Good morning. Welc ome to Mic ron's 2022 Investor Day. I'm Samir Patodia with Micron Investor Relations. Farhan couldn't make it today due to a family emergency. It's great to see so many of you here today with us in San Francisco. Thank you as well to those joining via the webcast, which is available on Micron's investor relations website, investors.micron.com. We really appreciate your time and attention. Before we get started, I'll read our safe harbor statement. This presentation and associated commentary contain forward-looking statements regarding our future financial performance, the industry, and other future events. Such statements are predictions, and actual events or results may differ materially. We identify important factors in the documents we file from time to time with the Securities and Exchange Commission, specifically our most recent Forms 10-K and 10-Q, that could cause our actual results to differ materially from those contained herein.
These factors can also be found at micron.com/certainfactors. Although we believe that the expectations reflected in the forward-looking statements are reasonable, we cannot guarantee future results, levels of activity, performance, or achievements. We're under no duty to update any of the forward-looking statements after the date of this presentation. Non-GAAP reconciliations will be posted along with the PDF of this presentation on Micron's IR website, investors.micron.com, at the end of this event. Today, you'll hear from Micron's senior leadership team about the tremendous opportunity for memory and storage, as well as our strategy. The presentations will be approximately two hours long and followed by a Q&A session. We'll get started today with our CEO, Sanjay Mehrotra, after a brief video. Let's begin.
Seize the momentum. Micron is seizing the momentum to establish a new era of leadership. Today, you will feel this momentum.
Good morning, ladies and gentlemen, and welcome to Micron's 2022 Investor Day. Micron has made tremendous progress over the course of last five years, establishing competitive advantages with respect to technology, product, and manufacturing, delivering excellent financial performance and creating value, significant value for all stakeholders. We are going to leverage these competitive advantages to build another period of ever increasing significant stakeholder value for all our stakeholders. Before I start my presentation, I just want to mention that I got a bad cough, starting yesterday. I have tested for COVID yesterday as well as today. I am negative, but please bear with me in case I cough during the presentation. I certainly hope that I get through it without coughing.
During the course of my presentation, I will highlight the importance of memory and storage. The demand drivers for memory and storage are tremendous. Memory, more content needed in all end market applications, and memory is enabling greater value for these applications as well. I will also talk about stronger industry dynamics, including improved profitability for the industry, strengthened supply environment and fundamentals, as well as increasing complexity of the industry. Then, of course, I will focus on Micron's strong position today, positioning us well for the next era of leadership. Data is transforming the world, and memory is at the heart of this transformation. Digitization of the global economy is creating more opportunities for data creation. Before the pandemic as well, of course, the world was going through digital transformation, but pandemic has accelerated this transformation.
Hyperconnectivity provided by 5G is not only connecting more people, it is also connecting machines and devices and creating vast amounts of data. Of course, AI and ML needs more and more data to draw deeper insights, more useful and transformative insights. These trends of digitization of the global economy, hyperconnectivity, and insights from AI and ML are driving the virtuous cycle of ever-increasing data creation across the data center, intelligent edge, and smart user devices. Look at how the world has really seen exponential growth in data. It was only 2010 when first zettabyte of data was created. Last year, 82 zettabytes of data created, and that translates into one gigabyte per hour per person on Earth. By 2025, amount of data created is expected to be doubled. The tremendous opportunity that is ahead of us.
Of course, this is being driven by the trends of digitization of the economy, hyperconnectivity, and insights from AI and ML, and enabling memory to extend the frontiers of what is possible. Let us just look at the example of AI. Over the course of last few years, AI algorithms for natural language processing have really advanced from 100 million trainable parameters a few years ago, and those required 5 GB of memory. Now we are looking at GPT-3 and Microsoft Turing requiring up to 1 trillion trainable parameters and requiring thousands of gigabyte of memory. This is leading to the transformation in terms of breakthroughs such as auto dialogue generation, auto code completion, translation, semantic search, and so on.
Of course, as AI algorithms get more sophisticated, they're working on more data, and they then need more memory to process and draw insights from all of that data. Today, smartphones have become video editors in your palms. The trend of 5G is driving many such more applications ahead and actually driving greater usage of DRAM content in those phones. The computational photography as well as advanced video editing capabilities that are now in your smartphones would not be possible without the AI algorithms, which are again operating on larger data requiring more memory. A fascinating example is in the autonomous vehicles. Increasingly autonomous features in the vehicles are making them what we call in the future to be data center on wheels. Even today, Level three vehicles announced that they're using more than 100 GB of DRAM and approaching 1 TB of NAND in them.
The features of autopilot safely requires more memory. It requires real-time decision-making. The trend is that just the autonomous vehicle will drive a trend towards greater than 30x DRAM content and greater than 100x NAND content. As AI gets bigger, as it gets pervasive, as it advances, it makes the demand for memory and storage bigger. Exciting part is that today the demand for memory and storage is across multiple large end markets. This diversity of large end markets has a favorable impact on the stability of demand environment. Used to be that PC was the majority of demand driver for memory and storage. Then came mobile, and of course, mobile became, over the course of last decade, mother of all markets, driving tremendous demand for memory and storage.
Today, the trend of data center has become the biggest market for memory and storage. When we look ahead, data center market will be growing faster than the industry average. This is because data centers, particularly cloud, is transforming the businesses across all industries. It's using AI as a tool to really deliver increasing value and unleashing new business models. Data center, a big growth driver for memory and storage in the years ahead. Of course, automotive market, while in terms of dollar terms, a relatively smaller market compared to the total industry, yet it has the fastest growth rate ahead for the very reasons that I just discussed earlier that are increasingly autonomous features as well as, of course, more advanced infotainment features. Industrial also growing faster than the average of the industry.
Today, all of these diverse end markets combined represent $161 billion of revenue in 2021. In terms of the demand growth for DRAM, we expect the demand bit growth to be in the mid- to high-teens over the 2021-2025 time frame on a CAGR basis. Data center growing faster than the DRAM CAGR for the industry, and of course, automotive having the highest CAGR over this time period. NAND, we estimate our latest estimate is that the 2021-2025 demand bit growth CAGR will be in the high-20% range. PCs have driven, in the past, lot of demand growth. Of course, average content in PCs will continue to go up. However, SSD penetration is now approaching 90%. Mobile, of course, average content continues to increase.
However, smartphone units have stabilized on a total global sales basis. What this means is that mobile and PC will be growing slightly below the industry average. While data center for NAND, as well as industrial and automotive, will continue to be a strong growth driver. What does all this bit growth trend translate into in terms of the dollar opportunity for memory and storage industry? We estimate that by 2030 timeframe, memory and storage will be a $330 billion opportunity. Think about it. It has taken the industry nearly four decades to get to $161 billion. Over the course of this decade, we will add another more than $161 billion, more than doubling of the market opportunity. Clearly, memory and storage is going to grow faster than the broader semiconductor industry.
Memory and storage today represents 30% of the semiconductor industry. Turn of the century, it represented about 10%. We estimate that by end of this decade, it will represent more than 30% of the semiconductor industry. Memory and storage is clearly at the sweet spot of the technology trends shaping our world today. Let me move on to stronger industry dynamics. First, let me highlight that the industry profitability over the course of the years has strengthened. If you look at over the course of last eight years, average industry gross margins across the cycles, they are at approximately 40%. Greater than 40%. The previous eight year period, from 2006 2013, average margin across the cycles was approximately 20%.
Clearly, the industry profitability, due to the strong fundamentals on the demand side, as well as the supply dynamics that I will talk about, has strengthened through the cycles. The last eight year period is 2,000 basis points higher in gross margins across the cycles versus the previous eight year period. Of course, the demand side is one side of the coin that is driving healthier industry fundamentals. The other side of the coin is the industry supply discipline. Slowing of Moore's Law is naturally slowing, creating increased industry supply discipline. Let me explain why. Slowing of Moore's Law means that what's happening in the industry is that with each successive technology transition, with each successive node transition, the bits gained per wafer are reducing.
In addition, the capital intensity required to implement the new node because of more expensive equipment involved, more advanced tools needed to implement more, increasingly more complex features, as well as all the materials and structural aspects, the capital intensity is also increasing. The combination of the two, slowing Moore's Law, slowing bit growth gains from technology transitions, as well as increasing capital intensity, has the effect of moderating the supply growth and reducing the cost declines that are happening. These are favorable factors in terms of the supply dynamic for the industry. Reduced cost declines make the risk of obsolescence of products to go down, which is an important factor for the suppliers, especially when you encounter periods of any supply-demand imbalances. Of course, the pricing trend line has also improved. Increasing capital intensity requires a sharper focus on return on capital investment as well.
Focus on return on invested capital is all leading, we believe, to increased industry discipline with respect to managing supply growth to be in line with the demand growth. Of course, capital returns provided by suppliers, DRAM suppliers, have all instituted dividends. When you have capital returns in the face of increasing capital intensity, of course, there's a greater discipline toward supply growth as well. This is all reflected, and one measure of it is that if you look at the industry wafer equipment spend as a percentage of EBITDA over the years, in recent years, it has come down to around 30% level, and it used to be above 50% level in the years past.
Of course, this is a function of increasing profitability, increasing revenue growth opportunity due to the demand drivers I discussed, as well as improving supply fundamentals in the industry. This is happening despite the increasing capital intensity of technology transitions. What is Micron's strategy in this environment? We absolutely stick to our strategy of growing with supply in line with demand. We have been disciplined over the course of five years in this regard, and of course, this will continue to be a priority at Micron. Return on invested capital is important, and we pay a lot of attention to it. Another important factor to understand is the rising industry complexity. This is actually also a factor that is helpful for the supply dynamic in the industry. Over the course of multiple decades, industry leaders have built formidable pillars of formidable strength.
These pillars of formidable strength for industry leaders are required to continue to grow and continue to be retained in order to be competitive. For any new entrant or a small player, each of these pillars, each of these five pillars have to be scaled successfully for those smaller players to be competitive in the industry. Let me take you through what these five pillars are. First, rising capital intensity means that the minimum scale required for manufacturing operations to be efficient increases as well. Second, the leaders in the industry have built tremendous amount of intellectual property over the course of last few years, last few decades, and of course, tremendous amount of cumulative knowledge as well that is needed to continue with the technology and product road maps that will meet the customer's requirements in the years ahead.
Of course, this intellectual property is continuing to grow as well, and the leaders are leading on that front. The third aspect is that as the markets have become more diversified, the requirements for memory and storage have increased. It used to be that it was just DRAM for PC. Today it is DRAM and NAND for data center, for PC, for mobile, for automotive, and all of these diverse end markets have varying requirements. What it means is that you are required to be a competitive player, you are required to have a diverse portfolio of solutions and features in your technology, and this is all increasing the complexity as well. This requires significant amount of R&D and expertise to be able to develop and grow the portfolio that is needed to address the diverse end markets.
In the face of increasing technology complexity, manufacturing and making sure that the manufacturing is able to scale up production of new technologies in managing defects, in managing yields, in managing the levels of quality which are ever increasing in our industry and various end market applications is really, really important as well. This too, of course, relies on cumulative experience in the industry of many, many engineers. Ultimately, the ability to deliver technology and product roadmap that meets customers' requirements as well as delivery that meets customers' requirements and quality standards, builds trusted relationships over very many years. These trusted relationships enable industry leaders to work closely with these customers to align their product road maps with our product road maps.
Micron has had, of course, these five strong pillars over time, and Micron, over the course of last five years, has further significantly strengthened each of these pillars. Micron's total replacement value built over the course of very many years through excellence in these five pillars, is resulting in a replacement value of approximately $100 billion. This replacement value of approximately $100 billion represents equipment in manufacturing and R&D, primarily. Primarily the equipment in manufacturing and R&D. Micron, of course, has built tremendous value in terms of R&D investments, $20 billion in R&D investments since 2010 timeframe, which are positioning the company with strong technology and product leadership, not only for what is happening today, but these investments have gone into preparing for technology leadership for the years ahead as well.
I'm very proud that through the phenomenal teamwork of Micron's team, earlier this week we announced that Micron now has achieved 50,000 patents. Micron is an innovation powerhouse. We have patents now at 50,000. That really exceeds the patents and intellectual property portfolio of companies that are far bigger than Micron as well. You see, Micron over the course of time, and particularly strengthening its fundamentals through the transformation that we have driven in the business over the course of last five years, through the team of 45,000 or more at the 45,000 approximately at the company, has really built tremendous value. We of course are focused on continuing to increase this value for the company as well.
I've talked about the demand drivers and the supply dynamic in the industry, and this is creating memory industry is becoming more and more important to our customer ecosystem. Memory is integral part of the offerings of world's leading technology providers. Micron is partnering with those leading providers to bring industry-leading solutions to the market at volumes that our customers need. Let us hear from Satya Nadella, President, CEO and Chairman of Microsoft, commenting on industry trends in today's data economy and the partnership with Micron.
