied strengths. Similarly, chips are reaching scaling limits, and we need to consider new ways to route the power and signals. Doing this could deliver chip area savings of over 30%. These results in new steps, information of buried and backside rail structures through silicon vias and backside processing. These inflections are enabled by our leadership products and solutions. Now let's change the gears and talk about solutions made possible by unique combinations of our technologies. Here is a framework to think about various level of solutions as we move beyond unit processes. The first type, co-optimization of processes is where we consider how to optimize upstream and downstream steps. About 40% of our products today have an element of co-optimization. The second type is Integrated materials Solutions. Interfaces are increasingly limiting PPACt. Any air exposure will impact device performance and yield. Our unique ability to connect broad leadership product in vacuum provides interface innovations that have huge PPACt benefits. About 30% of our products come designed with these capabilities. The third type of solutions connect our process systems with our metrology and inspection products, unique sensors and machine learning algorithms, which works as a tightly integrated solutions to accelerate PPACt. Now let us hear from one of our leading customers, Sanjay Mehrotra, CEO of Micron, on the increasing need for solutions.

As we continue to scale our technology to smaller dimensions, we find that many materials and process technologies that we used in the past are no longer viable. We appreciate how Applied Materials has been working with us to invent new film technologies that are co-optimized with etch technologies, which allows us to speed up development.

Thanks, Sanjay. As you just heard, co-optimization is very valuable and enables our customers' device road maps. DRAM chips are key performance enabler for data centers. And inside the DRAM, one of the most critical structure is the capacitor. The capacitor is an ultrasmall structure, about 30-nanometer in size which is almost 250x smaller than a blood cell. In the state of our DRAM today, there are 16 billion capacitors in an area that is smaller than a fingernail. Now let's talk about the challenge. There is a geometry change, combined with the further scaling of capacitor dimensions. This is causing a big increase in an aspect ratio, which then recurs taller hard mask for patterning and taller mask resulting higher etch process variability and defects. For example, varying mask hole sizes create issues with the final capacitor profile and cause yield problems. Current hard mask and etching no longer work and are limiting capacitor scaling. At Applied, we developed 3 innovation that helped to solve this problem. First, we developed a new hard mask material, Draco, with 30% higher selectivity. With this new material, We can lower mask thickness by 30%, which decreases aspect ratio and reduces the burden on etch process. Second, we delivered a new Sym3 etch technology that is tailored for the properties of this new material. Third, we are utilizing a unique nondestructive metrology innovation to measure bottom CD dimensions. This solution helped to accelerate our customers' time to market and gave us very significant growth in DRAM CVD and etch revenue. Bringing a new material to manufacturing is never easy. It needs to drop into an existing process flow and be compatible with upstream and downstream steps. By co-optimizing for process interactions between deposition and the etch, we accelerated the R&D learning cycles. And by using unique metrology, we got better and faster results. Overall, our solutions improved the local CD uniformity by over 50% and reduced the bridge defects by 100 times which provides a large increase in yield. Such improvements are only possible because of our breadth and the unique combinations of our capabilities. Our DRAM capacitor hard mask solution provides us with a $1 billion cumulative market opportunity through 2025. We are already generating revenue and expected to grow 4x more by 2024. Now let's shift gears again and look at Integrated Material Solutions, and hear a customer's perspective from Sanjay Natarajan, incoming Senior Vice President at Intel.

Hi, Prabu. Intel is improving our focus on process technology execution to accelerate our product leadership. We look forward to seeing the superior performance and the validation acceleration that comes from in-vacuum step integration. This is what Integrated Materials Solutions promises to provide as Intel continues to drive new ways to move forward on Moore's Law. We look forward to working with you, Prabu.

Thanks, Sanjay. Now let's look at an Integrated Materials Solution case study for interconnect in leading logic devices. Once again, I will begin by setting the context. A leading-edge mobile application processor has over 11.8 billion transistors, 10 billion via connections and almost 16 layers of interconnects of varying dimensions with ones closest to the transistor being very narrow. So what's the problem here? As seen on the chart, via resistance is increasing exponentially with the smaller dimensions and higher resistance equates to higher power consumptions. Almost 1/3 of the chips' power is consumed by the interconnects. Now let's look at what causes the resistance to increase. As the dimensions become smaller, interface properties become more significant than bulk properties of the material. And at a narrow via line connection, there are more than 5 materials interfaces. Interfaces dominate at small diamond sense and contribute more than 50% of total resistant. In summary, we need new interface engineering techniques to solve this critical roadblock to power, performance and scaling. Only a few atomic layer of materials are going to limit your cell phone's battery life. Here is the good news. We have been able to solve this challenge by using IMS. Before I show you exactly what we did, let me summarize the approach and the results. A few nodes ago, making an interconnect was less complex. You needed a couple of materials, you have few process steps and conventional process methods. Now we have to carefully engineer the interface, and we have to use our broad set of capabilities, including selective processing, metal reflow and many more materials. On the right chart, you can see the results. With our innovative IMS solution, we have been able to significantly lower the resistant to the level of previous node. This is a huge PPACt improvement. Each chip has over 1 billion vias per layer and at least 5 critical layers. So lowering the resistance can have a huge positive impact on power consumption. Here is a slide that explains what you are about to see in the upcoming animation for IMS. The integrated Endura tool on the left has many different technologies mounted on the same platform. ALD, PVD, CVD, interface engineering, surface treatment, reflow and onboard metrology are shown in different colors. All are integrated together in vacuum. The cartoon in the middle is the magnified image of copper via structure and the 7 steps listed on the right show the process sequence happening inside the tool. Now here is the video that shows how the IMS solution works. [Presentation]

Wow, that was an incredible sequence of step with interface engineering, gapfill with flowing copper at the end. To summarize, 7 sequential steps and all done in vacuum. And I want to say, only Applied can do this. These integrated solutions are unique and exclusive to Applied. Going from 20-nanometer to 5-nanometer and below, the process complexity is increasing, the number of steps more than double, increasing our opportunity by more than 3x. We have been the leader in interconnect for over 30 years, evidenced by a strong installed base of over 25,000 chambers on our flagship Endura system. And we have a track record of consistently delivering node-over-node power efficiency gains. We just discussed the power and performance challenges of on-chip interconnect. However, interconnect challenges extend to packaging, where they impact the overall performance and energy consumption of the system. More advanced packaging is targeted to address these off-chip firing issues. Let us hear from Babak Sabi, Corporate Vice President at Intel, who is driving many innovations in this space.

Thank you. Intel continues to drive advances in heterogeneous integration and advanced packaging architecture to deliver ever-growing product design flexibility, power efficiency and overall device performance. With Ponte Vecchio and Exascale GPU, we have brought together our Foveros 3D packaging technology and our embedded multi-interconnect bridge or innate 2.5 solution in a product for the first time. These technologies enable the package-level integration of leading-edge XTU silicon, application accelerators, audio components and high-bandwidth memory, demonstrating the system on a software capabilities required by high-performance compute application. Applied Materials is and will remain a key partner for Intel in continuing to drive technology advancements in advanced packaging.

Thanks, Babak, for giving us the opportunity to work with you on this exciting future technology. Advanced packaging is a growth area that the industry is pursuing with multiple approaches. There are 3 common goals. First is creating more and shorter interconnects because more IOs reduce latency and short-line lengths reduce power consumption. Second is 3D stacking. By stacking devices, we can simultaneously increase bandwidth, reduce power and shrink areas. This is now the strategic enabler for system integration and performance. There are many recent examples. Here is a GPU where advanced packaging helped improve the performance by 3x and lowered the power by 50%. Third is chiplet. It provides the ability to customize and use different building blocks to drive application-specific performance. It also enables smaller dies and improves yields. Advanced packaging is not a low-end market. It is becoming more and more like front-end processing. As the dimensions shrink, process complexity is increasing. The best solutions come from optimizing a broad portfolio of technologies and combining them in unique ways to achieve system-level benefits. What you are about to see now is an animation into the world of advanced packaging It is meant to introduce you to the building blocks for the off-chip interconnects and the various terminologies you will hear in a minute. You'll see that advanced packaging is entirely different from the low-end technology of the past. This is a new era of high process complexity and variable system PPACt enablement. Here is the video. [Presentation]

It is amazing to see the complex building blocks that make this new kind of system integration possible. Here is a system integration road map from bumps to micro bumps, to TSV and to hybrid bonding. There is clear step function increase in performance and step function reduction in power. Together, these building blocks enable the future integration of logic, memory and power management device as a system. Hybrid bonding, the newest reduces the interconnect lengths to the micron scale. Let's talk more about this. Hybrid bonding is about the capability to attack chips directly together and to a wafer. There are 2 critical modules for hybrid bonding, namely pad formation and bonding. First, pad formation is very similar to the interconnect inside a chip, which we saw earlier. We have a strong product portfolio and decades of expertise in this area. Second, bonding is about placement accuracy and metal to metal connections. We are excited to be partnering with BSI, an industry leader in packaging to jointly develop enabling solutions for our customers. To summarize, packaging now is an important PPACt enabling technology for chip and system performance and a new market for advanced process equipment and know-how. Applied is well positioned with our unique capabilities, a broad portfolio and a full floor lab where our customers can innovate with us. Additionally, we have a strong ecosystem collaborations and partnerships with customers, peers and research institutes. It's a great growth business for us with revenue of around $500 million in 2020. And we are still in early innings of growth. In summary, we reviewed 3 PPACt case studies for solutions-based approaches. We are combining our technologies to bring big improvements to our customers' devices. We have a strong pipeline of many such programs, and we are deeply engaged with our customers. We are excited about the validation results and customer pull for these programs. Now I would like to invite Raman to introduce you to something new and exciting, how we are generating actionable insight to accelerate the time of market of these valuable PPACt solutions.

