Welcome. Good morning. Thank you, everybody. We really appreciate your being here with us. But before I say another word, please bear with me as I read the Safe Harbor Statement. Please note that today's discussion will contain forward-looking statements based on management's current assumptions and expectations, which are subject to various risks and uncertainties. These forward-looking statements include expectations for our technology and product portfolio, our business plans and performance, the separation of our flash and HDD businesses, market trends and opportunities, our future financial results, and our financial model objectives. We assume no obligations to update these statements. Please refer to our most recent financial report on Form 10-K and our other filings with the SEC for more information on the risks and uncertainties that could cause actual results to differ materially from expectations. We also make references to non-GAAP financial measures today.
Reconciliations between the non-GAAP and comparable GAAP financial measures are included in written materials posted in the Investor Relations section of our website at investor.wdc.com. Okay, then. On to the agenda. We have two speakers this morning: Western Digital CEO Designate Irving Tan, followed by our CFO, Wissam Jabre. After that, we will have a Q&A session, so our one request is to please hold on to your questions until the Q&A begins. Following that, we're going to have lunch. And one housekeeping note is the exits are clearly marked. In the event of an emergency, please follow instructions from the hotel staff. Without further ado, it is my honor and with great pleasure to introduce and to welcome on stage our CEO Designate, Irving Tan. Irving?
Good morning, everyone, and thank you very much for joining us today. I'm very excited to be able to represent Western Digital to share with you the way forward for us as a standalone HDD company. But before I get into too much detail, I thought it would be very useful for me to first share who the new management team is, so you have a better sense of the people that will be running this business going forward. We've brought together a wonderful team, a wonderful combination of experienced WD executives with deep industry experience, the likes of Scott Davis, Vidya Gurbi, Sinta Teguh, and Don Bennett. And we've also combined that with the injection of new and exciting talent who have been leaders in their respective functions, be it in HR or IT, with individuals like Sesh Tirumala, Katie Watson, and Shantanu Sharma.
You'll see that we have a little gray blob at the bottom right. We're currently in the process of recruiting a new Chief Product Officer that will be responsible for leading our engineering and R&D function. Before I talk a little bit about myself, since it is the first time I'm having the honor of addressing many of you, I would like to take a moment to really thank Wissam Jabre, who has been our CFO for the last three years. Wissam has been a great partner to me and a great leader within the company, and we want to wish him very well in the next phase of his professional career. So thank you, Wissam. Maybe a quick few comments about myself. I'm so excited to be able to lead Western Digital because I am a mechanical engineer by training.
So when you look at what goes on in a hard drive, I always sit there and wonder to myself, this thing should not exist. It's a wonderful combination of electronics, physics, chemistry, material science, tribology, something I actually studied, and I figured out why you actually use it much later in life. But it's a wonderful fusion of different technologies that, as you'll see later, people have always discounted, but it's proven time and time again to be a mainstream of the storage industry now and for many years to come. I've had a career spanning energy with ExxonMobil, management consulting with Kearney, and most recently in technology with Cisco Systems, HP, and obviously, most recently with Western Digital as well. And during that time, I've had the pleasure of being in roles across operations, sales and marketing, corporate planning, M&A, and services as well.
Having also had some board experience, having served on the board of Stanley Black & Decker, and most recently, I chaired the board of SATS Limited out of Singapore, but it's not just about the management team. There are 40,000 team members that we have within the company that wake up each and every day, today and going forward, with a singular shared focus to be able to support the 140-plus markets that we serve our customers in through the 65,000 partners that we have. Developing new innovations every day, leveraging the 4,500 patents that we have, the over seven global R&D centers, and most importantly, which I'm sure why all of you are here today, and Wissam's going to cover a lot of that in this section, about creating long-term value for all our shareholders. It's also important to recognize that Western Digital has had a long history.
There are very few tech companies that have lasted over 50 years. We are one of them. We'll be celebrating our 55th anniversary this year, in fact, a few months from now. But throughout that time, we've been known for being a very ethical company, with good corporate governance and a strong focus on sustainability. As we look forward to being a standalone HDD company, those values and commitments will not change. So you should expect the exact same commitments across those three things from us. So let's transition into how we see the market shaping up into the future. Well, the good news is we are living in an age where data generation is expected to explode. IDC projects data generation between the periods 2023- 2028 to increase threefold.
Data, as we know, is going to get increasingly generated not only by human beings, as it has been done for over the last 100 years, but increasingly, it's going to be done by things, and it's going to be accelerated by the advent of AI that we're starting to see today. This massive opportunity of data growth will create a significant opportunity for data storage, as only 2% of the world's data that is generated is stored today. And we anticipate that number to go up in terms of the data generation to data storage ratio as AI makes data even more valuable going forward. So if you had ever any doubt, we are well and truly living in the zettabyte age. So if we take data generation, how does that translate into our business, the data that gets stored?
The good news is that's also going to continue to grow. Over the 2024- 2028 timeframe, data storage is expected to grow at a CAGR of 17%. But the biggest part of that data storage is going to be in the cloud. With the CAGR in the cloud going to be at 23%, a much faster growth rate. And when we talk about the cloud, we're talking about the core cloud and at the edge as well. So data storage is really about the cloud game. And that's where we really want to focus, where we've been focusing our energies around. And if you see the transformation that's been happening in Western Digital over the last three to four years, we've pivoted from being a bit of data center kind of consumer to being a predominantly data center-focused business.
If I dive deeper, specifically within the cloud space that we are very focused around, you'll see that actually the good news is that all data storage is actually growing. All boats are rising as a result of the increase in data being generated. The fastest growing of which will be enterprise SSDs, as the proliferation of more cloud-centric architectures and workloads that require high input/output, intense workloads such as AI will create more demand for it. But having said that, hard drives will still form the vast majority of the storage media within the cloud. It's estimated that over 80% of storage within the cloud will still be based on hard drives, and it will grow over the 2024-2028 timeframe at a 23% CAGR from an exabyte perspective.
Even tape, something that you have probably put to the very back of your mind, continues to grow, albeit at a slightly slower pace at 11% over that same timeframe as well. So again, it's a wonderful opportunity for everyone in the storage industry. All boats are going to rise. And I know one of the questions that you have on your mind is how much of that growth can be attributed to AI. It's not an exact science, but we took a stab at it to figure out what contribution AI is going to make. So of that 23% CAGR that I highlighted earlier, we anticipate that core growth within the cloud on hard drives will be roughly about 15%. And the additional 8% over that timeframe will be driven by AI, as AI efficiency gains and generative AI applications come to the fore starting in timeframe calendar 2026.
These applications, be it accelerated social media data, digital films, medical imaging that requires a lot more storage, will drive the demand for increased storage going forward. So if you take away one thing from what I just quickly walked through in terms of the market outlook, it is that the market for storage, and especially hard drive storage, is that of one of growth, where hard drives will be the primary beneficiary of. So how are we looking forward to realize this massive opportunity that we have in front of us? We have a clear vision of what we want to achieve, underpinned by a very well-defined strategy with specific strategic pillars.
Our vision is to be the market leader in data storage by delivering storage needs to our customers, not only today, but understanding their needs of the future as well, so that together we can help them unlock the value and power of data that we are starting to see come to the fore, especially in this age of AI. So let me dive deeper into our strategy and how we look forward to achieving this vision. Our strategy has six key pillars to it. First, enhanced customer focus. What does that mean? It means greater customer advocacy, deeper engagement with our customers across multiple functions across both businesses, and striving towards commercial and operational excellence. Second, ensuring we drive product and technology leadership. The ultimate goal is helping our customers to realize the best total cost of ownership that they have through IP-driven leadership and very disciplined product management.
Third, Wissam will cover this in a lot more detail, rigorous financial discipline. And that involves having a clear capital strategy, ambitious financial targets, and prudent capital investment. Fourth, operational excellence, something I know well, having had the honor of leading the operations team for the last three years, is about really driving best-in-class costs to achieve industry-leading margin profiles, coupled with very, very strong execution each and every day. And you'll see more of that later as we go through this presentation. Fifth, around innovation and growth. As a technology company, innovation is at the heart of everything that we do. So we continue to innovate every day. But it's not only about innovating to create new products and applications. It's about also looking at new market opportunities, be it in new geographic opportunities and market adjacencies as well, beyond the device.
So we'll have an opportunity to cover that with you going forward as well. Last but not least, it's about our people. It's about creating high-performance teams with a high culture of performance, making sure that our team members have capabilities and the skill sets to run the business and to plan for tomorrow, not only for today, but ensuring they're capable to meet the needs of the business going forward as well, and developing a strong leadership bench. So let me spend a few moments, if you would indulge me, to dive deeper into each of these six pillars, starting first and foremost with the most important thing, our customers.
