Marvell Technology, Inc. (MRVL) Earnings Call Transcript & Summary

Good morning and thank you for attending JP Morgan's 20th Annual Technology Investor virtual forum here at the Consumer Electronic Show. My name is Harlan Sur, I'm the semiconductor capital equipment analyst for the firm. Very pleased to have Chris Koopmans, Chief Operating Officer; Ashish Saran, Vice President of Investor Relations for Marvell here with us today. Marvell stock as you know is one of our topics in the semiconductor sector for 2022, leadership in cloud data center, 5G networking with their networking compute storage and custom ASIC solutions and emerging pipeline in the automotive space. So, thank you for joining us today Chris and Ashish.

I'll kick it off with the first few questions and then we'll open it up for some Q&A clients, again can use the digital conference book to type in their questions. So maybe to start off, the demand picture in calendar '22 looks quite strong for the Marvell team, 30% growth in the core Marvell business, 30% growth in the Inphi business. We'll get into the supply side dynamics in the next question, but help us understand what percentage of the revenue outlook is just the strong spending backdrop by your cloud and applies in 5G customers versus your product cycles, and what markets will outgrow your 30% target and what markets are going to be in line to below?

Sure. No, great question, Harlan, yes, good to see you today. I think you kind of go back to the construction of kind of how we look at our end markets. If you remember at our Investor Day, we talked about -- we're part of a market where the underlying market is growing at about a 13% kind of CAGR in our SAM, and you'd argue that number is probably a bit stronger right now based on your comments in terms of the end demand being stronger. So I'd say the outperformance beyond that, call it, somewhere in the mid-to-high teens is really a result of our own product cycle, right? So we've got an end market, which is growing perhaps better than our long-term forecast, certainly with the demand we are seeing in multiple end markets. And on top of that, we've certainly got our own product cycle. In 5G, you've got a big ramp in our processor business, right? You look in the cloud, we've got DPUs ramping. We have a 400ZR cycle on the Inphi product line coming up for us. So I think it's really a bit of both. I think if you look at enterprise networking, which has been very strong for us, I argue that's predominantly because of our own product cycles, right, whether it's our multi-gig on the PHY side or -- Ethernet set portfolio, so I think it's a combination of both. Part B of your question, which is, as we look at next year, what end markets are driving above and below that 30%. I'd say the markets one would expect to be our fastest growth drivers, which is 5G, auto and cloud. Those clearly would be well above, I would say, the average, call it, 30% we suggested as a target for next year. I think the only one, which would be below for logical reasons would really be consumer, I would say, and maybe enterprise that are on prime, right? That is still a growth market, by the way. It just wouldn't grow as quickly as 30%. But it's really being driven by cloud, in particular, 5G, we're ramping up quite nicely in Q4, and that obviously continues to grow next year. I think those are some of the good ones. But I mean, enterprise networking, we still expect will do well next year. I think we still have some legs in terms of both what we're doing from a market share perspective, and that is a refresh cycle, quite frankly, just getting started. I think the whole hybrid work environment is in very early stages of how the IT folks will outfit buildings to deal with this mixed environment, which will -- it's going to continue quite frankly. I don't see us going back 100% in either direction. So I think that's all in front of us.

Despite more supply availability back at the December earnings call, you guys mentioned that you've not made a dent in the unfulfilled backlog, which implies that demand continues to trend well ahead of supply and you don't expect -- at the time, you also said you don't expect to make a meaningful dent in the unfulfilled backlog even through this year, but things obviously can change very rapidly in this industry, as we all know. And investors have had some concerns on easing supply constraints. So has the Marvell team started to see any changes, pushouts, cancellations, lead time changes? Or is demand basically still trending ahead of supply at this point?

Chris, why don't you lead with that since you're very tightly managing that whole process?