Thanks, Prabhu. As Gary mentioned at the beginning of today's meeting, PPACt is the battleground. But the T in PPACt is the biggest value driver for all the companies in our ecosystem. In this section, I will talk more about the new applied AIx platform launched recently. Then I will cover case studies on how we employ this platform to accelerate customer solutions and applies internal R&D, speed to solutions and reducing are among the top high-value problems for our customers. Let's hear from one of our leading customers, Dr. Seok-hee Lee, CEO of SK Hynix Semiconductors.

It's hard to simply put a number on it, but we all know that improving the process margin is the key to enable technology node migration. In many cases, it's not only requires the adoption of advanced new technologies in a number of areas, including materials, process, and equipment, but also requires all those factors to be optimized for the integration of multiple process steps. Each change and process variable affect each other at multiple levels. So accelerating the cycle of learning to come up with an optimal solution is crucial. If Applied Materials develops new process technologies, which are already co-optimized with adjacent process steps, it will help reduce development complexity for chip makers. So I think our development activities can move at a faster pace if you work together to harness the power of sensors, Big Data and AI to map and predict the effect of many process variables.

Thanks, Dr. Lee, for emphasizing the importance of variability reduction, and we can agree with you more. Before we get into the details of how Applied is delivering solutions for PPACt acceleration, let's set the context. Everybody knows about Big Data and AI, so why is Applied talking about actionable insight? Our process tools produce over 1 petabyte of data in a year, for metrology and inspection tools also generate endless data. There is tremendous data verity, and the key question is, how much of the data is actionable for a given context or a given problem? The truth of the matter is that only a few critical data signatures have direct correlation to results or outcomes. We call this actionable data. Finding actionable data is like finding a needle in a haystack once the actionable data is identified, scaling it with speed becomes important. Speed in amplifying the actionable data can be a huge differentiator and is the prerequisite for developing actionable insights and better results. At Applied, we have unique actionable data generators such as proprietary sensors, modeling, in-vacuum and e-beam metrology. And we are introducing and deploying AIx, our actionable insight accelerator platform, along with innovative machine learning algorithms to deliver these insights. The applied AIx platform brings together process tools, sensors, metrology tools, analytics and machine learning capabilities and is designed to be scalable for new context and new problems. There are 3 elements to the framework. First, we cannot fix what we cannot see our measure, so having real-time ability to see into the process with [indiscernible] sensors and onboard metrology is the starting foundation. As you saw in the [Audio Gap] air exposure changes interface properties and hence, material properties need to be measured in vacuum. This is an example of critical insights from unique, actionable data. As I said before, process complexity is high. We can have more than 10,000 process possibilities for each tool and more than 1 million combinations for an integrated flow such as the interconnect. Hence, generating actionable insight is vital. Another innovation from Applied that provides actionable insights are our e-beam products. E-beam technology leadership enables us to measure 100x faster at a 50% better resolution and provide a map with millions of points on a wafer. This gives us wafer and chip signatures that help us learn fast and accelerate PPACt innovations. The last step is to bring all this actionable data into our Applied AIx analytics platform, which can be used across all Applied tools. Some of the key components in here include chamber AI; machine learning algorithms for process health; Applied Pro, our process recipe optimizer for a single tool or for matching across many tools; digital twin models; and integrated control schemes. Our goal is to accelerate every stage of product life cycle from R&D to ramp and to HVM. Speed and variability are key problems for our customers. And with applied AIX we can reduce the time by a factor of 2 and improve the process window by 30%. Now I'll give a few examples of accelerating time to PPACt solutions. First, let me refer back to the case study that Prabhu covered on the new capacitor, hard mask co-optimization solution and provide a little bit more color. Briefly, the whole size variations of the mask impacted the final result and cycle time for final result was long. Step 1 was to find what data was really actionable? Which was the CD or the whole size at the bottom of the mask? Conventional methods such as TEM are slow and destructive, sample volumes are small. Hence, scaling actionable data was critical. Using ProVision, along with its unique imaging and special algorithms, we were able to get bottom hole size measurements on a massive scale. Massive sampling enabled us to improve the signal to noise and measure the actual variation. With a 10x increase in sampling, we were able to optimize the process knobs or settings and CD uniformity improved 5x. We were able to deliver better results in a faster turnaround time by combining our leadership in integrated solutions with our leadership in e-beam and our differentiated Applied AIx. Another example is the recent product launched by our inspection and process control team, Extract AI. As geometry shrink, the signal-to-noise ratio for small defects gets worse. Optical inspection is fast for data capture, but has lower resolution, while has higher resolution but it's slower than optical. Submission can see and identify the actionable defect data and train the Extract AI software for defect and noise classification, combining Extract AI with in light has the ability to turn an unclassified map of 1 million potential defects into an actionable map of 1,000 yield impacting defects. Finally, I want to summarize by providing 2 case studies of how Applied's AIx platform has helped us accelerate R&D for critical applications and also transfer them to high volume. Etch has very high process complexity, There are more than 100 processed knobs, and you can have more than 10,000 ultrafast changes with transient FX. Chamber components also get eroded during processing and hence, real-time conditions are always changing. Every process recipe requires an associated clean recipe that needs to be developed to ensure process stability. As you can see on the charts, by combining the elements of applied AIx such as the Applied process recipe optimizer and chamber AI, we have been able to cut process development time and cleaning recipe development time in half. Variability reduction and process window expansion are critical to achieving high yield. Here, I have a case study of how Applied process recipe optimizer or Applied Pro was used to find a multidimensional optimum, in this case, using 7 parameters for an etch application. Multiple outcomes had to be simultaneously optimized, spatial signatures, at least 2 critical dimensions and defectivity. Applied Pro provides the ability to fingerprint the process space, which is important for matching, doing the transfer to high-volume manufacturing. In summary, Applied AIx Is a powerful and scalable platform that connects all the critical capabilities of Applied in new ways to accelerate time to market for PPACt solutions. We saw many examples of getting better and faster outcomes with this highly differentiated capability that will continue to scale in the future. Applied AI strongly positions us to solve bigger and more complex problems for the customers in the industry. Now over to you, Prabhu.

Thank you, Raman. PPACt enablement is the right strategy at the right time. It is highly valuable to our customers and is highly valuable to Applied. Applied is a company like no other. We have leadership positions in many areas. Our product portfolio is unique in its breadth and depth. We can deliver exclusive combination of technologies in new and innovative ways. And we accelerate these combinations with actionable insights. I have been doing product development for more than 25 years. And the customer pull for our PPACt innovations is the strongest that I have ever seen. Multiple future inflections give us growth opportunities. And by going from unit processes to the solutions, we can further expand our growth. It's truly Applied's time. Our best years of growth are ahead of us. And now I would like to turn the meeting over to Dan.

Thanks, Prabhu, and thank you to everyone for joining us. By now, I hope Gary has convinced you about the exciting growth opportunity ahead for the entire semiconductor industry. And that Prabhu has you understanding why the growth can only be achieved when we overcome challenges like the slowing of classic Moore's Law scaling and the limits of conventional materials and unit processing techniques. Major inflections are on the horizon, including 3D logic, 3D DRAM and advanced packaging. And enabling these devices plays to our strengths as a company. We appreciate the support of our customers today, and we value being their close partner on this journey. Now let's move into the second portion of the meeting where we'll discuss our strategy to drive subscription-like revenue. I'd like to link everything you've heard today to a business model inflection. As Gary explained earlier, we're focusing our strategy on enabling PPACt. We'll continue to deliver great products, but we can make our customers more successful faster by combining our products and technologies in unique ways. Prabhu and Raman described a continuum from unit process to co-optimization to integrated material solutions to digital solutions where we use metrology and AI to provide actionable insights that increase process windows, accelerate time to market and increase yields. This work depends on sensors, data and analytics, which we're now deploying throughout our business. It's our version of Industry 4.0. As you've seen from Prabhu and as you'll soon learn from Ali Salehpour, we're using industrial IoT and AI throughout the company to enable our customers in new ways. We're enhancing our products with sensors, connectivity, computing and digital twins. This allows us to digitize everything we're doing and give customers the actionable insights they need to drive the road map and accelerate time to money. With this new technology, we're able to give our customers better insights and better outcomes every day, week and month. And the best way to deliver outcomes is through subscriptions, which give our customers continuous improvement and more predictable cost structure. And this brings us to Applied Global Services. AGS changed its strategy in 2013. And here, you can see not only the growth which has accelerated but also the change in mix to subscriptions. Until around 2013, AGS was an aftermarket business, selling parts, services and legacy equipment upgrades on a transactional basis. There was a lot of low-cost competition and not a lot of growth. But the team adopted a true services mentality. It's not just about aftermarket, it's about what happens before, during and after our customers transition from R&D to high-volume manufacturing. Giving customers better outcomes during this critical transition accelerates time to yield, time to market, market share gains and the financial returns on multibillion-dollar investments. So AGS moved from transactions to assurances. First for basics, like parts availability and service response time and then to more sophisticated outcomes based on digital strategies, and the journey continues. With that quick introduction, I'll now ask Ali Salehpour to discuss how the AGS team is pivoting the business and most importantly, why this is valuable to our customers.