We today support a very broad range of customers, be they hyperscalers, some of the most demanding, sophisticated customers on the planet that are using our hard drive technology at a very cutting edge to deliver cloud-based services and increasingly more AI-based services. For them, it's all about getting the best capacity points with the best performance at the lowest total cost of ownership. We also service private cloud and OEM players that are delivering enterprise-grade solutions to the edge. And their focus is really on how do we ensure quality, reliability, and cost competitiveness for them to deliver those solutions to enterprise IT. These two customer segments represent the cloud for us. And just to give you a data point, we estimate by the end of fiscal year 2025, which we're middle of the way through, the cloud will represent 88% of our business.
So we are well and truly a cloud company. We've made that transition. We still have 12% of our business. That's sort of a channel consumer business where we continue to focus on delivering innovative solutions such as video surveillance, NAS, compute, and consumer storage products in partnership through our system integrators, distributors, and retailers around the world. As I mentioned, we're also looking at new growth opportunities. This presents themselves in things like sovereign clouds, where massive data centers are being built around the world. For example, in the Middle East, many large enterprises are also starting to build their own private clouds to be able to enable the delivery of services, especially, for example, in automotive, where autonomous driving will require significant amounts of data that enterprises want to be able to manage by themselves. So the good news is we service all these markets.
Our customers are central to what we do, but we are very focused on the cloud. Equally important, we understand what our cloud customers want. And what they really want is four-fold. First and foremost, ensuring that we deliver to them the best total cost of ownership. Second, how do we ensure that we're providing them with a clear, visible product lifecycle roadmap and the execution to back up that roadmap? I'll spend a few moments later talking about the roadmap that we have laid out. Third, how do we ensure we continue that spirit of partnership and predictability, saying what we do, doing what we say? And that's been something that has been the hallmark of this company for over 50 years.
And last but not least, increasingly the demand for sustainable solutions as energy consumption is increasing as a result of AI, how do we provide them products that have lower energy consumption profiles and can deliver better performance, especially as AI requires higher I/O performance in the data center as well? But don't take it from us. Why don't you hear from our customer? And we have a little video from someone from Dell sharing with us about what they're looking for from their technology partners in storage.
In today's rapidly evolving landscape, innovation is the engine that moves organizations forward. And it starts with people, their ideas, and it's fueled by data. There is little question that we're living in one of the most interesting times in human history. For years, customers have been on a digital transformation journey, and the underpinning of which has always been the data.
It's all about the data. And the advancements that we see in artificial intelligence will not only accelerate the value that the world's data brings to organizations of all sizes, it will also change the architecture of data centers and data flows. Silos of the past will be dismantled, and everything will be connected. Storage systems must innovate to reflect this. And being able to offer customers the best value storage is absolutely critical, which is why hard drive technology, specifically high-capacity hard drive technology, is important for our storage solutions like PowerScale for unstructured data and PowerProtect for data protection. It allows us to offer the right solutions to customers at higher value and is critical to help customers store and protect their most valuable assets, their data.
So hopefully that gave you a good sense or a glimpse into what customers are expecting from us.
We know what their wants are, and that really plays into the unique strengths of us, Western Digital as a company. What are those strengths? Well, first and foremost, our innovation, delivering a TCO benefit to our customers through innovation, be it through capacity points, scalable and reliable products, firmware innovation to continue to improve the system performance and continue to optimize them well beyond the initial insertion into the data center, and AI/ML tools to help them understand potential issues that they might face in their fleet before problems even happen. Next, our products. We have a very robust portfolio of products with a very clear defined product roadmap and execution. Our focus is always on creating no surprises for our customers, delivering them the right products at the right time. This also is driven through a very consultative partnership.
We have some members of the engineering team here to join us today who have spent much of their time engaging with our customers, understanding each and every day how they can improve the performance of their hard drive fleet, not only for today, but to understand the architectural changes that are required to support this new age of AI going forward as well. Third, our operational excellence. We've really moved the needle with our customers to have a much closer joint planning approach to how we see capacity demand needs over a 12-month horizon, then leveraging the vertically integrated supply chain that we have coupled with the global hubs and our distributor network to create a lot more flexibility within our supply capability as demand fluctuates and also better manage our inventory and optimize it going forward.
Last but not least, obviously the hallmark of what we've been delivering all these years, a high trust and dependable relationship, always being there for our customers during their times of needs and supporting their growth capabilities as well. Let me share with you what we are doing specifically. For the largest hyperscale customers that we have, we have pivoted to driving deeper customer engagements for each of these large hyperscale customers. We have created cross-functional teams across sales, product management, engineering to better align to the customer needs and ensuring that all those functions that are assigned to that specific hyperscale customer have common goals and metrics.
This will also enable us to have a much longitudinal type of engagement with the customer, not only supporting them with their technical supply and demand needs here and now, some of their performance requirements in the midterm, but really understanding their future state architecture in terms of what they're going towards and be able to build that back into our innovation roadmaps to better support them going forward. Last but not least, also much more predictable delivery supply planning through our build-to-order execution. We've shifted the needle quite significantly around build-to-order and long-term agreements. Today, close to 80% of the data center business that we have in Western Digital are under LTAs. And these LTAs are over a 12-month horizon. So that's given us a lot more visibility and predictability in the business to better support our customer needs. This brings benefits to both our customers and to ourselves.
It's really a win-win situation for all of us. So now let me transition from talking about our customers to an exciting piece about how we're delivering value to them through our innovation. Hard drives have had a wonderful run, 50 years of nonstop innovation. During that time, they've made significant changes in terms of form factor, in terms of capacity points, in terms of performance and application. I still remember the day not long after joining Western Digital, touring one of our facilities and seeing one of the first generations of hard drives. And my first question to the team was, "Is that a refrigerator?" And jokes aside, that literally was the size of the refrigerator. And then you take a step forward as you see the progression of what we've done. I looked at the iPod mini, if you remember that, right?
It was this small little one-inch thing that I still remember putting it on my sleeve when I went running. That had a hard drive in it, right? But the hard drive industry has fundamentally changed from being a client consumer business, as I've shown you earlier, to being the predominant storage media for the cloud. Today, we are a data center company, as I mentioned, and the capacity points that we have been delivering is well north of 20 terabytes per drive. And as you'll see later, we're going to push that envelope even more. Throughout that period of hard drive evolution, Western Digital has been responsible for many of the innovations that have enabled that consistent transformation of the hard drive industry. Most recently, if you look at the cloud era, we were responsible for introducing helium that had significant performance, capacity, and energy consumption benefits.
We also delivered energy-assisted perpendicular magnetic recording, OptiNAND, and UltraSMR, two very key innovations which I'll talk a bit more about later because that's really key differentiation for us as a company. This innovation that we've delivered, be it in the past or more recently within the cloud, extends to all different parts of the drive itself, from the casing to the media to the actuator to the head, and each has enabled significant steps forward, not only for us, but to the entire HDD industry as well. As you saw earlier, this innovation engine that we have is really built on strong foundations, a very large active patent portfolio, broad R&D capabilities.
We have over seven R&D centers around the world, but we also have 275 PhDs in mechanical engineering, materials science, software engineering, who spend their time each and every day testing, exploring how they can push the boundaries of HDD technology. Beyond those PhDs, we also have a very large group of engineers around the globe who have deep expertise in their respective areas that, as a combination, has enabled us to continue to innovate time and time again. But we're only getting started. So what does the future hold? We have a well-defined roadmap for HDD technology that is aligned with the industry. So let me walk you through left to right. It starts with where we are today in our ePMR portfolio. We've recently launched our 32-terabyte drives. It's undergoing qualification right now. It's expected to finish by the end of calendar Q1 2025.
Post-qualification, we're expecting these drives to ramp up to more than one million units of shipment per quarter for the second half of calendar year 2025. That shows you not only our leadership in capacity points, but our ability to drive volume at scale with the right reliability, with the right yields as well. Currently, two large hyperscale customers are already on UltraSMR at scale, and a third one is ramping as we speak. And in all three cases, these customers are not required to make any compromises in terms of their operating environments within the data center, be it temperature or the operating speeds of their drives. Close to 40% of the nearline capacity that we are shipping right now is UltraSMR-based and continues to grow. Next up, we have our 26 terabyte CMR, 28 terabyte CMR, 36 terabyte UltraSMR.