Thank you very much, Sanjay. It's great to join you virtually for Micron Investor Day. Without question, we are living through an era of unprecedented change to our society as well as our economy. Digital technology is the most valuable resource to help organizations differentiate, build resilience, and do more with less in this new era. Successful digital transformation hinges on an organization's ability to turn its data into predictive and analytical power. That's why we've built a comprehensive data platform to help organizations break down silos and manage their data estates to make informed business decisions. As the data economy continues to grow, so too will the demand for memory. That's why our strategic partnership with Micron is so very important.
Together, we are helping define the future of data and AI, combining Micron's world-leading memory and storage technologies with our own best-in-class databases, analytics, and governance tools, to deliver better insights for organizations around the world faster than ever before. This is just the beginning. As the world continues to rapidly change and digitize, I'm looking forward to how our partnership will grow in the years ahead. Thank you all very, very much.
Thank you, Satya. Now let us look at what Micron is doing to address the next phase of growth. First, investing in our people. I'm very proud of our 45,000-strong team across 17 different countries, building, executing, building leading-edge products and delivering excellent execution. Not only focused on business results, but also building an inclusive, collaborative, and a winning culture at the company that of course would be a lasting advantage for Micron. Our team and our culture is a competitive advantage for Micron. In fact, this has been recognized through very many industry-wide recognitions, including Fortune's Great Place to Work across most major sites in the world that Micron has operations in. This team is delivering world's most advanced technologies. Four years ago, at the Investor Day, we were looking at 1x DRAM and 64-layer NAND.
Today, Micron has led the world with leadership in both DRAM and NAND with introducing world's most advanced one alpha DRAM last year and ramping it into production well ahead of any other competitor. In NAND, we have led the way with 176-layer NAND well ahead of anybody else in the industry and ramping up in production. Today, one alpha DRAM and 176-layer NAND are already driving significant portion of our production. Again, there are many other companies that have not even introduced these nodes yet. We are continuing our technology leadership with one beta node, where we believe we will be ahead of competition as well. We plan to introduce world's most advanced one beta DRAM node into production, starting to ramp production by the end of calendar year 2022.
Similarly, we are leading with the next generation NAND as well. 232-layer NAND, we believe, will also be a leadership in technology and production for Micron well ahead of our competitors. We target to ramp the 232-layer NAND before the end of calendar year 2022. Let me point out that, of course, we don't stop here. We are committed to continuing our technology leadership beyond the 1 beta and 232-layer node as well. We have tremendous momentum in technology leadership. We are seizing this momentum for the era ahead of technology leadership in the industry. Not only are we leading the world in introducing most advanced technologies first, we are ramping them into production successfully as well. Micron is a powerhouse in terms of manufacturing. We build 6 million wafers a year, 3 billion units that we ship to customers.
We are leading in investing in smart manufacturing that is driving higher levels of yields faster, enabling node penetration to occur faster, which is really important. New technologies are really important in terms of driving the profitability of the business. They help grow, of course, the bits per wafer, they help reduce costs, and in manufacturing, it is important to bring up the yields on these nodes fast. Our manufacturing team has done an excellent job ramping these new technology nodes using smart manufacturing techniques at highest quality levels in the industry, as well as with strong yields and production throughput capabilities.
The confidence in our technology roadmap and manufacturing prowess, and the market environment that we project in the years ahead led us to announce that Micron will be investing more than $150 billion in leading-edge R&D and manufacturing on a global basis over the course of next decade. We made this announcement last fall. As we look ahead, of course, we will invest based on the industry demand requirement, always sticking to our mantra of growing supply in line with industry demand. As we project ahead, we think for second half of this decade, for DRAM, we will need a new greenfield fab in order to meet the 2030 era demand. We have been exploring our options in this regard, and I'll just give you some of the conside rations involved.
Those would be in terms of site selection, construction and operating costs, because cost competitiveness is critically important, and that includes government incentives as well. Existence of semiconductor ecosystem in the site involved for the next DRAM expansion of Greenfield fab and access to world-class talent. As a manufacturing powerhouse, we are responsible to our planet, and we are a good citizen in that regard. We just earlier this week announced that Micron targets to be net zero emissions by calendar year 2050 in line with the Paris Agreement. We are committed to environmental sustainability through our goals related to water, energy, and waste as well. Manish will take you through more of these details later. Micron is a strong partner with the communities that we operate in.
Our team members are giving to the communities, volunteering for the communities, and of course, Micron is always helping the communities and watching for the welfare of the communities where we have operations. I would now like to introduce AMD CEO Lisa Su. This is because we are bringing technology and manufacturing capabilities together to build unique competitive advantages that our leading customers are able to develop more products and solutions that we deliver to address their most critical problems and challenges. Let us hear from Lisa on how the memory and storage is important to their business and how Micron is partnering with them. Oops.
Thank you, Sanjay. It's great to be here with you all today. It's an incredibly exciting time for AMD and the industry as we are in a high-performance computing mega cycle. At the center of this, and one of the biggest growth drivers for both our industry and AMD is the data center. When we look at modern data centers, virtually every cloud, enterprise, and hybrid environment is looking for more performance, more scalability, more flexibility, and lower total cost of ownership. At the same time, emerging workloads like HPC and AI are dramatically shifting the compute demands in the data center. Our high-performance EPYC processors and Instinct accelerators must be paired with industry-leading memory solutions to meet these great demands. That's why we're so excited about our strategic relationship with Micron.
Through our long-standing collaboration, AMD and Micron have driven significant innovation at the intersection of processor and memory technology. We look forward to continuing our work together to deliver leadership performance for today's most important workloads, including the development of next-generation memory that will enable even higher performance for cloud and enterprise applications. We're also very excited about our work together on the upcoming CXL interconnect that will provide our next generation EPYC processors with even more memory capacity and bandwidth than is available with DDR5. AMD and Micron share a common belief in the importance of developing deep partnerships to deliver truly differentiated solutions. Together, we're unlocking new capabilities for our customers from the edge to the cloud.
Thank you, Lisa. Micron is focused on developing customer-focused products. We have been first to ramp advanced nodes across a broad portfolio of our products, and that is achieved through our vertical integration capabilities, going from the device to the design, to product and test, as well as understanding of the applications, and our deep customer partnerships that enable us to not only qualify products faster before they start shipping into production internally at Micron, but also accelerate the time of qualifications with our customers. Our deep customer relationship here is really important, and I would say that over the course of last five years, Micron has really deepened our customer partnerships. They are stronger than ever at Micron. Of course, they are helping us lead on new interfaces. Micron has established leadership in the market with DDR5, LPDDR5, LPDDR5X as well.
Leading with innovation, differentiated products, partnering with customers, great example is Micron's graphics GDDR6X, industry-leading with highest performance. In terms of quality, as I mentioned before, we are a leader in quality. Majority of our customers rank us as number one in quality. Of course, these deep relationships that we have built with customers have enabled us to have long-term agreements with them in the past that we have mentioned before, that more than 75% of our business is with long-term supply agreements. Now those partnerships are being taken to the next level by exploring a new business model innovation that will help improve the business dynamics in a win-win fashion. We'll address both supply as well as pricing considerations. Of course, this is in early stages of exploration, and Sumit will provide more details on this.
We have built our technology, product, manufacturing, leadership and deep customer partnerships into a strong execution engine that is producing strong financial leadership as well. The commitments that we made to you in 2018 investor day, we have delivered on those commitments. We are leading in technology. Our high-value solutions portfolio, we achieved more than 80%. We delivered $9 billion in operating improvements well before the timeline we had laid out in the 2018 investor day. Of course, we have been generating a strong free cash flow that is being returned to the shareholders as well. Over the course of last five years, Micron has delivered 16% revenue CAGR, 2000 basis points improvement in profitability.
We have invested $12 billion in R&D, positioning us well for the future and of course, have generated strong free cash flow. We have positioned ourselves to deliver a record year in revenue in fiscal year 2022, and we expect the strong trend of growth and profitability to continue in the future. It is our confidence in Micron's execution and the industry demand drivers that will lead us for the first time today, first time ever in the history of Micron, to introduce a financial model, a cross-cycle financial model. You will hear on this from Mark Murphy, our new CFO. I would like to take this opportunity to welcome Mark to Micron.
This is his fourth week and in a very short period of time, Mark is already demonstrating strong ramp up on our business and becoming a valued partner to me and to our leadership team. I strongly believe that Micron is entering a new era of leadership. Our industry is entering a new era marked by strong secular demand drivers across diverse end markets, and with industry fundamentals with respect to supply and pillars of formidable strength that are only continuing to strengthen further, leading to improved profitability and revenue growth opportunity in the years to come for industry leaders. Micron has a unique combination of product leadership, technology leadership, manufacturing and financial leadership that is better than any other company at this stage, underpinned by our strong team and our culture.
This positions Micron to really extend the frontiers of what is possible with memory and build, create value for all our stakeholders. Later today, you will hear from our leaders in how, in each of these areas we are excelling and driving that new era of leadership. Thank you. With that, I would like to invite Sumit Sadana, our Chief Business Officer, to discuss our product and business strategy. Thank you.
Thank you, Sanjay. Good morning, and it's really exciting to have all of you here. Welcome again to all of you who have joined us here in San Francisco and to everyone who is tuning in on the web. I wanted to talk about three things going to the next stage of detail of some of the things that Sanjay mentioned. The first one is our product portfolio. I'll talk a little bit about our end markets and then the business model innovation that we are continuously driving at Micron. We have been on a transformation journey, and Sanjay spoke about the leadership and technology both on the DRAM and NAND side, and it's really exciting that we have also been creating tremendous momentum in our product portfolio.
We have a huge amount of value proposition that we are bringing to our customers and driving co-innovation opportunities with customers to create more differentiation. This is backed by the industry's best quality. More customers rank Micron as number one in quality than all of our competition put together. A really strong franchise on quality. The best manufacturing diversified infrastructure around the world for memory and storage that Manish will discuss in more detail later. We are very focused on accelerating our time to market with our products. This is critical to our customers. It brings new capabilities, new value to our customers, so they can get to market faster, and I'll talk a little bit more about that, about our portfolio. Four years ago, when we spoke at Investor Day, we made several commitments about improving our product portfolio.
Today, I'm very happy to say that Micron has leading products in every single end market. We are focused o n time to market leadership, and you can see that with some of our leading automotive products, both on UFS 3.1 on the NAND side and with DDR5, LPDDR5 on the DRAM side, automotive qualified products. We also have leadership on the mobile front in terms of time to market. Our LP5 and LP5X mobile products are first to market, as well as our latest NAND technologies, and we are also first to market with MCPs, which combine DRAM and NAND into a single package that is tremendously valuable to our mobile customers. Besides that, we are also focused on differentiation. We have the world's fastest graphics memory with GDDR6X, which powers the most advanced graphics card in the industry.
This is obviously a very differentiated capability that we bring to customers. QLC leadership is another differentiated capability, and we believe we also have low power leadership in the industry, which is so important in mobile kind of applications across the board, both with smartphones as well as with laptops. Another important aspect I wanted to drive is we are focused on share in important differentiated segments of the market. We have number one share right now in D5 products. We also have number one share in automotive, and I'll discuss a little bit more about this ahead. We are strengthening our product portfolio with preferential share gains in important parts of the market.
Even as our overall bit supply share remains flat, because Sanjay said, this is a very important thing for us that we drive our overall supply growth in line with market demand, so our overall bit supply share will be flat over time. We are focused on driving preferential share gains in important portions of the market that foundationally strengthen our portfolio from year to year. The growth of TAM is slowing in areas like mobile and PC, and these consumer-focused segments of the market that are also more seasonal will naturally become smaller parts of our portfolio going forward, from 55% in 2021 to 38% in 2025. Meanwhile, data center is going to grow to 42% from 30%, and that is obviously a very attractive, fast-growing segment of the market.
The fastest-growing segments are automotive and industrial, and networking growing exceedingly fast with 5G and other new networking technologies in the data center as well. These segments also have much more stable gross margin profile over the cycle. We are very excited with our leadership in these markets. We have number one share in automotive, number one share in industrial, very high share in networking, and we are over-represented in these markets, in terms of versus the TAM, and consequently, it provides more stability to our portfolio, and this will increase from 15%-20% of our portfolio over the next few years. I wanted to now turn to speaking a little bit about our end markets, and I will start first with the data center and the server architecture. The legacy architecture for servers has been pretty simple and straightforward.
You have a CPU, DRAM connectivity, which has been increasing with time, the amount of DRAM content going into servers, and of course, HDD storage and then faster storage, which has also been increasing and displacing HDDs more and more in more workloads over time, which is through our data center SSDs. This architecture is not really capable of delivering to the needs of tomorrow's workloads. I wanted to spend a little bit of time talking about that change.
