Ladies and gentlemen, thank you for standing by, and welcome to the Second Quarter Fiscal 2021 Marvell Technology Group Earnings Conference Call. At this time, all participants are on a listen-only mode. After the speaker presentation, there will be a question- and- answer session. To ask a question during the session, you will need to press star one on your telephone. Please be advised that today's conference is being recorded. If you require any further assistance, please press star zero. It is now my pleasure to introduce Vice President of Investor Relations, Ashish Saran.
Thank you. Good afternoon, everyone. Welcome to Marvell's Second Quarter Fiscal Year 2021 Earnings Call. Joining me today are Matt Murphy, Marvell's President and CEO, and Jean Hu, our CFO. I would like to remind everyone that certain comments made today may include forward-looking statements, which are subject to significant risks and uncertainties that could cause our actual results to differ materially from management's current expectations.
Please review the cautionary statements and risk factors contained in our earnings press release, which we filed with the SEC today and posted on our website, as well as our most recent 10-K and 10-Q filings. We do not intend to update our forward-looking statements. During our call today, we will refer to certain non-GAAP financial measures. A reconciliation between our GAAP and non-GAAP financial measures is available on our website in the investor relations section. With that, I'll turn the call over to Matt for his comments on our performance.
Thanks, Ashish, and good afternoon, everyone. During the second quarter of fiscal 2021, we delivered strong financial results and achieved $727 million in revenue, $7 million above the midpoint of guidance. Revenue grew 5% sequentially and 11% year-over-year. Our GAAP loss per share was $0.24. Our non-GAAP earnings per share was $0.21, above the midpoint of guidance, driven by higher revenue, better gross margins, and lower operating expenses. We are now halfway through our fiscal year, a time period which has turned out to be very different than we had all imagined, impacted by a global crisis that has significantly disrupted our daily lives and altered our work environment. The Marvell team has met these challenges head-on and has been executing at a very high level.
I'm continually impressed by the relentless determination of our engineering teams in the face of a multitude of unprecedented circumstances, like having to manage a chip bring-up remotely. I have proudly watched our sales, marketing, and support teams embrace Marvell's new brand identity and use current circumstances to increase customer engagement. In times of uncertainty, I believe that we benefit significantly in building upon our existing deep-rooted engagements. When customers make forward-looking platform decisions, their experience with our products and trust in our team works to our advantage, and we have continued to close design wins at an impressive rate. While getting a design win is a critical milestone, getting our customers to production and volume ramp is the ultimate goal.
To that end, over the last few months, we have received customer approval to move into production several key programs, including a critical SSD controller and a cloud ASIC, which are both ramping now, and a 5G baseband processor for Nokia that is expected to start ramping later this fiscal year. Let me now provide an update on our process technology platform and our custom ASIC chip strategy. Underpinning the development of all our high-performance storage, networking, compute, and security products is our industry-leading IP portfolio, developed on the leading edge of process technology. Earlier this week, we announced an extension of our long-term collaboration with TSMC, the world's largest dedicated semiconductor foundry, to deliver a comprehensive silicon portfolio leveraging the industry's most advanced 5 nanometer process technology. This joint development positions Marvell for multi-generational leadership in data infrastructure technology.
This accelerated cadence in driving to the latest process node and jumping to five nanometers is the outcome of our multi-year evolution to focus on data infrastructure. In the infrastructure market, 5G, cloud, enterprise, and automotive customers are migrating to silicon partners who can solve their most challenging problems, requiring the highest performance to be delivered within a very tight power budget. This requires access to the latest process technology. The acquisition of Cavium, with their processing and compute-heavy portfolio, further crystallized the need to change our strategy from a fast follower to a technology leader. The later acquisition of Avera was done with the recognition that our advanced process roadmap would be a significant step up for Avera's ASIC technology platform and would also increase the volume of projects in advanced geometries with the subsequent benefits of scale across all of Marvell.
The leadership position we have established for our process technology platform is the culmination of the last two years of hard work by our central engineering team. This has been a key driver of our recent success in winning sockets in multiple upcoming platforms, such as next generation 5G base stations. Marvell's five nanometer platform solution covers the full spectrum of infrastructure requirements, including high speed, 112 gigabit per second long reach SerDes, processor subsystems, encryption engines, System on Chip fabrics, chip to chip interconnects, and a variety of physical layer interfaces. We have multiple five nanometer products in development now on TSMC's N5P process, an enhanced version of TSMC's five nanometer technology, which delivers higher performance and up to approximately 40% lower power compared to the previous seven nanometer generation. We expect to start sampling these products by the end of 2021, with volume production soon thereafter.
Our advanced technology is a key enabler of our new custom ASIC platform, which builds upon the Avera acquisition. While customizing silicon is not new to Marvell, this acquisition significantly expanded our capabilities to become a much broader custom silicon supplier. We benefit from Avera's decades of experience in developing over 2,000 complex, full custom ASICs as part of IBM and GlobalFoundries. We believe that access to our more advanced process platform, combined with exposure to Marvell's large customer base, opens up a host of new opportunities for the Avera team. Equally important, we are evolving the traditional ASIC model by including all of Marvell's leading standard product IP into Avera's custom platform. The breadth of IP we're offering for customization is what sets us apart. We aren't just providing high-speed SerDes and other foundational IP.