Thank you, Dan. In this section, it's my pleasure to talk to you about Applied Global Services, AGS. Earlier, you heard Gary and Dan described how the company is shifting more of our business to subscription style revenue. I'll share the story of how we're transforming AGS as we move from on-demand parts and maintenance to customer outcomes, deliver it through long-term service agreements that have subscription-like revenue. This has been an exciting journey that's required change in everything we do across the organization. This starts with developing a deeper understanding of the highest value problems facing our customers. Then we use technology in new ways to change how we operate, all the way from how we engineer and deliver parts to how we service our customer systems and deploy our network of experts. As we deliver more value, our customers are entrusting us with a larger portion of the money they spend to manage their capital investments. We are becoming a more tightly integrated part of their operations, and more trusted partner in their success. I'll share some data to illustrate this point later in my presentation. Our goal is to help customers solve much more than just their maintenance problems. Ultimately, our knowledge of our systems and fab operations is something we put to work, helping customers generate better return on their fab investments through cost output and yield improvements. We do this at the leading edge and in the specialty nodes. Today, we have over 14,000 tools under long-term agreements and the portfolio of services we offer customers are getting more and more advanced. Over the next 15 minutes, I'll describe how we're delivering better and more valuable customer outcomes, uniformly across the largest installed base in the industry. Before that, let me quickly describe some of the challenges our customers face today. This is a chart I've shown you before. It represents the life cycle of a process node. The x-axis is time. The y-axis is output and yield or better thought of as revenue and profit for our customers. New technologies and equipment are developed in R&D, both at Applied and at customer site, then transferred to small volume pilot lines, then ultimately progress into high-volume manufacturing. The green line depicts the ideal ramp of the output and yield through these phases of development. We refer to this as the S-curve. The dotted orange line represents the reality of the ramp a customer must manage. For a state-of-the-art fab, which can cost $20 billion or more, the difference between the orange line and the green line or what we call the area between the curves is worth hundreds of millions of dollars, if not billions of dollars. Time is incredibly valuable, a delayed ramp means longer time to return on these huge capital investments. And the opportunity cost of missing a product cycle or giving up a market share to a competitor is even larger than that. Equally important is the race to get to higher yields. Cost of processing a wafer is essentially the same, whether 90% of the chips can be sold or only 50%. As chip complexity rises, the ability to generate high yield is getting tougher with every node. Getting things right with a handful of R&D tool is not the same as getting results across a fleet of many hundreds of tools, each of which can have minor variation at the system level and even at a chamber level. ASPs are often a function of small differences in speed or power of these chips, and that's worth a lot of money. So the point is that winning companies managed to stay as close as they can to that green line at every stage of node development ramp. This creates a huge opportunity for Applied to create value for our customers. Our semiconductor customers have more than 40,000 Applied tools running, and we believe that installed base will increase to over 50,000 by 2024. If you add up our customer spending for engineering, maintenance to maximize output and yield, it will grow to over $20 billion by 2024, which is up almost 50% compared to today. Today, we only capture 25% of this spending, which means we have a lot of headroom to grow. When I started this role nearly 8 years ago, we saw the demand for yield output and cost improvements were underserved in the process equipment market. Gary encouraged us to think ahead about how we can fundamentally change AGS to serve this need. So next, let me tell you about the journey, the business model we created to meet the need, how the business is performing today and the new growth opportunities we see over the next 4 years. Central to our strategy in AGS is the drive to add more value beyond transactional parts and labor by focusing on better cost output and yield for our customers. Because we have the largest installed base of tools in the industry, whatever we do needs to be highly repeatable and scalable. So we created a whole new class of digital services to accelerate PPACt at every phase of the S-curve. In the beginning, it took us months to craft a service agreement for specific customers with mixed results, but we were learning. Over the past 5 years, however, we are using our technology and digital-enabled services framework. And we have scaled to over 14,000 tools that are covered by various levels of subscription agreements. Our retention rate is over 90% for these subscriptions. I think this really demonstrates that we're delivering great results for our customers, and they see these advanced services as highly valuable. In addition, Applied is capturing more value from our installed base and this can be seen in our financial results. Since 2013, when we started this transformation, we have more than doubled AGS revenue to over $4.2 billion. That's a compound growth rate of 11%. One question our investors sometimes ask is how much of the revenue is recurring and how much is equipment or WFE? The reality is that companies report differently. At Applied, we treat all 300-millimeter system upgrades and refurbs as WFE. That revenue is reported in our systems segment. AGS still manages 200-millimeter equipment upgrades and refurbs because these are legacy systems that are better served by the skills in our team. That represents only around 10% of the total AGS revenue. So our business is 90% services. In other words, 90% is truly recurring revenue. When we started the subscription-like service agreements, we're only a small fraction of that 90%. But over time, we passed 40%, and in 2018, we hit 50%, and in 2020, we reached over 60%. In fact, we've been growing our subscription revenue at 1.5x the rate of AGS overall, 2x the rate of our installed base growth and over 3x faster than our transactional business. Over the past few years, we've put a huge emphasis on digital services so that we can scale our business effectively and deliver more value to customers. We now have over 4,000 systems connected to our Applied AIx servers with the server attach rate growing by 35% in 2020 alone. To validate these numbers, we look at third-party analysis. In the 2020 report published by VLSI, just very recently, Applied was #1 in service recurring revenue, which is the highest amongst our peers. We've built a comprehensive cohesive strategy for subscription services that repeatedly delivers better customer outcomes and accelerates PPACt at every phase of the S-curve. This starts by understanding the major challenges of closing the gap between the green and the orange lines. Today, we have a strategy that is increasingly built around digital tools. We use hundreds of sensors and onboard metrology that generate Big Data. Then we use analytics and AI to turn petabytes of data into actionable insight for our service teams and our customers. This technology is reducing the time it takes to diagnose a problem from days to hours and minutes to seconds in real time. Here, we're showing a high-level framework that includes our fleet of systems with their sensors and metrology. It includes our AIx server. As you know, AIx stands for actionable insight accelerator, the firewall to ensure secure access, the applications and analytics, our customers and our experts and our data scientists have at their disposal, and the growing database we're building that guides our teams to best possible outcomes. All of the analysis is customer proprietary and remains that way. What I'll do next is walk you through specific examples of how we add value in each phase of the S-curve. I'm going to start on the right-hand side with how we optimize output yield and cost in high-volume manufacturing. Before I get into the example, I would like you to hear directly from one of our customers. Joel Hartmann is an Executive Vice President at STMicroelectronics. Joel, over to you.

Thank you, Ali. As you know, the characteristic of ST is the need to react very quickly to the market demands, to our customer expectations in terms of new products and in terms of new technology availabilities. We have adopted this strategy a few years ago in [indiscernible] of our 300 million to fab in France to grow by sequential phases in order to build additional expansion very quickly, ordering equipment and installing equipment very, very fast and also go very fast on startup, hook up, qualification and reaching to highest level of production as soon as possible. Of course, we are working with Applied Materials since many years. But a few years ago, in this context, we have expanded our partnership with Applied Materials for their ability to bring full support to our ramp-up and to our technology development needs, for example, forge and development programs. This goes through qualified resources, faster escalation, support of the factories, speed of logistics and part supply capability and also for the technology partners, strong support on technology and our differentiation technology program. We also work with Applied Materials on advanced data analytics to improve the performance and the stability of the tools and to improve the fab output. This goes for high level of equipment data being collected, both using ST, IT systems and Applied Materials specific analytic systems. And we are beginning to use some of these solutions, for example, exploring predictive maintenance improvement. The data collection systems gives a Applied Materials ability to analyze rapidly offline and to bring solutions to some of the corrective maintenance that we have to deal with. So in a nutshell, I would say that through all these programs, we are enjoying a lot this collaboration, and we see great promise on working with Applied Materials in particular in the future on this advanced analytics systems. And we thank them for that.

Thank you, Joel. We greatly appreciate your partnership and look forward to that promising future. It's been our experience that we can remarkably improve yield, output and cost on any fleet of tools we've touched. Here are 3 examples of the type of improvements we've done at leading-edge factories. First example shows how we reduced defects and tightened the range of outcomes at a leading foundry customer to meaningfully increase yield. At this leading DRAM manufacturer, we helped boost their output by 7% across a large fleet of tools. And at one of our NAND customers, we increased output by 4% and widened process window by 5%, which increases yield. We then shifted left, adding services that can dramatically accelerate transfer, matching and ramp. To give you a sense of what this means in the real world, I'd like you to hear from another of our customers, Buddy Nicoson, who is Senior Vice President of Global front-end Operations at Micron. Buddy is responsible for all Micron DRAM and NAND fabs globally. Buddy, the floor is yours.

Thank you, Ali. We've been delighted with the support from Applied Materials in our multiple fab ramps and max out initiatives. The dedicated Applied Materials team with cross fab matching services has enabled us to accelerate transfer, ramp and output at our high-volume manufacturing sites. Given the complexity and speed of our product development Applied's long-term partnership and services continue to be important for the success of our fabs. Applied Materials resources and supply chain insurance performance has been strong throughout the COVID-19 crisis. Thank you.

Thank you, Buddy, for letting us partner with you to improve the value of our services. Here's an illustration of an actual case study. Our customer dialed in their recipe on a development tool. Our job was to replicate the results of this golden tool to many tools as they transitioned to high-volume manufacturing. In this case, we were implementing a CVD hard mask used in patterning applications where the film uniformity needs to be kept in a narrow range. We were able to improve process margin by 61%. And we also worked on output, increasing that by 36%. If you look at the data from our most recent 50 engagements, our results show that we can typically get the desired output 2x faster and deliver 30% improvement in process margins. The next step, and this is where we're increasingly focused today, is to deploy solutions and methods to accelerate R&D. Our model is to work with both business unit partners inside applied and directly with our customers to accelerate their R&D programs. This helps Applied win more tool sales, and by proving our capabilities in early phases, we're winning more long-term subscription agreements. It is a virtuous cycle that begins with customer satisfaction, which then leads to higher equipment sales and then to higher subscription revenues and renewal rates. Here are a couple of real examples. In the first one, we used our digital process recipe optimizer in an integrated material solutions engagement for a memory customer. We delivered 50% better uniformity, and we did it 75% faster than our customers' previous best. On the right side, you can see how we engineered an advanced coding for one of our foundry customers, enabling them to develop a process with 4x reduction in particles, which translated to a 5% increase in output and higher yield. In fact, when we look at our most 80 recent R&D engagements, we've increased the rate at which customers sign up for subscription agreements by 4 times. To my mind, we're still in early innings, both in our transfer and ramp services and our R&D services. But judging from the value we're delivering and the business outcomes we're enabling, I fully expect these to become another strong vector of growth for us. Let me quickly summarize by bringing all this together and showing you how this adds up to profitable growth for AGS. Starting with revenue, when we compare average revenue in 2017 through 2020 to the earlier period, we've grown our revenue by $1.4 billion per year. We've also grown our operating margin by 300 basis points. Looking ahead, we plan to grow our revenue by $1.7 billion per year and increase operating margin by several hundred basis points. Our goal is to increase subscription to 70% of our services and parts revenue and to increase revenue per system by another 20% across the entire fleet of tools. I appreciate the opportunity to share the AGS story with you, and I'll be back a bit later for a quick update on our display business. Dan, I'll turn it back to you.