This is targeted to go to complete qualification sometime in the first half of calendar year 2026. That would mark the end of our roadmap for ePMR. Beyond that, we will transition to HAMR. As you can see, the plan is to transition to HAMR starting at 36 terabytes on the CMR recording technology and 44 terabytes on UltraSMR. We anticipate that we will complete qualification for this HAMR product by the end of calendar year 2026 and then transitioning to ship at volume into the first half of calendar year 2027. It's not that we just started on HAMR today. We have been engaging with our customers on HAMR for quite a while now. And we actually already have two hyperscale customers that are testing our HAMR drives today. But we're not only looking at HAMR.
We're also looking at how do we extend the life of hard drives and continue to be able to push the envelopes of capacity beyond that as well. Our R&D teams have already initiated initial R&D work on the next generation of recording technology called heated-dot magnetic recording or HDMR, for those of you who love acronyms. We have a ton of them that we use within the industry. But it's not just about recording technology. Our innovations also extend to silicon, firmware, and mechanics. And this is, to some degree, I feel very much our secret sauce. It's not just a race for higher capacity points. It's about ensuring that as a system, we're delivering greater capacity points, better TCO, greater reliability, and scalable transitions for our customers. So let me touch a bit on UltraSMR, OptiNAND, and ePMR. Why do I want to do that?
Because this is one of the key differentiators that has given us capacity leadership and reliability in the current portfolio of ePMR that we have today. If you look at ePMR, it's the first shift to energy-assisted recording that's using energy to support recording. That knowledge and learning is extensible into HAMR, which I'll touch about in a moment. OptiNAND and UltraSMR gives us tremendous capacity uplift, roughly about a 20% uplift over conventional magnetic recording. So it's recording type agnostic as well, so it's extensible into HAMR. And it utilizes common hardware, therefore giving us ease of manufacturability and the ease of transition as well.
Ultimately, these two technologies that have been at the forefront of creating our differentiation on the ePMR roadmap today will be extended into HAMR and support the ease of transition to HAMR and ensuring that we are delivering superior quality and reliability when they make that transition. All that sounds nice, technology, innovation, but where the rubber hits the road is what does that mean to the customer? For the customer, it's about delivering improved total cost of ownership, very important to them, rapid adoption. If you look at these two innovations, they've made our qualifications for our ePMR move at a very rapid pace. On average, we're able to move through the qualification process between one to two quarters with very little friction along the way. They obviously get that 20% capacity uplift that supports their total cost of ownership.
And most importantly, a very predictable and reliable product transition. We're not talking about a big leap, a very steady transition that gives them peace of mind. The value to Western Digital, it's obviously very investment efficient, and it provides us with a high degree of yields and reliability that translates into our profitability and to our bottom line as well. So I know many of you here have one thing on your mind. What is WD's HAMR strategy? So let me take a few moments to address that. First and foremost, I would say we have a very clear, well-defined HAMR strategy that we're executing to. We're not chasing the highest capacity point. We're really focused on ensuring that we give our customers a predictable, scalable, economic transition. It's the highest capacity point that can be delivered at scale with the best total cost of ownership ultimately.
Second, we want to be able to leverage the innovations that we have into the HAMR transition. I just highlighted two of them earlier, UltraSMR, ePMR. This will help us shorten the cycles of learning that we have. So when the time comes when we move into volume production, not only can we ramp up very quickly, we can ramp up with the confidence of the quality and reliability of the product as well. Last but not least, what are we doing from our investment standpoint? We've actually started to pivot the vast majority of our investment into HAMR. So let me touch on each of these in a bit more detail. First of all, as I mentioned, customers don't really care about the recording technology. What they really want is the highest capacity point with the best TCO in a predictable, scalable manner.
So our approach is really to deliver that to them. And we see the 40 terabyte point as the economic crossover when the TCO advantage of HAMR will actually be superior to what we have today. Second, we're focused on ensuring that we deliver that predictable transition to HAMR so that it's low risk, low friction. And if you saw the roadmap that I presented earlier, we give them intermediate capacity points in a very smooth cadence to enable to get them to HAMR towards the end of 2026. We're not only looking at innovation and capacity points and being able to make sure the technology works. We're also working very aggressively on making sure our supply chain is competitive in this new HAMR world as well, ensuring that we have manufacturability at scale. What does that mean?
I need to be able, we need to be able to deliver at least one million units of a product every single quarter with the right reliability, the right quality, and right yields that is equal or better than our ePMR products today. Last but not least, we're also focusing on supply chain security, making sure we have the contracts in place to be able to get the right supply at the right volume and the right cost structure to our customers as well so that as we transition, the margin profile of HAMR is at a minimum neutral to accretive going forward. Second, as I talked about, how do we leverage these transitions? The case for UltraSMR is very clear. You get that 20% uplift. It's extensible across ePMR into the HAMR realm as well. But I want to talk a bit about energy assists.
It's sometimes sort of lost in the whole equation. Our experience today with energy assist PMR will help us shorten the learning cycles of what we need to do within HAMR. If we look back when we first introduced helium drives, and also we looked at how enterprise SSDs have been adopted, this adoption typically goes through a four-phase approach. Phase one, which is about evaluation. What does that mean? Can the technology work? Is it viable? The second phase is around qualification. How does it test against the system requirements of the customer? Third, scaling. Can we deliver it, as I mentioned, at scale, i.e., one million units a quarter? Fourth, how do we get it to mainstream? So that's the majority of the products that we are shipping.
Our ability to leverage energy assist into HAMR will help us shorten phase two, the qualification phase, and phase three, the scaling phase, because a lot of that energy assist experience is directly extensible into HAMR. So that gives us a lot more confidence that we can move forward in a very rapid manner once we get into the phase two to three timeframes of what we need to do. As I mentioned, as we go about doing this, we actively engage with our customers to ensure that we are making the right decisions, and they're giving us regular feedback in terms of the adjustments that we need to make as well. Last but not least, the investments that we're making into HAMR. The majority of our R&D and CapEx today has been pivoted into HAMR.
As I mentioned, we already have a joint testing going on with two customers on our HAMR drives as well. So hopefully that gives you a sense of our HAMR strategy. We are confident that it's the right strategy, and it's one that we're executing on and will continue to execute on going forward. But it's not all about the drive. We also need to recognize that the world that we're living in, especially with the explosion of AI, is changing the requirements within the data center as well. The proliferation today of agentic and synthetic AI business models requires the ability to deliver data in massive amounts at very high speed. In order to do that, we actually have been working on delivering purpose-built systems for AI and the cloud.
This will include universal rack scale architecture for AI and solutions that will address storage needs from disk to flash to networking across blocks to files and to objects as well. So if you look at the product on the right, our Ultrastar Data102, which is a three-petabyte form factor dense HDD infrastructure that leverages up to 30-terabyte SMR drives. This will enable customers to store vast amounts of unstructured data, both source and output, for Gen AI, predictive AI, and large language model training. Next, our RapidFlex flash-based systems that leverage our Ethernet-based networking capability provide high-performance capacity flash-based storage up to three petabytes using a 128 TB QLC enterprise SSD. This will help utilize and maximize the performance of GPUs to enable AI workloads. All these products are shipping today. But we're not stopping there. We also have to think about the future.
We've gone from innovating at the drive level to how can we support the changing needs of the customer, especially in the AI era, through systems. We're also thinking about the future, leveraging the core intellectual property that we have as a company focused around servo mechanics, magnetics, and material science and system design. A lot of this is much further out, but I wanted to give you a glimpse into some of the work that we're doing in our R&D groups that's involving these core capabilities that we have and may translate into something interesting into the future. One that we are quite excited about is the opportunity in medtech. How can we use our magnetics experience to really look at magnetic biosensors and nanopores as well?
We're also exploring longevity storage solutions that may be an opportunity to store data from a much longer-term timeframe, but be much more accessible, be it through ceramic storage and to support DNA storage as well. So if you look at it, we're innovating not only for today, but for tomorrow and beyond, fulfilling the vision that I laid out earlier. So I know another important question that you have in your mind is, what is the ongoing competitiveness of HDD within the data center going forward? I.e., will hard drives get replaced by flash as the mainstream storage media? So first, I thought before I dive into it, it's probably good to give you a very simplistic, abstracted view of how data actually flows within a hyperscale data center. This is something that we work together with some of our hyperscale customers.
If you look at it, what does a hyperscaler really do? They are really focused on enabling intelligent data placement so that they can create the optimal economic structure for them with the right performance at the right scale. And in many cases, as I highlighted earlier, a lot of the new AI-centric workloads that require very high IO is starting in enterprise SSDs. But over time, a lot of the data that was warm becomes cool. And what we're seeing from our hyperscale customers is that data gets transferred from eSSDs into the cool storage tier, which is supported by hard drives. Similarly, if you look at the rightmost side of the workloads, which is tape, to some degree, people will say store and forget.