Yes. No, it's a great question, and it's something we watch very, very closely, of course. It's critical and important to make sure that we're putting the supply exactly where it's needed. So any push out or change in the bookings profile is something that I'm paying very close attention too. And no, to be honest, we haven't seen any changes. It's been pretty solid and pretty stable. We haven't seen pushouts or cancellations. We continue to see -- we increased our lead times earlier last year and at that point, but they've been pretty stable since then, and people continue to lay in -- companies continue to lay in bookings at that lead time right aligned with our forecast. So that we haven't really seen any changes in demand other than continued growth. I mean, as we mentioned, we continue to improve supply, but we haven't stopped winning designs. We haven't stopped gaining share. Our customers haven't stopped doing better. So ultimately, the demand continues to grow very strong and be very, very healthy.

Through most of last year, the team was focused -- very focused on trying to secure more supply and a strong demand profile. And it seems that, Chris, you and your team were successful in securing supply for this year because Matt and Jean were pretty confident enough to guide 30% growth this year with sequential growth every single quarter, right? So your operations and procurement teams have done an excellent job. So what changes, agreements, partnerships, have you put in place to motivate your suppliers to actually commit to your supply requirements?

Sure. Yes. So it's been my full focus for the past 10 months or so, it's really building strong relationships with our suppliers and securing additional capacity to support our customers' growth and demand. What I would say in terms of what we've done is, first of all, if you kind of think about the way that the supply chain typically works, it's like we're the customer, they're the supplier, and typically, the way a supply chain works is you try to get multiple suppliers and get the good better cost if you can and try to drive cost out of the system. We kind of flipped that on its end a little bit, and we turned into more of a sort of partnership. I went out to each of our suppliers and said, "Hey, we'd like to do -- this is what our company is doing. This is our growth profile, okay? These are what our customers, these are our end markets, this is our strategy, and this is where we're winning." Now if that's aligned to your strategy, right? And it's our profile. We're a company that we do business with. And ultimately, we said we'd like to do more business with fewer suppliers and build deep partnerships. When you're building leading-edge products like data center switches and high-end ASICs and the processors, these are just things you just send out to any old supplier, we need very tight technical relationships. So we went out to those suppliers, told our story, built executive relationships up and down and then found alignment amongst a number of suppliers, and said, yes, we want to over work with Marvell. And ultimately, some of those culminated in agreements. But I would say if the agreements were more of a culmination of the partnership and relationship rather than the sort of initial starting point, I mean initial starting point was let's align our strategies and let's build a multiyear growth profile. By the way, this was never about, hey, I need more supply next quarter. This was sort of line for the 3- to 5-year forecast of what I'm going to need based on what our growth profile and market profile looks like. Now how do we get from point A to point B because we all know sometime in the next 5 years, things will loosen up. And so that's kind of where we started and then move back from there.

Yes. And Harlan, just to add to that. I think similar to the conviction investors have right in Marvell's ability to grow in the long term, because I think we all agree that clearly the digitization of the world has accelerated and the demand is strong across the board in multiple end markets today. The question really is how many of these markets can sustain like year 1, year 2, year 3 and year 4 from today? I think that's the piece of the discussion where we started to really start to see our supply of the head started nod saying, yes, everyone is strong right now, but we actually -- once you laid out this road map, we actually have a lot of conviction that Marvell can sustain this growth for a much longer time period. You are already in very secular growth markets with sole-sourced positions with early marquee customers. I mean if you look at our customer list probably, it's really the top 4 or 5 customers in each key end market, right? I mean you can literally rattle off those names to the all what we would consider as part of in $100 million plus kind of annual revenue looks. And I think our suppliers see that as well. So I think adding that's been a big part of why we've been able to get more supply. Obviously, we still need to do more. So as much as Chris would like to say, yes, things are getting better, unfortunately, not quite.

Yes. I wouldn't say things getting better, I agree. I think the 2 things suppliers care about right now. Number 1 is the supply going to good use, right? They don't want it to end up in anybody's inventory, right? Number 2, is it sustainable, right? Is it just a bubble or is it sustainable? And I think everybody knows that some of the consumer demand is more cyclical in nature and tends to go up and down. And so they really like Marvell's infrastructure profile. It's sort of seen as a stable base and offset. So actually, some of the suppliers that have a very high exposure to mobile and consumer and PCs, actually like to bring Marvell more into their supply base or their customer base.

