Broadcom Inc. (AVGO) Earnings Call Transcript & Summary

Great. So I think we'll get started. Good morning, everyone. Thank you for joining us. I'm Stacy Rasgon. I'm Bernstein's senior analyst. I cover the semiconductor and semiconductor capital equipment space here at Bernstein. Today, it's my honor to host Broadcom's teach-in on their storage segment. Today from Broadcom, we have Hock Tan, Broadcom's President and CEO; Jas Tremblay, Jack Rondoni and Dan Dolan. Jas is Vice President and General Manager of the Data Center Solutions Group at Broadcom, responsible for developing silicon software and adapters for data center storage solutions to cloud service providers, enterprise customers and embedded OEM systems. Jas also manages the PCIe switching and Ethernet adapter franchises and joined Broadcom through the LSI acquisition, where he was Vice President of North America Sales. Jack is Senior Vice President and General Manager of the Brocade Storage Solutions division at Broadcom. In this role, he's responsible for leading the strategy and execution of the company's storage networking business. He joined Broadcom through the Brocade acquisition, where he was Senior Vice President of Storage Networking. Finally, Dan is Senior Director of Marketing of Broadcom's HDD storage business and is responsible for driving the strategy and road map for the company's HEV storage electronic solutions. Dan joined the company also through the LSI acquisition, where he was the Director of Marketing for LSI storage products and previously held product marketing positions at [ Azyra Systems ] and Lucent Technologies. Now today's teaching will last about 1.5 hours. We can go a little longer if we need to. We'll have Q&A following the presentation. [Operator Instructions] If you don't, you can feel free to e-mail questions to me at [email protected] and we can make sure to get those asked. With that, I'll turn things over to Hock Tan, Broadcom's President and CEO. Go ahead, Hock. Thank you.

All right. Thank you, Stacy. I'm pleased to be here today for our storage teach-in, the number 3 in our continuing series of providing more transparency, more details of the various segments and drivers within the Broadcom platform. Now having said that, I'm -- as Stacy said, I'm not planning to say much here today, but I'll start with a few words about Broadcom and the critical role Broadcom plays within the storage ecosystem. Then Jas, Jack and Dan will lead the discussion of our storage franchises. All right, next, to begin. Well, at the risk of repeating myself, and this is the same deck, same slide here I presented in the last 2 teach-in, but it's worthwhile perhaps to reinforce our business model that drives Broadcom. So today, Broadcom is a leading provider of technology to the global IT ecosystem and a critical partner to the world's largest enterprises, which includes most of the Fortune 500. And as we note here on the left, since 2009, which happens to be the year we went public, we expanded the Broadcom platform from 8 to 23 different product franchises. And we have stayed true to how we build these category-leading franchises. And let me describe what each is. There are 3 common attributes. First, each provides mission-critical technologies to end users and that ensures sustainable demand. Secondly, in each vertical, we have the technology leadership. And finally, we not only have technology leadership, we tend to be #1 market share. And you hear today how our storage franchises fits very squarely within this criteria. So looking again at this slide, since 2009 on the left, we have grown our revenue over this period of times, 2009 to 2020 through both organically. And I would say organically, we've probably grown to the extent I can estimate probably high single-digits compounded annual growth rate, but also we've grown through acquisitions. And you see here, revenue has grown to $24 billion from $1.5 billion when we went IPO, a 16x growth rate. But during the same period, consistent with our business model I described earlier and clearly, a contradiction to the -- some naysayers out there who view incorrectly that Broadcom is simply an industry roll-up, which we are not, but we increased our R&D spending 25x. You need to do that to grow and sustain our leading franchises. And what's even more interesting, the end result of that is that we grew operating profit 84x. You might say, how? Gross margin, product margin improved as we go through newer and newer product generations, and we have a platform scale on infrastructure costs. Right. Next, underpinning this base -- this technology leadership, are our basically routes, which comes from the great innovations of American technology icons, such as Bell Labs, HP, Hewlett Packard, LSI Technologies and even Broadcom Classic. We gained this heritage and the exceptional engineering talent that goes with it through our acquisitions, as you gather. And today, Broadcom has emerged as having the largest intellectual portfolio -- intellectual property portfolio, I'll say, in the semiconductor industry, more than 20,000 fundamental patents. And our storage business is no exception. And it's an example of the value created through our acquisitions during this period of LSI Technologies, Broadcom Classic, PLX Technology, Brocade and Emulex, acquisitions from which we extracted set of product franchises, which you'll hear more about today, with a history of proven technology, brought on innovation and execution and very deep customer relationships. Next, and as highlighted here today, the focus of the teach-in will be our products, hardware and software and technologies applied in the enterprise server storage connectivity segment. Together, these businesses represents over $4 billion of annual revenue in fiscal 2020. And as highlighted, this end market is driven over 80% from enterprise demand. And now traditional enterprise is very, very relevant today, have started to recover from the pandemic. Wasn't the case 12 months back over most of the last 12 months. But as we see over the next 12 months and we have great visibility these days because of tight supply constraints, we are seeing pent-up demand from enterprise customers, resulting in a very, very strong and accelerating demand trajectory in this business back half of 2021 and continuing now as we see, to most of 2022 calendar year. And next, please, my final slide just simply sets the stage for today's teach-in. Across all data center deployments as you gather, data is created in servers, computing that is, in storage and accessed through the network. Broadcom, its platform, enables this server storage ecosystem by connecting these elements in a very -- in a highly reliable and extremely efficient manner. And as you know, about roughly half of the 12 million service today go to enterprise. So today go to enterprises. And every one of these service does require some form of storage attach. In fact, 70% of this enterprise servers have direct attach storage controllers, which is an excellent franchise for us. And Jas will discuss that today. Now with the excitement, particularly and I would call it and the marketing hubris in the computing space between Intel, AMD, as well as future -- possibly future multi-based servers, connectivity of servers to storage has become quite exciting and game changing. Meanwhile, the other 30% of enterprise servers remains very stable, boring almost because they use very high-end Fibre Channels and attach, a purpose-built network for mission-critical servers and storage. This is another franchise Jack will elaborate more on today. And not to be forgotten, the primary choice of high-capacity storage in data centers today, both in cloud and enterprise continues to be nearline hard disk drives. Thankfully, we have not had much footprint on the client side, as Dan Dolan will explain later historically. But we have been able to leverage what we had initially as an emerging nearline hard disk drive business into today, a very substantial franchise within the hard disk drive storage industry. With that, I'm pleased to turn the presentation over to Jas.