But as data becomes more valuable in the AI age, they're also looking at how can they move some of that data that's dormant on tape back into their cold storage area, into data lakes that they can better leverage that data. So you're also seeing data flowing from tape into cold storage workload areas in hard drives and vice versa. So it's not static. That's the main message. There's this constant flow of data for hyperscalers to better optimize their scale and economics. And this gives us a lot more confidence as to why we feel hard drives will be growing and still be a mainstream media going forward. But at the crux of it is about total cost of ownership. And total cost of ownership has two core elements associated with it.
First, what is the cost of acquisition of the device, the hard drive or the enterprise SSD? Second, the ongoing operational cost, power, real estate, manpower required to support that. So let me dive a bit deeper into it. First and foremost, what does it take? What does it cost to invest to be able to build a drive or the NAND that's required to build an enterprise SSD? So if you look at this table, this is the amount of CapEx that was spent between the periods 2015 to 2024 in NAND and hard drives to deliver 4.6 zettabytes of NAND and 3.8 zettabytes of HDD. You can see that the CapEx required in NAND was quite sizable, $226 billion during that period. If you look at the CapEx for HDD, it was $4.7 billion.
This means the CapEx per exabyte of NAND was $50.7 million versus $1.2 million for hard drives. This translates to a CapEx per exabyte intensity of flash versus HDD of 42%. This is one of the reasons why, while NAND is actually very suited to big IO applications, there are some economics that have to be overcome as well, and if you look beyond just the capital intensity into the emissions intensity as well, semiconductor is a very large energy consumer, and as a result of that, hard drive production is also 7.5 times more efficient from an emissions intensity perspective as well, so that's just one data point, the investment required to actually be able to build those products.
One thing that we are focused relentlessly on is really what we can control, which is how do we continuously drive the cost per terabyte of hard drives down to maintain that competitive advantage versus flash. As you can see from this chart, we have a clear roadmap to maintain a 6x delta on a $ per terabyte basis between hard drives and enterprise SSDs. This ongoing cost reduction is really driven by some of the innovations I highlighted to you earlier, continuing supply chain cost improvements, automation that we're driving relentlessly within our manufacturing capabilities to improve productivity and efficiency, a focus on yield improvements, as well as leveraging AI and ML to better drive the efficiency of our production capability as well. This is about the cost side where we can control. The one piece that we don't have as much control on is obviously ASPs.
But what I'll say on that is, as hard drives are predominantly now a data center play, we have much more consistency in terms of ASPs within the data center market. Enterprise SSDs do have a degree of volatility because NAND is fungible into different markets as well. And it's a much more cyclical business. But the key takeaway is a strong focus on ongoing cost competitiveness versus flash to maintain the viability of hard drives within the data center. So if you bring that all together and you add the acquisition costs and the fully loaded operational costs, how do the TCOs stack up? So we've been working on this analysis with some of our hyperscale customers. And the conclusion of where we got to was that actually hard drives have a 3.6x TCO advantage versus enterprise SSDs.
As a result of that, if you go back to the first slide I shared with you around the movement of the workloads, we feel quite confident that nearline capacity hard drives will remain competitive for the foreseeable future as the primary media for cool workloads within the data center. But don't just take my word for it. Why don't you hear from our customer? Microsoft is a business that's successfully transitioned between a lot of different businesses. When we started Azure, we had six data centers. We have hundreds of data centers. The scale that we operate at is really mind-boggling. We used to deal with entities like as in sets of hardware in the tens and then hundreds. Now we're well into the thousands of sets and millions of servers. We expect that to continue to scale with double-digit growth.
Our current suspicion is that AI is an incredibly large workload. And depending on certain conditions, the amount of capacity required could be incredibly large. And we will struggle to figure out how to run the systems to store it cost-effectively. We run systems that have many different tiers that we use for storage. The bulk of the data remains on hard drives where it can be more cost-effectively retained. We have a substantially large fleet. There's some discussion in the industry about hard drives versus other media classes. I understand the other companies' technologies and the investment in what it takes to get to these types of capability. And I have yet to see any technology that will beat HDD for online storage. So everything I say about the future of storage and HDD, I always put a caveat unless something new comes along.
But my job is to monitor the some news that come along. And I haven't seen anything. So our expectation, barring something new, is that for the next decade, we're going to continue to see this exponential growth in storage needs. And the most efficient media, cost-effective media for storing it online is HDD. So you heard a validation from a very large hyperscale customer of ours in terms of how they see things are going to play out from a competitiveness standpoint between enterprise SSD and HDD. Again, important to remember, all boats rise as data continues to generate and more data gets stored. So I covered four of the six pillars so far of our strategy. And let me sort of bring it home by covering the last two, our operations and our people.
Everything that we do in helping manage the expectations of our customers to making sure that we're delivering impactful innovations is done with rigorous operational excellence at scale, and this really involves us pulling all the levers that we have. First and foremost, the vertical integration of our manufacturing capability from the heads to the media to the drive assembly and tests. This provides us cost competitiveness and control of the value chain and quality as a result of it as well. We've been doing a lot of work around platform commonality, and this is about consolidating multiple capacity points, excuse me, into a single platform, thereby reducing inventory, improving manufacturing efficiency, and productivity as well, and that's already completed.
Next, disciplined CapEx capacity management is about making sure we are prudent in adding new capacity into the marketplace and CapEx investments that's associated with it, really focusing on ROIC-based decision-making. And as mentioned earlier, the majority of CapEx that we're making today is really for technology transitions specifically into HAMR and for the next generation of our ePMR products. Supply chain management. This business is about managing a very complex supply chain and ensuring secure supply is actually very, very critical. We have over 80% of our bill of material in our drives today that are tied to long-term contracts. So that gives us a high degree of supply resiliency. And also, within those contracts, we have improved enhanced inventory management and terms as well. Digital transformation, something I was blown away at what we do within the company.
We have pervasive use of automation and analytics across all our manufacturing capabilities today. 70% of our HDD factories are automated. And many of them have been recognized by the World Economic Forum with Lighthouse Awards for automation and analytics. We're also started to use AI and ML to augment the capabilities of our workforce for faster issue resolution and to be able to raise the skill and capability of our teams as well. And last but not least, quality. Using, again, technology to make sure that we're not focused on quality after the fact, but shifting left to focus on preemptive quality to ensure that we are catching stuff before it even happens.
But bring all of what I described to you to life, supporting our customers, driving our innovation, engaging on our operations is our people, the 40,000 people that we have around the world that come to work each and every day singularly focused on ensuring we're delivering the best technology and delighting our customers. Our teams are spread across 22 countries around the world. And we have this nice mix of deep experience through many tenured employees that we have. We have been in Western Digital for a long time. And they have coupled that with a lot of new college hires that have come in with fresh ideas, new ways of thinking. And that amalgamation is really pushing us forward in terms of what we need to do as a company.
Last but not least, it's led by a very competent, deeply experienced leadership team across many levels within the organization. Whatever I say in this room will not do justice to what they do each and every day. We thought we'd play you a little video that shows how it all comes together.
In an era defined by unpredictability, it's more important than ever to have something dependable to rely on, something steady, something strong, something sure. Every single day at Western Digital, we tell a remarkable story, one of grit, one of precision, and one of best-in-class production that has proven itself time and time again as the premier solution for global enterprise, client, and data centers. That story begins right here in Silicon Valley with the production of our HDD wafers, the foundational material for our HDD components.
Once constructed, these wafers journey to our facilities in Thailand and the Philippines, where they are transformed into critical components, sliders, and head gimbal assemblies. Simultaneously, our sites in Malaysia and China produce substrates and media designed to perfectly complement the capabilities of our heads. Next, these components make their way to our advanced assembly plants in Thailand, where the heads and media are perfectly matched with the utmost precision to transform these components into HDDs that are among the highest performing on the market. With the inclusion of machine learning and AI, our manufacturing process ensures unparalleled reliability and efficiency, exceeding the ever-evolving demands of data storage. After our HDDs are assembled and tested, they move to our precision packing lines and prepare for final delivery. Once packed, our hard drives embark on their final journey with our global logistics network, ensuring timely delivery to millions of customers worldwide.
From factory floors to your hands, we prioritize speed, reliability, and care, delivering innovation wherever possible. Every step of our process is designed with one goal in mind, making Western Digital the most trustworthy, creative, and cutting-edge leader in storage solutions around the globe.
I hope you enjoyed the video. It always gives me goosebumps whenever I watch it because it really brings to life what our wonderful teams around the world do each and every day. But many of you are investors, so you're keen to see what does that all mean financially? It actually translates ultimately into the P&L and into the balance sheet. So with that, I have the great pleasure of bringing up Wissam, who's going to share with you what we're planning to do from a long-term value creation perspective. Wissam.