One important aspect that people have come to a realization to over the last several years is that the path to higher performance in a server really has to focus not just on the processing speeds and more cores in a processor, or not just about CPU versus GPU, but very crucially, the bandwidth between processor and memory. You heard Lisa speak about that on the video as well. That bandwidth between processor and memory is so critical, and there are a couple of ways to increase that bandwidth. One is to bring processor and memory closer together with new type of memory innovations. High bandwidth memory is one such innovation where very high performing memory is mounted on the same substrate as the processor, and provides that very high capability of memory and
Processor interface bandwidth, and can be used in AI and machine learning type of applications. In addition, content of memory is crucially important. The more memory you can put in a server, the more connectivity you can create, the higher the performance of the system. In large part, that happens because the processor is idle a lot of time waiting for data from the memory. Ensuring that we have adequate memory is hugely important. If you see how this architecture is evolving, you have HBM memory in the CPU, you have DDR connectivity as before, PCI-enabled HDD, as well as NVMe-enabled NAND SSDs. You have new types of processors coming in.
You have infrastructure processing units or IPUs, data processing units or DPUs, and these are offloading certain types of workloads from the processor in order to perform these tasks more efficiently. These, of course, need their own memory as well. Additional connectivity through PCIe can be used to connect graphics accelerator cards or graphics GPU-based accelerators. More specialized accelerators, which have specialized ASIC designs that have been done by our customers to drive those higher performing workloads. These interfaces, like PCIe, can be over time, replaced by CXL, and I'll talk about that in a minute. These accelerators also have their own high bandwidth memory and of course, graphics memory connected to GPUs. Finally, we are on the cusp of CXL being deployed in the industry.
We will have CXL 2.0 late in 2024, and that will really kickstart a whole new era of connectivity with memory and also enable new capability that never existed in the data center before. Memory disaggregation, memory pooling, and use of emerging memory technologies connected to CXL, where cost and performance trade-offs can be made, which today are not possible with DDR connected memory. If you look at this picture, there are several important benefits of this new heterogeneous architecture for our customers. One important thing you can see from all of the memory blocks in this picture is how much consumption of memory is growing in a typical server. In a typical AI machine learning-focused workload server, half of the overall BOM cost comes from memory alone. This is such an important part of the transformation of the data center architecture.
This movement is causing the average amount of memory and storage shipped into the data center to go up quite significantly. We are projecting doubling of DRAM and tripling of the NAND from 2021 to 2025, due to these trends that I just described. This will drive a strong double-digit growth of TAM in the data center far in excess of what the overall, TAM growth for the semiconductor industry is and also higher than the overall TAM growth of the memory and storage. That's a pretty big TAM growth of one of the largest portions of the TAM. I now wanted to switch to talking about DDR5 because Micron is a leader in DDR5, and we are on the cusp of another big change in the industry, which is the use of DDR5 over DDR4.
DDR4 is running out of capability and bandwidth, so DDR5 has to step in and deliver on the performance needs for all kinds of applications and workloads. Micron's DDR5 has 2x the data rates, 2x the concurrent operations, and 2x the bandwidth of its previous counterpart, and consequently, we have huge amount of demand from our customers for these products, and we are very excited to be having number one share. But also we are driving the ecosystem to enable a smooth transition to DDR5, which is very important to us. It's the only type of ecosystem program in the industry. We have engaged over 160 companies in this.
It's everything from silicon companies to system companies to OEMs and cloud vendors, IP providers, and that ecosystem is working together, put together by Micron to really enable a smooth transition on this very important change, that's coming up. We had been talking to you about high-bandwidth memory and our investments to enter this space as well. I'm very excited to announce that we are now in high volume production with HBM2E with the world's leading consumer of high-bandwidth memory as our customer. This was the commitment we had made to you that we would get to this stage, and we are in high volume production now. We are also driving a very aggressive and leadership roadmap when it comes to future high-bandwidth memory products, and you'll be hearing more about that as we are ready to introduce those products into the market.
This market is going to grow pretty rapidly. It was a pretty small market, only about $1 billion last year, versus, you know, the overall memory TAM, which was very large. We do expect this to be one of the fastest-growing product categories, again, driven by the growth in AI and machine learning primarily and associated workloads. Wanted to spend a minute to talk about CXL. We have CXL 1.1 that will start to be deployed, starting late this year, but mostly next year. That's going to be used mainly to do proof of concepts of CXL connected memory. The real deployment of memory on CXL will start late in 2024 when CXL 2.0 gets introduced and then will accelerate even further when CXL 3.0 standard enables memory pooling in a very real and practical way.
Memory pooling and memory disaggregation are critical aspects that our customers are looking for, but even more important, expansion of the amount of memory that can be connected to a processor, which is ultimately what CXL enables in a very material way. When CXL gets deployed, you will see even more memory and even higher amount of average DRAM content in a server, again, driven by CXL. CXL will enable emerging memory, as I said earlier, and Micron is, we believe, a leader in the industry when it comes to emerging memory technology through all of the IP we have developed, all of the experience we have developed, in this arena, and we are very excited with our roadmap, on this front. I wanted to spend a couple of minutes to talk about data center SSDs and NAND storage in the data center.
This is another exciting arena because certainly, we have been very focused on Micron's own share gains in data center NVMe SSD. Again, we have mentioned to you about new products that we are introducing, most importantly, our first vertically integrated NVMe data center SSD, which we introduced last fall in 2021. That was our first generation product. We introduced our second generation data center NVMe SSD just in March through an announcement that we made. This was built on 176-layer NAND. So in essence, our second generation product, vertically integrated product, already leveraging the most advanced NAND in the world, coming just a few months on the heels of our first generation product, leveraging a lot of that IP from that first generation product.
I'm very excited to announce that we are already just two months after that announcement, of our second generation product. We are already in volume production with that product, despite the fact that typical data center quals tend to be very long, taking many months. This goes to the quality of the products that we have been able to create. We are gaining share and momentum in the data center space with these SSDs, and it's a very exciting, very exciting journey for us. I did want to mention that the market in the data center for flash is massive. The opportunity is massive. Yes, a massive amount of data today in terms of bits is still stored on HDDs.
However, the cost gap between SSDs and HDDs in the data center has been reducing over the last decade and a half and will continue to reduce over time. There are certain important inflection points that these price points can deliver. Just like on the client and PC markets, when the SSD price hit $40, the adoption of SSDs in PCs became just exponential. Similarly, we see a very important set of inflection points coming later this decade when cost of SSDs enable a price point which is 4x that of an HDD, and later on, 3x that of an HDD. At those price points, there is so much overwhelming advantage that there will be a massive inflection of workloads moving from HDDs to SSDs in the data center.
That's what's driving the 3x and 5x bit growth that I'm describing here. Really exciting opportunities ahead in the data center from a market growth perspective. I did want to talk a little bit about mobile because that's another important market for us and a critical part of the overall TAM. The mobile market is continuing to be fueled both in DRAM content growth and NAND content growth by what is happening with the evolution of smartphones. You're seeing, as Sanjay mentioned, so much video editing capability in your smartphone. You're seeing the rise of the TikTok generation, 1 billion users, and a lot of content being created through short-form videos. A lot of this content with highly capable cameras are creating huge amount of need for storage, causing average capacities of storage to go up.
There is maybe more than most people appreciate, tremendous amount of AI and machine learning capability being integrated into mobile application processors. When you take a picture from these highly capable cameras, in that short while before the picture is stored onto your phone, a huge amount of optimizations are happening in the background to get the best kind of image capture and to enable computational photography. That's stressing the amount of DRAM that is needed and consequently driving the amount of DRAM up. We have mentioned in the past how 5G already has 50% more DRAM, 5G phones 50% more DRAM, double the NAND in terms of smartphone content. That movement should continue with over 80% of the units sold in the smartphone domain in 2025 being 5G units.
Micron's leadership here goes beyond the lowest power, LP5 and LP5X products. It also is about launching MCPs, these multi-chip packages that are so critical to our smartphone customers, launching them to market ahead of anyone else in the industry. You look at the mix of leading-edge DRAM and leading-edge NAND that we have been able to get into the hands of customers ahead of anyone. To do that with world-class quality, with our tremendously strong manufacturing and operations infrastructure is really a phenomenal competitive advantage for us. I wanted to spend a minute on PCs. You would think that, okay, how much, you know, how much innovation is happening on the PC architecture front?
All of a sudden you have new architectures coming onto the landscape, like Apple's Mac coming out with M1 Pro and M1 Ultra configurations. Take a look at the amount of memory in these configurations. It is just mind-boggling. An average PC with only 9 GB of DRAM and up to 32 GB in M1 Pro and M1 Ultra going up to 128 GB of DRAM. This is clear testimony that amount of DRAM is a critical factor to system-level performance. You will see more changes happening on the PC landscape, as ideas get thrown around about how to improve PC performance. Invariably, you will see more memory configurations in there. Of course, SSD content is increasing as well. Average content of SSDs is going up in PCs, and here our industry-leading QLC technology is so critical.
We have been doing really well with QLC technology. It's extremely profitable compared to TLC even. We are preferentially driving those type of products with our customers, and our customers are responding with more and more usage of QLC over time. Wanted to talk a little bit about the industrial revolution that's quietly happening in so many segments of the economy. There are tens of billions of devices out there that are being connected, and will get a new lease of capability and innovation through 5G connectivity that is ahead of us. So much value is going to be created in so many different segments.
The industrial market is the more fragmented market, but so many of these markets are growing exceptionally fast, and more and more of the industrial base is going to be driven off of intelligent decision making, data analytics, data-driven decisions. These are even penetrating areas like agriculture, which you would not have thought would lend itself so easily to technology. But you see the use of low Earth orbit communication satellites. You see the use of AI-enabled security and analytics and smart grid and clean energy and the use of technology there. I mean, it is application after application, tremendous amount of growth.
We have number one share in the industrial market, so really exciting position for us, and it's enabled through a long history of working with our customers in these segments, creating long lifecycle products, because customers don't tend to use new nodes of technology as often as in other segments. Long lifecycle support is crucial. We also do customization in our products for specific applications. That's what enables us to gain that leadership, again, with the backing of Micron's world-leading quality and reliability. Really exciting for us to leverage that. Now to talk about automotive, as Sanjay mentioned, the fastest-growing segment for the next decade in the industry. We are very excited about this because Micron is the clear leader in market share in automotive by a long mile, and we are continuing to build on that momentum.
We are bringing the industry's highest quality, highest performance products, the most expansive portfolio of products, and time to market leadership versus our competition to sustain that advantage and grow that advantage. We're also co-innovating with our customers, and that's a very exciting part of how we continue to build on that leadership. The market is in the midst of the most exhilarating set of changes on the technology front that the automotive industry has ever seen. It is super exciting for us because this is not just about autonomous vehicles and getting cars to drive themselves. That will happen in due time, even the improvements to create so many driver assist functions that use tremendous amount of DRAM and NAND for that intelligence to be deployed is critically important driver of the growth, vector here.
We have connectivity and communication modules that are getting deployed. We have in-vehicle infotainment and digital instrument clusters. We have high resolution displays. We have so many things that are improving the experience for the driver. Just like airplanes have black boxes, you're going to soon start seeing black box capability in your car using high capacity, high endurance SSDs. This will change so many things about diagnostics in a car, figuring out repairs that your car needs even before you know that your car needs those repairs, because so much of the activity of the car will be getting logged in these SSDs. It'll also be a sea change for the insurance business because all of the actions of the car and the actions of the driver at all times ar e also being recorded.
It's going to be a really big change in many different ways. When we talk about 100 million data center on wheels, as more and more technology gets embedded, more and more content of DRAM and NAND, you can compare that 100 million number with 14 million servers that get sold every year globally. Hundred million data center on wheels versus 14 million servers and average content obviously growing over time in cars. Really exciting moment for us. Let's hear from BMW, from Carsten at BMW. To have him talk about how he looks at automotive technology and the role of memory and storage and Micron.
Micron's outstanding support helps us to introduce groundbreaking new products like the fully electric BMW iX and the i4 to the market while the semiconductor market is under extreme tension. The underlying commitment to the automotive market is highly appreciated. BMW is looking forward to continue the collaboration with Micron in the development, infotainment, and ADA features of BMW's Neue Klasse, hitting the market in 2025.
Thank you, Carsten. You heard about so many different subsystems in the car that Micron is co-innovating and working with our customers on. This goes beyond, obviously, our partnership with customers in automotive. We have really deepened our partnerships over the last five years with customers and focusing on adding value that go way beyond just the delivery of products to our customers. We are doing joint R&D and joint labs with some of our customers, innovating in packaging technologies that are increasingly critical for our customers to differentiate their products. We are designing custom memory where appropriate to drive those advantages, and even focusing on how to improve qualification and monitoring of memory through telemetry capabilities, with our customers.
A whole host of capabilities that we are bringing to bear that ultimately are ensuring that our relationship with our customers is strengthened and made even better with time. With that, let us hear from Kirk at Lenovo about how he sees our partnership as well.