Thank you, Hock. So my name is Jas Tremblay. I've been with the company for 15 years. I'm responsible for the Data Center Solutions Group. And today, I'll be giving you an overview of this franchise and how we've achieved better than market share. So with this slide what I'd like to do is put things into context. Enterprises, their main focus is to deploy applications. And once the time comes to choose the infrastructure to deploy that applications, data protection and data privacy or core decision criteria. And they have options from an architecture perspective. The first option is on the left for mission-critical applications. Here, the data is stored on the drives located in external storage systems. And the data is accessed through servers interconnected through a specific storage networking, specifically Fibre Channel. This architecture provides the highest level of data protection and security. The second option in the center is for enterprise-grade applications. Here, the data is located in the drives, in the server directly interconnected to the CPU with a purpose-built storage adapter that provides protection and security. And this is the core focus of the division that I run and the storage connectivity franchise. The third option is moving your workloads and your applications to a cloud environment. And typically, the drives are located in the servers and software is utilized to provide data protection, typically replication where you simply copy the data from one server to the other. So again, when the time comes to pick your infrastructure to deploy your application, data protection, data privacy are very critical where you decide, do I keep this application on-prem or host it on the cloud? And in 70% of the cases for enterprise servers, they deploy the middle application or storage adapters. Next slide, please. So on this slide, I'll give you an overview specifically of the core product that we sell in this franchise and how we've achieved #1 position. So I'll draw your attention to the picture on the left. You have a storage adapter in the center. This is the product that we make, composed of silicon software and a board-level product. Its role is to interconnect the drives with the CPU and make this visible by the application. They are deployed in the server. So it's an option that goes inside a server. Three core functions for that storage adapter. The first one is the drives and the CPUs have different protocols. So the storage adapter converts the protocols from the drives to the CPU. The second one is drive failure protection. This is very critical. So imagine you have an application that's using the storage adapter and you have 3, 5, 10, 12, 24 drives. And one of these drives fails. Well, the storage after has the ability to reconstruct the data and provide seamless data protection to the applications. So drive failure protection is a second key function. The third function is server power failure protection. So think of a financial institution that has an application that needs to register a transaction, write a transaction to the drive, write the financial transactions. And while that transaction is in flight, the server loses complete power. Well, our storage adapter has an autonomous energy source, specifically with supercap technology that allows you to complete the right even when the server loses power. So again, the storage adapter has 3 core functions. The first one is drive protocol conversion. The second is it protects against drive's failure, and the third one is actually protects against complete power failure of the servers to ensure you don't lose writes or any recording of transactions. And these things are very important for financials, health care, multiple [ structures ] and this form of data protection is compatible and optimized for databases, ERPs, CRMs and many, many types of applications. And in fact, all the server suppliers offer our storage adapters as an option for data protection. So this franchise, the way that we're growing this franchise growing revenue is 2 aspects. The first is we're gaining share. We've been continuing to gain share over the past decade. And generation over generation, these servers require more performance. You need -- the drives are getting faster and the storage adapters are getting much faster. And I'll show you a little bit more details on that. So we're growing ASPs consistently generation over generation. And compounded, this business has grown mid- to high single digits consistently over the past several years. Next slide, please. Okay. So this franchise has existed for over 30 years. And let me go through a little bit of this heritage and how we've built up the #1 market share position. So again, 30-year history. In fact, my Head of Software Engineering, my Head of Silicon Engineering and my Head of Storage Architecture, all 3 of them have worked together and for the same franchise for the past 30 years. Multiple company names, but they've always stayed together. We've got quite a heritage and a team that the industry has learned to trust and know over decades. This really all started in the 1990s when we introduced the first storage controller. And then in the 2010s, we introduced jointly with the ecosystem, the SaaS protocol, a new protocol for drives, and this brought a lot more resiliency and performance to storage connectivity. And in the recent years, the focus of our investment has been to deliver the same data protection and trust that people have in a solution with NVMe-type drives. And NVMe-type drives provide better latency and more performance, and we're aligning our road map to deliver this level of performance. In fact, if you look at the slide and you take a snapshot of the past 10 years, over the past 10 years, we've increased performance by 20x more transactions per second over the past 10 years. And with the sustained innovation that we're -- an investment that we're making, we have a path to continue on that performance gain trajectory over the next several years. But at the end of the day, this is important in this heritage and the team and building this trust with customers is if you're an insurance company, over the weekend, you experience a series of drive failures or issues for your infrastructure and the Broadcom storage adapters protect your data, they recover it seamlessly. You come back to the office Monday morning and you didn't miss a beat there. That's how the trust is built, and we take this responsibility very seriously. Next slide, please. Okay. On this slide, I'll go over the 3 core components of our leadership. How have we achieved #1 position over multiple decades of sustained innovation. Well, the first core component is the software, the most trusted data protection software. It's taken decades to create. It's tens of millions of -- it's 10 million-plus lines of code. It's real-time, high-complexity, real-time software. The second component is we built our own silicon. So purpose-built silicon specifically for these applications. It's not general-purpose silicon, it's silicon dedicated for storage connectivity and data protection. And when you take the software -- the data protection software and the purpose-built silicon together, that's where you get the highest performance with the highest data protection. And the third core component of our leadership is the strong ecosystem support. It first starts with the operating systems. So if you're an enterprise and you want to deploy your application on, let's say, Red Hat Linux, you deploy Red Hat Linux and out of the box, our storage adapters and drivers are recognized automatically. And we do this for a very broad set of operating system. This just simplifies the deployment model. And second element of ecosystem support is drive compatibility. So what customers want is they want to make sure that the application operating system, storage adapter and drives, combined with the server combination works. And the ecosystem relies on us to ensure the drives are compatible with our solution. We have a very stringent drive compatibility infrastructure. In fact, just a few weeks ago, we had Samsung come to our offices in Colorado Springs, and they brought their latest and greatest drives to go through the joint testing together. They know -- and as multiple drive vendors know that for the enterprise space, having the Broadcom seal of approval for storage and being on our compatibility list is important. So we invest quite a bit in ecosystem support. So you take the software. It's taken decades to perfect. We continue to optimize the purpose-built silicon for performance, and we invest heavily in broad ecosystem support, both OS and drives. Next slide, please. Okay. Let's go through how we've created the broadest and most complete route to market for our products. So when it comes time to purchase and deploy the products, the enterprise customers want simplicity. And we have 3 pillars for our go-to-market strategy. The first one and most important is server OEMs. So imagine your customer, you want to buy a Dell server, you go online and you go through the configurator and then you're offered a series of options for storage adapters and data protection and you can pick and choose them. And then when you buy your server, it simply comes with it. It's a very simple process, and we have this process in place for all the industries' leading server OEMs. The second pillar is server ODMs. Similar approach, but their focus is hyperscalers, telcos and a few select large enterprise customers. But same approach and simplicity in terms of ordering. You pick your server, you configure it, you're offered options of a storage adapter and the solution comes with you fully integrated. It's not a separate purchasing decision. The third pillar of our route to market is channel. So this is a network of hundreds and thousands of system builders and integrators globally. These are typically companies that assemble software, hardware to create purpose-built solutions, either for specific verticals or actually for individual companies, they assemble solutions. And what they do is they buy, through distributors, our storage adapters and they fully integrate them. With this community, we'll often found some of our most advanced users. So just to put it simply, from a route to market, we wanted to simplify the purchasing process, and we've built up a very complete and thorough global that covers all segments route to markets for our storage adapters. Next slide, please. Okay. So we've covered storage connectivity. Let's step back and look at the different types of connectivity required in the server. So server, as displayed here on the left, there's about 12 million servers sold annually. Globally, about 12 million. These servers are composed of CPUs, memory, storage and a broad array of connectivity. And over the past 6 years, through multiple acquisitions, Broadcom has built the broadest, most complete portfolio from a server connectivity perspective. So let me go through a couple of these. The first one is storage, which we covered. Storage adapter is our leading server connectivity franchise, came to us from the LSI acquisition. The second one is Ethernet connectivity. So high-speed Ethernet connectivity inside the server, same business model, storage adapters that came to us from Broadcom. Thirdly is PCIe adapters, PCIe switching that gets deployed mostly in servers that came to us from the PLX acquisition. And finally, Fibre Channel adapters. Those came to us from the Emulex acquisition. And if you compare storage Ethernet, PCIe in Fibre Channel, they all share the same customers, the same route to markets and the same fundamental business model. And we've achieved #1 position overall in server connectivity. So next slide, please. Okay. So let's look at the competitive landscape a little bit and see where we stack up. So from a storage perspective, again, the franchise has existed for more than 30 years. We've achieved #1 -- we've had #1 position for quite a long time and are investing significantly given the scale that we have from a revenue perspective to keep that leadership position. On PCIe switching, we're also #1 with a very strong position. And Fibre Channel, we're also #1 globally. The competitive landscape for these solutions is extremely stable. And we're not seeing new entrants specifically in storage, PCIe and Fibre Channel, and we are growing quite rapidly from an Ethernet perspective. So for storage, server connectivity, we're #1 overall, and we're #1 in 3 of the 4 categories that -- where we participate. And this is important for our customers. They like having a company that is dedicated and focused on server connectivity. They want to interface with a team that offers multiple solutions. And they really like the fact that we're driving more investment than the competition across these product categories. So really from a product and investment perspective, no one else has this range or breadth of solutions, and customers are really appreciating the focus that we have and the commitment we have to this franchise. Next slide, please. So this is the final slide. So just to put things -- summarize things. From a portfolio perspective, we're extremely well positioned in the server space with #1 position overall in server connectivity. We're investing in 4 areas, and we've achieved #1 position in 3 out of the 4. Storage connectivity and servers is one of the flagships in this overall portfolio and 70% of enterprise servers utilize our storage connectivity. We've achieved this for building the trust with our customers over 30-plus years, and we're going to continue to invest in this space. And with this, we've actually achieved a pretty cool milestone. We've deployed now over 100 million of our storage adapters. And our strategy is very straightforward. It's focus, be committed to server connectivity. Because of the revenue scales and the #1 position we have, we're out-investing our competitors in these individual categories and delivering highly -- high differentiation from a performance perspective. And overall, we have a path to continue to grow this highly-profitable franchise. With that, I'll pass along to Jack. Thank you.