Thank you, Irving. Hello, everyone. Thanks for joining us today.
You've heard from Irving about our vision and strategy, about our deep customer relationships, as well as the great technology and innovation that our team creates every day, and also about our operations abilities and the great talented team that we have globally. What I'm going to cover over the next few minutes is how all this comes together to create value for our shareholders. So what I plan to do is do a quick look back over the last couple of years and how we navigated them, what this great leadership team, the actions they took to make sure that we position the company for future success, as well as cover the opportunities that lie ahead for Western Digital and the impact of the separation to our balance sheet. It's a great story.
And then, of course, before I close, I will cover probably the part that many of you have been eagerly waiting to hear, our three to five-year financial model and our shareholder-friendly capital allocation framework. So over the last couple of years, we've seen really a couple of major events that impacted our business. One, the cyclical downturn and what this team did over the time period. So this leadership team took great actions to navigate very effectively in the downturn. One, we've resized our fixed cost, as well as looking at our manufacturing footprint, totally rationalized it. By doing that, we managed to lower the fixed cost and enable the company to generate higher gross margins at lower revenue levels. The second thing that we've done is take operating expenses down, which basically allowed us to create a much better operating leverage from the revenue that will be generated.
By doing that, when you look at the last 12 months ending fiscal Q2 2025, we generated $8.4 billion of revenue and $3 billion of gross profit. When you compare that to fiscal 2022, where we generated $9 billion of revenue and $2.7 billion of gross profit, it just shows the beginning of the sort of the increased earning power of the company. In parallel, obviously, in the last 12 months, we crossed the 36% gross margin. And in fact, if we look at the first half of fiscal 2025, we're very much crossing the 38% mark already. The second major event that we saw over the last couple of years is a big shift in the product mix that we ship. We went from shipping higher volume of lower value products to shipping a lower volume of higher value products.
So our average selling price per unit increased quite a bit, which also gives us a good tailwind for gross profit and gross margin generation. So as you can see, we're emerging as a much stronger company with higher earnings power, but also a much higher ability to generate free cash flow. Coming out of the downturn, we've improved our free cash flow dramatically. Now, the few things that we focused on during that time are exactly the same that will help us accelerate that free cash flow generation over the next few years. So for one, it's our technology. It's our strong portfolio, our ability to continue to deliver best-in-class TCO for our customers, great value for them, which also translates as great value for us. Two, it's our focus on profitability, the discipline and execution.
You saw in the previous slide how we focused on the fixed cost from a manufacturing perspective, but also the operating expenses, keeping things completely tight as we develop great products and continue to invest in the business. Three, it's the disciplined capital investment with implementations of things like build-to-order, for example, which give us great visibility going forward. It allows us to not only focus on our cost structure, but also optimize our supply chain. And that translates in better cash flow generation. And finally, the blocking and tackling actions that we take every day to continue to improve our working capital and become much more efficient at turning our dollars and making sure that our cash conversion cycle keeps shrinking. So now I'm going to move to talk a little bit about the opportunities ahead.
Over the last few years, we also have seen some transformation in the composition of our revenue. In fiscal 2019, you could see we had a much bigger portion of the revenue shipped into consumer and client markets. And over the last few years, we saw that continuously sort of decline to sort of become a little bit more stable and being much more and our business being much more exposed to the cloud. Now, what's interesting about that is, obviously, the cloud is a large market, and it's a growing market. So in 2025, for instance, we're projecting to generate around 88% of our revenue from the cloud. And when you look at the nearline market and the growth of nearline from calendar 2024 to calendar 2028, excluding the effect of the AI adoption, we see that growth at around 6% compounded annual growth rate.
If you factor in the AI impact, that growth more than doubles to 14%. So this is really a growing market. It's a large market, and it's also a profitable market. What also this translates to, if you sort of want to take a deeper look into nearline and the markets that we serve from a cloud perspective, it translates to basically our business moving from being, as I said earlier, focused on three market segments to really being driven by nearline. So the nearline market and the trend of that nearline market and the growth of that nearline market, whether on the exabyte side or the dollar side, is what's going to define the revenue going forward for the hard drive business. And so, as you can see from the charts, the projections are very healthy. Exabytes are growing at around 23% CAGR.
The TAM in dollars is growing at around 14% CAGR if you factor in the impact of AI. So all in all, it's really a great story of the hard drive industry returning to growth after a few years of working down the consumer and client effects. Going forward, that portion of the revenue, consumer and client, becomes smaller, more stable, maybe declining at a slower pace. But the more important point here is it's nearline and the cloud that's going to define the revenue growth going forward. Before I go into the balance sheet and talk about the impact of the separation to our leveraging or de-leveraging story, so I'm just going to recap a couple of things. The last few years, we've created a much stronger company.
We also created a company that has higher earnings power and higher free cash flow generation abilities that is exposed to a profitable market that is large and growing. Okay, let's shift gears to talk a little bit about the de-leveraging story and the impact of the separation. So over the next 12 months, we're going to see some changes, actually major changes to the balance sheet. At the end of Q2 fiscal 2025, our net debt position is around $5.1 billion. In the next 12 months, which sort of really start with the separation, we're going to see a couple of step-downs. The first step-down will come from the separation itself. The proceeds that WD will get from the SanDisk transaction, which are around $600 million, will be a first step to reducing debt. The second step will come from the disposition of the 19.9% retained stake in SanDisk.
And that has to be done over the next 12 months to sort of qualify as non-taxable. And the way it's done is typically as an equity for debt exchange. So we have a couple of opportunities to work down the debt to de-lever the balance sheet effectively. And then, of course, as I mentioned earlier, our higher free cash flow generation abilities will help us to continue to voluntarily pay down debt over time until we reach our target net leverage, which takes me to the three to five-year financial model. The way to think of our financial models is we're modeling a revenue growth at industry compounded annual growth rates of mid- to high single-digit %. The way to think of it is, obviously, these numbers, of course, depend on what base year we're looking at.
If we're looking at fiscal 2024, those percentages could be higher than that. But if we're looking at the calendar year 2024 as a base year, then sort of the mid- to high-single-digit % is a reasonable number to look at. What's important, though, to note here is the way we think of our model is maintaining market share and growing with the industry and really focusing on our profitability and cash flow generation. So that's, I would say, the key probably takeaway from how we think of our revenue growth. And then next, with the focus on profitability, our non-GAAP gross margin is modeled at 38% or greater, so greater or equal to 38%. And as I mentioned a few minutes earlier, we just hit the 38% in the first half of fiscal 2025.
And sort of if you compare it to the previous model of 31%-34%, we're pretty much setting the floor seven points higher, which basically gives us the confidence that what we've achieved over the last year plus is very much here to stay. And what gives us even more confidence about being able to drive that gross margin from the 38% better is a few things that Irving mentioned. Obviously, our great technology and product portfolio that offers our customers great TCO, great value, and it's great value for them, but it's also great value for us. And then the focus and the relentless focus on operational excellence, on driving costs down, keeping very disciplined execution that will allow us to continue to maintain that sort of $ per terabyte cost declining at a very healthy pace.
So that should help us drive non-GAAP gross margins of 38% or higher. The next thing that would translate to on the operating margin is a non-GAAP operating margin of 24% or higher. So when you look at sort of the implied operating expenses, that says we're at around, let's say, 14% of revenue. But the way to think of it is this is not a fixed number. As revenue obviously grows, we don't expect operating expenses to grow at the same pace. And so we should see some of that operating leverage in the business translate to even better non-GAAP operating margin. On the CapEx front, we're still targeting 4%-6%. We think this is a good range for us to continue to invest in our capacity and our ability to transition to HAMR when HAMR comes.
That’s that we have ample room to continue to invest in the business, continue to sort of drive growth there. And lastly, we’re introducing a new metric, which is capital return. And here we intend to return 100% of our excess cash to our shareholders, of course, after achieving the net target leverage that I’m going to talk about next. Our shareholder-friendly capital allocation is based on three priorities that are very clear and simple. One is reinvesting in the business, continuing to develop the great technology and innovation that drives the differentiation as well, drives the great earnings and cash flow generation. Two, continue to invest in the growth while maintaining a disciplined approach to our capital investment. The second priority is reducing debt through voluntary debt repayments.
I talked about the couple of opportunities and the couple of steps we should expect in the next 12 months. And obviously, as cash flow generation accelerates, we will continue to voluntarily reduce the debt until we get to that target net leverage ratio of one to one and a half times. And then three is our commitment to returning capital to our shareholders. And we intend to initiate dividend in fiscal Q4 of 2025, which is really the next quarter. We will be talking more about the details of that as we get there. But the way to think of it is we plan to start small and grow over time.