Hello, Sumit. Thanks for inviting me, and I'm excited to talk about unlocking new edge to cloud innovations for our joint customers. Lenovo, working closely with Micron, has built and continues to grow one of the broadest, smarter technology portfolios in the industry. As the world's largest PC company to the other extreme, as the number one provider of the world's top 500 supercomputers, we're now expanding with one of the fastest growing edge device and edge server portfolios in the industry. With Lenovo now powering eight of the top 10 cloud service providers, we're now delivering a comprehensive set of edge to cloud solutions together. Lenovo's vision is smarter technology for all, and we're committed to doubling our R&D and doubling our net income over the next three years. To keep up with this crazy growth, we've doubled our spend with Micron over the past year.
We're expanding our memory and storage capabilities across our PC server storage and smartphone businesses, including new innovations from the edge to the cloud that are helping transform the manufacturing, retail, healthcare, and financial sectors, just to name a few. Lenovo prides itself on being the most trusted partner, empowering our customers' intelligent transformation and helping solve some of humanity's greatest challenges. This innovation with Micron is critical to helping us achieve that vision. Thanks for inviting me, and I look forward to continuing this momentum and growing our partnership even further.
Thank you, Kirk. I wanted to move to the last section, which is about business model innovation. When it comes to innovation, of course, we have tremendous amount of innovation happening across our technology, our engineering teams, in operations, in all aspects of our business, and we certainly use innovation to continuously think about what could we be doing differently about our business? What innovative ideas could we be coming up with to strengthen the foundation of our business? Some of these ideas do really well, scale up well, and have a profound and important impact on the company and the industry. Others are experiments that may or may not scale well. We continuously think about how to make our business model stronger. To that end, we have spoken about our long-term agreements.
These are written agreements that we have with our customers that ultimately enable us to get much better visibility into our customers' demand plans. These end up reducing the disconnects between customers and suppliers related to their view of demand versus what we are building towards because of our long cycle time investments in capital and manufacturing. They do improve revenue consistency. Even though these are not take or pay agreements, they do provide us volume visibility, quarter at a time over the next 4, 5, sometimes more quarters. We have taken about 10% of our business being done under LTAs in 2017 to over 75%. If you think about the remaining portion of our business that's not under LTAs, we do a lot of business with distributors and with consumers with our Crucial product line, etc .
There is always going to be a certain portion that we cannot do LTAs on. What this means then is substantially all of the business that we are doing with our large customers are under LTAs. Not only is this something that we believe helped Micron a lot, we saw a lot of our competitors follow in our footsteps, and now a large part of the industry work is now done through these LTAs. This has been a tremendously positive thing for the whole industry. I wanted to now talk about a completely new idea, and this relates to forward pricing agreements. Keep in mind, those LTAs don't have any kind of price in them. It's mainly just volume agreements.
If you look at a repr esentative model of, okay, here is a view of DRAM industry pricing over the course of multiple years. We obviously have this price curve. This is what we realize from customers. Our costs are declining over time at a different slope, obviously, because the costs don't have these ups and downs. They are going down pretty steadily. That creates a certain gross margin profile for our DRAM business. Obviously, in that gross margin profile, you have variability. Sometimes prices are strengthe ning, sometimes they're falling. The gross margin has that variability coming from the ups and downs of price versus a steady downward trend on cost. Now, what would happen if we could get the net price through this timeframe to be the price that we transact with our customers?
If this price curve that is achieved through this cycle in the industry is parallel to the cost curve, has the same slope as the cost curve, you would end up with pretty stable and flat gross margin through the cycle. Our goal now is to talk to our customers about these types of cross-cycle agreements where we set a price curve over time, call it a three year agreement. This price curve is not meant to be higher than what they would otherwise get, because then they would not be interested in a deal like that. It would not be lower than what we would expect to get through the ups and downs of the pricing cycle, because that's not something that we would want to do. It would have to be set at a fair level, and follow the cost curve down over time.
This kind of an arrangement can have certainly a lot of benefits to our customers, definitely has benefits for us. Our customers get guaranteed supply because we would give them that guaranteed volume and preferential allocation in terms of situations in the industry when supply is tight. They would not have to worry about should I be building inventory unnecessarily because we think the pricing is going up, etc , so they can have their working capital be much more efficient. They will get guaranteed price reductions that is following our cost curve down. Now, this sort of thing definitely has benefits in situations where our customers are creating consumer products that have to hit certain price points, and they don't want the volatility of DRAM pricing going up and down.
Because keep in mind, DRAM is becoming a bigger and bigger portion of the BOM cost for all of our customers in practically every single segment of the market. It is very attractive for customers from that angle. Customers who are in the cloud domain are signing multiyear deals with their end customers based on cloud capacity that they are putting in these contracts. I already mentioned to you what big portion of these server costs are coming from DRAM, right? Very, very large portion of their cost structure is DRAM. To have predictable pricing on that DRAM that locks in a known level of economic return for them with their customers ahead of time can be very attractive. For us, of course, we get planning predictability, stable gross margins. I don't have to tell you the benefit of this to Micron.
This is not just a theoretical exercise. I am super excited to announce that a top ten customer has already signed up on this model with us, more than $500 million a year of revenue for a three year agreement. Our goal with these agreements is obviously, you know, after the first year, make it a sort of an every year evergreen three year agreement. After the first year is done, you add another year to the agreement and you ensure that you're constantly doing course corrections on the price curve to ensure that it's a very, very fair kind of a price curve to both the customer and to us. Of course, this is an experiment. We are starting on this journey, and we are going to be continuing these discussions with our customers to see how this works.
We are super excited to continue pursuing ideas and innovative thinking to continuously strengthen our business in every way possible. Now I just wanted to end with a reiteration of some of the key points that I mentioned. Really strong industry growth dynamics, Micron's own focus on product portfolio and mix driving preferential growth in the most important profit pools of the industry. Our goal is to keep our bit shipment share constant, but continuously grow in the higher and higher versions of the profit pool, so we can continuously improve our profitability within that construct, and also to continue to innovate and think about ways to reduce volatility in our portfolio. Thank you. With that, I'm going to introduce Scott DeBoer, who's going to talk to you about our extraordinary momentum in both technology and products. Thank you.
Thank you, Sumit, and good morning, everyone. I'm disappointed not to be able to join you in person today, but unfortunately, after avoiding COVID for the first several years of the pandemic, it's caught up with me in the past week. I am excited to share the opportunity still with you today for the exciting path ahead for Micron's technology and products. We've had an outstanding situation over the past several years of progress in both our technology and our product performance, creating an environment of leadership. We're solidly positioned for the upcoming decade with leadership in our technology and products. As you've heard from Sanjay and Sumit, and certainly I'll reiterate here, we've made tremendous progress over the past few years. We've really built an exceptional foundation to build our future leadership over the next decade.
I'm gonna start out today with a description of the Micron development team across the globe that has delivered this amazing technology and product portfolio. As you can see, the team is focused in a variety of regions across the globe, and this provides us with the opportunity both for standalone disruptive technology development, as well as technology development directly inside of some of our manufacturing facilities, which allows us to have the opportunity to focus on quality and true volume manufacturing enablement as we finalize the technology efforts. A number of sites that are product-focused, where we create the software and the different aspects of the technology that lead to our end products.
We truly have what I think is the industry's strongest engineering team based on this level of diversity and the blend of talent all across the globe. At Micron, we have a strong culture of innovation, certainly highlighted in one way by the 50,000 patents that Sanjay referenced earlier today. In the past year, 2,500 patents, which was 15% more than the previous year. We have a number of innovation efforts across the company. One of those is Micron's Innovation Network for Women. It was something we started back in 2018, and that has led to almost a 4x increase in the number of patents with female inventors. Definitely proud of that.
Vertical integration is a key focus, and I'll talk about it a number of times today in different contexts. This is our focus from silicon to systems on our end product value creation. That goes all across our development efforts in ensuring that we are optimizing, whether it's in early technology development or end systems, we're optimizing all across that development process, the engineering efforts to ensure we wind up with a truly well-thought-out product that meets the end applications and creates the ultimate value for Micron and our customers. Our customer focus is brought up by Sanjay and Sumit previously also, is another key aspect of our overall development effort.
Certainly, our focus is on co-invention with our customers across a number of the more complicated system products, we're delivering today to ensure the end value for our customers is optimized. Micron's position in the memory industry growth overall, in terms of the importance, to multiple applications across the space, has really led to a much stronger partnership between our customers and our engineering teams in working towards that end product value. I'll start here with a little description of how we have managed through product transitions and major initiatives, over the past 10 years or so, and how that positions us going forward.
As we look back in time, we have at times chosen to lead with first of a kind innovations like multiple patterning and CMOS under the array, both driving industry transitions and providing Micron with significant industry value and leadership. At other times, we have chosen to implement the right technology at the right time. A couple of examples there would be our success squared three by two DRAM architecture transition, which is still the architecture we use today, and our most recent floating gate to replacement gate NAND technology transition, which we executed in strong fashion at the right time for Micron to create the most value for our products.
In addition to technology transitions, we've also had a number of initiatives that have helped us grow into the leader that we are today. Both in NAND and DRAM. A couple of those are around our technology cadence acceleration over the past five years, which has really led us to this position of technology on our one alpha DRAM and our 176-layer NAND, as well as our NAND solutions enablement initiatives internally really focused on vertical integration, which has led to a much stronger product portfolio on the NAND side that we'll talk about. Sumit's already talked about a bit, and I'll add some more color to in this section. These type of initiatives have absolutely positioned us with confidence to face the next set of future challenges and deal with those in a way that positions Micron for continued leadership going forward.
A number of them are listed on this slide, both in terms of technology and product architecture, and total changes in approach to how some of our major product portfolios are constructed. I'll go on from here to talk a little bit about one of those next initiatives that's coming up, which is EUV. We've talked for a number of years about Micron's approach to implementing EUV on the DRAM side. Absolutely, this has been a thoroughly studied and thought through technology strategy on the Micro n side for when the right time to introduce EUV was.
We're very confident both in our assessment previously of when the right time was and in what I'll talk about today as our strategy for implementing it and our ability to implement it on this schedule. We've targeted 2024 to be our introduction timeline for EUV in high volume manufacturing, and this is aligned with our one gamma node. I'll talk about our next node, which is one beta, and how we're enabling it with multiple patterning on the following few slides.
One gamma will be the intersection node and really based on the ability to time that with the most advanced EUV technology enablement, the tools, but also the infrastructure around EUV in terms of masks and resists and ensuring that the whole ecosystem for EUV is optimized at the right time for Micron to bring that along and have it be cost effective. Also, something I've talked about in the future or in the past is the ability to create differentiation in our products through the use of EUV. We've also determined that one gamma is the right time for EUV to provide both the cost structure and the innovation ability to move our leadership forward on the DRAM node.
On one gamma, we'll have multiple layers on EUV, and we'll choose those appropriately for where EUV can add the advantage of either differentiation or cost advantage. We'll continue to use our industry-leading multiple patterning technology for other layers, as appropriate. Our DRAM technology overall, as I mentioned, is in a solid place with leadership on our one alpha node. It really provides us with what our customers have come to expect in terms of excellence and quality, performance, density and cost. Our node overall for one alpha was approximately one year, nine months to one year ahead of our competition. This is a very strong foundation for us to build on, and our objective is to sustain that leadership over the next decade.
Our 1 beta node is our next node, and it'll be described in a bit more detail on the following slide. 1 beta node is our first node where we have integrated advanced CMOS in terms of high-K metal gate across the entire node portfolio. Our 1 gamma node, as I mentioned, is the following one beyond that, and that one is where we introduce EUV. For the nodes that follow, both advanced CMOS and EUV will be incorporated into each of them, ensuring that we extend the planar roadmap and maintain our leadership position through this period of time. We'll also be introducing High-NA EUV at the appropriate time in the future.
We've made investments relative to High-NA EUV tooling in the future and are deeply involved in ensuring that that roadmap intersects the DRAM roadmap at the right time. Again, Micron will make a decision on when to implement High-NA EUV . On top of all of this, we continue, as we have for a number of years, to invest in 3D DRAM R&D, and the 3D DRAM process flow will be introduced at the right time for Micron, again, to deliver return on investment appropriate for our product portfolio. A little bit more now about our next node, our one beta node. This will continue the path of providing the world's most advanced DRAM. This node will be ramping in late calendar year 2022.
As I mentioned previously, it does incorporate high-performance CMOS across the entire node, as well as makes use of our advanced multi-patterning lithography, which we strongly believe is still the right choice for this node to provide the kind of quality and performance that's expected by our customers. This particular node will go across our whole product portfolio, and it will extend our DDR5 and our LP5 leadership, in addition to providing next-generation graphics, HBM and automotive products. This will be a comprehensive node across our whole product portfolio and continue the path of leadership, not just in timing, but in power, performance and cost.
When we look across our DRAM product portfolio, absolutely we have leveraged our one alpha node and the prior 1z node to deliver leadership across that product portfolio. I'll add a little bit of color to some of the products that Sumit has mentioned in some detail previously on the technical side. On the compute products, we've absolutely delivered a cost leadership in terms of our DDR4 and DDR5 portfolio, leveraging these advanced technologies both in terms of performance and in terms of cost. On the mobile side, we have a number of first of a kind, first in industry technology products on LPDDR5, and those Sumit touched on a few of those. These absolutely have provided higher bandwidth operation than competition based on our advanced node implementation.