Thanks, Jas. Hello, everybody. My name is Jack Rondoni. I am the General Manager of the Brocade storage networking business unit. I've been with Brocade for about 15 years, came in obviously through the acquisition. Have been associated with Fibre Channel for almost 20 years. So first slide. When you think about the storage networking market, I always start with who uses storage? Who uses SANs? And it's enterprises that run mission-critical applications. And what a mission-critical application is, the way to think about that is, if that application that's running on that server, if that is not running, your business comes to a complete halt. So it's not like e-mail or something that's not important. Think about it as you're doing the ATM transactions. If that mission-critical ATM applications' not working, you're not doing business. And if you're an airline operations and that application is not running, you are not -- that business is not running. If you're processing payroll, and you can't because your mission-critical [ advocate ]. Your business comes to a stop. So it's very important when you think about SANs, those are the types of enterprise -- those are the type of applications that are deployed on storage area networks. And the reason is these applications require that the underlying infrastructure, meaning servers, network and storage, deliver the reliability and resiliency to keep the application running 24/7. It delivers very low latency to keep the applications performing as expected as well as being able to scale as well as being highly secure. The type of information, credit card numbers, social security numbers, very, very important. So security, obviously, is also very, very important. So because of the nature of the importance of these applications to these enterprise businesses, we see that these applications continue to be deployed primarily on premises, on infrastructure that is on-premises. Because the cloud, even though it's been successful in other models and other use cases, will have -- it's not suited necessarily for those types of characteristics. And there's certainly -- we've all seen in the press certain things that is happening with the cloud and where some of the outages have happened and other things have happened. So it's very important when you're thinking about storage area networks to always start with the notion of who is deploying it, mission critical applications or enterprises? And then as I talked through this teach-in and as I introduce myself as from Brocade, the important thing to remember, this is a Broadcom, the 2 franchises that deliver Fibre Channel networking, are Brocade and Emulex. And those 2 names are brands in the enterprise storage networking business. They are the trusted names in Fibre Channel SANs. So you'll hear me use those names quite a bit because we're one of the earlier inventors from the beginning. So going to the next slide. So now that I introduced the -- who uses these storage area networks, I want to make sure I'm very clear on what is a storage area network? What is a SAN? So it is a purpose-built network, meaning this network is built primarily for storage workflows, for storage information. And as I mentioned before, it connects mission-critical applications that are sitting on the server on the far left and connecting to external storage on the far right. Now on the far left, that red adapter there, that is a Fibre Channel HBA and that is the Emulex brand. They provide the connectivity to or through the network, which is where Brocade is in the middle there, to storage array, where there's also a Fibre Channel adapter, which is where the storage is. So again, the way to think about the SAN is the key applications running on the server, the actual data itself, social security numbers, key business operations, credit card, that flows through the network and then into the centralized storage on the right-hand side. Now these SANs that are delivered by Brocade and Emulex use a protocol, which is called Fibre Channel. And Fibre Channel, as I mentioned, is part of the purpose-built technology for SANs. And it differentiates itself versus other network technologies in a number of ways. First is Fibre Channel is a lossless network. And the best way I can explain that is think about 2 people that are playing catch with the ball. In a lossless network, before the person throws the ball, the person who is to catch the ball says, I'm ready to catch the ball and they're capable of catching the ball. So they send it over. And the ball in this case is mission critical information. In other networks, for example, general-purpose Ethernet networks. The way that would work is the person who's thrown the information, they don't -- they're not going to wait or ask if the person is ready to receive it. They're just going to send it. They're just going to throw the ball over. And we're going to keep sending it until somebody says, "Oh, I got it." Now in a general purpose, let's say, workload or applications that's not mission-critical, that's fine. Maybe 90% of the time, that's good enough. But as I think we can all appreciate, when if you're transferring, let's say, your 401(k) from one institution to another institution, 90% is not good enough. So the lossless aspect of how Fibre Channel was built is a key part of the differentiation of why Fibre Channel continues to be deployed with mission-critical applications. The other aspect of it is every generation of product, the speed doubles. And I'll talk about this a little bit more. And that's very important, as Jas showed the number of transactions per second on these drives, which is phenomenal. The scale of information and data is going up massively. So you don't want your network to be a bottleneck. Your network's got to be able to handle the volume of that information. And then the other key aspect of Fibre Channel that differentiates itself is integrated security. The best way to think about this is Fibre Channel is off by default, meaning you are able to go into a data center and plug into that Fibre Channel network, nothing happens. It's off by people -- you have to go and do further configuration. Now with some general purpose, you have networks that's different. We plug it in, it's connecting, it's ready to go. [ It started ] when you connect into WiFi and you're all ready to go. But you don't want that, really, when you're running mission-critical applications. And so that's another aspect of what really differentiates Fibre Channel. So I talk about the mission critical applications, hopefully give you a real clear view. When I'm talking about SANs, I'm really talking about the connection from the server all the way to the storage and why Fibre Channel has continued to be very well differentiated in the space. So next slide. So let me go through a little bit of kind of the history of revenue. So this is quarterly revenue for 20 years. Fibre Channel has been around for a while. And it includes switches as well as adapters. So it's combined. And the sources are the third-party analysts. So basically, this market for the first 10 years, let's say, from year 2000 to roughly 2008, was in a ramp phase. Basically kind of peaked out in 2008 or so, then there's the dip of the economic meltdown in 2008. But really what I'm going to focus on is the last decade starting kind of where that number one is. And basically, what we've seen over the last decade is that this market has remained stable for over a decade. And as you can see, sometimes quarter's up, sometimes quarter's down, we saw some strong growth in '18, '19. We did see some declines in 2020 because as Hock mentioned, the impact of the pandemic on the enterprise. Also, of course, at us since we're in the enterprise. Now we're certainly starting to see the second half improvement starting in '21 and expect a rebound as well in '22. Now more importantly, we expect this revenue dynamic to be sustainable for the foreseeable future. Let me explain why that is. Next slide, please. So this chart here is the exact same revenue chart, right? So it's the exact same revenue chart. But what I'm showing in the colors is really by generation of products because -- but I think it clearly shows is a refresh dynamic that happens within the Fibre Channel SAN market. So again, if I start with maybe the first 10 years as kind of the ramp-up phase of Fibre Channel, there is 1-, 2-, 4-gig products. That's all kind of the gray market. That's what was shipped of kind of that class of products. Now starting in kind of the middle there, in 2010 or 2009 just started to ramp, you see the 8-gigabit portfolio of products coming in. Remember, I said speed doubles every generation. So we came out -- the industry came out with 8 gigabit. And then what happens is it starts refreshing the installed base that is there. And then you can see with the darker blue that the 16-gig portfolio came in. And again, it's refreshing previous generations and previous generations of that. And as you can see in the red, as we get to the 2020 time frame, we're in the midst of a refresh with 32-gigabit products, we call that GEN 6. So you can see that this business, especially in the last 10 years, is really driven around refresh from new portfolios of products coming from the Fibre Channel industry. The other interesting thing is that as these new products come out, you get double speed, easier to use, lots of value for it. There is an ASP growth associated with the new portfolio, double the speed, as I mentioned. And so that ASP is -- that growth of ASP per port is also an important characteristic of how this industry is able to keep the Fibre Channel revenue sustainable over time. So next slide, please. So I went through kind of the snapshot over the last 20 years and maybe the dynamics, maybe the last decade and how tech refresh and new portfolios really drive the stability and the sustainability of the Fibre Channel market. So I'll talk a little bit more about this tech refresh process here. So first, it's required to keep up with the increasing application demands. There are more transactions being done, certainly more online transactions being done. Again, we talked about in the previous presentation, the number of transactions per second of storage. Storage is getting faster as well as it's getting larger scales per drives. And so these SAN product generations, on average, it's about 5 years. Some customers go a little bit longer, some a little bit less. And it's that 5-year refresh cycle. And as I mentioned that higher ASP per port because of the new portfolios that really drive customers to refresh. So they get better speed, they're able to scale their applications more, they get new capabilities within the new portfolio that makes operations easier. And that's why we see our customers refresh. Now if you look at just the last 2 generations, again, this is from their prior reports, there's about 43 million switch and adapter ports. In fact, there's probably another 20 million or so of the generation before that out there. So there's a very large install base of these Fibre Channel switches and adapter ports that are basically the source of the tech refresh phenomenon that I showed on the previous page. And the dynamics of it are actually pretty simple, and we commissioned an internal study in this. And 90%, roughly, of the refresh that we see is of existing footprint. A customer has a Fibre Channel SAN today, adds on to it brings in a new portfolio to replace the existing fabric. So the vast majority of it is the 90% range is really just refresh of existing Fibre Channel footprint. The other 10% are more greenfield opportunities that we've seen and I think of those as just maybe a new data center buildout, jump on, if you will. Now 5% of those, roughly, we've seen are -- we'll be considering alternative technologies. And depending on the mission-critical app, we've seen some small migration from Fibre Channel footprint towards that. But also, we see about half of these new greenfield opportunities are new Fibre Channel installs. And there, it's truly a jump ball. And we've done very well executing and we actually win the vast majority of those. So it's a very interesting dynamic. It's almost like a software renewals, right? At some point, the contract comes up and we got to go refresh or you got to go and renew it. It's very similar to what we see here. And then as I showed, we're in the midst of a 32-gig refresh cycle right now, and what Broadcom has done through Brocade and Emulex is we've launched our GEN 7 portfolio, which is the 64-gig product. So remember, the speed doubles every time. That was introduced just late last year. So the timing of the new portfolios for tech refresh for the foreseeable future is perfect because as the enterprise, as I mentioned, comes out of the challenges of the pandemic and certainly into next year, not only are we in the strength of our 32-gig portfolio, but we also have our GEN 7 64-gig portfolio at the ready. Let's go to the next slide, please. So I mentioned Broadcom is a SAN industry leader, and that's really driven through the 2 franchises associated with Fibre Channel, that being Brocade and Emulex. And the thing to remember also is Brocade and Emulex were there from the beginning. So there's 25-plus years of innovation. Both companies have one of the first products in the market in the late '90s. Brocade has been a switch market leader. Over the last decade, we've been about 70% market share. And as mentioned, Emulex also has as a market share leader as well for HBA. So very well positioned, both historically and currently and into the future with our GEN 7 portfolio. Next slide, please. I talked about the GEN 7 portfolio a little bit, and I just wanted to go into a little bit more detail. First is, this was the industry's first and currently only, end-to-end 64-gig portfolio, again, we call it GEN 7. And if you remember what constitutes a SAN, get the adapter for the server, GEN 7-ready, you have the Brocade switch, see a couple of pictures of the various switches that we have right in the middle. And then you have the other adapter that sits in the storage array. In addition to just the doubling of speed inside the fabrics, inside the SAN itself, we have what's called autonomous SAN technology. And essentially, the way to think about this is there is information flows that go over a SAN. Remember, I talked about mission-critical applications. There's a storage flow of information. And that storage is encapsulated in a Fibre Channel frame that set through the fabrics. Well, what we're able to do is because now the number of transactions has gone up massively, the amount of data is continuous to grow, we are able to basically profile or baseline an end user's environment. And actually, that's very important because, first off, every business is different. So how information flows through a SAN is different. And so having that kind of baseline of, this is what normal is, is very important. And that's what our autonomous SAN does automatically. Now once it baselines, it self-optimizes. And one of the way it self-optimizes is there's some business applications that run extremely fast maybe they're running some of those new NVMe drives or other things. And then there are some applications that are slower, maybe some legacy tape drives, other things. Part of the self-optimizing is our SANs are able to see how this -- these frames flow through there and they're able to put the, let's say, call it, the slower applications in the slow lane. Think of it as the highway and the high priority or maybe kind of more the money information goes in kind of the higher speed. So that self-optimizing is done automatically. And then there's a self-healing capability as well that if something goes wrong and let's say, an upgrade on a server or storage rate does not go correctly. And if you remember that analogy about the ball going back and forth and they say, I can't catch it. We can help self-heal that. And so this is a very important technology. And if you remember what I said is this is purpose-built for storage flows. A general purpose Ethernet network doesn't have that storage essential kind of analytics or view of what goes through a SAN. Only Fibre Channel has that. So this autonomous SAN technology translates into making it significantly easier to operate. So not only do you want your SANs to be high performing -- I mean very reliable and high performance, you want them easy to use. And the best way easy to use is the performance that is high automatically, and the resiliency is always there. So it's a very, very important part of our portfolio and differentiates us. The other thing is as part of the legacy associated with our business, we have a lot of highly-tenured Fibre Channel experts in our business units. And what they are able to do is also provide very proactive and automated support. So the smartest Fibre Channel experts in the world, there is no question, are within the Brocade and Emulex divisions. And they also provide them that support that enterprises and our storage OEM partners require to keep those applications exactly where they are. So this portfolio, we're very, very excited about. As I mentioned, we launched it in '20. And as we see -- as we get in the second half '21 and next year, we see 64-gig or GEN 7, adding on to the transition for our portfolio. So next slide, please. This will be my last slide. So the Fibre Channel road map, the best way to think about Fibre Channel road map is we always align it with how the enterprise storage is going. So when enterprises are updating or modernizing their storage, it could be all-flash, could be NVMe. We align our capabilities from the SAN fabrics with that. For example, our autonomous SAN technology. It doesn't matter if it's Fibre Channel drives, or [ seeing these SCSI ] drives or NVMe drives, all that autonomous capability that I talked about will work. So we align our road map to where enterprise storage is doing. The second thing is analytics is important, meaning as the storage flows, if you remember, I said storage flows are going through the fabric, Well, we put all these little sensors in the fabrics to basically look at the frames, never the data, we never look at payloads, just the frames, as they go through to understand so we can see how to baseline, how to optimize, how to self-heal. And then automation, the last is really about simplifying and enabling the SANs to participate in modern orchestration tools that enable infrastructure, IT operations to manage their environments as effectively as possible. So I'm going to wrap up as I've shown Fibre Channel is a very resilient market. As enterprises come out of the pandemic, coupled with the introduction of the new generation, the GEN 7 64-gig portfolio that it will support from the tech refresh, we are very much looking forward to continuing to drive this market for the foreseeable. And with that, I will turn it over to Dan.