We are also providing highly scalable multi-core Arm processors, Ethernet switches and controllers, storage controllers, and more. This IP can be integrated on chip or as a companion chip, tightly coupled with advanced system-in- package solutions. Customers can reuse Marvell's hardened and widely deployed IP and focus their engineering teams on only the unique aspects of their application, benefiting from faster time to market and lower program risk by using our proven technology. I'm very pleased to announce that our ASIC team has recently won a major design win under the Marvell umbrella with a Tier 1 hyperscaler. This is a new ASIC account for the Avera team. We hope that this is the first in a series of design wins with this important customer. Let me now provide an update on our server processor strategy.
Very much aligned with our growing emphasis on custom solutions, we are evolving our Arm-based server processor efforts towards a custom engagement model. Over the last few years, we have developed multiple generations of Arm server processors and have been working with customers to qualify our products for volume adoption. Our ThunderX2 processor was the industry's first Arm-based processor capable of powering dual socket servers, which could go toe-to-toe with X86-based solutions, and it established the performance credentials for Arm in the server market. Hyperscale data center customers represent the largest opportunity for Arm server processors. Having worked with them for multiple generations, it has become apparent that the long-term opportunity is for Arm server processors customized to their specific use cases rather than standard off-the-shelf products.
The power of the Arm architecture has always been in its ability to be integrated into highly customized designs optimized for specific use cases, and we see hyperscale data center applications as no different. With our breadth of processor know-how and now our custom ASIC capability, Marvell is uniquely positioned to address this opportunity. The significant amount of unique Arm server processor IP and technology we have developed over the last few years is ideal to create the custom processors hyperscalers are requesting. Therefore, we have decided to target future investments in the Arm server market exclusively on custom solutions. The business model will be similar to our ASIC and custom programs, where customers contribute engineering and mask expenses through NRE for us to develop and produce products specifically for them.
We believe that this is the best way for us to continue to drive the growing adoption of Arm-based compute within the server market. While we continue to invest in a number of initiatives to drive long-term growth, such as advanced process technology, new markets such as automotive, and the launch of our new brand, our team has also remained focused on driving operational excellence. The successful integration of Aquantia and Avera, continued operational discipline, and the change in scope of the Arm server project are collectively driving a significant reduction in our quarterly operating expenses from the current $300 million non-GAAP run rate to the $280 million we are guiding for the third quarter.
Our success in custom engagements, where we do not need to bear the cost and risk of new product development completely on our own, has also enabled us to manage OpEx tightly while continuing to invest in advanced technology. The increased willingness of our customers to co-invest with us is truly a testament to the trust and belief they have in our capabilities to provide them with differentiated solutions. The improvements in efficiency and the cost structure of the company allow us to improve our earnings power while we continue to invest more in R&D as a % of revenue than any other semiconductor company our size. We are investing heavily because we believe we can grow faster than the industry, and we have the conviction in our strategy to drive long-term growth. However, we remain extremely diligent about how we allocate the company's resources.
Last week, we held our annual strategic portfolio review, where we deep dive into each of our product lines, evaluate end market dynamics, and review progress against our goals. This review is the cornerstone of our annual business planning process and lays the strategic foundation for everything else. This was our fifth review since I joined Marvell in 2016, and it was by far the best one. It is clear how far we have come over that time period, and I could not be more impressed by the quality of our leaders and their depth of experience and technical capabilities. Every year, this process informs our capital allocation decisions and has driven a significant improvement in the quality of our businesses over time. I expect this cycle will be no different as we continue to drive long-term success for our employees, customers, and investors.
Let me now move on to discussing our two businesses in more detail. First in our networking business, revenue grew during the quarter was $406 million and grew 3% sequentially and 23% year-over-year. Growth was driven primarily by ongoing 5G deployments in China. In the cloud data center end market, we started ramping a new custom chip at a Tier 1 hyperscaler. Our LiquidIO programmable SmartNICs and LiquidSecurity HSMs continue to gain traction at cloud customers, and revenue for these two products more than doubled from the same quarter last year. We are very pleased with these results, which came on the heels of the very strong growth we had delivered in the prior quarter.
This was also the fourth consecutive quarter of sequential revenue growth from the wireless infrastructure market, as we benefited from the start of the 5G transition and our diversified design win position at four of the top five Tier 1 base station OEMs. In addition to the top five global OEMs, there's also a very active next tier of more regionally focused OEMs who are developing their own 5G base station equipment. We believe that our full suite of production ready 5G silicon solutions, including the industry's only merchant baseband and processor, provide a very compelling path to market for this set of customers. I'm excited to announce we have now secured design wins for our OCTEON embedded processor and our OCTEON Fusion baseband processor with a new 5G customer in this next tier of base station OEMs.
We expect this design to start ramping towards the end of next fiscal year. We continue to engage with this group of customers and look forward to updating you on our progress. We remain on track to start shipping basebands to Nokia and processors customized for massive MIMO applications to Samsung later this year. We believe that OCTEON can continue to improve its position within the large $4 billion market for embedded processors in infrastructure applications. OCTEON has a number of very compelling advantages, including our process technology platform, multi-core Arm architecture, optimized hardware accelerators, flexible programming model, and our ability to customize solutions. Compared to alternative products, our next generation processors, designed on the industry's most advanced 5 nanometer process using TSMC's proven technology platform, have become a very attractive solution for customers.