In addition, the quality and the capability of our leading OneAlpha node has distinguished itself from our prior nodes, in that we've been able to utilize it in the automotive space and have qualified it with a number of customers already, even at this earlier stage in the node than prior nodes. Great automotive capability as well as mobile product capability from these DRAM products. On the graphics side, we've got industry leadership in terms of high performance speed at 24 Gbps with our GDDR6X product. This is a Micron-only enabled technology using multi-level signaling to get the highest speeds in the industry.
This particular graph actually is a demonstration of 30 Gbps, just demonstrating some of the headroom that we have built into these products to ensure high volume operation at the 24 Gbps, where the product is actually offered. On high bandwidth memory, Sumit has already discussed some about the fact that we have now ramped this technology. The quality and the performance of our HBM2E is providing optimization for our customers to leverage it for future workloads. Our focus on the product development side is clearly now shifted to enabling leadership, performance, and power on HBM3 and beyond as we move into advanced high bandwidth memory product offerings. On the new memory architecture, Sumit touched on CXL and how it fits into the compute architecture space.
From a product development point of view, this is an area where we're definitely focused on customer co-invention and ensuring that we provide the right CXL products, whether they're DRAM based or future memory based, for our customers, and that we're meeting their end application goals. Shifting over now to NAND. We again have a roadmap that will deliver leadership over the next decade, building on our 176-layer technology, which we introduced roughly a year ahead of competition. The 232-layer technology I will talk about again in a bit more detail on the next slide will be our next node, followed by an extension of different nodes through the rest of the decade based on our replacement gate technology. All these technologies have a backbone built on a few common features.
One is ensuring that we have advanced CMOS to provide performance leadership, enabling the kind of mobile products and SSD products, that we have in the leadership space going forward. Each of these technologies is based on dual stack technology, wh ich we introduced first for cost effectiveness and process capability. Our focus going forward absolutely is on extending our QLC, our four bit per cell leadership path. Certainly QLC is a mixture of design technology and device and process technology to enable high quality, four bit per cell components and ultimately leadership products. A little bit more about our 232-layer technology, which we'll be starting to ramp in the late part of calendar year 2022.
This again will be the world's most advanced NAND and will be a technology platform that we build industry-leading products on over the upcoming 2 or 3 years. This technology extends our CMOS under the array and our two stack process architecture that I mentioned on the previous slide to provide industry leadership in terms of cost and performance. We've optimized the technology around what we need to make the world's best managed NAND in both data center and client SSD products. The combination of controllers, both internal and external, has been a strong focus and strong element of our vertical integration focus to ensure we have optimized both the NAND and the controller technology for what we need to deliver future leadership products.
Like our previous nodes, this node will deliver increased density, power, and bandwidth relative to our 176-layer node, and we'll provide more details on that as we get closer to the ramp. On the NAND product side, we definitely are leveraging our 176-layer leadership to ensure that we deliver the industry's best products. We have a number of first-in-industry products. Sumit mentioned many of them across our whole portfolio enabled by 176-layer NAND. I'll start a little bit with our mobile products, and one that we're very proud of is our UFS 3.1 product, which leverages our industry-leading DRAM along with our 176-layer NAND. In addition, our first mobile controller that was internally developed at Micron.
The combination of these leads to a unique balance of cost and performance that ultimately manifests itself in improved battery life for our customers and for the end users. On the data center SSD side, our second generation 7450 leverages our 176-layer technology and provides unique performance and reliability that meet the really demanding requirements of our advanced data center customers. This is a product that we're very proud of and that we have good feedback from customers, provides industry-leading performance on a number of real-world workloads that are important to those customers. An excellent product here that we'll build on in the future in the data center space based on our leading technology. Finally, I'll highlight our QLC client SSD.
This is the strongest SSD product in the client space at this time, based on a cost structure enabled by our leading NAND 176-layer, and the combination of that with 4-bit per cell QLC technology, ultimately providing the best-in-class client SSD. All this is enabled fundamentally by vertical integration and our end-to-end focus on delivering system-level products on the NAND side that are the most optimized in Micron's history. We've done a significant improvement in terms of our execution of getting from NAND qualification to end product qualification.
As you can see on the graphs on the right side of this slide, in both the client space and the data center space, we have a 30% improvement in the amount of time it's taken us to get from NAND qualification all the way to final product qualification. Finally, I'll conclude today with some key takeaways relative to our technology and product position at Micron. Over a number of years, we've executed through major transitions and with major initiatives and provided solid leadership in positioning Micron for the future. We have the team, the technology, and the products to build on this solid foundation and deliver leadership over the next decade. We truly have momentum for a new era of leadership.
Ladies and gentlemen. Ladies and gentlemen, I made a mistake. We will take a 10-minute break, and when we return, we will be hearing some more presentations from our executives. Thank you.
Ladies and gentlemen, Micron Investor Day 2022 will resume shortly. Please move towards your seats. Thank you. Ladies and gentlemen, please return to your seats. Today's presentation will resume momentarily.
The cooperation between Micron and ASML started three and a half decades ago. At the time, Micron made a bold move of ordering 10 little tools from ASML. On our side, ASML's first year profit after five years of struggling was a fact. This cooperation continues by both sides driving at the lowest manufacturing cost to be successful in each of those markets.
Micron is a leading adopter of our next generation co-optimized and integrated products. They are also forward thinkers when it comes to applying digital tools like our Actionable Insight Accelerator, AIX, to speed up R&D cycles and the transfer of new technologies into volume production.
As technical complexity grows, we believe that collaboration is essential to overcoming challenges and accelerating progress, and that's why we are fully invested in our partnership with Micron. By working together from early R&D through high volume manufacturing, we both gain valuable insight into the solutions needed to push the leading edge in memory technology.
The way the industry delivers this transformative innovation is changing in significant ways. The future is about developing new architectures, new 3D structures, new materials, new ways to shrink and advance packaging, and then ramping these technologies in volume manufacturing as quickly as possible.
Today, as we speak, we are about ready for the next transition to EUV. EUV will drive down complexity, drive down cost, cycle time, and energy use. With that in mind, I think we are facing the next level of our cooperation in the future of EUV and DRAM.
Delivering game-changing technology requires more than just excellence in process and hardware. It requires utilizing high volumes of data from every tool in our fleet. This ensures that Micron is reaching mature yield in a shorter amount of time while ramping to high volume production faster. Together, Micron's fab implementation of AI, combined with Lam's equipment intelligence solutions, are optimizing the fab for maximum productivity.
Well, welcome back from the break, and a big thank you to Gary, Tim, and Martin for those kind words. It's actually absolutely appropriate that that video played in between Scott's section and my section, because collaboration with the equipment vendors is absolutely central to the shared mission of Scott and my organization, which is to deliver industry-leading technology and products at scale. In global operations, we're focused on delivering cost, scale, quality, and cycle time across our end-to-end operations, to provide value for our customers. We do this across the industry's most diversified manufacturing footprint. This manufacturing footprint is a competitive advantage for us. It enables us to have structural flexibility in our supply chain operations that leads to supply chain resiliency for our customers, and that resiliency is absolutely something that our customers have appreciated over the challenging last couple of years.
At the highest levels at our customers, at the CEO levels, we receive feedback that we've been a model partner for them in delivering on their own end market requirements throughout this challenging last couple of years. The diversified footprint is also a competitive advantage because it allows us to tap into the global talent pool for the best and brightest engineers and operators to help us deal with the challenges of being at the leading edge of semiconductor manufacturing. This diversity of thought that comes from having such diverse teams around the world, much like in Scott's organization, absolutely allows us to have the best ideas flourish within manufacturing.
There is no manufacturing process on the planet that is more complex than leading edge semiconductor manufacturing. Thousand process steps, each with their own nanoscale variations and defects that need to be eliminated in order to ramp those technologies to scale. In order to address these challenges, Micron has made significant investments in smart manufacturing technologies to be able to help us ramp these technologies successfully. There are massive amounts of data that are being generated off of, across our manufacturing footprint. With 500,000 sensors deployed, we have a quarter of a billion different process control points across all of our factories, equipment, and facilities. We process 20 million images every week.
All of that coming together in a 30-petabyte global data warehouse that we're mining and extracting for to be able to predict operational challenges and respond to them in real time, and that 30 petabytes growing every single day. Micron began investing in big data infrastructure and end-to-end artificial intelligence and machine learning platforms almost a decade ago. We've been recognized by several third parties, including the World Economic Forum and multiple leading equipment providers and materials providers in the industry as being in the leadership position in implementation of these smart manufacturing techniques and methodologies in the entire semiconductor industry. The results have been remarkable. Faster yield ramps, higher tool productivity, and quality as recognized by our customers at the highest levels in the industry.
Scott talked about how Micron has been able to achieve leadership in DRAM technology and NAND technology at the same time for the first time in our company's history. Absolutely amazing. What's even more amazing is that while we've been able to leapfrog the competition to leadership with our one alpha DRAM and our 176-layer NAND, we've actually been able to enhance our manufacturing capability to implement those technologies even relative to their predecessors. We've been able to accelerate the production ramps of those in terms of yield, 20%-30% faster time to mature yield for those technologies relative to their predecessors. In terms of productivity and high volume ramp, 2-4 quarters faster ramp to high volume production milestones for those two technologies, again, versus their predecessors. Truly generating value out of that leading edge technology and implementing it at scale.
Today, while other competitors are just starting to ramp their equivalent technologies to these, we are mature, in high volume, and as you heard earlier from Sumit, shipping these technologies into a diverse set of end products, into a diverse set of end markets. We're even looking ahead, as you just heard the exciting news from Scott, that we are expecting to ramp our next generations of one beta and 232-layer 3D NAND towards the end of this calendar year.
The collaboration between the technology development organization, the manufacturing organization, and with our equipment vendors to be able to identify equipment variations, the sources of variation in the process and in the equipment early on during the production ramp, to be able to address those sources of variation so the ramp can be smooth and defect-free with high quality and enabling us to qualify quickly with our customers. This is absolutely a competitive advantage, a foundational competitive advantage for Micron, and something that we're going to be leveraging into the future. The semiconductor industry is investing at record levels of capital investment.
At times like these, it's actually even more important that we drive for higher productivity off of both the investments that we already have, the assets we already have in our manufacturing footprint, as well as the new investments we're gonna be making, whether that's new equipment or new facilities that we'll be building for the future. Micron's commitment to smart manufacturing has really enabled us to be able to drive that higher productivity, whether that's tool throughput improvements or increasing the tool uptime through implementation of predictive maintenance to be able to predict when a tool is gonna go down and avoid an unplanned situation, which can be extremely disruptive, and make sure the maintenance is done at the right time. Whether it's optimizing our clean room footprint.
We've been able to avoid more than $6 billion actually, in capital spending across fab equipment, assembly and test equipment, and clean room space in just the last two years. At times like this, when equipment delays or lead times are longer and equipment delays become more common, it's actually even more important to drive productivity of our equipment so that we can keep our production ramps for those leading-edge technologies on track and deliver for our customers. Quality is at the heart of everything we do. Across our end-to-end manufacturing network, Micron focuses on identifying sources of variability, on monitoring for defects, and being able to respond in real time to correct those issues.
In the last three years, we've been able to reduce the amount of non-conforming material in our manufacturing operations by more than 60%. In addition to that, we made significant investments in education and training across our 45,000 team member base on the importance of quality in their everyday activities. I'm really proud of the fact that Micron's rated the number one company in the semiconductor industry in Gartner's survey on the culture of quality across the semiconductor industry and multiple other industries. We're number one in the semiconductor industry and in the top 5% of all companies Gartner surveyed. This index shows measures how employees feel about the commitment to quality and the tools they have to be able to drive and the importance of quality in their everyday activities.
I'm really pleased that we've been able to see this score. It shows that we're not just talking about quality, we're actually walking the talk all the way across our end-to-end operations, whether it's in design operations or in process development or in manufacturing. I think this quote from one of our leading Japanese customers, a tier one supplier to OEM auto manufacturers, sums it up pretty well. "Micron is definitely best in class in quality and an indispensable supplier of DRAM." As technology ramps become more challenging, as our customers' applications become less forgiving, quality, now more than ever, is a differentiator for Micron.
Leading-edge technology, faster yield ramps, higher productivity, improved quality, less scrap, all of this together has enabled Micron to be able to outpace the industry in terms of cost reductions over the last several years. We've closed the gap in DRAM, where we now rank in second place in terms of cost at the yielded die memory level. In NAND, even more impressively, we've gone from fourth to leadership as we first transitioned from 2D NAND to 3D NAND, and then, as Scott described, transitioning to leadership with our 176-layer NAND. We are well positioned to be able to continue to have competitive cost declines in both DRAM and NAND moving forward. In DRAM, on an annual basis, high single-digit % cost decline, and in NAND, the low- to mid-teens %.