Thanks, Ali. To summarize this section of the presentation, we're using technology to give customers more than better products. We're giving them better outcomes. We're improving our solutions using technology, and we're delivering them on a subscription basis, which gives our customers continuous improvement and more predictable cost structure. The approach also gives Applied a more predictable and steady revenue stream. According to the latest report from VLSI Research, Applied Materials is #1 in both systems and services revenue. We're investing to grow our services business in more ways with innovations like integrated materials solutions and actionable insight accelerators that are even more valuable. These programs are in the early stages of growth today, but they're an increasing focus for Applied and our customers and a key part of our long-term strategy. Finally, opening up the strategy lens to see and address more of the industry's high-value problems is a great way for us to grow the addressable market for everything we do. Now we'll transition to the next section of the meeting, how we're using our materials engineering capabilities we've developed in the semiconductor industry into synergistic markets. We do this in large end markets where we can solve high-value problems and build sustainable differentiation with our technology. This increases our served addressable markets and gives us additional sources of profit and cash flow. Today, the best example of this strategy is our display business. Now I'll invite Ali Salehpour back to discuss how we built a synergistic business in display, how we've made investments to solidify our technology leadership in areas that are now poised for growth and how we're positioning the business to deliver higher through-cycle cash flow in the future.

Thanks, Dan. Since not everyone knows the story, I'll quickly describe how Applied ported our semiconductor technologies from silicon wafers to glass substrates, creating a large adjacent business for us in display. We contributed our expertise in both materials engineering and high-volume manufacturing, helping our customers to create the LCD display industry and grew the market to hundreds of millions of units per year. Our first products were CVD systems, which are critical to building the transistor that controls the pixels. With our expertise in uniform deposition, we were able to help scale the glass substrates through multiple generations of size increases, each time driving down cost. We helped move the LCD technology from laptop PC market to desktop monitors and then TVs and smartphones. Along the way, we brought our PVD technology to the display market, giving us more capability to enable faster and lower power transistors. We also adapted our e-beam technology for the display market, allowing our customers to review and analyze defects as well as test transistors. We enabled this to be performed in line, which improves productivity, yield and cost. A second big wave of growth came when smartphone road map moved to new materials and structures to enable higher resolution, better image quality, sleeker form factor and longer battery life. These new requirements in mobile display market resemble the PPACt needs in the semiconductor market. And we've worked to help our customers bring these technologies to the market faster. First, we helped enable new kind of transistors called LTPS that improves display performance. The higher capital intensity of LTPS more than doubled our CVD and PVD opportunity on the back plane. Then we adapted our CVD technology with a product for OLED thin-film encapsulation. This again doubled our CVD opportunity in the smartphone market. These technology changes created higher annual spending in the display equipment market, which averaged $9 billion per year until 2015, and has grown to an average of $15 billion in the year since. It's a cyclical market, and we've been bouncing along the bottom of the cycle for over 2 years now, awaiting the next upturn. We've been very active in R&D during this period, strengthening our product lines for the upcycle. Today, we see a number of green sheets in the market. End market demand is strong, capacity is tight, especially for the large displays. That's been positive for panel prices and for our customers, giving them resources to increase their investments as they look to 2022 and beyond. The pandemic hurt consumer spending. But it accelerated trends like work from home, learn from home and online entertainment and shopping. As a result, we've seen healthy sales of PCs and tablets. We've also seen growth in the average size of TVs. Unit growth and size growth are 2 of the 3 key drivers of the market. The third is capital intensity, which is also increasing because of the number of new display technologies that are right on the horizon. The end result is that we see a growing number of new fabs being planned and each fab will need more of our systems. To maximize our performance in the upturn, we've strengthened our core products in CVD, PVD, yield management and OLED thin-film encapsulation. In addition, we've borrowed the concept of integrated materials solutions. So we're now using a combination of our display technologies to help our customers accelerate the development of faster transistors that use less power. Since 2017, in fact, we've released 15 new products. On the back plane, our new products enable metal oxide transistors for OLED TVs. We're also using our LTPS technology to bring the benefits of OLED displays to notebooks and tablets. And we have LTPO transistors, which are even faster, lower power, which is perfect for foldable OLED smartphones. On the front plane, we've enhanced our thin-film encapsulation technology to support OLED notebooks and OLED TVs. We've increased our market share with these products, which will be great for us in the recovery. As you may know, we've also been developing new technology beyond our core products. The status is we've completed development and our customers are evaluating whether to use them in the OLED market. This could give us upside revenue in the future beyond the horizon of our new target model. Now that we've strengthened our core products, we're set up for more attractive returns. We'll deliver higher revenue and higher through-cycle operating margin. As you can see, our strategy has been delivering incremental cash flow. Our operating margin dollars increased to an average of $200 million per year into 4 years ending in 2016. As smartphones adopted OLED and as LCD TVs move to Gen 10 panel sizes, we grew annual operating margins to an average of $450 million in the 4 years ending in 2020. In our target model, we expect the operating margin to grow to an average of $600 million into 4 years ending in 2024. With many of our product enhancements now completed, we see a path to higher cash flow in the years ahead. Our operating margins will increase in 2022 as we transition to our new target model. In 2023 and beyond, we're committed to operating margins of between 25% to 30% when revenue cycles between $1.6 billion and $2.7 billion. This will give us our best ever through-cycle operating margins and cash flow. And with that, Dan, it's back to you.