Thanks, Jack. Good afternoon, everyone. My name is Dan Dolan, Senior Director of Marketing for HDD Storage. I've been with Broadcom and in the HD Storage market now for over 20 years, and I'm excited to be with you today to discuss Broadcom's growth and leadership in the nearline HDD industry. With that, we'll get started with my first slide. As you can see on the chart, the demand for nearline data storage is growing. To put this in perspective, more data is created today in 1 hour that was created in an entire year just 20 years ago. While there is a perception of the hard disk drive market declining relative to SSDs, which is happening in client disk drives, you can see the demand for nearline data storage is growing. If you look back to 2016 on the left side of the chart, you can see that the majority of the bits were stored on client disk drives. But there's a shift that's happened where now the far majority of the bits are stored on nearline disk drives. In fact, near-line disk drives are expected to account for more than 80% of bits stored in nearline relative to SSDs out through 2025. On the next slide, we'll talk about how Broadcom Electronics play a key role in disk drives for data centers. As you may know, in data centers, there are racks and racks and rows and rows of disk drives. Inside every disk drive, there are 3 components that are responsible for the operation of the disk drive. Broadcom provides 2 of these 3 components, the preamplifier or preamp as we call it for short, and the system on a chip or SoC. The preamp is located inside the disk drive very close to the head sensors and the media or disks to provide the highest single-to-noise ratio. The SoC is a large mixed-signal device that has integrated many functions into one chip over time. These 2 devices together comprise the data path that enables the approximately $10 billion near-line disk drive market that is growing. Next slide, please. Both devices are complex custom designs that are very difficult to develop. The preamp is a high-speed mixed-signal device that continues to add new functions and capabilities, such as laser drivers to increase the capacity of the disk drive. The preamp is primarily an analog device that uses decades of specialized design techniques to provide world-class performance. The SoC functions as the brain of the electronics inside the disk drives and has several IP blocks that are used to process information and control the operation of the drive, including the rechanneled core processor subsystems, and high-speed interfaces. The rechannel core uses sophisticated signal processing algorithms to increase the capacity of the drive by converting the analog information it receives from the preamp into digital 1s and 0s that can be used by the host to transfer terabytes of information from the disk drive. Next slide, please. We'll talk about why we win. There's 4 key reasons why Broadcom has maintained its leadership position in the nearline disk drive electronics market. and I can sum it up as follows: our leading process technologies and innovative IP designs have a time-to-market advantage that produces quality silicon that works right the first time. The preamp uses custom process technologies that are developed by Broadcom and specifically tailored for the HDD nearline preamp, providing significant advantages to our customers. The SoC leverages Broadcom's no geometry process technology leadership to provide our customers with the leading-edge benefits on the most advanced technologies. Through strategic acquisitions and our market expertise, we have been winning with this strategy for over 2 decades. Now on the next slide, we're going to look at where the HDD industry road map is headed. As you can see on the left side of the chart, the nearline HDD industry continues to innovate to increase capacity and depends on Broadcom electronics to make it happen. Not only do we expect to grow with the market, but we'll add silicon content to help continue driving solid growth. And finally, I'll wrap up on my last slide with some key points. Broadcom is a leader in the large and growing nearline hard disk drive electronics market. We have a long history and a proven track record with over 2 billion units sold. Our annual revenue growth of over 20% since 2016 also demonstrates the strength of this market and Broadcom's performance leadership. With that, I'll hand it back to you, Hock. Thank you. Hock, I think you're on mute.

