Ladies and gentlemen, thank you for standing by, and welcome to the Marvell Technology Group Fourth Quarter 2021 Earnings Conference Call. [Operator Instructions] As a reminder, today's program is being recorded. I would now like to introduce your host for today's program, Ashish Saran, Vice President, Investor Relations. Please go ahead.

Thank you, and good afternoon, everyone. Welcome to Marvell's Fourth Quarter and 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. Matt?

Thanks, Ashish, and good afternoon, everyone. Let me start with a recap of Marvell's highlights for fiscal year 2021. Our GAAP revenue was $2.97 billion, GAAP gross margin was 50.1% and GAAP loss per diluted share was $0.41. On a non-GAAP basis, our gross margin was 63.3% and non-GAAP earnings per diluted share was $0.92. Our consolidated revenue grew 10% year-on-year, led by our networking business, which grew 22% year-on-year. Revenue growth, combined with strong operating leverage from our business model, enabled non-GAAP EPS to grow 39% year-on-year. I am very pleased with our outstanding performance during what has been an otherwise challenging year. Our growth initiatives in 5G, cloud and automotive drove results in these key end markets, which collectively more than doubled in revenue from the prior year to represent more than 25% of fiscal year 2021 revenue. We are really pleased with the performance of Avera and Aquantia, who both delivered revenue above our expectations. We successfully completed the integration of both businesses and overachieved on our synergy targets. We also made great strides with our technology platform, announcing the jump to 5-nanometer and along with it, the industry's first 112 gig SerDes for cloud data center infrastructure. This move was a direct result of our multiyear strategic shift to focus on data infrastructure and brings our process node cadence to the cutting edge. The power of our 5-nanometer platform and the benefits it provides for customers is evident in our opportunity funnel, which has grown significantly since we adopted this new node. The activity level on 5-nanometer continues to accelerate, and we have already secured multiple leading edge design wins, each meaningful from a revenue perspective. Our advanced technology is also a key enabler of our custom ASIC offering, which has continued to gain momentum, particularly…

Thanks, Matt, and good afternoon, everyone. I'll start with a review of our financial results for the fourth quarter and then provide our current outlook for the first quarter of fiscal 2022. Revenue in the fourth quarter was $798 million, [ $13 million ] above the midpoint of our guidance. Networking represented 55% of our revenue, and storage contributing 41%. Revenue from other accounted for 4% of our revenue. GAAP gross margin was 52.8%. Non-GAAP gross margin was 63.9% of revenue, both consistent with our guidance. GAAP operating expenses were $423 million and include the cost of share-based compensation expenses, amortization of acquired intangible assets, legal segments and acquisition and divestiture-related costs. Non-GAAP operating expense were $283 million, just above midpoint due to slightly higher project costs. GAAP operating loss was $2 million. Non-GAAP operating profit was $226 million or 28.4% of revenue. For the fourth quarter, GAAP income per diluted share was $0.02. Non-GAAP income per diluted share was $0.29. Now turning to our balance sheet. During the quarter, cash flow from operations was $158 million. We returned $40 million to shareholders through dividend payment. During the quarter, we paid down $150 million of our term loan and exit the quarter with $748 million in cash and cash equivalents and total debt outstanding of $1.2 billion. Our net debt-to-EBITDA ratio was 0.5x on a trailing 12-month basis. We have temporarily suspended our share repurchase program due to the pending acquisition of Inphi. For the full fiscal year of 2021, we returned a total of $186 million to shareholders through $161 million in dividends and $25 million in share repurchases. In addition, we paid down $250 million of our term loan. During fiscal 2021, we also generated strong cash flow from operations of $817 million. Turning to our guidance. For…

[Operator Instructions] Our first question comes from the line of Harlan Sur from JPMorgan.

Appreciate the commentary on supply constraints. And I'm not asking the team to endorse higher revenues in the second half of the year, but I would expect China 5G spend to come back in the second half. 5G outside of China remains strong. Sony is going to double their PS5 shipments this year, and then you have continued momentum in cloud. But if the backlog supported higher revenues in the second half, let's say, 10% higher versus the $800 million in April, is the team confident that you can support quarter-on-quarter revenue growth through the second half? Or do you think that, that will be a challenge just given the tight wafer substrate and assembly and test capacity constraints?