Now, keep in mind there are many inflationary headwinds that are going to be impacting everyone across the industry in the next few years, including higher equipment and construction costs for new fabs and new facilities, higher costs for specialty chemicals and specialty materials and gases that we use in the semiconductor manufacturing process, and of course, as all of us know at home as well, rising energy costs. All of these together are going to be headwinds which make it, in addition to focusing at the memory level, we have to focus on cost reduction beyond the memory level.
As Micron has been enhancing our product portfolio, as Sanjay and Sumit described over the last few years, from wafer and components to more system solutions like solid-state drives and DRAM modules as a higher portion of our portfolio, the bill of materials costs become even more important. We have transitioned and will continue to transition more of our portfolio towards our own captive ASICs, which will help reduce our end bill of materials cost for system solutions. We're very focused on scaling and enhancing the cost competitiveness of our assembly and test footprint as well. We have invested over the last several years to be able to increase our captive assembly footprint to greater than 60% of our total requirements.
Now that, combined with the strong relationships we have with our subcontractor partners who support us on the rest, provides us a really good balance of both cost effectiveness as well as the scalability and flexibility to be able to meet our customers' requirements. On the test side, Micron has a unique capability where we design and develop our own high speed, high performance test equipment. This allows us to extend design for test from design to process development, manufacturing, all the way to testing the test equipment that is testing the silicon on the manufacturing floor. Another great example of vertical integration. It allows us to be more cost competitive, to drive faster volume ramps in production, as well as another one of the enablers of having the industry's highest levels of quality. Absolutely a competitive advantage for Micron.
I think across all manufacturing industries, and certainly our industry, inventory management has become increasingly strategic. At Micron, we've taken several decisions over the last several years in order to address some of the challenges. First, I wanna point out that as memory technologies become more complex for everyone, not just for us, as there are more and more process steps as we move from one generation to the next, the cycle time in the fab is getting longer. As Micron has transitioned again towards more system solution products, the cycle time and assembly and test is getting longer. More of our inventory is naturally getting held in work in pr ocess.
Additionally, we've taken the decision to increase our raw materials inventory as a portion of our total inventory to be able to provide consistent supply into our operations and provide continuity of supply for all of that product, all of our production facilities. We've more than doubled the amount of inventory that we hold in raw materials over the last couple of years. Now going in the other direction, we've actually focused very hard on reducing the amount of finished goods that we carry. We've implemented a postponement model, which means we postpone the final configuration of product and delivering it into finished goods inventory, holding it back within our manufacturing operations. In fact, we've cut our finished goods inventory levels by approximately half over the last few years.
What this really has done is allowed us to be much more agile and flexible, to be able to react to the changing mix of demand from our customers and optimize the memory inventory that we have in a much better way. Through that same time as we've dropped the finished goods inventory in half, we've actually raised our ability to deliver to our customers or our customer service level by 20 percentage points in the last three years. Absolutely a mark of supply chain excellence is to be able to lower inventory and increase service level to our customers at the same time. Now, moving from sustaining our business to sustaining our planet.
As Sanjay mentioned, really pleased to be able to have announced earlier this week a commitment to net zero emissions of scope one and scope two emissions from our operations by calendar year 2050. Inclusive of that is a commitment, a milestone commitment along the way that by 2030, we will reduce the amount of scope one emissions, the absolute amount of scope one emissions from our operations by 42%. Now, 42% sounds like a very precise number. It's actually the number that's specified in the Paris Agreement. We are doing our part to collaborate with the international agreement to be able to impact climate change.
We're also pleased to be able to, a few weeks ago, have announced that we've added Malaysia to the list of countries where we'll be able to source 100% of our energy requirements from renewable energy sources, in addition to the United States. We're continuing to work by 2025 with governments and energy providers in all the regions where we operate around the world to grow the pool of renewable energy sources, to be able to continue to achieve toward achieving our goal of net-zero Scope one plus Scope 2 emissions in 2050.
Last year, in our Fast Forward sustainability report, we outlined new goals for water conservation and waste recycling and reuse, and we're making solid progress towards both of those goals by 2030. We also outlined a commitment to spend $1 billion to achieve all of these goals over the next few years. We're well on our way to being able to use those funds in the right way. Just like everything else in the semiconductor industry, it's going to take innovation to be able to achieve these goals. We don't have all the answers today. There will be collaboration required between the semiconductor manufacturers, the equipment industry, the materials industry, energy producers around the world, and governments around the world. We absolutely feel that these commitments are the right commitments to make.
They are necessary commitments for Micron to make for our planet. We absolutely strive to be good corporate citizens and partners in every community where we operate around the world. I wanna, in closing, just reiterate my enthusiasm and my excitement for the opportunities ahead for Micron and the way that global operations is able to contribute and add value to that, to our stakeholders. Focus on cost, scale, quality, cycle time, as well as sustainable operations. I've been in the semiconductor industry, excuse me, I've been in the memory industry for more than 20 years, and I can say there's never been a better time to be in memory, and there's never been a better time to be at Micron. Now I'd like to introduce the new kid on the block, our newest member of our team, Mark Murphy, our Chief Financial Officer.
Thanks, Manish. Good morning. It's great to see everybody in person, and I also extend my welcome to those that are on the webcast. I'll start with maybe acknowledging the great transformation that's occurred by this Micron team over the past five years. As Manish said, this is an exciting point in the industry, and Micron is exceptionally well positioned for profitable growth and capital return. I haven't been here that long, but what I have experienced so far is I've met a lot of people, and I am very impressed by the quality and the tenacity of this Micron team. I'm excited to help contribute to the new era of Micron in the future. Before I get started on the slides, what I'm gonna cover is simply a.
You heard a lot today about the exciting markets ahead of us, and you also heard about Micron's leadership. What I'd like to do is translate how those things lead to sustainable financial results and financial results that we can build on. It's our ambition to seize the growth in front of this company, but do it in a way that balances it with return on capital and cash returns. I, again, I haven't been here that long, but what I'm thrilled about is how quickly I've aligned with this team and how we're all going to, I know, work effectively in doing that. That analysis starts with the industry fundamentals. As you heard from Sanjay and Sumit, we feel good about the industry fundamentals going forward.
Demand for memory and storage is outpacing broader semiconductor growth, and that's with tremendous growth in data center and auto. It's from more intensive memory applications with artificial intelligence. This is a multi-year, multi-decade growth trajectory, and we feel great about Micron's position in serving that. Secondly, the technology, scale, and quality required by customers is growing. This business is getting harder. These tech challenges, and also the focus on ROC, is helping the supply balance, and we expect that to stabilize things going forward. Building on these industry fundamentals is Micron's leadership and the pillars that Sanjay talked about earlier today. Micron is now a technology, product, and manufacturing leader. On technology and products, as you've heard, the company was the first to launch one alpha DRAM and 176-layer NAND, and the company expects to be first on the next node.
This tech leadership supports higher value products and more value per wafer, and that's manifest in a lot of different ways. It's manifest in the agility in being able to target markets that are more attractive, such as auto and data center and others that you heard about. It's less discrete products, more complex products. You know, there are again a number of other ways that it provides you agility when you have this leadership. It allows you to test new models like you heard Sumit talk about earlier today. Then finally, from what you just heard from Manish, the manufacturing leadership the company enjoys with state-of-the-art smart manufacturing and quality, I think all of us can directly see the benefits of that in the financials. Improved yields, less scrap, better inventory management.
I think, importantly, it's a critical part of the customer intimacy that lines up with the product side. Again, it's on these industry fundamentals on which Micron's technology, product, and manufacturing leadership is built that we think produces strong cross-cycle financials and capital return. Now, when the intro video was going and talked about momentum, and Sanjay mentioned momentum, if I look at the slide deck today, this is a gross margin slide on gross margin relative to the industry, excluding Micron. This is a pretty impactful slide on momentum. This highlights Micron's execution and financial progress since 2016. Now, there are a number of factors in this around manufacturing excellence and mix and so forth.
I think what's important, because this is not only was there absolute margin improvement by Micron, this is a relative improvement by Micron to the rest of the industry, a 1,500 basis point improvement. That's a remarkable achievement. You know, all of us, and especially this audience, is forward-looking. I think the most exciting thing is when you look at the fundamentals and you look at Micron's capabilities, those capabilities are self-reinforcing. We think that this margin sustainability is durable, and that's the important thing at this point. With continued leadership, execution, and disciplined investment, we would expect to maintain margin through the cross-cycle at the levels that we're seeing.
Another way to look at the structural improvements of the business is to look at the business cross-cycle financial performance, and that's what this slide's intended to do. It's using a four year average to view cross-cycle trends. As you can see, since 2014, Micron revenues have more than doubled, and free cash flow measures have more than quadrupled. There will be cycles, and, you know, maybe they'll be more muted with the supply balancing that's occurring. In any case, we're confident that cross-cycle, this company is capable of strong profitability. We would also expect strong free cash flow generation to support increasing capital return. More on cash flow. This slide represents fiscal year 2018 through the last reported quarter, the second quarter of 2022.
Over that time, Micron generated $70 billion of operating cash flow before R&D investment. Now, as you can see, Micron spent the vast majority, over $50 billion of that $70 billion, on investing in the business, investing in innovation, investing in leadership for node advancement, and of course, for bit growth. Half of the remaining balance was used to strengthen the balance sheet over the period, which is an important action because it strengthens the ability of the company to sustain investment cross-cycle. This business is about leadership and growing with customers in the right way, and that affords the company the opportunity to do that. With the remainder, and I think this is also an important point, the company picked up its capital return, and that's something that we expect to deliver going forward.
We talked about the majority of the free cash flow of the business being reinvested in the business. Sanjay mentioned this earlier, that reinvestment in the business has created tremendous value, and it's created a hard-to-replicate combination of manufacturing capabilities, vast intellectual property, and human capital, distinctive human capital advantage. We see asset replacement values near $100 billion, which includes about 500,000 wafer starts per month, state-of-the-art capacity. It's very difficult to build, and it's only getting more expensive, as we all know. You all saw the 50,000 patent milestone yesterday, which is a very impressive feat. As impressive as it is, it actually is only part of the picture. Only part of the picture of the vast know-how within the company and the continuous cycles of learning.
I think that came through with, especially with Scott and Manish's presentations. Then finally, the human capital distinctive advantage that Micron has. We're the most diverse memory and storage company in the world. 45,000 employees across 17 countries supporting customers, serving their communities, and advancing Micron. Here's another example of the importance of this investment in the business. Here's what it looks like when it's done the right way, when we deploy capital in an efficient and thoughtful way. You can see what's happened over this period of time, where Micron invested in technology to achieve node leadership and went from a deficit position or a laggard to now a clear leader and a leadership position that the company intends to maintain. At the same time, the company has doubled DRAM bit shipments and tripled NAND bit shipme nts.
I mentioned a part of the cash flow being used to strengthen the balance sheet, and it's this execution of the company and its financial performance that allowed the company to do this. The balance sheet has become a competitive advantage at this point. Gross leverage is under half a turn. It's in a net cash position, $5 billion in net cash. There are no near-term maturities that the company has. The weighted average tenor on the debt is 2031, so very stable balance sheet, investment grade across all three agencies.
This financial flexibility and greater financial flexibility that the company has now versus in the past is a very important point because it allows the company to sustainably and confidently invest in R&D and CapEx to both maintain technology leadership and be able to pursue the right type of products, and also to sustain CapEx for supply assurance for customers and growing in line with the rest of the industry. It's the strong industry fundamentals and the demand profile we see, and the difficulty in gaining supply in the space. It's the strong industry fundamentals, and it's Micron's leadership and our confidence in the ability to sustain that leadership that gives us confidence to introduce a financial model today.
Now, the years will vary on supply-demand factors, but across the cycle, we expect revenue to be in the high single digits, which will be above broader semis. We project the operating margin about 30% as we sustain critical investments in R&D. We still see DRAM being over 70% of our business, but I wanna note that we see very strong synergies between our DRAM business and our NAND business. As you heard several say today, we expect bit growth to be in line with the industry, and through the cycle capital intensity, we expect to be in the mid-30s% of sales. Now I'll leave this chart with free cash flow is a critical measure for the company, and we expect over the cycle 10% free cash flow margin cross-cycle.
In addition to this view in the business, we're pleased to provide you an update on capital allocation. Much of capital allocation's unchanged. The majority of CapEx will be allocated to our internal business priorities, high-priority projects, and to maintain the financial flexibility that we enjoy today and the investment-grade ratings. After those conditions are satisfied and other investments considered, we expect to return all free cash flow cross-cycle to shareholders. As part of that, on this topic, almost skipped the dividend increase. On this topic of returning cash, we are announcing our first dividend increase today. The quarterly dividend will increase by 15%, and annualized, that would be a total of about $500 million of capital return to shareholders. It's steady capital return.
