邁威爾科技 (MRVL) 2021 Q3 法說會逐字稿

Ladies and gentlemen, thank you for standing by, and welcome to the Q3 Fiscal 2021 Marvell Technology Group Earnings Conference Call. (Operator Instructions) Please be advised that today's conference is being recorded. (Operator Instructions)

I would now like to hand the conference over to your speaker, Mr. Ashish Saran, Vice President, Investor Relations. Please go ahead.

Thank you, and good afternoon, everyone. Welcome to Marvell's Third 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.

Great. Thanks, Ashish, and good afternoon, everyone. The third quarter was an exciting and exceptionally busy time for the Marvell team as we hosted our 2020 Investor Day and soon after, announced plans for our transformational acquisition of Inphi. These were both significant events, which continue to generate strong interest from the financial community.

So before I discuss our quarterly results, let me summarize the growth strategy we articulated at our Investor Day and the benefits we expect to drive from the combination with Inphi.

During our virtual Investor Day, we discussed the importance of choosing the right end market, building a larger base of key customers and driving technology platform leadership, aligned with the most exciting growth drivers in the industry. We discussed our focus on the data infrastructure market, a $110 billion total semiconductor opportunity for Marvell, of which we are currently servicing $16 billion and that we expect to grow 50% faster than the TAM over the next few years. We showcased the expansion we expect to drive in our base of Tier 1 $100 million plus customers to a total of 13 over the next couple of years. We outlined the evolution of our technology road map to being first to market with a 5-nanometer platform for portfolio-wide leadership. We concluded with our compelling long-term growth vision focused on the 5G, cloud and automotive end markets. At the event, we also disclosed a number of new design wins, which we believe will contribute meaningfully towards revenue growth over the next several years. In 5G, we announced a second customer for OCTEON in the radio head as well as 2 additional base station customers for our Ethernet products.

In the enterprise market, we announced a broader footprint for our Ethernet switches and PHYs at 2 Tier 1 North American networking OEMs. One of these OEMs also selected our next-generation 5-nanometer OCTEON processors to replace their current x86 solutions across their entire enterprise router platforms.

In our storage business, we won a cloud data center DIY SSD controller and announced a new Tier 1 NAND OEM for merchant SSD controllers. Consistent with the strategy we discussed at Investor Day, on October 29, we announced our plans to combine with Inphi in a highly complementary and strategic transaction, and expected to accelerate our growth and leadership in the fast-growing cloud and 5G markets. We discussed multiple benefits expected to result from this merger, including a larger and faster-growing addressable market, a broadening of our Tier 1 cloud customer base, greater scale for our 5- and 3-nanometer platforms, an exceptionally talented SerDes team and an even stronger long-term financial model. We also discussed significant opportunities for cross selling, which would represent upside to the target operating model.

Since the announcement, we have been able to discuss the benefits of this merger with employees, customers, partners and shareholders of both companies. These discussions have further strengthened our conviction in all the key strategic elements of this exciting transaction. We've had the opportunity to meet with the broadest cross-section of Inphi employees and continue to be highly impressed with their depth of technical capability, focus on innovation and seamless program execution. The Marvell and Inphi teams are well aligned from a cultural perspective with many shared values and core beliefs. Similar to the approach we have taken on our prior acquisitions, we are looking forward to integrating Inphi employees across all levels of the combined company. I can already sense that the sales, product management and engineering teams on both sides are chomping at the bit so they can start driving new revenue growth opportunities.

Let's dig a little deeper into these opportunities, which we expect will result from the strong positions established by Marvell in 5G and Inphi in cloud. We believe that having a leading product at the heart of the customer's architecture is critical, as it sets the cadence of technology adoption and pulls along additional products into the full solution. We saw this play out effectively following the Cavium acquisition and you are all now familiar with the significant increase in content and customer expansion we were able to drive in 5G as a combined company.

In base stations, the front, mid and backhaul connectivity is primarily optical. With the emergence of 5G, the speed and bandwidth of these lengths needs to increase substantially. Inphi's high-performance PAM4 and coherent DSP-based electrooptic solutions are perfectly aligned to the line to deliver these higher speeds. We believe that Marvell's leadership in embedded processors for 5G infrastructure will better position Inphi to participate in the optical connectivity opportunity.

In cloud, Inphi's electrooptics platform is critical to data center architecture, and they have established direct relationships with the industry's largest cloud infrastructure providers. Inphi's recent results and guidance illustrate the benefits of having a strong presence in the cloud market. For the third quarter of fiscal 2020, they reported revenue growth of 92% year-over-year, and the organic portion of their business grew 57% year-over-year. We are excited to see that they expect to grow -- growth to continue in the fourth quarter, with Inphi revenue guided to grow 82% year-over-year at the midpoint.

Marvell also has a growing presence in the cloud market, and merging with Inphi is expected to create multiple 100 million-dollar-plus cloud customers for the combined company. We expect to drive numerous revenue growth opportunities for ASICs and DPUs at these large cloud customers.

Within cloud data centers, we expect compute for AI and other use cases to increasingly require tailored solutions, resulting in an expansion of custom and ASIC opportunities for the combined company. In cloud, custom compute engines in ASICs are often connected directly to external optics, such as in a machine learning cluster. This presents a unique opportunity for the combined company to tightly integrate optics inside ASICs and custom compute processors through onboard and co-packaged optics solutions. Cloud customers are particularly excited about this capability.