We believe that our competitive differentiators will enable us to continue to take share in this large market. Shifting gears to our Ethernet portfolio. We've been working closely with our customers to design the next generation of Ethernet products tailored to meet the new requirements being placed on networks by mobility and cloud applications, which are extending the boundaries of the traditional enterprise. This quarter, we announced our latest Ethernet solutions for the emerging world of the borderless enterprise. The borderless enterprise is not just about the transformation of on-campus corporate environments such as Marvell's own infrastructure, but also includes new use cases in the rapidly evolving retail, manufacturing, hospitality, finance, and education verticals. The number of devices connecting to the network is expanding rapidly, and no matter where the users are physically located, enterprise IT organizations are being asked to deliver a secure high bandwidth experience.
The increasing amount of cloud access and remote work needs switches with more intelligence, visibility, security, and bandwidth. These requirements are driving the next refresh cycle. Access switches need to apply intelligent processing right at the edge and offload data to inference engines. The data needs to be encrypted and secured end to end with enhanced network visibility for a seamless experience. The adoption of Wi-Fi 6 is increasing the bandwidth requirements of wired networks. Marvell's unified Prestera Ethernet switch and Alaska PHY solution set is architected from the ground up to manage this complex and evolving environment. Our advanced telemetry capabilities facilitate network automation. Our switches have embedded cryptography-based security for Ethernet traffic. Intelligent workload management enables optimized data processing at or near the network access edge, improving the performance of hybrid cloud architectures.
Our Multi-Gig capable switches and PHYs facilitate the transition to 2.5 and 5 Gig connectivity, which is essential for Wi-Fi 6 enabled access points and high performance end devices. Our software development kit enables networking system vendors to easily develop new and differentiated products quickly on our silicon platform. As you may recall, we refreshed our Ethernet portfolio soon after I joined Marvell in 2016. That led to a sustained period of growth for our switch and PHY businesses as we took share with our differentiated products that commanded higher gross margins and outgrew the market. This year's refresh adds an even more impressive set of capabilities. Addresses a larger part of the enterprise market, further expanding in the access, core, and aggregation layers. We believe that we will continue to drive share gains.
In our emerging automotive business, we continue to make progress in expanding the customer base for our Ethernet switch and PHY connectivity products. We have recently closed design wins with eight new customers, adding to the 16 I have discussed in prior calls. We also continue to see a larger opportunity for us in automotive. Leveraging our growing Ethernet position, we are front and center in discussions around future architectures and have the opportunity to extend into the adjacent compute, security, and storage domains within the connected car. Let me now discuss the outlook for the third fiscal quarter for our networking business. We expect strong growth from the wireless infrastructure end market driven by multiple customers. We also project growth from the cloud data center market. We expect our automotive Ethernet products to start shipping into model year 2021 vehicles, which are starting production now.
However, given the softness in the enterprise market, driven by COVID-19, and recently telegraphed by multiple customers and peers, not surprisingly, we project demand for our networking products selling into enterprise applications to be weak. The net result is that we expect our overall networking revenue in the 3rd quarter of fiscal 2021 to increase sequentially in the mid to high single digits on a percentage basis. Turning now to our storage business. Storage revenue for the 2nd quarter was $290 million, growing 12% sequentially and 6% year-on-year. Strong growth was driven by the start of a ramp of a customized SSD controller for a do it yourself, or DIY program, and easing of COVID-19 production challenges we and our customers experienced in the 1st quarter. Supply chain improvements benefited both our storage controller and Fibre Channel products. Controller demand from nearline 16 TB HDDs was particularly strong.
The combination of our product cycles and supply recovery drove our outperformance as compared to the drive market, where demand was starting to get impacted from COVID-19. Our pipeline of SSD controller engagements continues to grow, driven by our commitment to deliver the most responsive and best user experience at the lowest power using the most advanced process technology. I'm very pleased that we have won the next generation of SSD controllers for data center applications at one of our key Tier 1 NAND OEMs, extending the multi-year relationship we have with them. This controller features our leadership PCIe Gen 5 technology. We expect to continue to drive long-term growth for our storage business. Looking to the third quarter, we are expecting very different trends between our storage controller and Fibre Channel businesses.
We project our storage controller business, which addresses both hard disk drive and solid state drive applications, to continue to grow sequentially. We expect this growth from the continuation of a ramp of a custom SSD controller, partially offset by weaker drive demand from enterprise data centers and some edge applications such as retail. Storage controllers shipping into cloud applications are also expected to continue to trend up in the third quarter. Conversely, in our Fibre Channel business, we project a significant sequential decline in revenue resulting from COVID-19 related weakness in enterprise server and storage system demand. The weakness in Fibre Channel is expected to offset the growth from storage controllers, and as a result, we project third quarter consolidated storage revenue to be approximately flat on a sequential basis.
In closing, our results continue to validate our strategy to focus on developing the most advanced silicon for data infrastructure. Our portfolio actions have also significantly diversified our end market exposure, and we have a much larger share of revenue today from fast-growing 5G wireless and cloud end markets. In 5G, we have racked up an impressive set of design wins, which we expect will drive significant revenue growth for us over a number of years. The product ramps at Samsung and particularly Nokia are largely in front of us. In addition to the strong momentum expected from our own product cycles, we are also turning increasingly positive on the likelihood of our diversified customer base to gain share in light of recent geopolitical events and market dynamics. Cloud has only recently become a larger part of our business, crossing over 10% of our total revenue.
In this market with long-term secular growth, we have multiple drivers through our OCTEON processor platform, our security products, full custom ASICs, merchant and DIY data center SSD controllers, and nearline HDD controllers and preamps. In enterprise, while there are near-term headwinds from COVID-19, it is important to keep a longer-term perspective that this is a large and diversified worldwide market spread across a number of industry verticals. The need for secure and intelligent access to bandwidth is not going away, and the number of endpoints trying to connect to a network are only expanding. We have introduced new solutions specifically designed to address these challenges, and we believe that we can gain share and drive revenue growth from our own product cycles.