Sure. Thanks, Harlan. Yes. No, I think you highlighted some of the growth drivers. So just a few comments. I think the first is we are -- we have been planning for growth in our FY '22. I think we're off to a good start for the year. If you look at our Q1 guidance alone, it's going to be up 15% year-over-year versus a year ago. We've got some very strong secular growth drivers. You mentioned 5G, and there's certainly some positive signs, especially for the second half there. We have our cloud business, which is also growing on a secular basis as well as our new product ramps. And actually, we also have automotive, which is really off to a good start this year, and we see that continuing. So we've got that as a tailwind behind us. I'd say we've been able to drive growth, even if you look at FY '21, pretty consistently. We've been growing each quarter, quarter-over-quarter and year-over-year. Q4, we were able to actually exceed guidance and make up -- service some of the demand that was there. So that was a positive side. I think -- so I think the message is we've been delivering well. There's been a new sort of increased surge, and that's really where we're focused is working with our supply chain partners to be able to capture that upside above and beyond the growth rate that we've already been achieving. So bottom line answer to your question is I'm very confident that we can grow revenue in the second half based on the capacity that we've secured and worked hard with our supply chain partners to put in place. And we're also going to look to see if there's upside opportunity to pursue that as well. And that's not only for this year, but as I look even into our FY '23, we're starting to work with our supply chain partners there as well to really give them advanced notice because we see those trends like 5G, cloud and automotive being even stronger in the out years with new products ramping, new customers coming online, and we're pretty excited about our future there. So yes, I think we'll be in -- we're working hard on the -- on all fronts.

Our next question comes from the line of Vivek Arya from Bank of America.

Matt, when I look back at this earnings season, most semiconductor companies reported and guided to trends well ahead of expectation, right, so much so that investors, right, have often complained about overshipments or pull-in of orders. But when I compare your Q4 and Q1 against expectations, it's been more measured. I'm curious, how much of this would you attribute to a different end market mix, right? How much of this is due to supply constraints? And then more importantly, what changes or gets better the rest of the year so you can start to accelerate your sales growth?

Sure. Yes. I think, Vivek, it's a couple of things. I think, first, it has a lot to do with the end markets that we're in. I mean, fundamentally, remember, we're one of the only pure-play infrastructure companies. So we don't have consumer demand that's sort of very volatile, maybe good 1 quarter, maybe not so good in 2 quarters. We don't have a lot of demand that sort of comes and goes. We tend to -- and this is sort of by design, by the way, the business model that we've put together is to bring together these various end applications and drive consistent and ideally predictable growth. So while we may not be as flashy in some quarters as certain companies that really have some explosive demand, but then maybe it abates later, we're focused on slow and steady wins today and continuing to execute and grow our company quarter-over-quarter and year-over-year. Certainly, there's opportunities out there to secure some upside, and we're working on that. But I think even if you saw in Q4, for us, we were quite pleased with the overachievement on revenue, and we're very comfortable with the first quarter guide as well. So I think it's more around we've been one of the faster growers in the last few quarters and also our end-market dynamics. Ideally, we'll deliver more repeatable and sustainable long-term growth with maybe less quarter-to-quarter variation. That's been the strategy.

And anything changed in the second half versus what you have seen in the last 2 quarters? What gets better from here?

In terms of the demand or the supply?

Yes, on the demand and the market -- end market mix.