Thank you, Ali. And now we're transitioning to the last section of the presentation, how we'll take advantage of market inflections, technology inflections and business model inflections to generate strong financial returns and how we'll allocate the capital to fuel profitable growth and return cash to shareholders. The semiconductor industry will drive the majority of our growth over the next several years. And I'll take a few minutes to show you how our markets have changed and evolved and why we can look forward to a healthy market in the years ahead. Applied Materials grew up alongside the semiconductor industry. It's been a great driver of our business over a long period of time as semiconductor sales have increased. The semiconductor industry's first $200 billion a year was in 2000. It took the industry 40 years to get there. The journey from $200 billion to $400 billion took about 17 years, which is less than half the time. And we could reach the $600 billion milestone by 2024, again, in less than half the time. Prior to the year 2000, the semiconductor and equipment companies both enjoyed significant growth. But the next 13 years weren't as kind to the equipment industry. In 2000, equipment capital intensity peaked, and we had a no-growth cyclical industry until equipment intensity bottomed in 2013. Since then, we've returned to growth. Let me explain what happened between 2000 and 2013. Over that period, a number of efficiencies were fully absorbed. One, the productivity of moving from 200-millimeter to 300-millimeter wafers, which gave our customers 2.3x the number of chips per wafer; two, the greater automation and productivity of these same systems; three, customer consolidation in leading-edge technology when a handful of companies came to lead the industry in DRAM, NAND and foundry logic. This accelerated during the 2009 financial crisis when subscale players exited the business and put their factories up for sale. This allowed the remaining companies to buy used fabs instead of building new ones; and four, the foundry model enabled many chip companies to outsource their manufacturing and stop buying equipment. In fact, equipment intensity slowly declined from around 17% in 2000 to about 9% in 2013, and that was the bottom. By then, all of these efficiencies were fully absorbed and equipment intensity gradually increased. Specifically, we've experienced new demand drivers. Chip designs have become more complex, requiring more process steps. 2D scaling has slowed and the promise of 450-millimeter wafers has evaporated. More recently, we're seeing a trend towards localized production. It's a better setup for long-term growth. But many investors wrote off the industry between 2000 and 2013. Others used it as a trading vehicle. And today, many investors are still not ready to trust that this is a growth industry. Here's an observation I shared with you on our November earnings call. If you plot equipment spending on a rolling 2-year basis, starting with 2012, this is what you get. I like that growth trend. In fact, equipment spending has more than doubled over the past 8 years, from about $28 billion in 2013 to over $70 billion this year. That's a compound annual growth rate of over 12%. Not every year was a record, and not every year will be a record, but I like the consistency. It's been a time when investors could go along the sector and make money. In fact, if you invested in Applied Materials at the start of this period, your stock gains would have been over 1100%. That's over 2.5x the returns of the NASDAQ and over 5x the returns of the S&P 500. I believe the semiconductor market will keep growing at a healthy pace. Not every year will be a new record, but I believe there will be higher highs and higher lows, we will deliver growth. And I'll wager that the equipment spending in 2021 plus 2022 combined will be between $150 billion and $160 billion. That will be the ninth consecutive increase in the 2-year rolling average. And for everyone who's been trained to get a little nervous when they hear the word record, yes, for the past 5 years were records. And 2021 will make it 5 out of 6. But instead of overcapacity, we're seeing the most dramatic global shortages the semiconductor industry has ever seen. The shortages tell us 4 things: One, there are more demand drivers than ever. Gary showed us this in his presentation when he talked about 4 eras of computing. In fact, we've been anticipating this growth for years. This slide is from our 2017 Analyst Day from PCs to smartphones and tablets to even bigger, more permanent wave that we call the AI era, the biggest wave yet. Second, this new wave is permanent. It's not consumer discretionary. It's business critical. You adopt the technology or you fade away. Our company is just 1 example. We're embracing the technology. It's a catalyst for our strategy. Third, semiconductors are increasingly strategic. Every advanced nation in the world wants to have an assured supply, because you can't have an advanced economy without it. And fourth, semiconductors are cool again. We went through a long period where semiconductors were seen as a commodity and the lion's share of value accrued to software. I think the shortages have taught us an obvious lesson, Great software runs on great hardware. And with Moore's Law slowing, there is no more free lunch. And one last observation in this section.While our customers are investing more than ever, they're also more profitable than ever. They're making disciplined investments that they have to make to capture the growth they expect to see in the years ahead. In summary, will every year be a record? No. But is this a growth industry? Yes, absolutely. It's higher highs and higher lows. And with Moore's Law slowing and more of the road map being driven by materials engineering and a new playbook, is Applied in a special place? I strongly believe the answer is yes. Next, I'd like to talk about the financials. How our strategy is designed to deliver high ROI, cash flows and shareholder returns. I'll start by showing you a journey we've been on and then layer on our new target model. This will give you a sense of Applied's growth trajectory in the years since Gary was named CEO in 2013. If I had to describe Gary's philosophy in a couple of words, I'd say it's inflection focused innovation. The ability to see around corners and the conviction to make strong bets aimed at major inflections to create differentiated technology and outsized financial returns. What you see here is that Applied grew revenue by almost 100% between the time Gary joined as President in 2012, and 2020. We grew operating profit by more than 200% and EPS by more than 400%. Comparing 2012 to 2016 and 2020, you see the benefits of the product development engine with revenue momentum accelerating and increasing by 59% in the latter period. And we're talking about profitable growth. We've expanded gross margins by 230 basis points through 2016 and by 420 basis points through 2020. We increased operating margin by 590 basis points through 2016 and over 1,000 basis points through 2020. We think of R&D as an investment. It's fuel for profitable growth. And we've increased it from 55% of OpEx to nearly 70% today. This drove growth in share gains for Applied, and it gave us the most balanced share profile of the major companies. Here, you can see that our share used to be high in foundry logic and significantly lower in memory. Today, our share is relatively balanced across foundry logic, DRAM and NAND. We don't need to lose sleep at night when 1 spending category is higher than another. Sometimes investors get concerned when foundry logic spending is strong. They think it's unusual and that there will be a mean reversion and an increase in memory spending. Well, it really doesn't matter to us because of our balanced share profile. Here are the facts. On average, foundry logic spending has been over 55% of equipment spending for the past 10 years and for the past 20 years. In fact, there have only been 3 years in this entire period when memory was above 50%. In 2007, and also 2017 and 2018 during the transition from 2D to 3D NAND. So high exposure to foundry logic spending is valuable. Enabling the foundry logic road map and accelerating PPACt is even more valuable. Now I'd like to summarize the momentum in our reporting segments, which is the foundation of our new financial model. Prabhu and Ali showed you the individual trends earlier in the presentation. What you notice when you put things side-by-side is that each segment has been generating revenue growth over the past 8 years. And each has excellent opportunities and well-funded strategies to keep driving revenue momentum into 2024, which is the year of our new target model. Comparing 2012 to 2020, you also see profitable growth with operating margin increases of 670 basis points in semi systems, 310 basis points in AGS and and 290 basis points in display. Now it's time to reveal how this translates to the new growth we expect to deliver through 2024 from the revenue line through to margins and to earnings per share, and especially to free cash flow and shareholder returns. This is our new target financial model. You'll notice 3 scenarios: our base case and also our low and high-case scenarios. You may be wondering about the WFE assumptions. In our base case, calendar 2024 WFE is around $85 billion. Our high case is around $100 billion, and our low case is around $75 billion. In our base case, we expect to grow the company by more than 55% over the next 4 years. We expect our highest growth in Semi Systems at over 60%. Next, AGS at over 45% and then display at over 35%. We plan to increase gross margin by 340 basis points and operating margin by 610 basis points. This should produce earnings of $8.50 per share at a tax rate of 12% and around 875 million shares. And our EPS bridge, you'll see that the revenue plan generates around 55% of our earnings growth, with margin expansion driving over 35% of the EPS growth and buybacks, the remaining 9%. We expect to grow revenue by about $9.5 billion with $7 billion coming from Semi Systems, $1.9 billion from AGS and $600 million from display. I'll double-click on Semi Systems, which drives over 70% of the growth plan. We dedicated a lot of today's meeting to demonstrate the value of PPACt enablement to our customers. It's going to be a powerful secular growth driver for Applied for many years in the future. PPACt enablement represents over 55% of our Semi Systems' revenue in 2020. And we expect it to grow more than 80% through 2024, becoming around 2/3 of our mix. We're excited about the unique opportunity we have to combine technologies to solve higher value problems worth billions of dollars to our customers. We feel confident about the momentum we're having with Integrated Materials Solutions and actionable insight accelerators to unique and proprietary capabilities that we're constantly improving on to accelerate PPACt for customers. These exclusive technologies are exactly what the industry needs to overcome roadmap challenges and deliver better logic chips and memories. These innovations give us confidence that we can continue to earn healthier margins over the next several years. We've also been working on important cost components of our margin strategy. We've deployed new systems and processes and our teams are operating with discipline, generating margin improvements even while we're experiencing cost headwinds from the pandemic. But R&D is our best fuel for generating profitable growth. And that's why we'll continue to focus around 70% of OpEx on research and development. Now let's talk about cash flow. We saw momentum through 2020, but our new model puts us on a higher trajectory. The revenue momentum and margin improvements are setting us up for profitable growth. And we expect to generate around $6.5 billion of cash in 2024, which would be up more than 90% compared to 2020. And that brings us to our capital allocation philosophy. Our first priority is to keep making disciplined investments to grow and strengthen the business. At the same time, we want to preserve a strong and flexible balance sheet. Beyond that, our goal is to return excess cash to shareholders using a mix of buybacks and dividends. Looking at our track record, you see ratable dividend growth, especially in recent periods. We've increased the payout in each of the past 4 years. Last month, we announced a 9% increase, but most of our capital returns have been through stock buybacks. We anticipated that our opportunities were improving, and we took advantage of that by repurchasing a large portion of shares outstanding. We still believe buybacks are the more attractive way to return cash. The board recently announced a new $7.5 billion buyback authorization, which supplements the $1.3 billion we had remaining on the prior authorization. And today, we're announcing something new. We're now committing to return 80% to 100% of free cash flow. You've heard a lot of numbers today. So let me recap the financial section before inviting Gary to summarize the meeting and begin the Q&A. By now, I hope you share my excitement for the opportunity we see as our end markets accelerate and need more of the technology we uniquely provide. I hope you see why we're making strong targeted investments to fuel growth and outperform in our markets will push ourselves to improve execution, to expand margins, grow earnings and generate cash. And we'll continue to be disciplined in how we allocate capital while following up on our firm commitment to return cash to shareholders. Thank you for listening today. And now let me welcome Gary back to the meeting.

Thanks, Dan. Before we move to the Q&A section of the meeting, I'll quickly summarize the key takeaway messages. First and foremost, Applied's markets are more attractive with more opportunities than ever. Digital transformation of the economy and the AI inflection is driving secular market growth and a new era of innovation. What comes next is not going to be like the past, and the new PPACt playbook that will shape the industry's future will be enabled by advances in materials engineering. We have aligned our strategy and investment to make applied the PPACt enablement company. We have the broadest and most enabling portfolio of process technologies, and we're able to combine these technologies in unique and highly enabling ways that no one else can. In addition, as you heard from Prabhu, Raman and Ali, we've developed a proprietary methodology and suite of solutions to provide actionable insight acceleration and significantly improve the time to market of new PPACt innovation. On top of this, we're shifting more of our business to subscription-style revenues. Over the past 5 years, we've moved a significant portion of our installed-base business to recurring revenues, and we're starting to monetize new products and services using subscription-style approaches. In addition, we continue to optimize our portfolio to drive profitable growth and higher free cash flows. This all adds up to a high ROI operating model that generates attractive shareholder returns. We expect to grow our earnings per share at 1.7 to 2x our top line growth and as you've just heard from Dan, we're also making a commitment to return 80% to 100% of our free cash flow to shareholders. Now let's move on to the Q&A section of the meeting.

Thanks, Gary. [Operator Instructions] Operator, let's please begin.

[Operator Instructions] Our first question comes from the line of John Pitzer from Crédit Suisse.

I guess, Gary, I want to talk a little bit about the industry backdrop is clearly very strong and I think supportive of good growth in WFE over time. But I want to talk more specifically about your strategy around integrated solutions is sort of best-of-breed point solutions. Applied has always had sort of a broader swath of IP than any other equipment player. And you've always been striving towards these integrated solutions. Yet, if you look out over the last sort of decade, your share of WFE really has been pretty constant, hasn't gone up all that much at around kind of high teens to 20%. And yet the industry -- the semi industry has gone through a bunch of inflections over the last decade. And so the industry has seemed to have chosen kind of best-of-breed point solutions over integration. And so I want to get a better sense of why you think there's a tipping point to more integrated solutions. And if you're confident, why not have a more sort of aggressive stance on your ability to gain WFE share? I think at the midpoint, of Dan's model, it only implies about 1.5% of WFE share gain between now and 2024.