In our edge end market, we have barely scratched the surface of the current Ethernet and future compute opportunity in autos, which we believe will become another key driver for long-term revenue growth. We have assembled under one roof a critical mass of scarce and unique IP with a very flexible and customized engagement model, which is proving very attractive to our customers. We are accelerating our adoption of advanced process technology, which has already started to pay dividends in the form of design wins and new sockets and customers. Our borderless enterprise and ASIC announcements received strong validation from the industry, and we have seen an increase in inbound requests for collaboration from customers as they become more aware of the depth and breadth of technology at Marvell.
We believe that we have built a business for the long haul and are confident in driving revenue growth and managing through any transitional challenges in some of our end markets. We continue to invest in technology while reducing overall operating expenses by driving higher levels of efficiency with our platform. This operational excellence enhances our ability to deliver operating leverage and drive earnings growth. With that, I'll turn the call over to Jean for more detail on our recent results and outlook.
Thanks, Matt. Good afternoon, everyone. I'll start with a review of our financial results for the second quarter and then provide our current outlook for the third quarter of fiscal 2021. Revenue in the second quarter was $727 million, above middle point of our guidance. Networking represent 56% of our revenue in the second quarter, with storage contributing 40%. Revenue from other accounted for 4% of revenue, declining 25% sequentially and 41% year-on-year. As a reminder, this business consists of a product we have stopped investing in. We expect they will continue to decline over time. Our guidance for the third quarter anticipates a small sequential decline in revenue from these products. GAAP gross margin was 49.4%. non-GAAP gross margin was 63.3% of revenue, better than expectations, reflecting the hard work from our operations team to drive operational efficiency to improve our product cost.
GAAP operating expenses were $511 million, they include the cost of share-based compensation expenses, amortization of acquired intangible asset, acquisition and divestiture-related cost, as well as the impairment and other related restructuring charges as a result of the changing scope of the server processor program Matt discussed earlier in the call. Non-GAAP operating expenses were $297 million, $3 million lower than expected, primarily because our continued focus on OpEx management. GAAP operating loss was $151 million. Non-GAAP operating profit was $163 million, or 22.4% of revenue. For the second quarter, GAAP loss per diluted share was $0.24. Non-GAAP income per diluted share was $0.21, above the middle point of the guidance. Now turning to our balance sheet. During the quarter, cash flow from operations was $226 million. Our team continues to drive the improvement of working capital metrics in the second quarter.
We improved our days of sales outstanding to 61 days and days of inventory to 90 days. We returned $40 million to shareholders through dividend payment. We have temporarily suspended our share repurchase program as we believe it's prudent to further strengthen our liquidity and increase our cash balance during the uncertain environment. We'll continue to evaluate the business conditions to decide when to restart share repurchase program. We exit the quarter with $832 million in cash and short-term investment, an increase of $164 million from the prior quarter. We continue to have $500 million of liquidity available from our undrawn revolver. Our net debt to EBITDA ratio was 0.8 x on a trailing 12 months basis. As of 1st of August 2020, the $450 million term loan is due within 12 months and has been classified as short-term debt on the balance sheet.
We expect our business to continue to generate a strong cash flow, and we intend to repay this amount with the cash flow from operations. I'm pleased that the revenue growth, combined with our strong business model and operating expense discipline, continue to drive improvement in operating results and is strengthening our balance sheet. Now moving on to our current outlook for the third quarter of fiscal 2021. Please note that compared to second quarter results, our outlook for operating expenses reflects the significant improvement Matt discussed earlier from the successful integration of Aquantia and Avera, continued operational discipline, and the change in scope of the Arm server project. As a reminder, our operating expenses can vary quarter to quarter, affected by factors such as the number of tape-out within a particular quarter, and these tape-out are becoming more expensive in newer process geometries.
The cadence of NRE payment, which are primarily treated as contra-OpEx, can also add variability. We now have a higher level of NREs following the acquisition of Avera with their customer ASIC model. Due to the typical seasonality in payroll taxes, our OpEx in the first fiscal quarter tends to increase sequentially. This effect then dissipates in the rest of the fiscal year. Here is the specific guidance for the third quarter. We're forecasting revenue to be in the range of $750 million, ±5%. We anticipate our GAAP gross margin will be approximately 51.4%.
The non-GAAP gross margin will be approximately 63%. We project our GAAP operating expenses to be approximately $368 million. We anticipate our non-GAAP operating expenses to be approximately $280 million. We expect net interest expense to be approximately $15 million. We expect non-GAAP tax rate of 5%. As a result, we anticipate GAAP results in the range of a loss of $0.04 per diluted share on the low end to an income of $0.04 per diluted share on the high end. We expect non-GAAP income per diluted share in the range of $0.22-$0.28. Operator, please open the line and announce the Q&A instructions. Thanks.
Certainly. As a reminder, ladies and gentlemen, to ask a question, you will need to press star one on your telephone. To withdraw your question, press the pound key. Due to time constraints, we ask that you please limit yourself to one question, and if time permits, you may reenter the queue with any additional questions. Please stand by while we compile the Q&A roster. Our first question comes from the line of Blayne Curtis with Barclays.