You know, completing the capital allocation framework, I talked through the CapEx. I talked through the importance of maintaining strong liquidity, investment-grade ratings. I also, you know, the core principles, building on those, leadership and sustainability and performance will allow greater capital return. Beyond the dividend increase today, we would expect to increase share repurchase. In fact, I do wanna mention that quarter to date, so this third quarter to date for Micron, we've purchased $700 million of stock, and we will continue to buy at these levels, based on what we see today. In summary, I'm going to end where I started, and that is Micron has undergone a remarkable transition or transformation, I should say, and Micron is now a technology product and manufacturing leader.
We enjoy strong industry fundamentals, we enjoy this leadership, and we're firing on all cylinders, and we believe that will deliver excellent financial results. These financial results allow Micron to invest to sustain that leadership and capitalize on the durable growth trends that we see in the memory and storage businesses. Micron's focus on delivering sustained financial excellence and providing greater capital return. Thank you for joining us today and your interest in Micron. This concludes the presentations. After a short video on our recent patent milestone, we will star t Q&A. Thank you.
All right. As we begin Q&A, please, if you'd like, state your name prior to your question, and please use good judgment as you ask questions to limit the duration of the question.
Then for what? 2-3 years, if I remember correctly. That's what you said. What should we be modeling for cost down over the next 2-3 years before you get to the long term? Kind of related to cost down, how should we think about the back end? Because you only talked about the front end. Thank you.
Sure. Actually, the cost reduction targets that we gave for that next 2 - 3 years, they're inclusive of our projections today of those inflationary pressures that I talked about. Now, basically, the technologies can deliver somewhat better than that, but input costs are going up, which is why we've included those in the high single-digit and low- to mid-teens numbers that we provided. High single digits% for DRAM and low- to mid-teens% for NAND. In terms of the back-end assembly test, I gave some color on how we're, you know, reducing our bill of materials costs and how we're making our assembly and test costs more competitive. At the same time, we're also undergoing a mix shift towards more and more high-value solutions.
There are some areas where costs will be going up to be able to provide. One example is a DDR5 module costs more on an assembly and test and component basis than a DDR4 module. There are going to be some areas, but I think what's important on the assembly and test to note is that we are going to be competitive or even in some areas, best in class going forward, given our mix of strong captive assembly to drive cost leadership as well as our vertical integration to be able to improve the bill of materials for our end products.
Got it. Then for Sumit, on the forward pricing approach that you're taking, what % do you think of the 75% LTAs could get to that? Then can you cite some examples of where that approach has worked with some other companies? Thank you.
Yeah. We just announced our first customer agreement with that principle today. As I mentioned, three year agreement, $500 million, greater than $500 million a year. We are not, due to confidentiality reasons, mentioning the identity of the customer. And we're also not making at this time any projections of how much of our overall business can be done under the scope of these agreements. Because as you can see, it's a very new concept. It'll take time to socialize with customers, get them to assess if this is the right thing for them. We view this as a very important initiative, but one that has to be viewed under the construct of an experiment that we're doing to see if we can get customers to get more comfortable with this.
As more customers do this over time, we can definitely provide appropriate updates at the appropriate time. At this time, we're not making projections.
Thank you. Vivek Arya from Bank of America Securities. Thank you for the presentation. Question for you, Mark. You gave us some kind of cross-cycle metrics. What should we think about the cycle duration? Are you thinking about the same four year type of cycles? Then more importantly, what is the variability of margins in that cycle? Because I think it's more the variability that often right makes investors concerned about the cyclicality of the space. How are you thinking about the variability of gross margins and EBIT margins across what you are defining as a cycle? Then I had a quick follow-up.
It's a good question. You know, I think that we purposely have not tried to define the timeline on the model because it's hard to predict the timing of the cycles and duration of the cycles. I think a four year is a reasonable assumption. We also purposely provide a cross cycle because we acknowledge there could be variability and, you know, we're focused on
How does a business perform over a period of time through cycle? Is it providing a strong financial performance and a predictable financial performance that we can build on? I would leave it at that.
All right. On DDR5, Sanjay, Sumit, you mentioned that you see DDR5 as roughly about 11% or so of the server mix by the end of this year. Has that changed over the last 6-9 months, just given how tight the industry situation? A kind of follow-up to that is, how do you see the price convergence, you know, you mentioned DDR5 trades at a premium, right, sells at a premium. How do you see that price converging over time so you can increase the adoption of that new standard?
Sumit, go ahead.
Yeah. In terms of the 11%, that number has come down over the last year, mainly driven by the timing of the launches of certain industry platforms that better leverage DDR5 in the data center. Again, a lot of the data center adoption of DDR5 will remain dependent on the timing of those launches. As those launches shift, those numbers can shift, which is why we also provided a 2023 number to show that, you know, we feel like half of the DDR5 transition would have occurred, more than half by the end of next year. That seems to be a more, I would say, confident view than the exact timing of the shipments of some of the server platforms.
Having said that, I think in terms of the actual price premium, DDR5 is still at a considerable premium to DDR4. Definitely there are challenges on the availability side for non-memory components that go on these modules, including power management ICs and things of that nature. Also there is general tightness in the overall industry supply chain as well. DDR5 we think is going to retain a decent level of premium for a good period of time. That premium will reduce as the volumes pick up, as they often do in the industry. Keep in mind that customers who are selling these products right now and will continue to do so over the next several months and quarters are not doing that just watching the price premium alone.
They get considerable value by deploying DDR5 because it dramatically improves system performance, and those system gains can be monetized by our customers in very effective ways with their end customers. While that premium exists, you know, we think that, you know, it's also delivering dramatic value gains for our customers.
Thank you. It's Mehdi Hosseini from Susquehanna International Group. Two questions. One for Sumit Sadana. You highligh ted t he 88% CAGR in memory revenue growth for over 8 years, reaching $130-some billion. What is the underlying assumption for bits versus ASP? I have a follow-up.
In terms of the bit CAGR for DRAM, you know, we have mentioned, Sanjay mentioned in his presentation that it is going to be, from our viewpoint, a mid- to high-teens bit CAGR in the industry. Of course, there are mix changes happening in terms of the overall TAM within these mid- to high-teens TAM level bit growth numbers and different segments are growing faster than other segments, and those things do impact pricing. So the pricing aspect is a much more complex picture because like for like pricing is very different from the impact on pricing due to mix. Same thing happens on the cost side as well, of course, you know, cost gets impacted by mix, not just like for like.
That's the number on the DRAM side and on the NAND side, it's high 20s where we look at the bit growth rate of the NAND business. We told you that the overall TAM should grow in that high single-digit % range on a dollar basis.
Great. Then for Sanjay, there's been an ongoing debate as to how much of a ASIC and software IP could memory companies incubate internally. Is there like a rough estimate in terms of like for SSDs? Is there a minimum run number of SSD unit shipment that would justify having an internal ASIC design team and internal software team for to develop the firmware?
You know, certainly, we are well past any such number and because, you know, our volumes in SSDs, client as well as data center are high, and there's tremendous value in having internal ASICs. I mean, they provide, of course, the cost benefit, but more importantly, really enable that vertical integration. Memory is getting more and more complex. The reliability, the endurance, the performance aspects, the device physics, all of this is getting more complex. It really needs features that can be implemented through ASIC and firmware to really ultimately deliver the solutions at the reliability and quality that the customers need, but also with opportunities to differentiate ourselves. That's why ASIC is really, really important. I mean, clearly, we will have a mix of internally developed ASIC as well as external controllers.
You will see Micron delivering more and more value through internally developed controllers and differentiating our roadmap in the years ahead.
Great. Thank you. C.J. Muse with Evercore ISI. I guess first question on DDR5. I think you spoke to a lot of work with the ecosystem. Curious, how does that position you in terms of market share, looking out to that kind of 50% penetration into 2023? And how should we think about the implications to pricing and cost downs, in that world?
In terms of the market share, of course, we are number one right now. We were the first to market with DDR5, and we also got our DDR5 maturity, reliability, and quality to the required levels much ahead of our competitors based on feedback from our customers. We definitely intend to sustain that momentum. Obviously, when DDR5 becomes the majority of the market and ultimately, you know, the penetration becomes very high in all of these large segments, it's not going to be possible for us to sustain that high share because our overall bit share in the industry is, you know, just under the quarter of the industry volume is from Micron, on the DRAM side. It's not really mathematically possible to sustain that kind of share when the overwhelming part of the volume does become DDR5.
Having said that, you know, all of our focus on the product side has been to accelerate time to market, because you can see now DDR5 sells at a considerable premium to DDR4, so we end up having higher profitability. These are the time to market advantages that we are able to bring to our portfolio and our financial performance by capturing those premiums in the time when those premiums exist. Over time, those premiums will come down to similar levels as the cost premium plus some extra premium because it's a better performing product. Those premiums over time will come down, but you know, our goal is always to capture the bulk of those premiums that exist due to our time to market advantage. We are doing that in a very effective way.
Very helpful. Just a quick follow-up on capital intensity, you reiterated the mid thirties. Curious was whether there was any internal debate to change that, considering your growth expectations, as well as rising capital intensity for the equipment industry.
I think it's important for us to understand that over the years, the number can change. You know, sometimes in a certain year, the number may be above mid-30s%, in other years, it may be lower. Of course, it's a function of revenue, industry dynamics, but our best projection and our focus on free cash flow and managing our business in a disciplined fashion is to maintain this number in mid-30s%. Again, some years higher, some years lower. This is, of course, taking into account expectations over time of greenfield facility as well. I wanted to mention that Scott DeBoer, our Executive Vice President of Technology, is available on audio in case there are any questions that you may ha ve for him.
Hey. Hello, guys. Eddy Orabi for Krish Sankar at Cowen. You showed an interesting slide of declining capital intensity for DRAM by nearly 50% over the last decade. I'm curious, because obviously it's different in NAND. I'm curious, when do you think NAND will achieve similar kind of efficiency? Should we expect that like over the next 3, 5 years, or is there a certain timeframe you guys have in mind for the NAND industry? Thank you.
Manish, you want to comment?
Certainly the chart you're referring to is in Sanjay's deck, where I talked about the declining capital intensity relative to EBITDA margin. There's actually multiple factors ongoing in that. There's the improved industry profitability, right, which improves the EBITDA. There is the scaling of Moore's Law, the slowing of Moore's Law, which makes capital investments more expensive, but everyone, you know, invests more responsibly. Then there's also the commitment that the DRAM players have towards shareholder returns. There's actually multiple elements there that, you know, impacted that line improving.
For the NAND industry, you know, there are, you know, still some structural elements that are not quite where the DRAM industry is, and those will, you know, certainly need to be in place in order to have, you know, continued improvement there.
Hi, it's Toshiya Hari fro m Goldman Sachs. Thank you so much for hosting the presentation. I had two questions. The first one for Sumit Sadana. I think in your presentation you talked about your expectation for enterprise SSD costs to be 4x that of HDD by the end of the decade, if I caught that correctly. When you make that statement, I think today it's 8 or 9x. When you make that statement, what kind of cost down assumptions are you making for both your flash business as well as your peers on the HDD side? What kind of penetration rates are you assuming by the end of the decade?
Yeah, that is the regular cost decline that will happen in the NAND business that we have already spoken about today, in various presentations. I think the part that is also important to keep in mind is, when we look at the second half of this decade, we do expect ongoing innovation, to look at different types of media that can, continue to replace, hard drives. Those kinds of media are going to be part of the equation in terms of very cost effective NAND or variants of NAND that can address that opportunity. It is a combination of that. I think the other aspect to keep in mind is, a lot of what is happening on the hard drive side, you know, customers are not able to use all of that capacity very effectively.
Certainly a lot of wastage occurring with those kind of very high capacity drives and definitely huge amount of power consumption, et c. That was what was driving that 4x number. You don't need cost parity in order to make the switch happen. That was what was behind that. Because infrastructures that are built with SSDs are fundamentally more efficient, significantly lower power consumption, far smaller footprint on the data center you can have. You know, SSDs are an order of magnitude more reliable than hard drives. There are all of those advantages which cause customers to switch. We believe that the switching of workloads driving more bits towards SSDs will continue to happen over the next few years.
Later in the decade, based on even lower cost ideas that we are working on, I'm sure others in the industry are working on the NAND side will create new options for customers that will cause that 4x kind of threshold to be hit, and that would escalate pretty significantly the switching of workloads from hard drives to SSDs in the data center.
Great. That's super helpful. As my follow-up question for Sanjay, curious how you're thinking about any industry consolidation on the NAND side. I think it's pretty clear that on the DRAM side you've sort of hit a steady state, very concentrated, and you've got great profitability. On the NAND side, I think most people would agree that in five years' time, the industry is gonna look somewhat different from how it looks today. Do you expect any consolidation to be organic? And, you know, people like yourself with cost competitiveness kind of gaining share over time. Or do you think, you know, there could be M&A in this space? And to the extent you do expect M&A, what kind of role can you play there? Thank you.