Yes. Thanks, John, for the question. So if you look at what is enabling power, performance, area, cost and we talked about time, how fast do you move forward with the technology road map, you have the whole portfolio that you talked about of all of our different technologies to create, shape, modify, analyze different types of structures. But it's really the combination of all of those things that I think right now, matters more than ever. For many, many years or decades, you had Moore's Law driving the industry forward with 2D shrinking. That clearly has ended. And you can see from many companies in the ecosystem, talking about a new playbook, TSMC in September and also recently talked about the 5 elements of the playbook going forward, new architectures, new structures, new materials, new ways to shrink and new ways to connect chips together. So I definitely don't believe what enabled the past will enable the future. And this combination of all of these technologies in the broad portfolio and connecting those together to enable those new materials, those new structures, new ways to connect chips together, has never been more valuable than what it is today. We gave an example earlier in this meeting of an interconnect solution that improves the contact resistance by 50%. That's an enormously important innovation from Applied. And if you look at that platform, we're combining ALD, CVD, PVD technologies to managed all of those interfaces, the interfaced engineering, into a single platform to enable the power and performance for the future. So that's just one example. We've given other examples around transistor innovations, where we can improve leakage current by orders of magnitude. So John, I think that it's very clear, 2D scaling is not going to enable the future and this combination of technologies and integrated solutions where you're managing a few angstroms of films as you're building these different structures and shaping them and this interconnect example we gave today is really the best example of that in this particular meeting, that absolutely is the future, and no company is better positioned than Applied with those combinations of technologies. And our ability to bring all that together under vacuum and an integrated platform, we see tremendous pull. You can also hear that from all the customers that were part of the event today, the pull for that, John, has never been stronger. And the playbook going forward, I believe, we have it right with those 5 elements.

And then to address the second part of your question, John, as we think about where we set expectations for the purpose of this meeting, let's break it up into 2 buckets. There's momentum that we have in the current environment, and then there's where we set those expectations for the purpose of an event like today. So from a momentum standpoint, all of the innovation that Gary has been talking about, leading the strong outperformance in the market. We saw it in 2019. We saw it again in 2020, where the company significantly outperformed the DRAM market, significantly outperformed the foundry logic market and showed strength in the NAND market. We would expect that market momentum and outperformance to continue again in 2021. And we don't see any reason why that strong momentum that's being delivered by these innovations and combinations, unique combinations of innovations that we're delivering to market to solve our customers' highest value problems. We don't see any reason why that momentum would slow down. When it comes to an event like this and where we set expectations, we think it's prudent to set those expectations in a way that give the company a strong probability of exceeding expectations. So it's balancing the near-term momentum and the confidence we have in the unique solutions we're bringing to market, the way they're solving our customers' problems and the strong momentum that, that's delivering with a prudent set of expectations that we feel very good about meeting.

Our next question comes from the line of Vivek Arya from Bank of America.

And thanks for a very informative Investor Day. It's kind of a longer-term question for Gary or Dan. I'm curious to get your insight on the potential risks of more cyclical swings and stranded capacity when more fabs or more foundries come up globally. I know it's not a near-term issue, but over the next 3 to 5 years as more foundries come up and every country and every region wants to be self-sufficient in semiconductors, is there any downside risk to that? And how are you making sure that you're managing your customer demand more carefully when that situation happens?

Yes. So from an overall market perspective, one of the great things we see in the current environment is a diversification of the end market drivers. We've seen that hand off -- in the early stages of that handoff of end market demand from consumer discretionary, to non-discretionary spend by the largest companies exposed to the largest industries. That aspect of demand diversification, I think, serves this industry well, and it's one of the key drivers that's going to take us from the environment we're in now to structurally larger over time, defined by higher highs and higher lows. As we think about this trend in the current environment to more localized supply chains, semiconductors is one example of it, we'll see others. What I'm encouraged by in the current environment is, I do think global supply in this industry is going to meet global demand. And so while the location of that supply may change, I don't think it's going to dramatically alter the long-term supply-demand balance in our industry. And so I do think capacity still is deployed in an efficient way. But I do think that regionalization of those factories means that the capacity is going to be deployed in smaller increments than what is currently done in the market today. That's going to lead to an upward bias, in my view, of capital intensity over time in our industry. But it's not going to dramatically alter that balance in a meaningful way, one way or the other. The other aspect of this trend that we see developing in our markets that we think is going to accrue benefits to us long term is, it's going to be a nice adder from a service standpoint. When we see our customers build capacity outside of their home geographies where they cannot leverage the ecosystem that they've built, it's led to higher service entitlement on that capacity deployment. And as the capacity gets deployed in different geographies over time, we think it's going to be a nice adder to our service business. We will be there alongside with our customers in those different geographies, doing everything we can to make them successful over time. So we see this trend in a very favorable way, and we're really encouraged by what we see developing in the market.

Our next question comes from the line of Stacy Rasgon from Bernstein Research.

I was happy to -- I'm glad to see you gave information on your mature node business. The visibility of the ICAPS is very helpful that's up $3 billion today. How do you see, I guess, the growth trajectories of leading-edge versus the mature node as we go forward and build towards your 2024 model? Can you give us any thoughts on growth rates or adoption of that mature node business versus the leading-edge? Do you think one grows faster than the other? And are the economics for you any different whether you're selling mature node versus more leading-edge tools?

Yes. Thanks, Stacy. So let me take a step back and talk about the contours of how we see the overall market-shaping up. And then come to the economic side of the equation. So we talked about a compound growth rate of the industry at about 8.5% over the forecast period. We also talked about the rough shape of the market being greater than 55% foundry logic and less than 45% memory over a very long period of time in our industry. We don't see that general construct changing. We think the memory market and the foundry market grow roughly in line with each other and maintain that relationship of greater than 55% foundry logic, less than 45% memory in this forecast horizon. Then when we break down the memory market to DRAM and NAND, I would say both markets will grow, and DRAM probably outgrows the NAND market by a bit. Within foundry logic, if we break it down into leading-edge and trailing-node geometries, we would expect both of those markets to roughly grow in line with each other. And so as you think about leading edge today being about 2/3 to 70% of the spend in our industry and trailing-node geometries being roughly 30% to 1/3 of the market, we see that relationship generally holding in this forecast horizon. So I would expect both of those markets to roughly grow in line with each other. Against that backdrop, I would expect our business based on the momentum we've been talking about, the key problems, high-value problems that we're solving for our customers and the unique combinations of technology, we would expect the outperformance against that opportunity to continue. And we feel really good about how well we're positioned against those markets. From an economic standpoint, I would say the trailing-node geometries are accretive. The margin -- operating margins are accretive to the overall company operating margin. And so as that market continues to grow, and we deliver that differentiating technology to our customers to drive their road maps, we would expect it to be value creative for investors. So we feel good about how we're positioned against these specific market opportunities.

Great. Stacy, I would add one more thing on this topic. Relative to the leading edge, of course, I personally and our company spends a lot of time driving power, performance, area, cost for all of those leading customers. But we also, a couple of years ago, formed an ICAPS organization, IoT, Communication, Auto, Power, Sensors. And the response from our customers has been really tremendous. As you said in your question, last year was $3 billion relative to the overall size of that business. And as Dan talked about, it is accretive from a company perspective, but the really more important thing for me by bringing together all of these technologies and all of the talent across Applied Materials, we have a really great focus, enabling the road map for all of those ICAPS customers also. So as Dan said, both of those markets are going to grow. If you think about a 0.5 trillion or 1 trillion connected devices in the future from the edge to the cloud, just a really tremendous opportunity. And to capitalize on that, we changed our structure, and we have a strategy to get closer in enabling that customer base.

Our next question comes from the line of C.J. Muse from Evercore.

I guess to focus back on the market share of WFE, it looks like you're implying at the midpoint roughly $1.5 billion of incremental share gains. And I guess, if you take a step back and you look at that number and you think about your growth areas where you're really starting to accelerate, you think about cooptimization, IMS that you discussed today and then some of the emerging growth areas within ICAP and advanced packaging, I guess curious how much of that $1.5 billion would be incremental kind of spend by the industry as opposed to just pure share gains? And then if you could kind of rank order how you're thinking about which parts of the business should really shine in the coming 4 years would be very helpful? And then just one last kind of follow-up. On the implied EPS, it looks like you're talking about buying back around $10 billion of stock, whereas you said you're going to do about $6 billion free cash flow a year. So if you could talk about anticipated kind of share count declines and what you put into the model would be great.

So let me take a stab at this, C.J., and then Gary, see if you want to add anything. So from a product momentum standpoint and a share momentum standpoint, one of the great things about where we're at today and the momentum we've showed in 2019, the momentum we showed in 2020, significant outperformance, and you'll see follow through on that again in 2021. What I like about it is, is it's very broad-based. If I look customer-by-customer, and I look at our node-over-node opportunity expanding with each successive node and whether that's memory or foundry logic, the company's opportunity is increasing. And then if I break it down to the product portfolio, and it's almost top to bottom, in our product portfolio, you see strength across the board. So whether it's customer, whether it's end markets, or whether it's product portfolio, the company has got a broad base of strength. And then when I look at the unique combinations of the technology that we bring together and our ability to solve what we know are the highest value problems in our customers' road map. It's hard to take one piece of the portfolio and say that $1.5 billion of incremental revenue comes from any one piece, it really is broad-based strength again, whether it's customers, end markets or product portfolio. If I were to rank order the end markets, I would probably put strength in DRAM at the top followed by foundry logic, followed by NAND. But it's a real tight, closed grouping at the top of that list. It's almost a force ranking, but you're going to see significant strength across the board, particularly in DRAM and foundry logic. So we feel really good about how well we're positioned against this opportunity. Then from a stock buyback standpoint, one of the great things is, the company is performing well. We've got a lot of momentum. We're generating a lot of cash, and we're going to deploy that cash for the benefit of shareholders. We've committed to 80% to 100% of free cash flow coming back to investors. From a model standpoint, we've got an embedded share count assumption of 875 million shares. Embedded in that -- in 2024, 875 million shares. Embedded in that is assumption that 80% of the free cash flow comes back to investors. If we do more than 80%, then I would expect to do better against that share count. And then there's a stock price assumption that supports how much stock we can purchase by deploying that cash. We've got a pretty strong point of view of where we think our company is going from a stock price standpoint. And if we can buy our shares back at a lower price than it's currently modeled, then I would expect to do better than the share count that's embedded in our base case model. So again, companies performing well, generating a lot of cash top to bottom momentum, and we're going to give the cash back to shareholders over time. And we would hope to do better than the embedded assumptions in the model and where we're setting those expectations.