Yes, we got -- okay. Stacy, you hear me now?

Oh, yes. We can hear you now, Hock. Thank you.

Sorry about that. Just -- I want to leave you guys with just 2 thoughts before we open up to questions. One is, as you can probably gather today, in this space of server storage connectivity Broadcom is pretty much ubiquitous. That doesn't stop innovation, technology from upgrading in the hardware, software in this space. We out-invest our limited competition in this space. And because we do that and we continue to do the out-execute, we have been gaining share quite dramatically over the last 5 years. Second point, deep customer relationships, strong ecosystem support you've heard about are very important characteristics. And we have those. We're trusted, we're practically branded in this space. The message I'm coming to you is price sensitivity is not a big thing in this space. And the combination of this 2 makes this segment of our market -- of our business, our platform extremely attractive. So with that, Stacy, let me turn it over to you for questions.

So we are going to go to the Q&A portion of today's teach-in. [Operator Instructions] We actually have a ton of questions though. Individually, we probably won't be able to get to them all. I do have a question of my own. One of the biggest questions I always get on this segment is growth. People tend to think of it as an X growth segment. It doesn't sound to me, from what I'm hearing though, that you guys share that point of view. And I know you've given some growth targets for some of these. You talked about Brocade being stable. But just in general, can you talk about how you see normalized growth across these 3 segments going forward? And maybe how that rolls up into how you guys view for normalized growth for the storage segment overall if you look out several years? Although I'd say I get the near-term cyclicality and some of those other drivers that are going on. But how do you think about that?

Well, yes, we have history of 10 years, 5 years. But if you look at it over the last 5 years, as I indicated, we have -- organic growth in the whole market is about mid-single digits. Couple that with share gains in this segment, we have been growing closer to high single digits on a compounded annual growth rate keeping in mind there have been some level of cyclicality, volatility through -- as we go through years and quarters. But over a compounded 5-, 10-year level, we see ourselves continuing to grow probably high -- mid- to high single digits.

Got it. And one of the top questions in the Pigeonhole actually sort of relates to it, but specifically for the Fibre Channel business for Brocade. And I get the drive when you sort of talked about stable revenue with ASPs going up on new technology. But that chart that you showed -- I believe that was for the whole market, showed ASP per port doubling on stable revenues, which implies units or, I guess, at least ports potentially cut in half over that time. Is that actually true? And I guess how does that trend going forward? How do you -- what do you expect for like, I guess, unit or port growth versus ASP trends going forward to maintain these kind of like stable relationships? Do you think like ASPs can like double again over the next 5 years and ports get cut in half again or what?

Jack, if I could take it on first, then you give them an elaboration. What you're missing in that and what Jack may not have elaborated clearer is this, let me do the math, simple. There are, as you said, Jack indicated, there are over 43 million ports installed today, the dynamic chart that Jack was showing. But not all 43 million refresh in any 1 year, Stacy, keep in mind. Translate to an annual cadence of refreshment, you're talking about 4 million, 5 million ports refreshing or coming up on an annualized basis, not 43 million, all 43 million. 43 million is installed base. Annual line is 4 million, 5 million. Of that 4 million, 5 million, to use Jack's example, 90% of it goes into the same Fibre Channel vendor because there are only 2 Fibre Channel vendors in this world. We're the leading guys, of course. And they are -- they do not interoperate in the data center. So you refresh the data center, you refresh with the incumbent. 90% goes in. And Jack indicated only 10% is a jump ball. And on that 10%, which are mostly greenfield site or they totally rip out and expand a new data center, half of it goes back to data centers and the other half possibly go into other technologies. So you're really talking about 5% of a refreshment of around 10%. That's how the math really works. And that's how you end up -- we call it stable. It's actually slightly growing. I will call it low single digits to even possibly in some years mid-single digits, is what Fibre Channel is. And it's driven by the fact that even if you have a slow attrition of the port, keep in mind, attrition is 5% of probably 8% of an installed base. Your price is going up pretty dramatically with new generation.

Got it. Jack, did you have anything you wanted to add to that? Or did Hock steal your thunder?

No, I think it was well said.