Then finally, as we reach our target net leverage ratio of one to one and a half times, we would expect to return 100% of the excess cash to our shareholders in the form of either growing dividends or share repurchases, which basically creates a great value for our shareholders in the long run. Before I close, the future of Western Digital is one of great story of value creation. We have a portfolio that is extremely compelling. The company is very well positioned to continue to innovate in a large, growing, and profitable market. We have a leadership team that has a proven track record of strong execution, as well as creating higher earnings power and increasing the capabilities of generation of cash. We're also committed to driving long-term shareholder value with growing capital returns over time.
So with that, I will hand it over to Irving for some closing remarks, and then we'll move to Q&A. Irving. Thank you.
Thank you.
It's hard to come up and beat what Wissam just shared with you. But just to reiterate, as a business, what do we want to stand for? What is our commitment to our customers, to our shareholders? It's really around ensuring we have a laser-like focus on our customers, helping them on the journey, especially with our large cloud customers. A strong focus on innovation. That is at the heart of what we do. Not only ensuring we're just innovating for innovation's sake, but being able to translate that into the right economics for our customers that's reliable and scalable for them as well. Continuing to have a very strong and intense focus on our operational excellence, be it in our strength of execution and in creating flexibility within the supply chain as well.
Having a committed, experienced and leadership team that's going to be that you've seen help navigate and restructure the business over the last 12 to 24 months. Most importantly, translating all that into the long-term value creation for all our shareholders as well. Thank you. With that, I'm going to invite Peter Andrew, our VP for Investor Relations, to join Wissam and I up on stage as we host a Q&A session for you.
So before we get started, I just want to remind everyone that this is also being webcast. So please wait. We have three mic runners around the room. Please wait for the mic to come to you before asking your question, just so everyone on the web can also get it. And the mic runners are you want to bring it up here? One in the center. Aaron, do you want to take the first one?
Yeah, thank you. Thank you. Aaron Rakers at Wells Fargo. Sorry. Aaron Rakers at Wells Fargo. Thanks for doing it today. Appreciate all the detail. I guess the first question, Wissam, I want to go to you. And yesterday, your SanDisk presented and they talked about the inputs of the P&L for the fiscal third quarter. I'm curious, can you give us an update of how we're thinking about standalone Western Digital and maybe walk us through some of the inputs there, if it's revenue, if it's gross margin assumptions, anything you're able to share? And then I'll ask my second question right away. When I think about your growth rate that you outlined in your long-term model, and I juxtapose that relative to the nearline capacity shipment CAGR you're talking about, to me, it just looks optically as maybe conservative.
Maybe you can tie in that discussion of how you see ASP trends in that scenario, assuming your revenue growth when we look at that capacity shipped number and now close to 90% of your business being driven by that nearline cloud business? Thank you.
Thanks, Aaron. So let me talk about Q3. I'm going to refer back to the guidance that we've provided at the earnings call. So the way to think of the hard drive side of the business and sort of what Western Digital go forward for Q3, the revenue guide translates to $2.2-$2.3 billion of revenue. That sort of gets us there. On the gross margin side, we did say that sequentially we expect 50 basis points growth relative to the second quarter. So that still holds. OpEx, it's when you look at a total company in what we guided at earnings, the split is around 45%, 55%, 45% being hard drive, 55% being flash. And then maybe I'll offer one more thing in the sort of the below operating income lines. OI&E, I think we had guided around 100.
I would say it translates around $90 million for the hard drive business, so this is how we think of or this is how sort of the numbers that we guided imply for the hard drive business. Now, on to your second question about the future, sort of when we look at the growth, and look, the key takeaway from the revenue growth is the model and the way we plan to run the business is to grow with the market, so maintaining market share and growing at the speed and the sort of growth rate of the industry, and so if that over time translates to a higher number, that's great. If it's going to be a lower number, obviously, we're going to have to operate within those parameters, and so that's the way to think of it.
Now, we said it's mid- to single-high-digit percentage, but that also depends on the base year. Because fiscal 2024, for instance, was a much lower year. So if you translate, if you sort of take fiscal 2024, then you're going to end up with a much higher number. But if you take calendar year 2024, that mid- to single-high-digit percentage translates well. And then to the last part of your question, which is around ASPs, we don't typically project ASPs and we don't talk about ASPs in the future. And so probably I'll make some comments, and I'm not sure, Irving, if you want to chime in. The way we think of the market and the business going forward, the volumes aren't going to change dramatically. The volumes will probably move, let's say, flat to up.
And so a lot of the improvement could be coming from better average selling prices because we will be selling bigger capacity units that sort of translates in an average selling price.
No, I think you covered it very well, Wissam.
Thank you.
Okay.
Asiya, I think you got the mic. Do you want to take the next one?
Great. Thank you, Asiya.
And then could we get another mic runner over here in the front?
Asiya from Citi Research. Just given your largest competitor is on this race to HAMR, how much confidence do you have that in this timeline for your roadmap for HAMR to be in calendar 27, if I heard that right? How much confidence that you have that you're not going to be losing market share here? I think that seems to be the biggest question investors have that market share is not at loss. And I see the PMR and the UltraSMR roadmap that you have. But just what kind of confidence that you have that you're not going to be losing market share in this time period while you ramp your HAMR technology?
Asiya, thanks for the question. I think just to recap, we look to complete qualification from HAMR products at the end of calendar 2026 and then ramping up at volume in the first half of the year 2027 timeframe. As I mentioned earlier, we have a lot of confidence in our strategy because ultimately what we are providing our customers with a very clear low risk frictionless path to be able to get to the economic benefit of HAMR in that time through that Ultra SMR, ePMR roadmap that I shared with earlier. But it's not only just about capacity points. I think what we really wanted to highlight was being able to deliver the best total cost of ownership, not only at the right highest capacity points, but through the reliability and the scalability as well.
That's why we focus a lot on our ability to deliver at least one million units a quarter consistently with the right yields, with the right reliability. The other point to your question about market share, if we look back to our helium experience when we were first into the market with helium a few years ago, what we saw that actually when customers, even as they started to adopt it and move into phase four, which is the mainstream phase, they never went beyond 15% market share. They actually sort of gave you that head start, and then they would sort of cap the market share to create a second player to come into the market. There was surety of supply within the marketplace. The main thing is we have confidence in our strategy. We feel it's the right one for our customers.
We are continuing to work closely with them, share with them our progress, testing with them the timeframe, and the feedback given to us is we're on the right track.
Okay. Why don't we go to Wamsi on this side of the room, and then can we get a mic over for Karl on the left after?
Thank you so much for doing the presentation. Wamsi Mohan, Bank of America. I was curious to get your thoughts around two things. First, on the growth rate differential that you have between baseline storage and AI applications, the incremental growth is actually larger coming from the AI opportunity over the next several years. So can you maybe just give us anecdotally what you're seeing currently at some of these AI instances, if you're seeing the kind of product uptake there is, what you're expecting specifically will be the type of workloads, or maybe is it very heavily inference-driven at the edge, kind of any color you can share around that AI opportunity if you could break that down?
And then secondarily, just to follow up on the HAMR point that you made about a million units is where you want to be, at least per quarter, to be able to make this economically viable. Would you say that the HAMR margins, as you're going through that ramp, would be accretive to overall HDD margins given what the capacity points we are starting at, or is there a transition time as you're managing sort of the two portfolios and scaling if there's a margin impact? Thank you.
Thanks for the question, Wamsi. Maybe I'll answer your second one first, and I'll come back to the first one. Yeah, I mean, as I highlighted in the presentation, our focus is to ensure that we make that transition point to HAMR, which we've indicated where we feel the economic crossover from a TCO perspective. It's at roughly 40 terabyte capacity point, is to ensure that we not only have the capacity point, but we have the reliability and the yields within the operations and the cost structure to make sure that when we make that transition from our financial perspective, it's neutral to accretive. And so that's why that sort of consistent step change in terms of the roadmap we deliver will enable us to deliver that consistency of margin equation going forward. That's on the second piece.
In regards to GenAI, we're actually starting to see some of that benefit even today. I think the social media companies, it's less about GenAI, what they're selling as a service for GenAI, but where the immediate use cases we are seeing is them using GenAI within their own operations itself. It's actually helping, for example, in social media and advertising, creating much more targeted capability and driving better performance and yields in terms of what they are seeing in terms of their advertising and being much more targeted to the consumers that they are supporting. So this is where we see the early stage, less about specific application, but how they're using it within their existing environment as well, and as we project into the future, and you see, as I mentioned, you start to be a bit more material in 2026, scaling up into the 2028 timeframe.