Our next question comes from the line of Toshiya Hari from Goldman Sachs.

Gary, Dan, given your recent experience with Kokusai and given the current geopolitical backdrop, is it fair to say that M&A is no longer a priority for Applied? And to the extent you continue to entertain potential deals, given your focus on delivering material-enabled solutions and driving higher levels of recurring revenue, would you ever consider making an acquisition in the materials space?

Yes, Toshiya, thanks for the question. We're always looking for opportunities that are synergistic with our strategy and financially attractive. I would say, certainly, given the geopolitical situation today, it's very, very difficult to see anything large getting done in this current environment.

Our next question comes from the line of Harlan Sur from JPMorgan.

And thank you for hosting this event. The AI framework is a great example of using AI-based methods, the speed of yield learning and the high-volume manufacturing environment. And another big potential opportunity that you guys highlighted today, which is using an AI-based framework for helping process engineers and process integration teams, more efficiently develop your -- more manufacturable processes. Looks like you're monetizing your AI(x) analytics platforms via your services business, and it looks like the digital services business grew about 4 to 5x faster than overall AGS in 2020. Is this how we should think about the growth of this key capability going forward? And what percentage of your services business today is software and advanced analytics capabilities?

Yes. Thanks for the question. So first, maybe let me take a step back and talk about the strategy for this actionable insight accelerator. As we've talked about in this event today, accelerating the road map for the entire electronic ecosystem, the PPACt focus for our company is -- we're in a very, very unique place as a company in terms of the capabilities, but it's also tremendously valuable. Winning the PPACt race is incredibly enabling across the entire electronics ecosystem. So then if you think about our services business or our subscription revenue, there is a part of our business that is break fix. And then there's a part of our business that is agreements. The agreements part of our business has been growing at a very high rate. If you go back to when I first came to Applied, our service agreement growth at that point was 0. We weren't adding any more than we were losing. And since that time, we had a big shift in strategy around service agreements and subscription-style revenue, where we grew to service agreements being more than 60% of our total services business. And then if you think about where we focused, we've certainly focused in high-volume manufacturing to optimize yield, output and cost for customers, and with digital services, as you've talked about, and we gave some examples today tremendous value in working with customers, over 90% retention rate around that subscription-style revenue. But we've also shifted focus. If you think about PPACt, the value of time, to moving faster and better in R&D, transfer and ramp beyond the high-volume manufacturing. If I can dial in a unit process or an integrated solution, 2x faster with a 30% bigger process margin, that's worth an enormous amount of money. I gave an example earlier today of what it's worth to accelerate time to market by a year in the cloud part of the industry. And again, this is the biggest focus for the entire industry, time to market on power, performance and a cost advantage. So accelerating R&D, accelerating transfer, accelerating ramp is enormously important for Applied Materials, accelerating R&D is important and for our customers and for the ecosystem. And what we've done within Applied is bring together many unique capabilities, unique sensors, unique in situ and integrated metrology. And we showed that earlier on this -- in this interconnect example that we gave. Unique offline metrology capability with our leadership in e-beam, where we can look at orders of magnitude, more data across the chip to look at micro loading, across the wafer signatures that we can then tune all of the parameters that we have on our systems to accelerate R&D and increase the process margins. So that's really the focus. And it's really a tremendous drive for me personally and for our entire company to accelerate our digital services. We've had a lot of success there. But I really believe, if you think about the value of time to market, increasing those process margins, it is really tremendous. And we have a number of unique capabilities within Applied. We're bringing together in this AI(x) platform certainly, when I'm talking to CEOs or R&D leaders on a regular basis, just really tremendous pull for these capabilities. We have a number of examples where we've been successful but we're really in the early innings of that opportunity as a company.

Our next question comes from the line of Krish Sankar from Cowen.

And thanks for the very informative session. The question on the clarification I have is for Dan. I understand that there might be some conservatives under your long-term forecast. But if we're going to have over 55% foundry logic in the WFE mix, and AMAT gets more process control traction, more digital services and subscription revenues, why can't the gross margin and off margin and therefore, earnings be higher in FY '24? Or are you encumbered in some -- in those businesses? And then just a clarification, Ali mentioned on the services side, you're moving to an outcome-based subscription. Does this mean you're going to charge customers on a per wafer basis?

Sure. Thanks, Krish. So as we think about the company's progress from a margin structure standpoint, since Gary took over the company, in the prepared comments, you've seen the steady progress we, as a company, have made with respect to gross margin and operating margin. Since Gary took over, I think gross margin is up over 400 basis points, and operating margin is up over 1,000 basis points. It's led to strong profitable growth and value creation for the company. No matter where our margin structure is, we'll never be satisfied. There's a number of initiatives underway at the company to drive these higher. And we're going to continue to work very, very hard to capture more and more opportunity over time. If there's an opportunity in this forecast horizon to do even better than the target margin of 48.5% from a gross margin standpoint, we'll obviously take advantage of that and do better than the model. So the company is confident about our momentum and the progress we're making with respect to margins, but we want to balance that with, again, being prudent about where we set targets and the achievability of those targets and the probability of exceeding the targets. And so we feel this is -- just strikes the right balance between where the company's momentum is and setting the target in the right way. From a services standpoint and monetizing through subscription-like services, we've made a lot of progress with respect to long-term service agreements, taking it from 40% of revenue to 50% of revenue to now over 60% of revenue, and we're going to drive that to over 70% of revenue in this forecast horizon. Other aspects of the business model transformation, we're in the early stages of taking the full suite of capabilities we, as a company, have, leveraging it on our company -- our customers' high-value problems to drive better outcomes for them. And we will look to monetize those technologies and investments in different ways going forward. Beyond the forecast horizon, we do think there's an opportunity to see more and more traction in these areas that are independent and disconnected from purely WFE from an end market standpoint. So we think beyond the forecast horizon, there's an opportunity to create a leg of growth, that's independent from the core traditional model. And I would say that we would have a very open mind on whether it's a per wafer basis or other type of model, just evolving from what is today a very traditional monetization of innovation and technology through a hardware sale. That will always be a part of the company, but we're going to look to complement that business model with other aspects and innovative ways of monetizing technology that reflects the outcomes that we're going to drive for customers. And so we feel good about the progress. And as you can imagine, this stuff takes time to develop. We would expect to see not only early traction and momentum from a company standpoint, but more momentum beyond the forecast horizon that creates that vector of growth independent from WFE.

Our next question comes from the line of Patrick Ho from Stifel.

Thank you very much for the event and the insights you provided. Maybe a question for you, Gary. You've been one of the biggest proponents of R&D spending to foster innovation and delivering next-generation products and solutions. Your new AI(x) platform is another example of this part of your strategic vision. I see how it plays out to your PPACt strategy and how you can help customers in accelerating time to market in their respective product development cycles. How do you take that AI capability and a lot of the analytics and data insights to the high volume manufacturing? And once you get -- once you help the customers get their products out to the marketplace.

Thanks, Patrick. So let me talk about the sequence of how things happen with our customers. First, we're engaged with them in R&D. All of this broad portfolio of products, our etch products, CVD, PVD, epi, implant, everything, all of our entire portfolio. And as we're moving forward, complexity is increasing. We have those unit processes. We have those integrated systems like you saw earlier today, in the interconnect example where you have CVD, PVD, ALD, all on a single platform. So dialing in the optimum recipe and settings for that tool or integrated solution is incredibly complex. I mean, the image I have in my head is a 30-year PhD engineer with his hair all over the place and thick glasses, by the way, I used to be that person many, many years ago, versus a data scientist. And so you have this -- and a lot of this concept around AI being actionable insight. So you have the actionable insight data generators in R&D. And you're bringing that together inside our chambers, inside our platforms, with unique e-beam metrology, imaging and algorithms to dial in this multi-dimension optimization that's incredibly complex. My goal is actually 10x faster inside Applied and with our customers. So then -- and the other -- another part of that is making that process margin as wide as possible. That is incredibly valuable from a yield perspective for our customers. So you drive that, make a one chamber golden in R&D, then Patrick, you go to transfer and ramp that new device technology in high-volume manufacturing, you have a factory that's maybe $18 billion. And cutting that time is worth an enormous amount of money and then making all those tools and chambers and integrated solutions, golden, so that they're all matched, that is incredibly valuable. And then you'll get into the high-volume manufacturing. So the way that works from our perspective, it accelerates Applied success with all of the new integrated solution, all the new products that we're bringing to market, but it also is very sticky when you move into high-volume manufacturing. If you have an advantage in time to market with all of those unit processes and all of those integrated solutions and you can move faster and better process margins across a single tool and then entire fleet of systems in high-volume manufacturing you have that infrastructure there to continue to optimize yield, output and cost for customers and high-volume manufacturing. And we gave some examples of some of those digital services earlier today. So that capability, from my perspective, is enormously important. And I strongly believe we're really in the early innings of using this AI, Actionable Insight technology inside Applied and with our customers to accelerate the time to market, also better results for our customers and from a digital services standpoint, really in the early innings of that opportunity.

Our next question comes from the line of Tim Arcuri from UBS Securities.

Dan, there were some numbers that Ali gave. And I guess if you strip out the 200-millimeter refurb and you divide by the amount that customers are spending to basically maintain your tools, the capture rate sounds like it's about 25% to 30%, which is quite good. But I guess the question is sort of how to think about where that number can go? And like what are the big pieces of the 70% to 75% that you're not getting that you can get? And sort of how to think about what the ceiling might be on that capture rate?