Sure. Yes. Well, I think a couple of things are going on. If you -- as I indicated in my comments earlier with Harlan, we certainly have -- 5G in the second half will continue to accelerate. We have -- there are new geographies coming online. We believe that the U.S. will -- shows very good prospects for the second half. We believe that China as well will go through their digestion phase, and then you have Japan and you have new countries, even India is talking about potentially pulling in their spectrum auction. So that's all positive. So we expect that to continue. Our cloud business continues to grow, and we see that continuing, both on a secular basis as well as new products, including our ASIC storage and our DPUs, which are doing quite well in terms of smartNICs. Automotive, the same thing. We've got a good ramp going with our calendar -- our model year, sorry, '21 ramps. We've got more to come in model year '22 with additional OEMs coming online as well as expanded content. And even some of the new automotive, I call it, the new breed of automotive OEMs, the EV-only focused companies, really aggressively adopting our solutions. So we're positive there. And then we've got custom SSD controllers, both for cloud as well as for our system OEM. And finally, on top of that, our enterprise business has done extremely well if you look at -- which was surprising, but if you look at even our results in last year, that business with our switch and PHY products grew significantly, primarily on our own product cycle, and we also are very optimistic about that as well. So there's a number of good things that are happening in our business, Vivek, both from an end market as well as a product cycle standpoint. And so as Harlan pointed out, we've got to go, just continue to make sure we've got the right supply profile to meet that, but we're working hard on that.

Our next question comes from the line of Timothy Arcuri from UBS.

Matt, I just had a question on the storage business. It was well above seasonal Q4 and a little below seasonal for fiscal Q1. I guess I'm a little surprised that it will be below seasonal for fiscal Q1 while all your customers are talking about controllers being on allocation. So I guess I wanted to just talk about that. And then I was wondering if maybe you can talk about what sort of you see as normal season in fiscal Q2. I think, typically, it's up like mid-singles. Do you also have flatter tail happening as well? So can you just sort of talk about those 2 factors?

Sure. Maybe, Jean, I'll have you answer the first part on storage. And then, Tim, I'll have you repeat your question, you broke up a little bit on the second part of it. But Jean, why don't you go ahead and give us some comments around storage?

Yes. On storage side, in Q4, we really had a really strong quarter, and the performance really is a combination of our own product cycle and also strong end-market demand across the board. So in HDD and the fiber channel, certainly, we see the demand in nearline, which our business performed really well, and also just broad demand for -- on the client side, too. On fiber channel, we do see the demand bouncing back and also the customers are restocking because of the supply challenge in the last year has caused a lot of problems in fiber channel. So fiber channel is a little bit lumpy. And then if you look at the Q1, frankly, the storage controller business, when you look at HDD and SSD, they are actually quite seasonal. It's declining quarter-over-quarter, very consistent with the broad market. The fiber channel is where it becomes really lumpy. The Q4 performance is pretty good and strong, but the customers also are restocking. So we do see Q1 fiber channel, again, is declining much more than seasonal. But if you look at the overall, in the longer term, fiber channel business actually is quite stable. So the quarter-over-quarter lumpiness is definitely exaggerated last year by the supply chain constraints. I think you need to repeat your second question because we did not hear it clearly.

Okay. You can hear me now. So I guess the question was just given all that movement, how you think about what normal seasonal is for fiscal Q2, like it's typically up sort of mid-singles. Is that the right way to think about it into fiscal Q2?

Yes. I think we'll -- I think, Tim, so first of all, I think we're -- we have a -- I don't even know what the new normal is in storage anywhere -- anymore, but there certainly is a seasonality effect to that, but we're not really guiding out that far at this point. What I want to emphasize, though, is that even on the sort of lumpiness of fiber channel, which has definitely started, I think, with the shelter-in-place orders that went into effect earlier in 2020, there's just been a -- it's been a lumpier business than normal, and so it had a very good Q4. And then, obviously, we've guided it down for Q1. But if you just sort of step back, storage overall for Marvell is up 10% year-over-year in Q1 is what we're guiding. So -- despite some of the lumpiness. And this is a higher sort of beta business than the rest of our businesses. We're still pretty pleased with the trajectory of this business. And we do expect the fiber channel portion to normalize throughout this year, assuming that there's no more supply chain surprises.

Our next question comes from the line of John Pitzer from Crédit Suisse.