Hey, guys. Thanks for taking my question. Nice results, obviously given the backdrop in enterprise. Matt, I'm just kind of curious, you mentioned as you walking through, you spent a lot of time on ASIC. You mentioned a new design win with a hyperscaler. I didn't know if you'd give any color with that. Maybe just thinking broader, if you can kind of just give us a flavor as the types of wins that you mentioned you're picking up and types of segments and, I think, in both across wireless and networking, that'd be helpful.
Sure, Blayne. Hey, thanks for the question. Yeah, on the hyperscale win, as I said, we're pretty excited about that. I think bringing the Avera team into Marvell has created a much more relevant portfolio and broader portfolio for all the hyperscale accounts. I think it was a great testament to the capability of the team to actually break into a new one. As you know, due to the nature of the ASIC business, these are highly confidential type of engagements that we have, so we can't go into a ton of detail. What I would say is that broadly speaking, our five nanometer platform is being extremely well received, both from an ASIC as well as a standard product offering.
What I mean is when you go out into not only hyperscale, but also the 5G market and enterprise, we're migrating the entire Marvell platform across multiple product lines, jumping into 5G. It's much broader than just an ASIC engagement. In fact, you should assume that our Ethernet business, our OCTEON platform, and others are all going to migrate. It's just created a very rich and compelling set of IPs and engagement model that's being well received. This is one example of that, where we're actually gaining new customers for new types of technologies. Hopefully there'll be more to come on that as we head towards the investor day that we just announced in early October to give even more details around our technology platform and the various engagements that we have, and talk about them in as much detail as we can at that time, Blayne.
Thanks, Matt.
Yeah.
Thank you. Our next question comes from the line of Vivek Arya with Bank of America.
Thanks for taking my question and congrats on the good outlook despite all the headwinds. Matt, I'm just curious. You mentioned two kind of headwinds. One, because of all the restrictions in terms of shipping to customers in China, and I think separately, you also mentioned some headwinds because of COVID and Fibre Channel. I was hoping if you could address and maybe quantify how much of those headwinds, and I think specifically on the China customer headwinds. Do those sales just go away? Is some of that recoverable with some license? I'm just curious, what is that China exposure now in terms of kind of wired, wireless, and enterprise mix?
Sure. Yeah. I'll answer it separately because I think they're two different dynamics. I think the China situation obviously is broadly impacting people and certainly in all of our disclosures, we always call it out as a risk factor just in general. Just to be very clear in terms of our guidance for the third quarter, there are no China headwinds included in that. That was not something that we're saying is a big part of any headwind we're seeing there. I mean, it could be minor, but it's not material. Really the central issue in terms of just the growth sequentially, which is still up quarter-over-quarter, certainly would have been better if enterprise had performed maybe where we thought they would have been a few months back, was really around the enterprise market.
We saw it in both our networking business, really where we sell Ethernet solutions into things like Enterprise Campus and SMB, also our Fibre Channel business, which is pretty broadly deployed in a number of enterprise applications, including storage. That also is projected to be down in the third quarter. That business, just to note, is generally pretty stable. It doesn't fluctuate around a whole lot, we definitely are noting it for Q3.
I would say, ex those trends in the enterprise very specifically, the rest of the business is obviously performing very well in terms of the strong growth we're seeing sequentially in our 5G customers. If you look out to Q3, assuming we achieve what we believe, that would be five quarters in a row of quarter-over-quarter growth in that market. In Cloud Data Center, again, another up quarter after a really strong Q1 and Q2. That's really the nature of the headwinds that we're calling out. They're very focused on the enterprise set of issues that are out there.
Okay. Thank you.
Sure.
Thank you. Our next question comes from the line of John Pitzer with Credit Suisse.
Hey, Matt, Jean Hu. Congratulations on the solid results. Matt, I was wondering if you could just spend a little bit more time on the 5G drivers. There's been some of your peers, after several quarters of very strong growth, that are kind of characterizing September, October as sort of a digestion period for the 5G wireless CapEx. I know you've got a lot of company-specific drivers, but I was hoping you could help me understand what % of revenue is 5G today, and as you look out to October, what % might it be? I guess importantly, how much runway do you have of company-specific drivers before you become a little bit more dependent upon the overall macro backdrop for spending?
Sure. Yeah. Great question, John. You're right to note that the performance we're seeing in that business is really very specific to us. Because we didn't have very large share in 4G, we're not really seeing that counterbalance that maybe others were that are more broadly exposed to both. As you've seen, we have new products ramping. We have new OEMs that are going into production. Certainly regionally, there are regions like China, as one example we called out, that's obviously very aggressive in deployment. We have a lot of wind at our back in terms of the 5G business. Of course, in front of us really is still all the other major geographies, which are projected at various stages to roll out 5G. We have new content with new customers also in front of us, with Nokia being an example.
I think the longer view is it's going to take us a while, which I think is a good thing, to be at a point where we'll kind of represent the market. I think the market needs to develop, 5G would need to be a bigger portion of the total wireless CapEx out there, our design wins would have to ramp. I think that's something that while it is a choppy market, we certainly feel like, at least at this point, based on our own product cycles and our own design wins that we achieved, that's why the business is performing that way. It continues to grow every quarter in terms of revenue as a % of total and just as a net amount.
When you think out to the investor day, we typically take that opportunity to frame the various markets and our sizes and our shares. Sort of look forward to the investor day to get a more comprehensive view from a market standpoint of where we are. We're very pleased at the growth rate of that business. Similarly to where we were happy with the growth rate of our cloud business and its representation inside the company. Both of those are working really well for us right now.