Yes. Thanks, Tim. Maybe let me try to break it apart into a couple of pieces to give you a bit of context and perspective of how we think about the business. So I think there's a continuum between the older generation legacy installed base that's been out in the market for several decades, and then the new systems we're shipping to today. Because I would say the service entitlement from a penetration standpoint is going to be a function of that embedded complexity. If you take the legacy installed base, I think there's going to be elements of that spend that are probably not going to be a realistically addressable bias. It's just a bit less differentiated. If I window forward to the systems we're shipping today, that's more current generations of technology, I'd like to think 100% of that spend over time can be penetrated. Now it's not just going to fall on our lap. We've got to go earn that business from our customers. And at the core of earning that business or a lot of the digital services techniques that we're talking about today, it's creating a set of capabilities that can be permeated across our entire tool suite. It's collection of sensors and capabilities and algorithms that get to better outcomes for our customers and speed time to market, enhance process windows, get better throughput and performance, get better on wafer performance, better yields for our customers. All of these outcomes create better ROI for customers. And so I think we're putting the building blocks in place, and we're in the very early innings of it today. We're putting the building blocks in place to be able to target the vast majority of the spend on those systems that are being shipped in the market today. So today, I think there's probably an entitlement over time. I would expect that to encompass the vast majority of all of that spend over time. And then we'll just have to work hard to demonstrate the value, earn that right to capture that spend from our customers and grow the business at an outsized rate that's better than the growth of our installed base.

Yes. Tim, one other thing I would add, when you think about our opportunity, there's certainly maintenance that's done in high-volume manufacturing. And we shared those numbers earlier today. But when you think about the value of accelerating time to market for devices, that's enormous value. If you can move the device time to market ahead 6 months or 1 year, that is enormously valuable. So I would not just focus on the high-volume manufacturing. Certainly, we've sized those numbers. But if you think about the value of accelerating PPACt, in R&D, in transfer and ramp, I just think that the TAM is much bigger than just the parts and service and high-volume manufacturing.

And our next question comes from the line of Joe Quatrochi from Wells Fargo.

Maybe one for Dan. When I look at the target 2024 model, I was wondering if you could kind of talk about how you're embedding -- embed incremental EBIT margin in terms of -- as we go to -- maybe from the base case of, you call it $85 billion WFE, I think you said to $100 billion in the high case, what are you embedding in terms of your just incremental margin? Just some more color there would be helpful.

Sure. Thanks, Joe. The way we really look at the margin profile let's break it out into gross margin. Let's come back to operating margin and give you a bit of insight on how we think about it. So when we think about going to a higher overall market size, and we think about the momentum that we see across the product portfolio, there's clearly an end market mix, and there's clearly a product mix, but also embedded in that are all of the initiatives that the company is driving internally to drive efficiency into the core of our operation to get more flexibility into our operating footprint to hone the warehouse and supply chain infrastructure and to think about continued progress and how we design our products and how we bring them to market in a way that the initial set point of gross margin is higher than it was previously. And then also adoption of digital techniques, not from a servicing of a customer standpoint, but from an optimization internally as a company. All of these initiatives not only are linked to size of the end market, but they're also linked to the time dimension. It just takes time to continue to drive these. And so this is the progress we think we can make within a realistic time horizon. And again, setting targets in a way that's prudent that allows us to exceed expectations is an important aspect of the event today. So we feel good about the gross margin progression, not only over a longer period of time but within this window. Then when we think about operating margin, at the core of it is OpEx and specifically R&D. This is a company that spends about 70% of our OpEx on R&D. It's really the fuel for the company's growth. We keep a tight lid on the discretionary spend and it's a tough thing to get more in corporate support functions to fund the operations. Vast majority of our spend is going to go to R&D. And Gary's philosophy, and he's had this philosophy for his entire career, spanning 3 companies and over 3 decades in the industry, it's really what are the opportunities to see the inflections coming in the market and invest in innovation that allows us to intercept those inflections or catalyze those inflections so that we can grow in a strong, profitable way. That formula has served this company extremely well since Gary took over, and it's nothing more than looking at those opportunities, opportunity by opportunity, seeing what makes sense for us to invest in, draw a disciplined cut line that balances near-term discipline with long-term value creation and set the right model for the company. It's not a ratio of spend. It's not an incremental fall through methodology. It's really opportunity by opportunity on where we think we can innovate, differentiate, solve difficult problems and drive value for shareholders, and where we see an opportunity to serve our customers, make them successful and drive value for shareholders, that's what we're going to invest in. It's really more of a bottoms-up philosophy on what we see happening in the market and where we want to strike the balance between the near-term discipline and the long-term value creation.

Our next question comes from the line of Joseph Moore from Morgan Stanley.

I wanted to ask about how you're thinking in this multiyear forecasting about the regulatory environment in China? Do you see the Export Controls Act of 2018 resulting in any restrictions in your ability to ship to Chinese customers? And just how are you thinking about that? And then on the same topic, do you think Applied is sort of well represented when it comes to government lobbying and sort of having influence over the direction of some of those policies?

So here's what I would say. It's hard to speculate about the future. You can always run scenarios and be prepared from a flexibility standpoint. But all we can do is deal with the situation as it currently exists. This is a company that operates around the globe. We've got great relationships with the governments and officials in each of the geographies we operate in. As you can imagine, most industry leaders do, and we're certainly no different. We've got a seat at the table. We've got a point of view that free, fair trade is the basis for strong economic development. Everybody wins when -- that's what defines global industries and economic flows. And so we certainly do have that seat at the table and express our point of view. And then with respect to the current environment, there's places where we need licenses. And as you would imagine, being fully compliant with the laws in all the geographies we operate in, we'll -- we intend to get those licenses. Those conversations are ongoing. And so based on the current set of circumstances that we see defining the industry, this is our best view of how the company will progress against a variety of market opportunities. $85 billion WFE, upside case of $100 billion WFE this is our best assessment of how we think the company will perform in that environment. We'll continue to stay close to the dialogue. We'll use that seat at the table we have to keep our finger on the pulse of where things are trending one way or the other. And then stay as agile as possible to respond to whatever environment we happen to be operating in. I think the company has done a really nice job in the last several years. And I would fully expect us to continue to do that going forward.

Our last question comes from the line of Quinn Bolton from Needham & Company.

For the past 20 years, WFE intensity has held at about 11.5% of semiconductor sales. You guys have highlighted a number of factors today growing process complexity, the transition to more 3D structures and advanced packaging and the requirement for new materials. My question is, do you expect WFE intensity to stay at about 11.5% and of semi sales? Or do you see that WFE intensity increasing? And Dan, what's built into the 2024 model that you've provided?

Yes. Thanks, Quinn. I think the best way to describe it -- let me take the modeling question first and then talk about the long-term trends from an industry standpoint. So in 2020 -- I mean, sorry, in 2020, WFE intensity was about a 13.7%, give or take, it was just shy of 14%. What we've done from the modeling perspective given the strong growth trajectory we see on our industry, alongside strong growth from a semiconductor standpoint is, is we've kept that ratio roughly in this forecast horizon. So there's no change in the capital intensity from the environment we're in, to 2024. And again, we think it's a prudent, conservative estimate on what we see developing from an industry standpoint. When I take a step back from the current environment and the model assumption, and we just look at what's driving the industry, traditional is Moore's Law, 2D shrinks. That has run its course, and that's no longer driving the power performance road map for the industry. The industry is inflecting. The PPACt road map is going to be driven by a new playbook. That new playbook has multiple vectors. The materials engineering is going to be a key enablement of each of those vectors, and we specifically are going to have a material role to play in driving this industry's road map going forward. All of that, when we combine it to a perspective of what we think is going to happen, we would naturally see an upward bias from a WFE intensity standpoint over time. So we think that math, that equation, that quotient has an upward bias, but we're not counting on that upward bias playing out in the model where we're setting expectations. If it plays out as we think, then I would expect there to be an upward bias to the results that we're putting forward as part of our model.

Yes, Quinn, maybe let me also add, if you look at the overall opportunity for Applied Materials, the pandemic really has accelerated the technology transformation of every industry. How we work, how we learn, how we shop, everything is accelerating. All of these technology transitions are accelerating. And many people talk about this decade, this technology transformation with AI, Big Data, being worth over $10 trillion. And then if you think about all of those connected devices, whether it's 0.5 trillion or 1 trillion connected devices, generating data, storing data, processing data, that's why we have this focus around PPACt, power, performance, area, cost and time to market. And that infrastructure for this new data economy, you see it in every company, you see it in every country, how everyone is recognizing the strategic importance of this industry and then also applied materials. That's why our focus is around enabling the acceleration of PPACt. And I deeply, deeply, deeply believe our industry is going larger, because it's at the foundation of the infrastructure for the future. There's no question. And what's enabling that power-efficient infrastructure, the performance per watt, are the new architectures, new materials, new structures, new ways to connect chips together, and no company is better positioned than Applied Materials in enabling that future.

Great. Thanks, Quinn, for your question. And I do have a couple of final housekeeping items to share with you in a second. Before I do share those with you, Dan, would you like to give us your closing thoughts.

Sure, Mike. First of all, we appreciate you joining us today. We've been looking forward to holding an investor meeting and giving you a closer look at all the programs we've been driving to position this company for continued strong profitable growth. I hope you've come away as excited as we are about the market inflections and the technology inflections that we are uniquely positioned to drive. PPACt is ready for prime time. And also the business model inflection to solutions and recurring revenue, we have high confidence in the target model. We're going to generate strong profitability and a lot of cash. It will fund our future growth. And we will return 80% to 100% of free cash flow to shareholders. Gary and I look forward to keeping you updated in our progress, and we hope to see you at conferences throughout the year. Mike, let's go ahead and give everyone the information you promised.

Okay. Thanks, Dan. So just as a reminder, the next investor event will be our Master Class on memory technology on Wednesday, May 5. After our next earnings call, we'll have the second Master Class on logic technology. That will be on June 16. Right now, on our website, you can already find video replays and PDF copies of all of today's presentations, also all of the customer quotes along with a copy of our prepared remarks. So with that, I want to join Gary and Dan in saying thanks for joining us today, and we hope we'll see you again very soon.