Got it. Got it. Another question for you on Brocade. So this is the segment that you actually don't include storage anymore. You obviously put it -- it's in the infrastructure software segment, within storage before you bought CA and formed that software segment. Why do you include it in software? Like is the go-to-market or is the customer strategy different for like Brocade versus some of the other storage segments? Like what's the reason for that differentiation in your reporting?

It's the end user. The end user with -- it's a system. Jack is selling software and with an appliance virtually. It's -- the product is really, that Jack described very strongly, the software. And you sell it as a system.

Okay. But it is the same customers, correct? It's still the same enterprise customers.

Well, JPMorgan saying is a huge customer for mainframe -- CA mainframe software; Symantec, cybersecurity protection and proxy as they are for Jack's Brocade Fibre Channel SAN systems.

Got it. Got it. Have another question here. Let's -- if we can stay focused on the Fibre Channel for a minute. Somebody would like to know like how you see the growth opportunity in 2022 and beyond for that business as we get -- I guess, sounds like 2 different cycles. We've got just the general enterprise rebound and a move to [ 66 Dwich ]. Obviously, should have higher ASPs and we'll see about the units. But do you expect like into 2022 like growth -- and I know like things are cyclical. But do you think growth into 2022 and beyond can actually be higher maybe than, call it, like the mid-single-digit kind of growth levels that you've been talking about historically given those 2 cycles that kind of hopefully [ we find you in someday or something ]?

Jack?

Yes. So certainly, as I mentioned -- I mean, you can hear me, right? Is 2022 because of the rebound from the pandemic, we would certainly expect a rebound in a growth, maybe single-digit kind of number. And as you get into that refresh cycle, I think the way to think about it is 90% is existing installed base, 5% is kind of this jump ball, which we win the vast majority of the time. So if we can execute as we have been, on that beyond even 2022, with the shift to more 32 gig and more 64 gig, certainly, I think -- single-digit growth is a realistic goal for us as a business.

Got it. I guess the 132 would be 5 -- you said these have something like a 5-year cycle. So that would be 5 years from now to 132.

Yes. Yes. I mean it's kind of -- you saw from the chart maybe halfway through and 64 is going to start to add on to it. Just keep going.

Got it. Got it.

Yes, it's very -- go ahead. Sorry.

No, go ahead. Finish.

No, no. I'm done. No problem.

Okay. Got it. I don't know if you guys can give us any sort of like high-level view for how the storage segment splits out between -- I guess in your reported segments, it would be primarily service, storage, connectivity and hard drive. Can you guys give us any sort of like -- even just a ballpark estimate for how the storage segment roughly splits out between those 2? And then Brocade, we kind of know a little bit like here and there. But...

Well, I would say it's about -- you know roughly the size of Brocade and the Jas biz segment, which is the enterprise-grade connectivity for storage and luxury enterprise again, is about roughly the same size as Brocade. And so if you take this $4-plus billion revenue segment I described in one of my slides earlier on, think as it as broken down to almost like -- I'm trying to say, close to 40% for both Jas and Jack, roughly close to 40% of that total $4-plus billion. And about over 30% is on our hard disk drive business that Dan Dolan went through. In other words, our hard disk drive market is almost $1 billion, it's just about $1 billion, put it in round numbers.

Got it. Do you guys do sort of similar stuff in SSDs as you do for HDDs? I saw the chart. I know it's smaller. But it -- obviously, it's growing. It's growing off of a small base. But on a percentage basis, it should be growing.

Well, I'd love to answer it, then I'll give -- pass it to Dan to give you more elaboration. Hard disk drive, as Dan indicated, there are only 3 electronic components, some chip -- semiconductor components in a hard disk drive. One is the SoC Dan talked about, which is really the brains, the rechannel. It used to be the rechannel. It's gone beyond rechannel now as things like HAMR come into play, it entering various other items. Then there's the preamp, which also has taken on more as more disks comes in for nearline and in complexity of the architecturing. Enhancers and multiple heads show up, and you start going to 10, 11 disk with more preamps. And the third item, which we do not participate is controllers, motor controllers. That's somebody else. Those are only 3 components, but we have 2 of them. And in each of the 2 we are in, there's differentiated, deep differentiated technology in the semiconductor space we own and keep enhancing. In flash controllers, which is in SSD to create solid state drives, that -- one thing, the technologies, we do not see -- there's 2 reasons why we're not deep -- considering it as the core franchise. One is the differentiated technology is not that obvious and deep, even though we do touch on that. They are not that obvious and deep. And secondly, it's just the dynamics of the market, right? If you have a fab that produces in an SSD $10,000 of memory, flash memory, a $20 flash controller can be given away for free.

Okay. Got it. Dan, did you have anything to add?

No, I didn't. Hock covered it very nicely. Thank you.

Okay. Got it. I had a question on like -- you talked about some of the differentiated technologies, especially process technologies within preamps and even in SoCs. My understanding, I thought you guys were fabulous on these parts, but from what you said, now I'm not so sure anymore. Like how do you actually handle the manufacturing of those parts?

For the SoC, that's true. It's a fabless model, Stacy. On the preamp, it's a process technology that we own. And it's custom and developed by us. So it's more of that kind of internal model.

Okay. Do you actually make them yourselves in your own fabs? Or do you like outsource them using a process that you like transfer?

No. We make our own fab.

You make them. Okay. Interesting.

To keep more of the differentiating strategy.

Got it. Can you talk a little bit about what makes it special? I gather it's a mixed signal, so like analog, digital kind of thing. But is there anything specific that you want to call out?

Specific for the preamp or both the preamp and the SoC?

Specific to the preamp.

For the preamp, as I said, it's a process technology that's developed by Broadcom, so it's a proprietary process technology. But you can imagine if you're developing the foundation that our design team works with to provide the best products in the industry, we focus on the parameters inside the process, and we can tune those specifically making other trade-offs in the process to provide the best performance. So that kind of gives you a flavor of how we do that. We've done that for several generations.

Got it.

Stacy, if I can add to that. I can add to that because I know why you're driving it. Now there are people out there who talk about doing preamps. You don't do it outsourcing, not if you want to compete against us. Because the process we developed is unique. It's a unique process, and it's not available easily out there in the foundry industry. And that's what creates the differentiated performance. And not to mention, we own the fab. So the cost to us per wafer is half the cost if you try to do a foundry with suboptimal process.

Got it. Interesting. Another question on HDD on the Pigeonhole here. Somebody would like to know what happens to HDD and preamp content as the industry moves to HAMR drives? And what your view in general is on how much of nearline HDD shipments HAMR will represent over time?

That's a very good question. So the HAMR technology requires a laser driver, which is an entirely new IP block that we add to our preamps to drive the lasers that are in the disk drives. The amount of capacity that can be increased with the addition of laser drivers and the HAMR heat-assisted magnetic recording technology is significant. And we do absolutely expect a transition to this technology. But it's a whole system solution that's being worked out. We think the technology is mature and it will absolutely ramp in the future. I can't provide an exact date, but it will ramp.

Got it. That's helpful.

And Stacy, just to follow on to that question. On the storage connectivity -- because Dan and I go hand in hand, the drives in enterprise and data centers attached to storage adapters. Multi-actuator technology and innovation in the drive space is actually driving refreshes on our side. If you go back 5, 6, 7 years ago, the industry was so maniacally focused on SSDs that you kind of saw a little bit of innovation in the HDD space not as much in the limelight. But the top hyperscalers have made it clear that HDDs are going to be core to their storage infrastructure for decades to come. And we're seeing actually a wave of refreshes for the top hyperscalers and in enterprise of HDD systems, purpose HDD systems. And they want to make sure from a connectivity perspective, they're future proof for the multi-actuator drives.

Got it. Got it. One quick question just popped up here as well. I know you mentioned that, like client isn't a huge driver of this business. But you do have some client exposure in the HDD business, I assume?