Perfect. Congratulations.
Thanks, John.
Thank you. Our next question comes from the line of C.J. Muse with Evercore.
Hey, C.J., are you there? Why don't we go to the next question, please?
Okay, one moment, please. Pardon me, C.J. Muse, please check your mute button.
Andrew, let's just go to the next question. Operator, do you want to just move us to the next question, please?
Hey, operator, can you just move to the next person's question?
I just got a message. I think the operator's having technical issues, so just hang on for a second.
Pardon me. This is the conference coordinator. I do show our next question comes from the line of Tore Svanberg from Stifel. Please go ahead.
Yes, thank you, and congratulations on the results. Matt, I had a bit of a broader question for you. You talked a lot about five nanometer. You talked about custom ASICs. I know recently you talked a lot about edge processing. It just seems like the world is changing right in front of us, and I'm just wondering what that means for Marvell as a company, but perhaps more importantly for the whole business model, because it does seem like we are a bit at an inflection point on all those fronts.
Well, I think you're right, and you've certainly looked downstream, Tore, in the design chain all the way to the OEMs. I think the level of digital transformation that's occurring, certainly in companies, in the equipment that needs to get designed to service these new needs is quite dramatic. We're aware of that, and I think that really is playing into our advantage, to be honest with you. I think when you look at what our customer base wants, it's a unique blend. They want access to the leading-edge IP, and maybe that sounds like table stakes, but they have their own very critical applications that they're trying to serve, and they're trying to do that, in a lot of cases, with much higher performance than they ever had to deal with before, or much lower power, especially as applications move closer and closer to the edge.
Increasingly as well, the other trend I would note is that the intersection of computing, of networking, security, and even storage, these key IPs for the data infrastructure, they're all tending to blend together in a lot of ways in terms of what customers want to go do. What that really speaks to is you need to have flexibility in your model in order to service these opportunities. I think one advantage we have, Tore, at Marvell, is that we've, by design, developed this extremely flexible business model from merchant offerings, where customers can prove out our IP because they can just go buy the part, to this partner model we innovated, where it's a combination of our intellectual property and our customers to drive time to market, as well as just being able to service this huge opportunity for custom silicon.
I think when I look forward, we're very enthusiastic about all the opportunities in front of us and also the way that the platform we are really able to leverage. When I mean platform, I mean our design win platform, our technology platform, where we don't have a lot of random adjacent businesses. What we do for a living is we do this data infrastructure silicon really, really well, and we do that in a very concentrated manner. You see that both in our now five nanometer platform, which we announced this week with TSMC, and the translation of all that investment and the focus we have, actually, surprisingly, even driving lower operating expenses than we had before. I think it's a unique combination. Tore, I would just say you nailed it. There's a big opportunity for Marvell on the go forward here.
Thank you.
Thank you. I'll show our next question comes from Karl Ackerman from Cowen. Please go ahead.
Hi. Good afternoon. Thanks for taking my question. Jean, it's great to see the solid execution on OpEx. I'm curious, though, what's the argument for OpEx to not trend toward 30%, even if we assume flattish revenue from here, particularly given the maturity of the storage TAM and the co-investment from customers for processors? Thank you.
First, we're really pleased our team has done a great job to reduce the OpEx to $280 million quarterly run rate versus our original expectation, it's more like $300 million, when we entered into fiscal 2021. That's just a tremendous effort by realizing more synergies from Avera and Aquantia integration and also just the operational efficiency across the board. If you think about that, we're investing the right level and with the portfolio adjustment, the portfolio optimization we constantly review as a company, we do see actually with driving the revenue ramp, if you look at our Q3 guide, it implies revenue year-over-year growth of double digits.
Continue the revenue ramp with the current operating expense level, we do think our model are going to show tremendous leverage and continue to drive the earnings expansion. If you look at our Q3 guide, the operating margin at the middle point of our guidance actually is close to 26%. I think we are setting up a great business model with the OpEx level to invest and to continue to expand.
Thank you.
Thank you. I show our next question comes from Quinn Bolton from Needham. Please go ahead.
Hi, guys. It's Michelle on for Quinn. Thanks for taking the question and congrats on the results. My question is on the storage business. I know you guys don't break out the storage controllers from Fibre Channel, but I was wondering if you can give some color on how the two segments are broken up within that bucket. Because I'm just trying to figure out the puts and takes for the Fibre Channel decline, and trying to determine kind of the magnitude?
Hi, Michelle. I'll help you, give you some color. We're typically going to give all the details and report all the details. To help you out, the market opportunity of SAN is about $500 million each year. It's quite a stable market, actually. It's a split between Marvell and Broadcom. Largely, we're very much similar size. If you think about our Fibre Channel business, it's all selling into the enterprise data center on premise. Current weakness has certainly impacted tremendously. Sequentially, the revenue decline actually is 20% quarter-over-quarter, it's quite significant. Hopefully, that will give you some color. We don't break down the details within the storage category generally.
That's helpful. Thank you.
Thank you. Our next question comes from Gary Mobley from Wells Fargo Securities. Please go ahead.
Hey, everyone. Thanks for taking my question. I wanted to dig a little bit deeper into the ThunderX Arm-based server development. To what extent was that $121 million charge related to sort of restructuring the development of ThunderX? How does it impact your joint development, I guess, cost sharing relationship with Arm and as well, the related option for them to take some Marvell equity? The ThunderX3 that you highlighted at Hot Chips last week, should we think about that as being funded by maybe one or two specific customers?