That's really when video comes to the fore. Today, a lot of AI is still around text. And if you've seen what OpenAI has done with Sora, the ability to convert text into a video, if you extrapolate that going forward, the ability to convert entire scripts into movie-length video, that's one of the big drivers that we see that's going to happen along with some of the medical imaging and diagnostics capability as well. And as customers are developing new models and the associated data, they're actually storing each of those iterations of the models and the associated data onto our platforms as well. So that's also driving the growth that we see.
Okay, we've got Karl next, but anyone else over here? Can we get a mic? Karl, do you want to go?
Yes, thanks. Thanks, everyone. Karl Ackerman from BNP Paribas. I was hoping you could address where do you think the mix of your nearline business on long-term contract agreements goes next year from the 80% today? And as you address that question, do you believe LTAs are valuable in market downturns too, or are they perhaps most beneficial during market tightness in upcycles?
Do you want me to answer that one?
Well, 80% is a really nice number. I would say a lot of that's driven by the large hyperscale customers that we have. And I think what we've seen is definitely an attitude where they really want to work collaboratively with us. They do not want to be in a situation that we both had to go through 12-18 months ago. And again, our LTAs, as I mentioned, are a combination of volume and price commitments. So in that sense, we actually view it positively because it gives us better visibility. It helps us to better plan our supply chain and to be able to make the right CapEx decisions going forward. Because I think one of the things that the industry has done in the past is to overinvest and then get caught up.
I think this will help us with the planning. Most importantly, we will continue to work very closely with our customers, Scott Davis, who runs our sales and marketing organization, has a wonderful team of salespeople who are working very closely with our customers every day, understanding the changes that they're seeing within their environment, and will support them as they have to make those adjustments as well.
Okay. We got CJ over here. I guess, do you want to go in right there?
Yeah, good morning. Thank you for hosting. My name is C.J. Muse with Cantor Fitzgerald. I guess, Irving, a question for you, and then Wissam, a follow-up. In terms of planned supply, can you speak to how you're going to hopefully add measured capacity just to follow on Karl's question in terms of kind of sustaining these long-term agreements with your customers? Would love to hear more about how you plan to meet growth, but also make sure that you stay under supply or balanced. And then Wissam, essentially the target model that you've provided suggests roughly $5 free cash flow per share, 18% at a minimum. And so once you do get to that kind of net leverage ratio that you'd like, what would you anticipate kind of the percentage being for dividend versus buybacks? Thanks so much.
Thanks for the question, CJ. It goes back to, again, first and foremost, that LTA visibility that we have really helps us to plan long-term. And as you saw the slides from Wissam, we've gone through quite a significant transformation where our business today is predominantly nearline. So it's less about a volume of units game. We see that the number of units will be growing at a relatively low rate, low single-digit figure going forward. But it's really the capacity points that we're able to deliver into each drive. I mean, if I go back three years ago when I started within the company, the average capacity into the data center was roughly about 18 terabytes. Today it's roughly about 24 terabytes. So we're able to add a significant amount of exabyte capacity into the market without actually increasing the number of units that we're producing.
Hence why I mentioned our focus around any CapEx investment is really to support technology transitions and to invest into HAMR, not so much into increasing the volume of capacity of the units that we're producing.
Thanks, Irving. And to the second part of your question, CJ, look, I don't want to get into the numbers on free cash flow percentage of revenue and so on, but obviously it's a big focus because of the accelerating nature that I described in the prepared remarks. The way to think of sort of the capital return, the dividend initiation that's happening in the next quarter is sort of the beginning. And as I said, we're going to start small and grow over time. And so over time, meaning over a few years. And I mean, the company had at one point a great capital return program. So that over time, again, three plus years out, you could probably look at that as a good indication or an initial indication.
But the point is the plan, once we get to that target net leverage ratio, is to return 100% of the excess cash to shareholders because we think it drives a great value for our shareholders and also helps with, obviously, driving that sort of long-term value that we are very much focused on. So I don't know if this gives you a bit of unsatisfactory answer, but at least it sort of gives you something to start looking at.
Okay, we got Krish over here. Then can we go in the back?
Yeah, hi. It's Krish Sankar from TD Cowen. And thanks for the meeting. And I just had two quick questions on HAMR for Irving. Number one is just a clarification. If the end of 2026 HAMR qual is that for one cloud or multiple cloud customers? And number two, when you look at your competition, the timing they gave for HAMR to the actual qual was actually much more longer. And it seems like failure rate was one of the biggest issues. So I'm just kind of curious your confidence on the HAMR qual timing and how do you look at the failure rate and any lessons learned from them. Thank you.
Thanks for the question. So first and foremost, as I highlighted, we actually have HAMR drives being tested within two hyperscaler customer environments today as we speak. And that testing will give us a lot of feedback in terms of how we make sure we accelerate the development process and the focus around reliability and quality and performance of the drives as well. In terms of the number of hyperscalers that we get to, I mean, that's really the transition plans that we work with each customer, and they'll be unique to each customer depending on where they are on that journey. So we'll be able to share more of that with you as we get further into that timeframe. You're absolutely right in the sense that we're very focused on making sure that this transition is one of confidence for our customers.
That makes economic sense for us as well, and that's why we spend a lot of emphasis on the reliability of the product, the yields of the product. The focus for us is not only in just delivering the best capacity points, but the right TCO as well. Could we produce a 40 terabyte HAMR drive today? Yeah, we could. Could we do that at the right scale with the right yields? Does it make economic sense? The answer is probably not, and that's why we have the roadmap that we have, where we feel that going through the ePMR transition up to the 36 terabyte capacity point on Ultra SMR is the right play before we make the transition to HAMR going forward.
Let's go over here on the right.
Yes, it's Mehdi Hosseini, Susquehanna. Two follow-up questions. One is diving into the gross margin, sorry, 38% or higher. How should I think about unit economics? As Wissam, you highlighted HDD units probably not going to grow as much. So as we think about the next couple of years, how does the system cost or unit cost change, especially with HAMR added? And obviously, with increased content per system, that's also going to help you with ASP. But in terms of the units, how does the unit ASP and the unit cost trend? And is that difference what's going to give you that gross margin expansion? And I have a follow-up.
So, look, I don't want to get into the details that way out. I don't think I'm as good as projecting a number of units. But the way to think of it is really for the gross margin to work effectively. And we're still anticipating, obviously, continuing improvement in gross margin that's gradual over time. So 38% is very much for our model three to five years out. We think this is the floor. The way to think of it is with the great development of products that will drive TCO to our customers that will allow us to keep more value for us as well as provide great value to our customers. But also the dollar per terabyte cost reduction will continue to be on a good path down.
So the cost curve that we've seen over the last few years should continue to sort of on the same trend down. And so between these two, you'd expect to see gross margin improvement. Now, if you sort of want to model a little bit more in detail how this could work, obviously, volume will probably increase a little bit from here, but not dramatically. And I mean, we showed at least the data that the industry analysts are forecasting for exabyte growth and revenue. And so if you have to make some assumptions on where the capacity points will be over the next few years, that will give you a good idea of what the average selling price per unit would do.
Because you have done all the resizing. So the reason I asked the question is your footprint has reduced, so has your competitor done. There will be some retooling required by HAMR, but we really don't know to what extent. So that content growth or capacity per system growth becomes very incremental in driving both system ASP and gross margin expansion. That's what I was trying to figure out.
Yeah, and it's a great question, Mehdi. This is why we keep talking about the economics, and we keep talking about our focus on the economics for us and for our customers as to at what point or at what capacity point we think that sort of break-even plus that makes sense, which is what we've always talked about around that 40 terabyte point, and that's where we think it'll make it more neutral to accretive from a margin perspective as we introduce HAMR.
Just one quick question. How should we think about the number of plates per system in HAMR compared to the existing technology? How does that change?
The number of plates? I don't think we're getting into that. The platters.
Yeah, competitor has referenced that level of details. That's why I ask.
I think with the recent product that we just launched, we've moved to 11 disks. Our goal is obviously to improve areal density of the drive, so the more we can improve areal density, the less the number of platters that we will require. So there's still a lot of work that we're doing in that. The more we can reduce the number of platters, that makes a lot of sense, so we can push the areal density capacity a lot more. So it's something that we will continue to work on as we move forward.
Let's go on the left. Anyone else here? Do you want to go up front, Eric? Let's go.
Matt Bryson with Wedbush Securities. Just two questions, one structural, one short term, if you will. Structurally, I just pulled up a Gartner report from 2019. It talked about nearline exabyte CAGR of 39%. I think you had the IDC DataSphere consistently talking about 30% plus data growth. Clearly, AI is a significant driver of exabytes. So I guess, what changed structurally that we should only be expecting 23% exabyte CAGR? Or are we going to wake up at some point and find out that the industry is underestimating what exabyte growth is? And then the near-term question is, after you finished your report and both Seagate and Western Digital had talked to decline in nearline shipments, Q on Q, number one question I got from investors is, doesn't that mean we're at the end of the cycle?