Dan, go ahead.

Yes. The answer is -- it's a good question. We do have some client business, but it's not the majority. The majority of our business is nearline.

Got it. And I guess given the trajectory that you put for nearline growth, it sounds to me like -- again, going back to the growth story in general. You do think the HDD business is a growth business. It can grow. Again, more in the cyclicality and everything that can be there. But...

Yes, I absolutely think it's a growth business for us, yes. Like Hock talked about it, it's maybe single-digit kind of growth, but it's a growth business.

Got it. There's a question here on the Pigeonhole ask -- people asking about CXL, Compute Express Link. Is that a protocol that's important for this business? And do you have any plans to support it? And are there any consequences if you do not?

So Hock, I'll field that question.

No, please. Yes, Jas.

So there's -- if you zoom into a data center, there's the -- connectivity landscape is quite complex. The pillar for connectivity inside the data center is Ethernet. You've got Fibre Channel. You've got multiple HDD and SSD technology, SAS, SATA, NVMe. And you've got specific memory interconnect like DDR and so forth. And then there's another category of basically AI, ML, accelerator interconnects. Historically, that has been PCIe as the interconnect of choice. And we're actually the market leader in PCIe interconnect inside the data center. That's an area that we invest quite a bit. Now what has happened with -- there's been a whole series of different connectivity protocol, Gen-Z, CXL, NVLink for that type of application. And many people in the industry have been chasing the holy grail of data center connectivity, which is, let's have one connectivity fabric that will cover storage, networking, accelerators, memory, the whole works. Personally, I don't believe that that's going to materialize. And I think CXL is going to be very successful in the memory disaggregation space, where basically, companies want to serialize and disaggregate DDR and other types of memory, which historically has not been interconnected. That has been dominated by DDR as a protocol directly interconnected to the CPU. So I think CCX -- CXL is going to be good initiatives. We're very supportive for memory disaggregation. Now for AI, ML interconnect and things of that nature, whether it replaces PCIe or not, we'll have to see. But it is -- it's very interesting to follow the dynamics around protocol, races and different interconnections inside the data center. And there's been many that have come and gone, but Ethernet, Fibre Channel, PCIe storage connectivity protocols, DDR have been there to stay. So we'll see what happens.

Got it. That's helpful. Question on one of the connectivity areas you play in. So you obviously mentioned you're #1 in 3 areas. What about Ethernet? What is your competitive positioning in Ethernet?

Jas?

Yes.

Go ahead.

So Ethernet is a very important segment, and we have #1 market share in Ethernet switching, both in hyperscale and enterprise. That's an area where separate franchise -- and there was a teach-in session on that. We -- historically, Broadcom has been #1 in Ethernet connectivity, Ethernet adapters. Since then, we've invested in different areas. And right now, we're really focused on investing and leading in simple enterprise NIC technology. And we're seeing a lot of refreshes in that space. We're seeing transition from 10 gig to 25 gig happening in the enterprise, and that's driving ASP gains and the refresh -- ASP gains and refreshes. And in the hyperscalers, we're seeing the transition from 25 gig to 100 gig, 200 gig. And we're also seeing in HPC in different subsegments and enterprise, adoption of 50 gig, 100 gig technology strengthening and adapters. So that's an area where we're focusing our investment. And we are -- we have aspirations to regain the #1 position in enterprise space.

Well, Stacy, let me expand on that. We have not much invested in that space. What you're saying NICs, network interface cards is what you're saying, right? Cut to the chase. We've not considered it as a very strong differentiated space previously, historically, when it was 10 gigabits and below, not a big deal. And we do it, we have the capability to do it, but we have not focused on it. Since it's gone up above 10-gigabit connectivity, the bandwidth, 25, 50, now 100, we have stepped up our focus on investing and basically creating products in that space. And as Jas said correctly, we gave them the level of investments and given the fact that it's, in a way, looked at as really a key part of connecting to the server, of a server connecting to the network or connecting to storage for storage control for storage connectivity sometimes. We have made a -- we have basically taken an initiative that we want to be going to be #1 in this space. And we are well on the way to doing that with the level of investments we have made, especially in the 50 and 100 gigabit bandwidth NICs.

Got it. How long did...

And Stacy, just to add on to that. The recipe for success is very common across the different storage adapters that go into servers. It starts with leading process and leading IP. So typically, on these technologies, we're one process node ahead of our competition, which derives better power, better performance. The second thing is the SerDes, which is the underlying technology where you get the reach and the connectivity piece -- because at the end of the day, this is all connectivity. You need to have the best SerDes with the best reach, best power, best performance. And this is an area we really out-invest the competition, and it differentiates significantly on SerDes. And this is shared across the different storage connectivity. Then the second aspect is a mindset around low-power hardware accelerated technology to deliver performance. And the third recipe for success, which is common, is the system mindset, is delivering the purpose-built silicon with the software, with the adapter and having an execution machine that basically outperforms the competition when you integrate all these things together. That's the secret sauce to our success. And we're going to apply this, like Hock said, to high-performance Ethernet NICs.

Got it. That's helpful. That's helpful. There is -- there are other questions here on -- hang on, there was one I just saw that I wanted to grab. Hold on a minute. Here we go. Somebody is asking about cloud storage adapters. They wanted to think through how content per drive for Broadcom varies, I guess, between cloud and enterprise. In particular, the question is, is there lower Broadcom content per drive for cloud storage adapters since data protection is performed at the software level? Or is that just not an area where you guys play anyways?

Stacy, exactly right. And I think as Jas put up in one of his early slides, with the exception of Dan Dolan's hard disk drives, where a big part of hard disk drives is not in; enterprise, but in the cloud. And certain cloud or large hyperscalers like -- well I won't mention names, have used hard disk drives as their core storage. What Jack and Jas largely focus on their business is driven through enterprise. Cloud use software replication, as Jas said, for really data protection and data recovery. And in Jack's case, totally enterprise, not cloud at all. And in most of Jas' case, I would say most of it, very little cloud.

Got it. There's another question, I think, around some of that. Again, Jas talked about the [ voyage of ] data protection software. Somebody would like to know if there are indeed synergies with Symantec, especially around data loss prevention products and other capabilities that you can or would bring with future products along those lines.

So security is key for our customers, but it's different type of security than what a Symantec would offer. So the security aspects that we're really focused on is protecting the data from a physical perspective and protecting the server infrastructure from external attacks. So we're very focused on security, but it's a different, complementary security than application-level security protection.

Got it. That's helpful.

But overall, I would say that we differentiate on security. And overall, as a company, we have a mindset around being leaders in security at the right level for the technology that we build.

Got it. That's helpful. We're bumping up on 1.5 hours. So I'm going to, I guess, open it up to Hock or any of you if there's any closing remarks, anything else anybody wants to say at this point as we bring things to a close.

Stacy, not at all. And if you have any further questions you really want to pick by all means, continue. But no, we don't have anything to add.

Okay. There are more questions here. We can keep going on some of these.

Happy to do that, Stacy.

Yes. There's a question on market structure. So I know you said -- and they were pretty -- talking specifically about Marvell. I think in -- across a number of these segments, you do compete with Marvell. Are most of these primarily duopolies? Like hard drives is, I think, is you and Marvell. You talked about a Fibre Channel like HBAs. I think it was Emulex and QLogic as you and Marvell. Can you talk maybe just a little bit broader around the market structure, not just in the connectivity, but maybe more broadly across these different things. I think...