In addition to the ASIC opportunities I just discussed, we expect similar benefits to Marvell's cloud DPU business. Today, Marvell sells DPUs or smartNICs as standard product to cloud customers. In the short term, we see a greater opportunity to win more business for our existing products. In the longer term, we can provide higher value DPU solutions with tightly integrated onboard and co-packaged optics.

Since announcing the deal, we have completed a number of initial meetings with key customers. While any detailed joint road map discussions will need to wait until receipt of all regulatory approvals, customer reactions from the meeting so far have been extremely positive, and they have already identified multiple new opportunities for Marvell. In fact, I've been personally involved in a number of these discussions. And I am now even more confident in the potential for new revenue growth opportunities.

Let me move on now to our quarterly results and expectations. Revenue for the third quarter of fiscal 2021 was $750 million, growing 3% sequentially and 13% year-over-year. Stronger-than-expected revenue growth from our networking business offset weaker-than-anticipated results from storage. Our GAAP loss per share was $0.03. Our non-GAAP earnings per share was $0.25, growing 19% sequentially and 47% year-over-year. I am pleased with the growth in revenue and earnings we have been able to drive in the third quarter. We are projecting strong growth in the fourth quarter as demand from customers continues to increase.

Our operations team is continuing to ramp production with our global supply chain partners. However, we have begun to experience a number of industry-wide supply constraints affecting the type of high complexity products we provide for data infrastructure. The rapid recovery this year in the semiconductor industry appears to be stressing significant portions of the supply chain. These supply challenges are currently limiting our ability to fully satisfy the increase in demand for some of our networking products.

Let's now discuss our 2 businesses in more detail. First, in our networking business. Revenue during the quarter was $445 million and grew 10% sequentially and 35% year-over-year. Revenue growth was stronger than our expectations of mid- to high single-digit sequential growth. And overall, this was a very strong quarter across a broad base of our products and end markets, including 5G, cloud, automotive and enterprise. In 5G, we delivered our fifth straight quarter of sequential revenue growth. In the first half of this year, 5G growth was driven primarily from our ASIC business, benefiting from the rapid deployments in China. In the third quarter, while the wireless infrastructure ASIC business remained strong, the sequential growth was driven primarily by standard and semi-custom product shipments to Samsung. The rollout of 5G outside of China is starting to pick up. And we expect consumer demand for 5G services will continue to grow worldwide, following the launch of new 5G phones from Apple. As we look forward, we expect 5G momentum to accelerate through next year in multiple regions, including the U.S. and Japan.

Our 5G customer base continues to expand, and a second regional customer has selected Marvell's industry-leading OCTEON fusion baseband processors to power their new 5G base stations. They plan to engage with Marvell on a variety of RAN architectures, including emerging ORAN initiatives, enabling them to flexibly address the needs of network operators, regardless of the network topology being deployed.

Let me take a moment to preview Marvell's efforts in the emerging open RAN and virtualized RAN architectures. These are commonly referred to as ORAN and vRAN, respectively. Global operators are driving toward open standards-based interfaces with the RAN to enable multi-vendor interoperability. In addition, the use of virtualization to separate hardware and software has begun making its way into the radio access network.

Now whether virtual or physical, the complexity of 5G processing, real-time latency requirements and a tight power envelope necessitate optimized hardware acceleration to be deployed at scale. We address this need with our OCTEON fusion processors, the industry's leading silicon platform for 5G baseband and radio processing. We are adding more capabilities to this proven technology to further optimize it for open virtual environments. We will soon unveil details of our new vRAN accelerator, which is based on OCTEON Fusion, designed to support an open and interoperable ecosystem with a full set of ORAN standards-based interfaces. By adding ORAN and vRAN capabilities to our existing 5G offerings, Marvell will be the ideal semiconductor partner with a complete 5G platform, capable of supporting all RAN architectures on common hardware in software framework.

This is a critical differentiator for Marvell, given that beyond a few greenfield deployments, most 5G networks will have a complex hybrid architecture to support a diverse set of deployment scenarios. I look forward to discussing the continued evolution of our 5G platform.

Moving on to our ASIC business. We benefited from a strong ramp in a new cloud program, and we are excited about this opportunity. I would note that similar to other cloud programs, we expect purchasing patterns for the new ASIC to be somewhat lumpy with demand fluctuations in any given quarter. We recently announced our 112 gig, 5-nanometer SerDes solution that has been fully validated in hardware and is ready for adoption by customers. We believe that this technology, with industry-leading performance, power and area, will help drive 100-gig as the interconnect of choice for next-generation data infrastructure.

Our ASIC business is the first beneficiary of this leading edge IP, and I am very pleased that we have secured a new custom ASIC design win that will embed our 112-gig SerDes in next generation, top of rack and spine switches for leading hyperscale data centers.

Turning to our automotive business. Industry production appears to be recovering from COVID-19 impacts earlier in the year. We are also seeing record bookings and have started ramping multiple Ethernet design wins in model year 2021 vehicles. These ramps drove strong revenue growth in the third quarter, and we expect growth to continue in the fourth quarter and into fiscal 2022.