Matt, glad to see the 10% sequential growth expected in networking in the April quarter. But I'm kind of curious, when you look back at the January quarter, I'm a little surprised that networking was down sequential. I mean, it was essentially flat in line with your guide, but just given the level of delinquencies, did the supply problem get worse? Or is this maybe a function that when you look at the networking business, there's portions that are just tied to campus and enterprise? And then there's sort of the growthy parts of your business. I guess, that the growthy parts grow sequentially and there was an enterprise overhang. As you look out throughout the balance of the year, how do you think about just a reopening trade around enterprise and the impact it's going to have on your networking business?

Yes. No, I think as we've said, the supply impacts we've seen have been more pronounced in the networking business. So that has limited our ability to deliver upside, like we want to. Even then, you got to remember, too, our -- the compare is a little bit tough because we did have a very strong Q3. I mean, you've got to now go back to sort of what are you comparing it to. If you remember, Q3 was up very significantly from our Q2. And then year-over-year, it was up a ton. So the flat compare, while we wish we could have done a little better, it's still, on an annual basis, still growing really well. I wouldn't really break out that there was sort of weakness in one or the other. I think we're coming off a big Q3 and then also the 5G was up and enterprise has continued to be up, and we've actually seen, from a year ago, everything growing. So I think it's more of a sequential issue. And certainly, we're pleased with the Q1, 10% up, and that business continues to do very well for us.

And then, Matt, just on the cost side, how much are you absorbing versus being able to pass on? Is it material to gross margins? And is that something that sticks around for most of this fiscal year? Or is that something that could reverse itself in the second half?

John, this is Jean. Yes, I'll take this question. So definitely, as Matt mentioned earlier, we are working with the customers to share the increased cost. And the way we're looking at it is our first priority is to really meet customer demand, especially with the increased demand for our product. As far as the gross margin, we are very pleased with our Q4 performance and the Q1 guide. And the Q1 guide, if you look at it, it's primarily driven by mix and we'll continue to improve our gross margin to our target level. We discussed during our Investor Day, which is 65%. I think the pace of that improvement is probably different in today's supply-constrained environment versus a normal environment because we do want to make sure we meet customers' demand, increase the demand on this current supply-constraint situation. So the pace will be a little bit different, not as we expected during our last Investor Day.

Our next question comes from the line of Blayne Curtis from Barclays.

Just I just want to ask, I know you answered a lot on the storages and said it would be short all year. You did miss some shipments in January. And I thought at the time you were thinking of catching up in April. So as you look to the guidance of 10% for networking, one, I just wanted to know, you're catching up on some of those January shipments within that 10%. And then I think I heard you say basically everything but 5G ASICs would be up. I was wondering if you could give us any more color as to what the primary drivers of that growth is.

Well, yes, on the second part, you're right, other than 5G ASICs, which were mostly tied to the China deployments in 5G, which were extremely strong in calendar '20. And we're seeing that digestion now. Yes, the rest of it is up, but there's not necessarily a catch-up per se, in Q1. I mean, we are seeing demand increase. And I think we've -- we see revenue going up and we see -- so I'd say it's more demand-driven than just, oh, well, we've caught up and now the demand is flat. We actually have a pretty robust demand picture for that particular business, both year-on-year and sequentially.

And overall, to just add to what Matt said, overall, we continue to have a similar delinquency, right? So it's certainly supply chain constraint.

Our next question comes from the line of C.J. Muse from Evercore.

I guess to clarify, Jean, are you implying OpEx in fiscal '22 is growing roughly 2.5%? And then I guess for the primary question, Matt, can you discuss your engagement with hyperscale customers? Can you kind of walk us through the initial work on perhaps discrete ASICs and how that kind of relationship is evolving to perhaps more custom solutions across increased areas of processing?

Yes, it's our OpEx year-over-year, you can think about, it's around 3%, the typical inflation year-over-year. We are managing our resource allocation to ensure we continue to invest in all the growth platforms we have. So that's how you can model it.