Gary, we'll divide this question in two pieces. I'll answer the impairment part, then Matt can talk about the strategic side of the Arm business. The impairment actually is quite straightforward. As you recall, we acquired Cavium, and we closed the transaction two years ago. There's a very small piece of develop the technology intangible related to older ThunderX2 Cavium developed before the acquisition of Marvell acquired Cavium. That's primarily, if you look at the impairment, the primary impairment is related the intangible develop the technology of ThunderX2, how Cavium did that business in the past. With our new focus going forward, definitely those old development technology intangible is going to be impaired. That's the impairment part of it. I'll let Matt answer the strategic side of the Arm server business.
Sure. Yeah. Hey, Gary. I think at the highest level, and you know this well, we've had a long-term history with Arm. They're used in almost all of our products, and we cooperate on a number of different applications, like the Arm server opportunity to 5G, automotive, et cetera. Two companies work together, and the way you should think about, again, what we're doing with the Thunder product line is really just focusing it. The market that we've always called out for this has always been the hyperscale customers. What we're basically doing is just really acknowledging that and also acknowledging the fact that it turns out they seem to really want their own thing, their own special chip.
This notion of maybe if you go back a few years to where you would have a broad-based Arm server platform that you would drive and you could sell to everybody and maybe put it on like a tick-tock type of cadence and sort of run it like a normal CPU business. I think just so much has changed since we acquired Cavium and this market's developed. Arm in the data center and Arm in servers has actually continued to get traction in the market, whether that's with us or that's with customers doing their own developments themselves in-house with partners.
We expect that to continue. We're just going to do this in a much more focused and targeted manner and with a business model that looks more like our traditional customer, semi-custom model versus funding the whole thing on our own kind of in perpetuity. We've said basically we think it's better for us and better for the customers actually to do this in a more focused manner. Hope that helps. Next question, please.
Our next question comes from the line of Harlan Sur with JP Morgan.
Hey, good afternoon and good job on the quarterly execution. Good to see the diversification in the storage business driving flattish growth in a, what I would consider to be a very tough storage environment. You guys have been talking about the ramp of your DIY SSD controller into the large gaming opportunity. That's starting to fire and looks to be pretty strong here in the second half. Even on the HDD side, I believe that your second nearline HDD customer just got qualified on its 16 TB platform and is ramping here in the September quarter. Are you guys benefiting from this ramp here in Q3, and maybe mid to longer term, you guys also have some DIY SSD controller wins with tier one hyperscalers? When do you guys expect these programs to ramp?
Yeah. Great question, Harlan. I think, if you think about the storage business, the storage controller portion is performing quite well into Q3. As Jean sort of laid the breadcrumbs earlier on the Fibre Channel business, how to think about that market size, our share of it, how much it was down, we're basically offsetting that by growth in both the SSD side and to your point, in our custom SSD engagement, as well as in both of our nearline customers that continue to ramp with our solutions of SoCs, and then one also using our preamplifier solution as well. Yeah, the traditional, call it Marvell storage controller business, is performing quite well in the near term here, coming off of the lows through the COVID-19 supply issues that were going on. That's doing well.
It's really this very targeted slowdown in enterprise we're seeing within the storage areas offsetting that growth. The growth drivers are very much working in storage, is what I would say. The areas where we put a lot of R&D and investment in cloud applications as well as in DIY. It's nice to see the revenue coming in, and I think there's more to come on discussions of future DIY engagements, which we're very active in right now. Some of those have already turned into design wins with hyperscalers, and that is also part of our future growth strategy to continue to keep storage growing is expanding with additional customers that we didn't have before. More on that later.
Great. Yep. Thanks, Matt.
Thank you. Our next question comes from the line of Timothy Arcuri with UBS.
Thanks a lot. Matt, I think you alluded to a design win with another 5G base station customer beyond the two that we already know about. Can you talk a little more about that and maybe help size the content for that win, maybe versus the two that you've already talked about? Thanks.
Sure. Yeah, I think there was a lot of content we went through today, but the main point was that while there's a lot made, certainly of the top five vendors in the base station market, and typically those five are, call it 90% of the total. It's a big number historically. There's also the remainder is currently quite active in this 5G transition. I would say in almost every region, whether it's in parts of Asia, certainly North America, and other geos, there's a lot of activity there in what we would call the second tier. The nice thing is that the products that we develop for all the top guys, those are all going to be available as standard products for the rest of the market.
For example, the win that we got that's in the second tier, it's using the existing parts we already have. Obviously a lot of leverage from that because we didn't have to go and design a brand new chip and spend a bunch of money on it, and they're able to leverage the investment that we already made. Without getting into dollars, what we did say is that the content is both the Fusion baseband, and you guys have a sense of what those go for, as well as our OCTEON embedded processor for the transport and Layer 2 processing. We have very solid content from a processing perspective there.
I would say is with the last thing, Tim, is with all the disruption going on and all the geopolitical events, I think there's an opportunity for this second tier to actually become more relevant in 5G, whether it's taking share from traditional incumbents or even with new standards like O-RAN, where you might see new people come in and actually be meaningful participants in some of the regions. Again, that's one that we're watching, but it's nice that we're able to take, kind of benefit from all the hard work we've done to put these IP platforms together, to be able to then go in and sell those and enable other customers of ours to participate in the 5G market.
Got it. Got it, Matt. Thank you.
Yeah, thanks.
Thank you. Our next question comes from the line of Ross Seymore with Deutsche Bank.