In our enterprise end market, we continue to refresh our Ethernet products and recently introduced the second-generation of our Octal multi-gig PHY family. These are the newest additions to Marvell's borderless enterprise portfolio, a comprehensive set of switches and PHYs architected to address the specific requirements of emerging mobility and cloud applications that are extending the boundaries of the traditional campus environment. The new PHY family offers a clear upgrade path to multi-gigabit technology. We believe that an increase in customer interest for our multi-gig products and our recent design win momentum are strong leading indicators that the industry is getting ready for a much broader multi-gig adoption.

During the quarter, we closed multi-gig PHY design wins with 6 customers across access points, gateway, switches and firewalls. We are also getting very strong traction with our refresh switches in PHYs with multiple security hardware customers.

Now let me discuss the outlook for the fourth fiscal quarter for our networking business. We expect growth from 5G to continue with Samsung and the start of ramp into Nokia, partially offset by a decline in 5G ASICs, as deployments in China take a pause. We project a strong quarter for our automotive business with quarterly revenue expected to cross in double digit millions. We expect enterprise switching to continue to improve as we ramp new products. We also expect growth from our LiquidSecurity DPU shipping to cloud customers. Offsetting this growth, we project a sequential decline in revenue from a cloud ASIC.

As a result, after a very strong third quarter, we expect networking revenue in the fourth quarter to be approximately flat on a sequential basis. And on a year-over-year basis, projected to grow approximately 25% compared to the fourth quarter fiscal 2020 results adjusted for the divestiture of WiFi.

As I mentioned in my opening remarks, we are currently supply-constrained on some of our networking products.

Now turning to our storage business. Storage revenue for the third quarter was $276 million, declining 5% sequentially, below our expectations of being flat. During the quarter, demand for our fiber channel products from our enterprise server and storage customers was impacted more heavily than anticipated, although we expect this level of weakness to be temporary. On the other hand, the sequential revenue growth we had expected from our storage controller business did materialize, led by a ramp in our custom SSD controllers. Our cloud HDD business also grew sequentially.

At this year's Virtual Flash Memory Summit, our storage team had another strong show, and alongside HPE, we were named a Best of Show Award winner for the Marvell-based NVMe RAID Boot SSD. HPE is the first of Marvell's partners to support this new accelerator. The product lowers data center total cost of ownership by offloading RAID 1 processing from costly server CPU resources, maximizing application performance. In addition, we demonstrated our latest data center flash controller for building large capacity, ultra-low latency, high-performance SSDs and our low-power 12-nanometer PCIe Gen4 controller for client SSDs.

Looking to the fourth quarter, we expect our storage business to rebound strongly and project sequential revenue growth in the low teens on a percentage basis. We expect this growth to be driven from multiple products. We project a strong recovery in our fiber channel business towards a more normalized run rate. We also expect our cloud storage revenue to continue to grow, which includes more meaningful contribution from our preamplifiers. In addition, we are projecting the start of a product ramp of a cloud DIY SSD controller design win I mentioned in my opening remarks.

In closing, our results and guidance continue to progress in the right direction. We are projecting fiscal 2021 fourth quarter revenue at the midpoint of guidance to grow approximately 9% year-over-year and grow 13% year-over-year for the ongoing business as compared to last year's fourth quarter results adjusted for the divestiture of WiFi. We continue to work closely with our supply chain partners to alleviate constraints and meet the growing demand from our customers.

Through this fiscal year, which has certainly had its share of COVID-19-related challenges, I'm very pleased with the growth we have driven from 5G and cloud, and it is still early days for us in these 2 critical end markets. In addition, our automotive and preamplifier businesses are now starting to contribute more meaningfully. Our custom SSD controller programs have recently started to ramp, and we will benefit from a full year of shipments in fiscal 2022. We also expect that our refreshed Ethernet switch and PHY products will enable us to take share in the enterprise and carrier end markets.

From a broader perspective, there has been encouraging news on the vaccine front, which bodes well for a macroeconomic recovery next year. Overall, the setup for fiscal year 2022 looks very promising. Equally exciting, we are looking forward to combining with Inphi to accelerate our joint vision to lead the ongoing transformation in the fast-growing cloud and 5G markets.

And with that, I'll turn the call over to Jean for more detail on our recent results and outlook.

Thank you, Matt, and good afternoon, everyone. I'll start with a review of our financial results for the third quarter and then provide our current outlook for the fourth quarter of fiscal 2021.

Revenue in the third quarter was $750 million. Networking represented 59% of our revenue with storage contributing 37%. Driven by growth in 5G and the cloud and the relative revenue contribution from our networking business of nearly 60%, reached a new record in the third quarter. Revenue from other accounted for 4% of our revenue, declining 5% sequentially and 35% year-on-year. As a reminder, this business consists of product we have stopped investing in. So we expect they will continue to decline over time.

Our guidance for the fourth quarter anticipated sequential growth from other as we expect some customers to complete last-time buys on certain components.

GAAP gross margin was 50.8%. Non-GAAP gross margin was 63% of revenue, consistent with our guidance. GAAP operating expenses were $390 million and include the cost for share-based compensation expenses, amortization of acquired intangible assets and acquisition and the divestiture-related costs. Non-GAAP operating expenses were $280 million. GAAP operating loss was $9 million. Non-GAAP operating profit was $193 million or 25.7% of revenue. For the third quarter, GAAP loss per diluted share was $0.03. Non-GAAP income per diluted share was $0.25.