Yes. C.J., I'll take your other question. So I think the -- if you even go back to our purchase of Avera, there was really 2 -- we had a thesis there on 2 fronts. One was to be more relevant in the 5G market and be able to offer full custom design capability along with the semi-custom that we had developed with our lead customer. And as you've seen, that's gone really well. The second was that the recognition that in the cloud hyperscale market that more and more of the -- particularly compute-based solutions, were going to go to full custom or even semi-custom, but a need to really have a full ASIC capability. And these are -- there's a range of applications, by the way. These things could be anything from AI or ML chips to server-class CPUs to networking products, smartNICs, so there's a variety of applications. And as I said in my prepared remarks, this trend is here to stay, and we think it's good for us. I think there is a world that, that's post x86 or certainly coexist with it where more and more of the compute and more and more of these solutions are moving to very, very customized products. And we have all the ideal pieces to support that in terms of our process technology, our in-house IPs, like our SerDes, both for short reach and medium and long-reach at the highest frequencies, including 112 gig. And our ability to really partner closely and deeply and be flexible in our business model as well. And what I mean by that is we can do standard ASICs that are the traditional model you think of. But increasingly, what I'm seeing is the engagements with the cloud companies are really more partnership-oriented where it's solutioning together and it's proposing architectures and proposing 4 different ways to do something and then coming up with a model that really suits their needs. And so I think sort of our go-to-market plus our technology road map and then having all these key pieces, whether it's networking IP, very, very strong compute portfolio with all of our history from the Cavium side and ARM-based processors to security, to storage. And then within PHY coming in, obviously, the optical connectivity. So I think we're going to -- we can be an ideal partner here, and we think this is going to be a very big opportunity, and it's going to be one that's here to stay.

Our next question comes from the line of Ross Seymore from Deutsche Bank.

Just a clarification, then a question. The clarification, just, Jean, usually, you give a little color on what the other segment is going to do. But the main question for Matt is on the 5G side of things. I think everybody knows the digestion period that's going on in China now, some of the transitions going on geographically and the optimism people have in the second half, but you're really outperforming that -- during that transition period that we have right now and it sounds like you'll continue to in the back half. So I guess what I'm getting at is when you think about your company-specific drivers, separated from the market as a whole, improving in the back half, do you think that outperformance you're delivering now accelerates or decelerates? And what are the drivers of that performance above and beyond what the end market itself is going to do?

Sure. Why don't I take the 5G question, then Jean, you can comment on other. So yes, Ross, I think you're right. I mean, we've we performed really well on the cycle. We had 6 quarters of sequential revenue growth in 5G, and we were able to power through as new products ramped. And new customers came online when some others maybe felt different paces of digestion. So I think we're pleased with that outperformance in that run. The China, in particular, deployments last year were very large. And so as you pointed out, I think a number of companies are going through a similar digestion, but we do expect this business to accelerate and continue to outperform for a couple of reasons. I think one is our lead customer has certainly gained momentum in some of the markets that we'll be ramping in the second half plus in calendar '22. So that's a positive sign. And there's also -- within them, there's programs where we have new sockets where content is also -- it's new content for us. And then we have our second customer, Nokia, which this year will be our first full year -- not even a full year. It's ramping this year into production with our baseband product. It will be a full year in '22. And then we have additional content with that customer, which was all publicly announced last year that will start to phase in after that. So I think there's -- yes, there's a number of Marvell -- very specific Marvell drivers with some of the most important companies in this space. And then even looking beyond that horizon, we've got really good traction with our ORAN initiatives. And we're in the mix on a number of very important sockets across all the OEMs that will keep driving growth here. So we feel very good about our 5G position that it can continue to outperform and accelerate going forward.

Yes, Ross, on the other question, just a reminder, other is largely for our printer business today. And quarter-over-quarter, it's actually going to be just up a couple of million dollars sequentially.

Our next question comes from the line of Tore Svanberg from Stifel.

Yes. And I'll keep it to the one question. So Matt, you referenced some strategic meetings with Inphi. I know the deal is not going to close for a few more months. But from those meetings, could you perhaps give us a few examples on how the 2 companies are combining the IP to go out and target some new opportunities?