I've said before that NAND can certainly use consolidation. It will provide the benefit to customers of greater R&D efficiencies as well as greater scale benefits, providing cost benefits. Certainly NAND industry can use consolidation and become more efficient in serving its customers. We don't really comment on any M&A on part of Micron, so I would refrain from that. Certainly I can say that for Micron, our combination of DRAM and NAND, as Mark had pointed out in his presentation, presents strong synergies as well as enables us to address end market opportunities that require both DRAM and NAND. The combination of having DRAM and NAND for Micron is a competitive advantage. We feel really good about our NAND roadmap ahead as well, as you saw from Scott.
Hi, Joe Moore from Morgan Stanley. Wonder if you could talk about the long-term agreements, and I guess to the extent we've seen those in the past in the more consumer-centric markets, it's been a struggle because the enforceability angle. You know, I could see why this is good for everybody to have the certainty of supply. If they're paying more than a market price in a year, do they try to renegotiate? Just how much conviction do you have that the terms of these can stay steady over the agreement?
Yeah, I mean, these agreements do have volume and pricing and, they are, you know, written contracts. In terms of the enforceability, obviously part of it is having conviction with our customer who is signing the agreement that, you know, we both understand there are going to be ups and downs. One part why we are not committing as to how much, you know, how fast this will grow is because, like you mentioned, one party or the other is always on the wrong side of this market price on any given day or any given quarter or any given year. You just have to have the conviction that over the cycle, the price is fair and that, you know, there'll be ups and there'll be downs, but the benefits ultimately outweigh the risks of actually doing an agreement like this.
On top of that, you know, every time, you know, we want to think of this in three year timeframe type of windows. So if we have the first year go by, we want to add a third year to the agreement again. And again, we have this clear thinking on our side that our goal is not to have that net price that a customer pays cross-cycle to be more advantageous to us or to them. It just needs to be a very close approximation of what both sides believe will occur in the industry. And if the industry curves change over time, you know, we'll obviously adapt to those as these agreements continue.
Part of the whole discussion around the depth and breadth of our, you know, stronger customer relationships relates to this idea that, you know, we are constantly in discussions with them and, you know, think through a lot of these things ahead of time with them. Again, as long as both sides are on the same page that this is not meant to structurally disadvantage either side, I think a good solution can be found.
Morning. Thank you. Harlan Sur from J.P. Morgan. I've actually got a question for Scott. Actually, maybe I'll ask a different question. You articulated a roadmap for NAND that extended 10 years, but on the planar DRAM roadmap, I might have missed it, so I apologize, but how far forward does that extend in years? I think the last note that you had on your slide was on Epsilon. Just want to get a sense for the potential transition to 3D DRAM.
Sure. Hopefully you can hear me. We extend the planar DRAM, and we don't really see.
Thanks, thanks for that. Just maybe one question for Manish. With the adoption of EUV on one gamma, continued integration of more EUV layers, one delta and beyond, I know you said long term, but I just wanna confirm that the manufacturing team can drive high single digits cost per bit declines post-EUV adoption.
Yes. You know, in the long term range that we gave, it does include the initial implementation of EUV in the one gamma. You know, we haven't, as I said, given projections out beyond that. Absolutely, we expect that, as Scott described in his presentation, that EUV adoption by us is actually going to be an enabler of cost reductions in the future, both because of special application that we can have to certain modules that we wouldn't be able to have otherwise, as well as because the industry's more mature, the ecosystem's more mature, the tools that we'll be buying will be more mature, higher throughput, higher availability.
All of those areas together, I mean, you know, give us confidence that we're gonna be able to utilize EUV to be able to maximize cost reductions moving forward.
Thank you.
Thank you for taking the questions. Aaron Rakers with Wells Fargo. I had two as well, 'cause I think I have to. Everybody else is. I guess the first question's on a longer term model, and maybe it's directed towards Manish. You know, when I look at it looks like over the past four years, you've outlined kind of a through cycle model that's consistent with what we've seen over the last four years. Manish, relative to your presentation, you talk a lot about leveraging efficiencies in manufacturing, maybe discrete components going to systems. Do you think that's kind of played out at this point, or is there incremental opportunities in the model to drive more margin leverage? Basically, are you being conservative with the long term through cycle model? I have a follow-up.
You keep looking at me, but he's the one that presented the long-term model.
Yeah.
I'll start answering, and then Mark can finish. You know, look, we are alw ays driving improved productivity and looking at how we can implement those leading-edge technologies. Yields have to ramp faster, deploy them across the portfolio faster, and of course, driving productivity and cost decline or cost improvements in all of our manufacturing operations. You know, the cost declines that I outlined are our best judgment of both what we're gonna be able to implement, including improvements in productivity and ramps of those new technologies, and also our projections for some of the, you know, inflationary headwinds that are going to be there.
I do believe that our, you know, investment in end-to-end smart manufacturing technologies and platforms that allow us to be able to optimize the productivity of the fabs is actually gonna continue to be a differentiated capability for us moving forward, and should enable us to be, you know, cost competitive.
I would just add that, you know, this is the first model that the company's provided. The management team spent a lot of time on it. You're correct that, depending on the periods you use, you can certainly see that it's sort of in line. I would also add that, you know, it's a view in which, you know, the management team will always aspire to do better. It's important to understand that's our view of the business and, with the volume growth and the end market growth that we'll see, we believe that model produces substantial value over time. When combined with the balance sheet where it is, we have more flexibility on capital return than the company had before.
Yeah, I appreciate that. Then the follow-up real quickly is on the NVMe enterprise SSD side, that's a big market opportunity, Micron's relatively small share today. I'm just curious of where do you think we're at in terms of really seeing that hit a true stride from a volume perspective? Where are you at in any kind of qualification cycles? Just curious how we think about the ramp there.
Yeah. We are doing very well with the introduction of our two products that are vertically integrated data center NVMe SSDs. The first one that we launched last year, it's been deployed and in the process of ramping at large hyperscale companies. The rate at which the latest product that we announced in March is getting reception at customers, going through quals, the quality with which we are seeing some of these quals happen, the rate and pace of progress through these quals gives us a lot of optimism that we have a really robust product that customers are eager to get their hands on and deploy in volume. Over the course of the next several quarters, you're going to see an ongoing ramp in our data center NVMe business.
We have already gained share over the last couple of quarters. We expect to continue to gain share in that arena through this year and next year.
Thanks. Thanks for your presentation. Sidney Ho with Deutsche Bank. Just wanna follow up on that forward pricing agreement, that makes a lot of sense. In the past, pricing arrangements didn't really work for the reasons Sumit just talked about. For it to work this time, does it require your competitors to do the same for it to work? Do you have to accept a lower margin for that arrangement, maybe through the cycle? I'll follow up.
We are not intending to have a lower gross margin with these agreements. That's why I spoke about, you know, a fair price in these agreements. A fair price to us means a price that, you know, we feel is likely to be realized on a through cycle basis over time. To the extent that other suppliers go down that path, you know, certainly would be helpful to the industry. You know, like I said, the long-term agreements were something that we did in a very material way and got our revenue share of those agreements from 10% plus to all the way to over 75%.
The rest of the industry followed and ended up also, you know, causing the entire industry structure to move to those long-term agreements, which almost create a lot of momentum at all of the key customers to do very good planning on a by quarter basis of their demand projections, and consequently allow us to do our planning on the supply side. It has been a very positive development. Of course, these forward pricing agreements are very new. We are going to continue to work on them with our customers over all the different segments of the market over the next few years. If this spreads and becomes an industry-wide thing, I think it's just a hugely positive outcome for the whole industry.
Great. Maybe one quick follow-up. This one is for Scott. On the CXL side of things, I know the real opportunity is late 2024 with CXL 2.0. Does Micron benefit mostly from selling more DDR5 memory, or are there new products that you are developing just for the CXL interface? Thanks.
Sure. I'll add a little bit to that, and I'll pass it to Sumit to follow up on the selling side. We're absolutely focused on a variety of different products with CXL, not just DDR5, but certainly on the DRAM side, but also on the future RAM side. As Sumit pointed out, this architecture.
Combination of memory optimized along with controller technology to ultimately wind up with products that meet different kinds of customer needs that we have. That's why both Sumit and I highlighted really the co-invention aspect to CXL, which really opens up some unique opportunities. Maybe, Sumit, you'd add something to that.
I think that was well said. In terms of CXL connected memory, keep in mind this is not your typical DDR connected memory type of characteristic because there will be a controller, there will be different types of latencies involved. On different levels of latency with, you know, for example, traditional memory, DDR4, DDR5, versus emerging memory type of capabilities that can be deployed behind a CXL bus, can provide different levels of cost performance trade-offs to customers. And the kind of controller and varying levels of performance, bandwidth, and latency that these products provide would enable the customers to make this, those trade-offs.
We are in the process of creating a portfolio of products that span the gamut of allowing customers to make those trade-offs, and working closely with them to ensure that our solutions are hitting the sweet spot of their deployment needs as they see the architectures evolving.
Thank you for the presentation. Nik Todorov, Longbow Research. First question for Mark. I think in your prepared remarks, you mentioned a muted cycle expectation going forward. Certainly, I think the current cycle we can characterize as a more muted. I guess, what are the assumptions behind those expectations for a muted cycle going forward? Is that more so on the DRAM side or the NAND side or in general? And do you think that's just a sign of maturing industry? And then I have a follow-up.
I have to check the transcript, but what I believe what I would have said was, there will be cycles, or may be cycles, I should say, supply-demand imbalances. Hopefully, they're more muted given the conditions we talked about on the industry structure, and I think that was the intent.
Okay. As a follow-up for Sumit, a question. On the forward pricing model, how do you think about the trade-offs of market share versus stability of gross margins, particularly on a down cycle, when there's more discounts or competition from peers in the industry?
Sorry, can you repeat that question?
How do you think about the trade-offs of market share versus stability and gross margin with the forward pricing model, particularly in a down cycle environment?
Yeah. You know, there are different moving parts here, and there isn't a single template that will work in these agreements for every customer. A lot of the, you know, by nature, these agreements are going to be sort of customized to each individual customer. Consequently, the way we think about it is, you know, when we step back and look at our portfolio, we are already very focused on moving the portfolio to a place where we have more weight in the portfolio on those markets which are less volatile. The automotive market doesn't have these multi-year pricing agreements, but there is a general agreement between us and our customers about a sort of a price curve that we are going to be on. Consequently, that allows for the automotive margins to be much more stable.
There are similarly, you know, parts of the business where there is, an industry, trend for just, you know, the way these, pricing discussions are done for there to be a lot more stability. I don't want to say independent of what's happening in the market, but, far more sheltered from all of the same level of volatility that you may see in other markets. That's already a move on our part to move more of our revenue into those type of market segments.
As it relates to these type of agreements, these can only help because, you know, so long as the cyclicality is, you know, within expected bounds, which is, you know, what typically tends to happen most of the time, you could get some odd outlier years, like 2018 was, you know, incredibly big on the upside in terms of both pricing and margins. You can always get those kind of outlier years, but the goal is that over the timeframe of the agreement, you know, both sides agree that the pricing that is being realized is fair. If things overshoot one way or the other, there are always ways to compensate for that in the new years that are added to the agreements. It's not a completely, you know, rigid structure. There are things that can be done.
These new agreements have the element of pricing as well as supply volume, both in them.
That's right.
I think we have time for one final question.
Thanks for squeezing me in. Chris Danely from Citi. I guess first one, just a clarification. You know, DRAM upturns and downturns used to be 1-2 years, and over the last, you know, 2-3 years, they've been like these 2 or 3 quarter events. Are you guys saying you expect this to be more the norm going forward, or do you think we will get back to these, you know, 1-2 year upturns and downturns? I just had a quick follow-up based on this.
I would comment on it. You know, given the demand and supply fundamentals that we discussed today, we definitely do expect that the cyclicality in the industry would moderate. Yes, I mean, we would expect that when there is any down cycle in the industry would rebound faster as well. Less volatility certainly, given the supply and demand dynamics that we discussed today.
Okay. As my follow-up, two of the big questions we get from investors are, number one, what do you think the normalized EPS or operating margin can be for Micron? Thanks a lot for throwing that out. Then the other one is, you know, can Micron get back to the previous peak in terms of the 60% gross and 50% op margin? Would it take a 2018-type upturn to get there? Or do you think that if these, I guess, mini cycles persist, that you can get back to those peak margins?
We have projected cross-cycle margins. We are not projecting peak or trough. The point, again, is with less volatility, there will be greater stability in the industry as well. With our execution on technology, product portfolio, and certainly manufacturing excellence, we certainly expect to continue to lead the industry with respect to progress on the profit, profitability front.
Great. Thanks, guys.
All right. Thanks so much. That concludes our Q&A. Sanjay, over to you to close things out before we head over to lunch.
Just want to say that, you know, thank you for being here. I think we presented Micron's strategy for driving next era of leadership, and we are really looking forward to the exciting opportunities ahead, working closely with our customers to leverage our technology, product, and manufacturing excellence, and absolutely produce strong financial results in the process as well. Thank you, and we'll stay engaged.