Now turning to our balance sheet. During the quarter, cash flow from operations was $258 million. In fiscal 2021, we have driven strong free cash flow conversion of 114% of non-GAAP net income on a year-to-date basis. We returned $40 million to shareholders through dividend payments. During the quarter, we paid down $100 million of our term loan and exit the quarter with $832 million in cash and short-term investment and a total debt outstanding of $1.35 billion.

We continue to have $500 million of liquidity available from our undrawn revolver. Our net debt-to-EBITDA ratio was 0.6x on a trailing 12-month basis. We have temporarily suspended our share repurchase program due to the pending acquisition of Inphi.

Before I provide our outlook for the fourth quarter, a reminder on our OpEx items. Please note that due to the typical seasonality in payroll taxes, our OpEx in the first fiscal quarter tends to increase mid-single digits sequentially on a percentage basis, and this effect then dissipates in the rest of the fiscal year.

Here's the specific guidance for the fourth quarter of fiscal 2021. We're forecasting revenue to be in the range of $785 million, plus or minus 5%. We expect our GAAP gross margin will be approximately 52.8%. We project our non-GAAP gross margin will be approximately 64%, which anticipates a favorable product mix during the fourth quarter. As we discussed during our Investor Day, we expect our non-GAAP gross margin in the near future to be in the range of 63% to 64%, depending upon the relative mix of certain 5G and cloud products in any given quarter. We project our GAAP operating expenses to be approximately $379 million. We anticipate our non-GAAP operating expenses to be approximately $280 million. We expect net interest expense to be approximately $15 million and non-GAAP tax rate of 5%. We expect our basic weighted average shares outstanding will be 673 million, and our diluted weighted average share outstanding will be 686 million. As a result, we anticipate GAAP earnings per share in the range of a loss of $0.03 per share on the low end to an income of $0.07 per diluted share on the high end. We expect non-GAAP income diluted share in the range of $0.25 to $0.33. Please note that our GAAP diluted EPS is calculated using basic weighted average shares outstanding when there is a GAAP net loss, and then calculated using diluted weighted average share outstanding when there's GAAP net income. Non-GAAP diluted EPS is calculated using diluted weighted average shares outstanding.

Operator, please open the line and announce the Q&A instructions. Thank you.

(Operator Instructions) Our first question will come from Toshiya Hari with Goldman Sachs.

Matt, you talked about supply constraints having some impact on your business in the quarter. Can you help us quantify how big the impact was in the quarter? And what sort of assumptions are embedded in your guide? And then I've got a quick follow-up.

Sure. Yes. As we noted, we have seen a number of constraints, I would say, over the last few quarters, by the way, in certain process node bottlenecks as well as I think it's fairly widely known, there's significant increase in demand this year for complex substrates as well. And so we've seen this tightness. We're managing through it. The way I would think about it is, we've had a, historically a fairly steady level of delinquency that we enter every quarter with, with delinquency being defined as the amount of product that we have got request dates for in the current quarter we're in, that we can't supply within the quarter. And that's something that we track on a regular basis.

Heading into Q4, that number is significantly larger than we've had. And we've -- part of that is customers, due to the constraints out there have also placed longer lead times on us. So I think we're pleased to be able to guide up sequentially. And certainly, if you look year-over-year, the business is up quite strongly overall and especially in our networking business. So we're happy about that. But I guess what I -- all I would say is you should think about it as we're, like many other companies, I think, carrying a significantly higher level of delinquency than we have historically.

Got it. And then as my follow-up, on the storage side of the business, the quarter came in a little bit below expectations, and you spoke to the decline in the fiber channel business. Going forward, it's good to see that you're guiding the business up low teens in the current quarter. But beyond that, should we assume kind of growth to be in the low single-digit range, consistent with what you talked about at your Analyst Day with some of the drivers like the DIY, SSD controller and the preamps? Or should we expect it to remain pretty lumpy as it relates to storage?

Sure. Yes. No, I think the thesis from Investor Day is certainly still intact, and we are pleased to see a nice upswing in the fourth quarter, understanding that Q3 was a little bit weaker. And we did have this somewhat unusual lumpiness in fiber channel, which, I think as you know, historically, has been a relatively stable business. But given some of the impacts earlier this year with supply chain on boards and some of the COVID-19-related impacts on enterprise, that business did have a lot more volatility to it than we've normally experienced. Offsetting that, obviously, has been very, very strong growth in our traditional Marvell storage controller business, which is, by the way, the business that normally has been historically a little bit more lumpy. That one's performing extremely well due to this DIY flash controller you mentioned and also our shipments of controllers that end up in cloud applications. And so I think the fourth quarter is a strong increase, and we certainly see all those growth drivers intact. Next year, we get the full benefit of that full year of the DIY shipments to our lead customer there. We certainly see cloud computing growing. And then we've also got share gains in preamplifiers.

So I think -- and by the way, I think from a fiber channel standpoint, we do expect that in Q4 and beyond, that, that business trends to a more normal run rate. And I would expect it would not have the same level of volatility we saw earlier this year, just given that, that was primarily caused by some external shocks relative to COVID 19.

(Operator Instructions) Our next question will come from Vivek Arya with Bank of America Securities.