Yes. So yes, let me give you a little background on what we did, how we're thinking about it and maybe some broad brush strokes on the opportunities. Certainly, we're not -- we're still 2 separate companies. So I think more to come on that, Tore. And some of the opportunities as well, some are pronounced and some are nuanced, but the bottom line is first of all, Ford and Lloyd, the founder of N5, put together just an incredible team. I mean we've now gotten exposure through this process, which was great. Their key engineering leadership, their key product marketing people, their business unit leaders and also same on our side. So it was really a very, very powerful, I'd say, joint meeting where we reviewed all of our various businesses, and we were able to identify a number of exciting areas where we certainly can work together. In general, the technology is obviously very complementary, but we do sit in the same systems. And certainly, there's things like process road map and customers where we can go in together. So I'd say the net -- and the way we -- the reason we did this is we're really approaching this as a merger with the Inphi team. And so we're going to bring them in, and we're going to look at really the combined company spending and identify jointly the best opportunities to get the highest ROI and drive the best top line growth. And so what we found is we believe this is going to give us an opportunity to allocate some additional spending in some very targeted areas within the Inphi team to actually accelerate some of their ambitions and growth. And I can tell you from our side and my entire management team, multiple layers…

Our next question comes from the line of Christopher Rolland from Susquehanna.

Going to the ASIC side of the business, maybe piggybacking on some that have already been asked, but maybe you can talk about where this new design pipeline is coming from? Is it storage? Is it networking? Is it compute? Where are you getting the most traction there? And then as we look out, let's say, 5 years, Matt, what percent of your business do you think or expect or would like ASICs to be?

Yes. I think -- let me give you the backdrop. So I think the exciting part of our ASIC opportunity, and I would even broaden that to say our broader, I'd call it, advanced technology opportunity, of which ASIC is certainly an important piece. This would be what I would call our 5 -- just think about our 5-nanometer platform as its own potential growth opportunity. We publicly announced this in August of 2020, but we have been securing designs in this technology as early as end of 2019 type of time frame, early 2020. And so since that time frame, we've grown this design opportunity funnel to be very significant size relative to anything we've seen before on a single sort of process node. And it's very broad, Chris. I think that's sort of the point I want to make. I think there are -- if you go by market, certainly, there are 5G designs that we've already won and secured as an example there, and there's a large pipeline in front of us as people move from sort of 12-, 14-, 16-nanometer type of products, and they want to go all the way down. There are significant cloud hyperscale opportunities. Some of those I mentioned earlier. But also in enterprise, we've had good success in promoting our OCTEON-based CPUs into that market. That would be more of a standard product, by the way, not a custom. And then even in storage, we've seen some very important customers in the storage market take a leadership role and work directly with us to develop state-of-the-art flash controllers for things like data center and enterprise applications using our 5-nanometer technology. So it's -- this is much broader, Chris, than just, oh, well, Marvell has got a 5-nanometer thing and they can go offer ASICs with a standard sort of model, and that will generate some revenue. It's really a transformational platform for us, which we're -- we could use for traditional ASICs. But increasingly, we're finding that these custom opportunities that people come and ask us for, several of them have now converted into Marvell products where we're designing the entire product based on a spec. And designing the product for them rather than -- and then, of course, there's more value that can be captured there. It's a more stickier engagement. And these are all very significant type of opportunities, in particular, in the cloud market. So I hope that's helpful context to sort of broaden it because certainly, we had a view maybe going back to when we bought Avera. Hey, this will be great. We can have a nice big ASIC business, but it's turned into much more than that. And yes, and we're excited to talk about that in the coming quarters as we close designs and we take advantage of the pipeline that our sales team and our business units have developed, which is extremely strong.

And what percent do you think it could be in 5 years?

That's hard to say. And I think it depends on what you mean, like is it -- how much is ASIC versus how much is semi-custom versus how much is in 5G, but this is clearly a -- this is the future growth of the company. So you should assume these things are multibillion type of opportunities for the whole platform going forward, right? I mean, this is where we're going to be the growth engine of the company. So it's very significant. These are not nichey one-off designs that we're winning that will kind of be a nice sort of headline news product, this is going to be really the driver of the company's revenue growth. So in aggregate, all of this becomes a very significant portion of our future company revenue.

Our next question comes from the line of Joe Moore from Morgan Stanley.