Hi, guys. Thanks for letting me ask a question. Matt, I wanted to go back to the enterprise side. I think everybody understands that the demand is weak right now. It's nothing Marvell specific, but I wanted to dig a little bit into your views of the duration of that weakness and some of the sub causes of it. In the past, we've seen the macro demand weaken and there also be some significant inventory burn that exacerbated that for a period of time. I guess the two parts of this question would be: Is your fiscal third quarter enterprise guidance being weak exacerbated by that inventory digestion? Do you think that continues into your fiscal fourth quarter? Similarly, are there company specific offsets, either share gains or losses within enterprise that would either offset or magnify that weakness?
Sure, Ross. Yeah, I'd say certainly in the short term, we're seeing the impacts you said. I think some of this may be certainly inventory related. I do think that when we had the supply crunch and countries were shutting down left and right, I think certainly OEMs wanted to make sure that they'd have inventory. I would say that is also very much coupled depending on the segment you're talking about, with just straight up demand weakness, just given that certain segments of enterprise companies aren't spending on. I think those are very well known. What I'd say that the inventory part, unlike sort of, I'd say, prior cycles which I've been through, I'd say this one is just more on the demand side at this point. That's really in the short term, and we have to deal with that.
We're still very encouraged, and bullish about our enterprise offerings, once we're through this cycle we're in, especially as we look out to calendar 2021. I referenced in my remarks, as well as you probably saw some of the announcements we made around our borderless enterprise. I mean, that whole thing is really code for Marvell just refreshed its entire Ethernet switching and PHY portfolio. The last time we did that, which was a few years ago, we jumped a process node or two at that time. We added a whole bunch of features. We optimized the products. We had great success. It was part of the initial Marvell turnaround story, was that networking growth.
We see the same type of opportunity starting really next year with some of our new products, which are, again, in the latest process nodes that are relevant for those markets, the latest sets of features. Even beyond that, as we look out, I'd say beyond next year, and you start looking at our five nanometer portfolio, which we're sampling our first products next year on, but all the design wins that we're in the hunt on right now are really in that node for the large sockets. We're extremely well positioned. I would say that even goes beyond Ethernet and also looking out to our next generation OCTEON products, which have gotten a lot of press because of their success in base stations recently, but don't forget about the wired market.
The wired market, I think if you look at all the networking OEMs and communication OEMs and what they're going to need in the future, and you look at us with a best-in-class infrastructure processor in five nanometer, which would be the 10th generation Cavium processor, we're very excited about that business on a number of fronts. I think the shorter term, we certainly have some chop. Fortunately, it's being offset by cloud and 5G. We're still very optimistic on enterprise, which drives great margins for the company, great stickiness, and I think we'll benefit from the hard work the team has put in to really refresh the portfolio. We're seeing great traction on those products.
Thanks, Matt.
Yep.
Thank you. The last question we'll take comes from Srini Pajjuri with SMBC.
Thank you. Thanks for squeezing me in. Jean, I have a question on the cash usage. I saw that the share count has perked up a little bit in the quarter. Just wondering how you're thinking about the cash usage as we go into the next few quarters. Maybe along the same lines, Matt, you guys have done a great job with M&A, integrating some of the acquisitions. Now that the cash flow is improving, the balance sheet is pretty strong. I'm just curious as to how you're thinking about M&A for the next 12 months. What do you see out there in the market? What kind of opportunities you're seeing, and how are you thinking about it? Thank you.
On the cash usage for next few quarters, certainly our business will continue to generate a very strong free cash flow. If you look at the first half of fiscal year 2021, our cash flow conversion is very significantly higher than the more than 100% of our net income. I think certainly we're going to be able to pay down our short-term debt, which is $450 million. Additional excess cash, we have always been quite committed to return cash to shareholders. Certainly, the macro environment is still uncertain, but we definitely will monitor and assess if we should restart the share repurchase program. That's definitely what we're considering.
Yep. I'll just conclude on the second topic you mentioned about M&A. Yeah, I think first of all, we're very pleased with all the effort we put in in 2019. We bit off a lot, as you remember. We sold our Wi-Fi business to NXP, and then we purchased Aquantia and Avera. We made those decisions candidly right kind of in the midst of the trade war and a lot of uncertainty. I think in reflecting back, we're thrilled, right, that we made those decisions. I think both of those acquisitions we did last year are performing better than we had anticipated when we announced them.
As you saw from our results, I think the benefit of the Marvell platform from a cost standpoint has really enabled us to run these two businesses that we acquired very efficiently, while keeping key technical talent and key management talent, which we were thrilled to get. We like those kind of deals. I think the prices paid were very reasonable and certainly they've really helped our company. Look, we're in execution mode at this point. We've got a lot of design wins in front of us. We're executing a five nanometer platform. We're always going to keep a lookout and certainly very encouraged by the guidance we gave in terms of revenue growth. Also, if you even look at last quarter, was the cash flow generation of the company.
I think you're starting to see that the company's got a tremendous leverage in its operating model and can generate significant free cash flow as we grow, and that certainly sets us up well if we were to find opportunities to go after. As you know, we've been very disciplined on this front. Multiples are definitely high at this point. We've got a lot going on. Of course, we'll look around, but right now, I know myself and the team are plenty busy with executing our current initiatives that we've worked really hard to put together. We will see. Thanks, Srini.
Thanks, Matt.
Thank you. Ladies and gentlemen, this concludes today's conference call. Thank you for participating, and you may now disconnect.