Matt, let me press you a little more on the supply constraint issue, but I -- just to get a sense of whether it's like $5 million, $10 million, $15 million? Just some ballpark would help. And then importantly, what does that say about your visibility heading into April? When I look at the current kind of expectations around April, therefore, sort of flattish trends versus January. But given this visibility, how should we -- and these unsolicited orders, what you're referring to as delinquency, how should we think about seasonality heading into April? So just some quantification and just some color around trends heading into April would be really helpful.

Sure. Well, let me first comment. I think we're certainly pleased that Q4 -- Q4 is up. And that's a positive. I think if I just go back up more to 30,000 feet, we're very encouraged about our fiscal '22. The -- as I mentioned earlier, we've got the full year of DIY storage controller. We've got 5G kicking in. There's obviously more growth in cloud and a number of growth drivers. So I think that we feel very good about for next year. And then I think the slope of that and how that all rolls in, some of that is going to be related to demand, and some of it's going to also be related to supply and how much progress we can make. At this point, it's a little bit difficult to quantify exactly the amount. But what I would say is that directionally, the way the business is heading and the growth drivers we outlined at our Investor Day would certainly support the type of growth that we're anticipating next year. So I think all that's positive in terms of heading into the fourth quarter and beyond. But where Q1 lands, we'll have to get to that when we go ahead and guide that quarter. I think we'll have much better visibility on the supply dynamics. But certainly, in the short term, you can see from our Q4 guide, demand has been strong, and that's also backed up by strong bookings that we've seen as well as, as I mentioned, we've got a lot of orders sitting out there that we're working very hard to go fulfill.

Our next question will come from Timothy Arcuri with UBS.

Matt, I guess I also wanted to ask about the same topic. And I guess my question is more on timing. And really, when you think you'll be able to ship to demand, it sounds like maybe you're getting a little bit of a trade off here, that you're getting a little more longer term visibility. But I guess my question is really, when do you think you'll be able to ship to demand? And in the fiscal Q4 guidance, does that assume that delinquencies go up versus fiscal Q3? Or does it assume that they stay about the same?

Yes. No. Look, Tim, I -- trust me, I wish I could be a little bit more precise. But I think the way to think about it is on the supply side, at least when we -- when we talk to our supply chain partners, there is an anticipation that certainly within the first quarter or 2 in calendar '21, that we will see some improvements there. I think a lot of that, though, is -- depends on how strong the demand environment is. Certainly, it's not just us, I think, signaling that demand is good. I think other companies as well are indicating that. So I'd just say it's a fluid situation. But certainly, we hope within the first quarter or 2 in next year that we would be able to get up to that run rate. We take our customer satisfaction very seriously, and we're all hands on deck in terms of how to meet the ramp. And I'm confident we will meet the ramp ultimately. It's -- we've just got a -- and I've been through a few of these since I've been in this industry. And ultimately, these things work themselves out, and we've got a world-class operations team and pretty tight communication with our customers at this point. So it will get there, and I think we'll start seeing improvements in the first half.

And just to be clear, Tim, just to be clear, in Q4, I mean and maybe for Vivek as well. I mean, we would have guided higher in Q4, obviously, had we had better access to the supply that we needed.

Our next question will come from C.J. Muse with Evercore.

Sorry to beat a dead horse, Matt. But just curious on the supply constraint side, are there any particular products within networking that are tighter than others? And then I guess, if we can get off of this subject, perhaps you could speak to, now that Inphi is expected to be in the fold at Marvell, what are your conversations trending to now with your lead customers?

Sure. Yes. Happy to finish the meeting, C.J. But really, what -- if you think about it, in our networking business and whether that's -- whether that's in ASICs or in some of our switches or even some of our PHYs, some of these products are extremely high pin count devices in very complex packages. And as you go to more and more complex substrates, especially, you could say with 5G products or processor products, there's just been a number of constraints as these substrate factories try to ramp up production on boards and PCBs that are multiples more complex in terms of number of layers and number of pins than in previous generations. So I think that's putting some constraint. Additionally, the large fabs, the large foundries out there are full. And in some cases, you have legacy nodes where there's constraints that are a little bit unique. It might be because of who knows last time buys, I'm not sure from other vendors, but you also have some legacy nodes that actually have some impact which are a little bit acute right now. So it's across the board. But again, the trend is more on the complex products and more around substrates.

And then your second question, C.J., was on Inphi?

Yes, that's right. In terms of your conversations with your leading customers post the deal announcement, similar to what you discussed, I guess, a few years back with Cavium, would love to hear the reception and whether engagements are accelerating because of bringing this asset into the fold?

Yes. No, thanks. Yes. We've had, I think, very strong joint engagements since we announced the combination. And what I would say is, is pretty much across the board, there's very strong support from the customer base for this combination. I think one is, certainly in the 5G area, they're very happy to have another strong product line to consider and we've already got some introductions made there. And then on the cloud side, which is really where Inphi was -- is very deep with a number of the large OEMs. I think there's very constructive discussions right now about a number of opportunities. A number of those are around ASICs. And I think the timing of our 5-nanometer platform and the traction we're seeing there, actually, that resonates really well. And in many cases, some of these high-performance custom ASICs for the cloud, they have direct interfaces to Inphi optical modules. And so there's actually a fairly tight synergy there between the dense digital logic type of solutions we can provide, connecting to these high-performance optical modules that Inphi is enabling. So it's going very well. And I think, certainly, as we clear the various regulatory approvals, we can engage even more deeply, but that's all going extremely well. And it does remind me a lot of the synergy that we saw very early on when we did Cavium in terms of the customer reception and the ability to go sell a very joint solution. So I'm pretty excited about it. And I think it's meeting all expectations or better than what I anticipated that reaction would be like.