In terms of the constraints, I wonder if you could talk to -- you mentioned substrates as being maybe the biggest factor and then certain geometry wafers, which geometries are those? And then just as a bigger picture aspect to this, it seems like from Marchex, that networking chips in general are tighter than other things, even though there's tightness in graphics and that you have to be able to grow quite a bit. Is there something different with substrates and things like that, where this is affecting the infrastructure businesses more than it's affecting other areas?

Yes. I think on the first question, Joe, it's hard to say. I mean this is such a dynamic environment, right? We've been reviewing every part of the supply chain as a team for months, right? And we're finding that we solve one issue and then another one seems to appear. And if you even go back to our last earnings call, I remember there were some questions there like, tell me about the supply chain issue and you fast forward 3 months and the U.S. automotive industry is in disarray because they can't get chips, right? So things are moving at a very rapid pace. I will note that we're supplying quite well actually on our automotive products, so don't make that comment with Marvell. That's more of a general one. But we do see point capacity stress with complex substrates, for sure. We see some of the "older nodes" and what I would mean by that is this isn't sort of 5-, 7-nanometer type of things. These are more going back to, call it, 28- and above, right, that have different pain points that companies are working with. And that's where a lot of this automotive, I think, tends to sit. So it's -- there's not a silver bullet. It's just going to take a lot of hard work. And even things like -- I think I noted in the last call, things like wire bonders have -- those lead times have stretched. So our OSATs are just having a hard time keeping up. Now they're spending money, and they're certainly adding capacity, and we're engaged in that planning process to make sure we're okay, but that's maybe a little bit more detail on that side. And then what was your second question?

Yes. No, that was basically the question is just, is there something different...

Oh, on the networking chips, yes.

Yes. I mean, graphics chips are in short supply, but people are kind of growing 20% faster than they thought and then constrained at that level where it seems like the networking shortages are just as severe. We're hearing about them from your customers, but demand doesn't seem to be on that kind of crazy trajectory.

Yes. It's hard for me to benchmark since we're not in the graphics business. I know that's a little bit of a different scale. But certainly, the complexity level of these products when you look at just the complexity of the substrates alone, I mean, I think the number of layers, if you go to these advanced products in, call it, 14-, 16-nanometer and below, have probably doubled in terms of the complexity of the substrates. And they're just extremely large die. These are thousands of IOs and some of these die are virtually the size of a credit card. So you've just got very low die per wafer, and you got a lot of demand. And it's been difficult for people to keep up. There's also test constraints that the test times on these things are very long because of the complexity of the product. So there's just -- maybe that's the reason. But I run an analog business and now I run Marvell, but I've never run a graphics business. I don't...

Yes. No I'm just trying to compare and contrast.

[indiscernible] about that. Yes.

Just comparing and contrasting. That's very helpful.

And our final question for today comes from the line of Srini Pajjuri from SMBC Nikko.

Jean, I have a question on OpEx and also on the margins. I guess as you ramp some of this 5-nanometer designs and as you engage more in 5-nanometer, I'm just wondering how you're thinking about OpEx, if there's going to be more volatility as we go forward? And also, I guess, as we go to second half of this year and beyond, and 5G and ASICs become a bigger portion of your revenue, do you see any further opportunity to improve your gross margins from these levels?

Yes. On OpEx, right, our team has done excellent job to manage operating expense. But you're right, there are some variabilities quarter-over-quarter due to the project cost and also we are working with our customer together, sometimes the customer portion of it may accelerate or push out. So there are variations, but I feel quite comfortable. Look at the overall for fiscal '22, we can manage the OpEx at the same time investing. So for instance, our Q4, our OpEx is slightly higher because of project cost, but we'll manage it overall. On the margin side, I think our #1 objective is really to meet the demand. We have significant opportunities for the company, and we have increased the demand across all the growth platforms, as Matt mentioned, the 5G, cloud and also automotive. So that's actually our #1 objective, and we certainly will continue to improve gross margin and make sure we get the most margin dollars out of our business.

This does conclude the question-and-answer session as well as today's program. Thank you, ladies and gentlemen, for your participation in today's conference. You may now disconnect. Good day.