Well, the most interesting one, you talked about -- you allude to and you're touching on earlier in your question, Stacy, is hard disk drives. And I mean we made a point of saying, our legacy, as Dan indicated, has been where we started from years back is when client was a bigger part of data storage in enterprise than were nearline hard disk drives. Marvell has always been in the SoC. They did a large -- they were a large SoC player in client. Nearline had been more LSI from which it became Broadcom. And we use that -- and as we're lucky as nearline expanded dramatically, we are very well positioned to strengthen where we are. And today, by far, we are nearline SoCs. And if you take the broader market, we have a larger share on the SoC side in nearline hard disk drives and enterprise drives. Client, we never had much position in the SoCs, as Dan indicated. In preamp, we're basically the only guy. Regardless what you may hear out there, Stacy, we are the practically only guy. All right, no matter what you may hear out there. Now in host bus adapters, Fibre Channel, you're right, that only 2 players. And that's Emulex and QLogic, which is -- which has been -- which was acquired under the Marvell umbrella. And you saw the market position, relative market position of 2 players only 2 players. In Fibre Channel, when it come to the switch storage area network, as Jack indicated and Jas put out in his one slide, there are only 2 players. That's Brocade with over 70% market, and there's the other guy, which is a subset, a subvertical within Cisco, MDS, and that's where it is. And as I said, let's put it bluntly, the competition in server storage connectivity when it comes to enterprise is interestingly limited.

I guess given that enterprise focus, like there's another question I have around synergies with some of your other franchises, for example, particularly networking and Ethernet switch. You talked a little bit about that. But just in general, are there broader synergies or complementary nature between aspects of storage and some of your other franchises? Or are they still pretty siloed?

They are starting to converge, Stacy. And you hear a lot of marketing talk out there that as part of running -- trying to manage networking connectivity better, you take -- as Jas start to touch on, you touched that little beachfront property on the server called network interface controller or network interface card, called NICs, from being just a plain translator from PCI Express from the CPU to Ethernet on the network to perhaps creating some intelligence in those NICs. And they call them Smart NICs, which I have indicated in the earnings call before and Jas is saying. Or else not -- people heighten it by calling them data processing unit, DPUs. All that nice stuff is coming down to -- hasn't emerged yet until more than a hill of beans at this point, Smart NIC. Because it's too customized on workloads, but also this: Running it in a server, it's a totally different game with totally different requirements, understanding from running a network where you had to understand the characteristics of network. And the key thing is congestion control, management of workloads in the network to ensure, as Jack start to touch on, historic, how you want to transfer data without congestion. Whereas in a server, no such thing happens. And right now, to be honest, there could be a convergence where the beachfront property called NICs might emerge as an interesting content increase for our networking vertical or networking segment. As of right now, it's very, very server focused, as Jas indicated very clearly. And we are now focused on just becoming the #1 player in that space.

Got it. I have another question here. I guess to get to the growth targets that you've talked about, obviously, pricing seems to be playing a very large role across a number of these different segments. Can you give us some idea of which segments are maybe more dependent on increasing ASPs for growth versus units? Or are they all dependent on pricing?

No, not all, but this is a very price insensitive market, as I indicated earlier, as you gather from it. We're ubiquitous and it's mission-critical and it is -- and it's -- a lot of it is very trusted. Because you think about it, right, for Jas as an example, just using example, the rate controller for data protection, data recovery, sitting, attached to a server barely represents 8% of a server value total ASP. But you bring -- but if it goes down, all hell breaks loose. So it's -- more important is quality, reliability and performance. Jas, do you want to add to that?

Yes. I'll add to that a little bit. So it's -- I think the important thing, the ASP increases are linked to technology refreshes. So example, the underlying interconnect inside of server is PCIe. To go from PCIe, Gen 3 to Gen 4, it took the industry 8 years. Now we're going from Gen 3 to Gen 4 and potentially Gen 6 in less time than it took to go in 1 generation. And it's a doubling of performance every time. And so that creates opportunities, but it makes it very difficult if you don't have the investment scale to keep up with this pace of innovation. And it all goes back when you're doubling these performances and these technology refreshes, you've got to have the core IP, the core process. You've got to have the architecture innovation to deliver the performance. So it's making it more difficult for new entrants to come in or for competitors that don't have the revenue scale to invest and keep up with that technology refresh. So really, the ASPs are linked to technology refreshes. And if you look at an Ethernet, 1 gig, 10 gig, now we're going from 25, 50, 100, 200 gig, the cadence of these refreshes is actually accelerating. There's so much appetite for data services inside core data centers that the industry needs these technologies.

Got it. Is it fair to say -- like I'm curious like how -- like on a qualitative basis, how the investment levels vary across the segments? I mean like looking outside, it would seem to me that probably like the storage connectivity space might have like higher different or higher levels of investment than, say, HDD. But like is that how it is? Or is it more uniform across the different segments? How do you think about allocating the R&D?

No, it's not allocated. Each of them are run very different, very uniquely. We have our own -- in each of these product segments, as we do have in the other 20-odd product segments out there -- product verticals we have out there, each runs their own R&D team. Each runs their own marketing team. And the level of R&D varies very dramatically. And obviously, the longer the product cycle, the R&D percentage of revenue is actually lower obviously. Because you spread it out over a longer time period. The area on the curve doesn't change that much by the end of the day, just spread it out further. So as you probably know, in the hard disk drive side, this nearline capacity drives changes now every 2 years, virtually, right Dan, almost every 2 years.

That's right.

So when a new set of products -- we have a new set of products, for 14 terabytes 2 years ago as it is for 16 terabytes, which is a sweet spot almost 6 months ago to 18 terabytes today. And next year, second half will be 20 terabytes, all new set of products. So the cadence of investment speeds up very dramatically versus, say, in the case of Jack, where the product cycle is more 5 years. And in the case of Jas, it's more like 3 years, right, Jas?

Yes. Somewhere in the middle.

So the percentage of revenue correspondingly matches the speed of innovation.

That's interesting. So like what I'm hearing you saying is that again, I get your concept of integrating under the curve over time, but it sounds to me like the hard drive business might actually have a higher percentage of revenue of current investment versus some of the other areas. I would argue that's not how most investors probably think about it.

No. Yes, isn't it fascinating? There's a lot of NICs running around the marketplace. It isn't. I mean we're investing in both SoCs and thinking about innovation happening in the hard disk drive industry to create this monster terabyte drives, which will give -- why it keeps growing in throughput capacity is because hyper cloud guys, whether it's Google or AWS, is using that as a long-term lowest cost storage, total cost of ownership, and they get a low and low TCO the higher the capacity. And it becomes a challenge, not just for the hard disk drive guys, which you probably -- many of you hear in your calls and analyst days from Seagate and Western Digital, where they not only use energy-assisted -- and HAMR is definitely one energy monster energy assisted. They also put in a higher number of disks, and then they go into multi-actuators as I said. All that impacts the content we put in our preamps and our SoCs, step by step. And it's pretty fascinating because some of the leading SoCs for -- I mean, the highest capacity drives are pushing almost to leading-edge silicon to create the SoCs. As we have to do that, Dan indicated, in updating our process technology, which is an interesting process in the sense that it's basically -- it's some form of a hybrid process from silicon and germanium and a bunch of other stuff, to press the kind of performance, which is on reading and writing especially on the media that hard disk drives require.

Wow. Very interesting. Really. We're running up -- we're 15 minutes past. I think we'll close it out here. Most of the questions, at least on the Pigeonhole, have been answered, and a lot of mine have as well. But if anybody else does have questions, please feel free. You can follow up with the company, you can follow up with me. And I'll just see if there's any other closing remarks. And if there's not, then I think we can close it out here. Thank you so much. This is really interesting and fantastic.

Thank you. Thank you.

Thank you, Stacy, for hosting.

Yes. Thank you.

My pleasure, my pleasure. I think we'll close it out there then. Thank you for joining, everybody. And again, if you have any questions, please feel free to reach out to me or to Broadcom. And with that, have a nice day.