Our next question will come from Blayne Curtis with Barclays.

Just kind of curious, Matt, on the outlook on the enterprise side, it was up, and I'm kind of just curious, in your conversations as you look into next year, whether we could see a bit better cycle on the enterprise side.

Yes. Sure, Blayne. Yes, I think a couple of things. I mean, certainly, there has been a lot of concern about the enterprise market in general. I think we have started to see some positive -- more positive commentary out of some of the large OEMs. So I think that's good. But really for us, it's sort of a continued story of our own new product introductions and share gain. And so we recently announced an addition to our borderless enterprise portfolio with our Octal multi-gig PHYs. We had strong design wins on PHYs in the most recent quarter. Our new switches have got great traction as well. And even in our traditional enterprise NIC and fiber channel business, we're also seeing strong design activity. So I think all of that bodes well for us. And it's certainly, even in the short term, helping offset what many might consider to be a more challenging overall enterprise environment. We actually we see positive trends into Q4, and we anticipate that's going to continue next year. And then as we even highlighted at the Analyst Day with some new, more strategic wins, 1 example would be in our Octeon processor area, there's even growth beyond that. So I think we're very excited about the enterprise business as a key part of our portfolio to drive growth and also be overall accretive to the gross margins of the company. So it's a good business for us. We're investing there, and we're very focused on product leadership in that area.

Our next question will come from John Pitzer with Crédit Suisse.

Matt, clearly, the strength of the story here are all the bottoms-up product cycles. And you mentioned a new win in your preamble, and I'd love to get a little bit more color. The 5-nanometer Octeon win against x86. Help us understand why you think you won that socket, when it starts to ramp, and kind of what the revenue potential there is? And given all of the product ramps you have, those ramps still live in the real world, and we're still kind of in a macro environment that's kind of just crawling along. I'm kind of curious, as we get into calendar year '21, if we do see that sort of cyclical recovery in the macro, how much of an accelerant to growth to some of these product cycles do you think that can be?

Sure. Yes. So look, I think if you go back and you look at -- and let's even go back all the way to Cavium days, there's been this ongoing -- first, it was MIPs and then transition to ARM battle in the networking processor area. And Cavium and Marvell has had a strong position there, great products, but not the share leader. And even in this sort of embedded world, x86 has always been very strong. We're very encouraged by where we sit from a product positioning as we look at our next generation, our tenth generation OCTEON platform that's going to be coming out in 5-nanometer. And I think that their -- I think it's well understood, beyond well understood, some of the challenges, the largest player in that market has in terms of their process road map. So I think we line up very well from that point of view, certainly. And -- but on top of that, we have leadership and architecture for these types of applications, for data playing processing and very deep relationships with our customers. And so the win that we highlighted, I talked about today, but we had actually talked about it at Investor Day as well, was for one of the leading OEMs, not only in the United States, but the world who typically makes these types of platform decisions much earlier than others, in terms of how they plan. So we're -- I would say this is sort of a very positive leading indicator that we got selected after very extensive benchmarking, by the way. And we said at the Investor Day, I think it was in Raghib's presentation, that this piece of business alone would be an incremental $100 million a year type of annual socket for us, which is huge for a company like ours. We don't have that many, many sockets of that nature. But I think that, that's a very strong leading indicator, John, that in this next cycle with 5-nanometer ramping, and again, this is going to take -- this is going to manifest itself over the next several years, we think we're in a very strong position there. So I think that -- and I think there's a chance Octeon could have a much stronger place in the market than it has historically.

And then you're right on the -- even in this environment, at least many of the chip companies are performing well. We think we are. And certainly, if we get the economy on a macro level moving in the right direction next year, that's really not in our assumptions today. I mean we sort of assume that, at least as we planned, we're planning that there's still going to be some turbulence. But certainly, if the macro improves and there's strength in enterprise spending as an example, our businesses are spending again, people are going back to work, et cetera, then I do think it's going to be a lift for companies like us. But I think it's a little bit early to tell, given all the transitions happening in the world today.

Our next question will come from Ross Seymore with Deutsche Bank.

Hey Matt, I wanted to dive a little bit into the networking guidance for the quarter and then some longer-term implications for it. It sounds like pretty much everything is going to be up sequentially except for 2 ASIC sides, ASIC on the wireless infrastructure as well as the cloud ASIC side. So are those ASIC pauses? Are those things that are due to the end market weakening, just taking a pause? Is it the -- I know it's not end of life, but have we already hit kind of a plateau in those businesses? And then the other side, where everything else is up, how much of that do you think is macro getting better, sort of a systematic increase versus Marvell-specific product cycles driving that sequential growth?