Obviously, that's not what you're thinking, but would love to hear you articulate again why that's not the case and why demand remains strong and will remain strong moving forward. Thank you.
Maybe I can take the first question. First, and Wissam handles that. Look, I think when it comes to AI, even as we highlighted that the cloud growth will be 15% with the additional 8% potentially driven by AI, it's still very early days. We're quite uncertain on how it's going to play out. The confidence that we have in the projection that we gave was a bit to Wamsi's question earlier, where we're seeing AI have immediate value in terms of existing operations to customers today, where they can see that through AI, they deliver better targeting, they're getting better yields on advertising. We haven't yet begun to factor new applications and new workloads that may be created going forward. We clearly have some visibility on the models that are being created, especially in medical diagnostics and imaging, how those models and the associated data need to be saved.
It's more driven by the line of sight of what we think can happen. Is there upside to that? Potentially, yes. But we'll have to see how it plays out because it's still early days and it's a market that has a tremendous amount of potential but needs to be able to deliver to it.
Yeah, and to the second part of your question, Matt, with respect to the Q3 revenue and how we sort of are seeing a small dip and we anticipate things to improve from here. Look, we started talking about this in October. In the October call, we did say that we anticipate the next quarter to be a little bit down for hard drive, but we see it as a one-quarter event. We said the same thing in January, and we still think this is a one-quarter event. And the reason, and I don't want to be sitting here to guide the rest of the calendar year, but the reason I feel confident that this is a one-quarter event is because we do have orders and outlook for the next few quarters that indicate that this is a one-quarter event. We don't think we're at the peak.
We continue to see improvement in the revenue from here.
Okay. Eric, and then I want you to go to Ananda right in the back.
Awesome. Eric Woodring, Morgan Stanley, thank you for hosting us today. Maybe Irving, I'll start with you. If we look kind of pre-downturn, the competitive intensity in this industry was very high. Obviously, we've seen end market share shift, but obviously in your three- to five-year model, you're talking about maintaining market share, which kind of strikes me and obviously seems more rational. I guess my question is, how would you respond in a scenario where you are losing share to competitors? What do you do in that type of scenario, whether that is addressing volumes, whether that's ASP actions? Just how in that hypothetical scenario would you respond? And then just a quick follow-up.
Yeah, maybe I'll start with someone who wants to chime in as well. Look, you've highlighted it really well. I think the industry has really shifted to a model where there's a lot more rationality and the focus is on driving profitability and cash flow generation, and that's the guiding principle by which we will continue to operate, so even if there's a situation where there's potentially share shifts in any given quarter, our focus is always on driving profitability and free cash flow generation for the business.
Okay, helpful, helpful. And then maybe Wissam, as a follow-up, is there any way you could just maybe double-click on the competitive, or excuse me, the capital intensity required for HAMR investments? Ultimately, what I'm getting at is you've given us that four to six target range. Where do you expect to land within that range over the next few years? How should we think about that trending? Meaning, do you start at six and trend down to four because you need to make HAMR investments? Can you maintain it four? Just curious what that curve might look like because obviously it influences free cash margins. Thank you.
Yeah, of course. So let me first start by looking at the last couple of years. The last couple of years we were running at sort of the lowest end of that range, maybe even lower than that. And so the way to think of it is as we ramp HAMR capacity, we'll probably be going a little bit closer to the upper end of that range. But I think the range itself, 4%-6%, is a good range. We don't anticipate to be higher than that. We don't see the need for it. And then the other thing to think about, Eric, is it also depends on what the top line does. If the top line follows the model that we think is going to follow, we may probably end up being closer to the middle of that range.
Thank you. Let's go over in the back there. And then anyone else on the right-hand side? Okay, if not, let's go back over here afterwards.
Thanks, guys. Ananda Baruah with Loop Capital. Thanks for all the great information today and for taking the question here. Really going back to the Gen AI opportunity and some of the remarks that you made a few moments ago, I was interested in getting any context you guys have, your view on the potential for Gen AI in video in coming years. Video seems to have tremendous potential with Gen AI, both in consumer and then even in commercial applications of things like AI factories begin to truly manifest and the potential to become fairly ubiquitous. And we're in sort of nascent starting point today. So in the context of, first of all, any context that you might have, any perspective just about that generally, and if good case were to manifest from that, how incremental could that be, at least anecdotally to the forecast that you gave?
You spoke a few moments ago about upside potential to that, and that's it for me. Thanks.
Okay, I'll take it. Look, as we've highlighted, I think video, especially in terms of Gen AI, where even today, if you look at what Sora is doing, the ability to create a 20-second video from text, I mean, imagine the day, which I don't think is that far out, converting a whole script into a video is that far out. So as I highlighted, that potentially would be a big driver of storage because when you talk about video, that's a big driver of data ultimately, and you need that appropriate storage. So it remains to see how it pans out. We think that that's a massive opportunity going forward, and that will be potentially a kicker to the growth story from AI going forward.
Well, thanks.
Okay, why don't we go to Steven, and then I believe Vijay had one, and then we'll break for lunch.
Thanks, and congratulations on getting to the finish line on the separation. I had two questions. First of all, can you describe a little bit more in detail your areal density roadmap, say, over the next three to five years, how much can you improve on areal density? And then depending on what you say on that, I was just curious on the gross margin improvements, how much it sounded like you were mainly talking about the value per exabyte shift, but I would imagine that there's an incremental margin that you're getting that's higher than you've gotten previously on just units of exabytes out the door. Can you give us a better sense for incrementals? Thanks.
Yeah, I'll take a stab at the areal density improvement. Obviously, I have my colleagues from engineering here who will be able to answer it or correct me if I get it wrong. Our focus is really on driving a 15% CAGR areal density going forward. If we are able to hit that trajectory, we'll deliver on the roadmap that we've highlighted and beyond. That's what we're working towards. We're very much on track from that perspective.
Yeah, and then from the incremental gross margin, so one of the things I didn't necessarily touch on, which we've always talked about, is also that sort of UltraSMR benefit. So UltraSMR, when you get into the hard drive itself, UltraSMR gives us a 20% capacity benefit. And that may not translate into 20% dollars, but that translates into sort of a better margin per product and better ASP. And so as we move into higher capacities, you'd expect that 20% to become bigger as well because now you're measuring 20% off of 36 terabytes as opposed to 26 terabytes. So there's an element of that as well in the gross margin story.
But I think at a very high level, the way we think of it is very much as we want to continue to drive better products that allow us to generate better revenue because we're delivering better TCO to our customers. And then we will continue to work on the cost curve. But as we step up into capacity units, we should expect some incremental benefit. I don't want to make it sound like with everyone we're going to get 100 basis points because obviously the last year we've pretty much went almost to 10 percentage points improvement. But gradual improvement over time is a reasonable thing to expect.
Okay, why don't we take the last question here from Vijay?
Vijay from Mizuho, just a quick two questions here, Irving and Wissam. On the AI side, I think you guys showed about 25% of revenues from AI by 2028. Just wondering what that mix is today and if you can give us some color on what the content is per AI server. And the second one is on HAMR. As you go to the hyperscalers trying to have your design wins, what's the value proposition, I guess, in terms of TCO or areal density versus your peer HAMR suppliers, I guess? Thanks.
Yeah, I guess on the AI question, what is it today? First of all, from applications that are AI specific that we can directly measure, it's still fairly nascent. But as I mentioned, if you look at how AI is being used within the core operations of hyperscale customers today, there's a meaningful impact. So the ability to separate it out specifically to measure it, it's a bit harder today. We anticipate as new AI workloads come on board that are specific, that are measurable, that will start to be able to be measurable in a material way starting in calendar year 2026. Back to the video sort of applications, medical diagnostics that we talked about as well. And to your question on HAMR, areal density and capacity points are one thing. I think there's an overfixation on them.
Obviously, it's an important aspect of it, but delivering the right TCO benefit to our customer that they can rely on so that they know that I can supply them a million units every quarter consistently with the right yield, with the right quality that makes economic sense for them and that makes economic sense for us is really what we're focusing on. So it goes back to the HAMR strategy that we laid out that we have very high conviction around.
Okay, back to you.
Thank you all very much for joining us today, and it was really an exciting opportunity to share with you sort of our coming out party as a standalone hard drive company. I really appreciate all the questions that you've asked us as well, and hopefully that gives you a better sense of who we are and what we're going to be going forward as the new Western Digital going forward. Thank you very much for joining us, and I think we have lunch served across the foyer. Thank you.
Thank you.