Sure. Yes. Thanks, Ross. So on the first part of it, I actually think it's very consistent with the question you've asked me in the past. I think you -- during one of our -- one of the investor road shows you hosted, where I did a call with a broad set of investors, you said, hey, with some of these markets you're getting into, like carrier or like cloud, how do you expect the volatility of those markets to be in, and those lend themselves to be somewhat lumpier? And so that is the case. I mean, you did nail that in that -- in the case of the cloud side, it had nothing to do with demand, but the pattern in which new products ramp and you've probably seen this with other semiconductor vendors, it isn't exactly a linear process. They tend to buy a larger quantity early, then there might be a quarter of sort of digestion and then they come back again. So this is actually, the strength we saw in Q3 was the beginning of a ramp on this program. And it just -- they took a large quantity in Q3. Some deployments, they're going to do in their own data centers through Q4. But certainly, we expect that overall program to be up quite a bit in next year because that's a brand-new program, and that's just ramping.

And then on the 5G side, China has been the most aggressive in terms of 5G rollouts globally, and we've seen that strength actually in China for the last 2 or 3 quarters, including Q3 being a strong quarter. And I think it's pretty natural for them to do some absorption as well. And that commentary is, I think, been out there somewhat broadly. We certainly think, again, overall, next year, that China 5G deployments should continue. There's still a whole bunch to go. And we'll benefit when that happens.

So yes, I think in those cases, Ross, it's really more just new products ramping, in the case of cloud. And in the case of 5G, it's just, I think, a normal fluctuation in how the carrier demand lines in.

And then I'd say for our Q4, you're right, pretty much across the board, all of our product lines are up. And I would say it's mostly product driven for us. I mean, let's take automotive as an example. We indicated it was crossing over into double digit millions. We've seen car production pick up. We had noted we were in a whole bunch of model year '21 vehicles. That's all now booking. And that's, again, a very unique Marvell product cycle in terms of automotive Ethernet that continues, not only through our fiscal '22, but certainly beyond that, as these programs go into high volume. So I would attribute this to be more our own product cycles. But certainly, I think if you just talk to other people in the chip industry right now, the sentiment is quite bullish in terms of other CEOs and other companies feeling like it is an overall stronger demand environment out there. But I think we're guiding up maybe a little bit more than others just because we do have some unique product cycles that we're a part of.

Our next question will come from Tore Svanberg with Stifel.

Yes. And congratulations on the results. And Matt, thank you for the sort of preview on ORAN, vRAN. I was hoping you could elaborate a little bit more on the company's positioning there, maybe not as much on the technology side, but kind of more on the business model and how that supply chain works. My understanding, it's going to be more of a white box market. So any color you could share with us there on the positioning would be great.

Sure. Yes. I'll continue a little bit of my mini preview, but I think, Tore, you should expect relatively soon, we'll be doing a pretty comprehensive release on ORAN and vRAN, by the way, in terms of how we're going to play, how we fit in the market, and it will be a combination of technology and also segmentation. But I think to keep it at a higher level, it is an interesting potential disruption that's out there, and it's certainly gained momentum in terms of the conversations we're having over the last 6 to 9 months. I do think there will be -- I think it will get deployed in some areas. And I view it in a couple of ways.

One, I think, I think there are smaller emerging players who would be more in the white box model. And there, we -- you should assume we're going to have more of a standard offering that leverages all of the IP that we've already developed and is a fairly efficient way to get merchant products out into the market. And that's available today. As you know, we're shipping macro class 5G baseband processors today. So I think we're in the best position of any company quickly to go do that.

At the same time, the large incumbent players in RAN are not asleep at the wheel. And I think each of them will have their own go-to-market strategy around oRAN and vRAN. And many of those customers were also a key provider to them already for the macro. And so I think we have an ability to support the traditional players who probably have to grapple with some level of innovator's dilemma on this market, in terms of how they go about it and how they play and how they manage it. And then there's a set of upstarts that we can also address.

So more on that later, but I think it won't be just white box and sort of startup companies. I think it will be a combination of the 2, and I think it'll take some time to play out, Tore. I don't think this is an overnight sensation, and probably we'll go -- could sort of map the hype cycle curve on it, like we've seen with a lot of other emerging technologies, ORAN probably goes a lot like that. And so you can pick your place on the hype cycle. It's probably on the way up, and there'll be a point where it's probably got to pause, and then at some point, it'll make an impact.

But we certainly plan to be there, but in a way that enables the broader market and leverages all the development we've already done. I think that's the key part, is doing this in an R&D efficient way, which enables us to participate in it.

Our next question will come from Harlan Sur with JPMorgan.

It looks like capacity optimized our nearline HDD shipments are looking or starting to pick up this quarter, ahead of what looks to be a reacceleration in cloud spending, starting kind of first half of next year? I think your 2 lead HDD customers are well into the ramp of 16 terabytes, but they're also starting to qualify 18 as well. And as we progress through next year, it sort of looks like 18 is going to be the sweet spot for volume. So what's Marvell's dollar content opportunity on 18 versus 16, with your 2 lead customers? And is the team starting to get visibility on continued strength in your nearline business over the next several quarters?

Sure. I think, Harlan, I think the corollary around cloud demand and cloud strength certainly ripples back in the nearline. So we do see that in the short term. As you probably know, the big move for us is really the ability to get preamplifiers in terms of the content. And so we have certainly 1 account that we're shipping with today along with controllers.

So beyond that, the content from a controller and even plus preamplifier standpoint doesn't change all that much from these generations. I think the big mover for us is if we can get more preamp engagements in the nearline side, we certainly have 1 that's already publicly announced, and we're working on more. But I think ASP's content, as you get to these high capacity drives, is going to be fairly similar from generation to generation. And that goes for 18, it goes for 20, it goes for -- and beyond.

Unfortunately, and that's always the challenge is, there's just tremendous exabyte growth and tremendous data storage growth happening. The controllers, unfortunately, don't get the same type of ASP uplift to them, although we've managed to get a little bit more value generation to generation, but it's very incremental relative to the bang for the buck, the cloud customers get by going from a 16 to an 18 or an 18 to a 20. We don't see that in our ASPs. But it's good business, and we certainly have a lot of value there.

Our next question will come from Christopher Rolland with Susquehanna.

I also thought the top-of-rack switch announcement at a hyperscaler was interesting as well. You've previously focused on the enterprise market rather than the market where we're really talking about speeds and feeds. But it looks like there could be more of a shift there. And I was wondering if that is the case. And perhaps you can describe your capabilities there? Is this all coming out of Israel right now? And what else would you have to do to try to catch up with the incumbent here?

Yes. So let me be really clear, Chris. For the Marvell-designed business, that has been classically focused on enterprise switching, campus switching, Soho, core and aggregation switching, and that's been our sweet spot, that's where we've been growing. When you go to the speeds and feeds and cloud side, we don't have a development internally on that. So the reference, I think, that you're pointing to is an ASIC win that we have. And we actually have a number of these that we're working on from the Avera business, okay? And now it's Marvell business because in some cases, we won these since we joined forces.

So the way that we think about accessing that kind of high-end top-of-rack switch market, actually, we think we're probably better served at this point supplying custom ASICs to people who want to compete in that market. And I would say that for global usage, whether it's across the Super 7. And that also gives us a nice strategic entrée as well as we think about when we can join forces with Inphi, and the road map really looking more and more to go to copackage optics, our ability to be, I would call, more open source there and really support the ecosystem and support the broader set of companies that are doing these that aren't the market leader, I think is good for us. I think we sort of think the returns to do that are probably better than if we just tried to go head-to-head, which we've said for many years, it's just not our strategy.

I do think the large competitor there who is the market leader, they just announced a new product. I think that's actually good, because it's showing from a, from an optic standpoint that, that -- it's going to continue to drive the need for optics, and that's going to be a tailwind ultimately for people like Inphi.

So we have to think it through, Chris, in terms of how we go in there. But today, just to be very clear, the win that we announced was a custom ASIC with another company who we're doing a high-end switch for.

And today's final question will come from Srini Pajjuri with SMBC Nikko Securities.

Hi Matt. I have a follow up to one of the previous questions. Matt, if I look at your networking guidance, you said Samsung will be up, Nokia will be up in 5G, but the ASIC business is down. So if you could please clarify if 5G in total is growing or declining or flattish? And then as we kind of head into the next year, first half of next year, if you could qualitatively talk about what you're thinking -- or how are you looking at Samsung and Nokia continuing ramps in the first half. And also, as far as ASIC is concerned, is the rebound of recovery in ASIC, is that pretty much a function of China coming back? Or do you see any other regions helping that market to recover?

Sure. Yes. So I think -- and kind of starting at the high level first. The nice thing about our portfolio, Srini, is we have both these ASIC opportunities that are -- have now materialized, mostly driving China, and then we have our other 2 lead customers for the Marvell stuff. And this is a great example of where we had a lot of strength early on, through the China ramp on our ASIC business. That's somewhat moderated. And now the good news is we see -- we see Samsung showing a lot of strength, which is great. We invested a lot there. And I think it's very encouraging to see that really start to go. And then we've got new wins at Nokia that are, I think, well understood. And so that's just starting.

If you look at Q4, it will be our -- we're guiding it to be our sixth straight sequential quarter-over-quarter growth quarter for 5G. So while in aggregate -- so in aggregate, it's going to continue to grow, and it has been growing. And I think others in this communications market have seen a little bit more volatility during different parts of the last, call it, 4 to 6 quarters because, in general, the other players are more established. They were bigger in 4G. And so our play is really -- we're emerging. We're gaining share in 5G. We have new programs ramping. We have new customers ramping. And we've got this nice mix now and diversity such that, even if 1 region or geography has a digestion quarter, we've got the others just kicking in. And certainly, we anticipate that next year, we'll see growth across the board. We think China should resume. And we see our lead customer continuing to ramp, especially as the U.S. starts to deploy, I think there's going to be a lot of need to go do that, especially with the 5G iPhone selling as well as they appear to be. And also, even if you look at Qualcomm's forecast next year of 500 million 5G-enabled phones, I think that's going to drive a lot of demand for networks. And between the design win positions we have with the leading players who will supply the leading early geographies, we think we're in very good shape for our fiscal '22 next year.

Ladies and gentlemen, thank you for participating in today's question-and-answer session. I would now like to turn the call back over to management for any closing remarks.

Thanks, guys, as we look forward to talking to all of you again next quarter.

Ladies and gentlemen, this concludes today's conference call. Thank you for your participation. You may now disconnect, and have a wonderful day.