So is it fair to assume given the success of your team, but the demand profile, your product cycles, is it still fair to assume that even with the supply situation that you secured that demand, in general, is still tracking, trending ahead of supply?

Absolutely. Yes. I mean, like I said, we haven't stopped winning designs. We haven't stopped gaining share. Ultimately, our customers -- this crunch has also forced our customers to do the same thing, right, which is to decide with who we want to partner with and talk to us in a more strategic partnership type way, recognizing that supply is not just a given and understanding the relationship and the partnership and the way Marvell treats our customers, right, in the way that we approach our business. And a lot of times, they've looked at us and said, okay, we want to do a long-term agreement as well with Marvell in order to be able to secure our supply. And so ultimately, demand continues to be very strong and continues to grow.

Let's talk about the design -- the unfolding of the design and pipeline, especially the 5-nanometer pipeline. In answer to one of the questions on the last earnings call about the unfolding of this pipeline. You said that it would commence in this calendar year, starting off with your 5G programs. You've got 5-nanometer wins with Samsung, Nokia, Ericsson, you started ramping Samsung first, right, at the 16-nanometer, the 14-nanometer nodes on your baseband products for their base stations. So is it fair to assume that your first 5-nanometer customer to ramp this year is going to be Samsung and their next-generation base station platforms? Or will it also be 5G programs with some of your other Tier 1 customers? And is it both for baseband processing? Or is it also for radial DSP processing where you have 5-nanometer engagements?

Yes, Harlan, I think as you correctly pointed out, right, we made a very big strategic shift, right, accelerating to 5-nanometers. It's been a very successful strategy. Certainly, leading with processors first, right? And of course, given the activity on the 5G side. You should expect that, yes, processes is one of the platforms we've taped out. But in addition to that, you should also expect that our switches, our 5s, some of our customer ASICs are right behind it, right? And some of them already taped out. So the reality is, as we get later into this calendar and into next year, you should expect to see a ramp in 5-nanometer across, quite frankly, multiple product lines at multiple customers. In the 5G space, specific to your question, yes, I think we've got a very successful ramp with really multiple customers. Baseband is one of the key products. But remember, we also have the same OCTEON platform extends into both the embedded processor as well as massive MIMO, which is your other question in the radio. And you should expect, we're seeing activity across all of these things. So our next generation OCTEON, we announced that a while back, that's really the basis for these multiple processors. So I mean, quite frankly, we're driving our entire portfolio towards 5-nanometer. Even in areas where you may say, hey, is that already 5-nanometer, but we've actually found it to be very successful. So as an example, historically, the storage market, typically, you would stay on lagging-edge nodes, only because you really had no need, right? You just didn't have a platform. But since we already have a 5-nanometer platform for advanced businesses, we announced an SSD design win, an SSD controller on 5-nanometers, right, with PCIe Gen6 technology, which is -- I mean, leaps and down ahead of almost everybody in the side. So I think the reality is over time, Harlan, you should expect the majority of our platforms, including storage will actually move towards 5-nanometer. I think it's a big differentiator really ready for Marvell in the marketplace.

So beyond -- so if I would -- at the Analyst Day, you guys talked about taping out 25-nanometer chips over the next 18 to -- 18 months or so. Can you just give me a sense of the mix of those products, as you mentioned, it's across the board, but is it more networking? Is it more service provider? Is it more compute acceleration, things like AI and video transcoding storage?

Yes. I would say cloud meaning, call it, customized solutions would be a big chunk of them. If you remember, we discussed the acceleration of our set of cloud design win revenue from 1 year ahead, we thought it was originally $400 million in fiscal '25, we pulled that into fiscal '24, which is calendar '23 roughly for folks on the call. And then we said that number is going to actually double, right, the year after that, right? So all of those, right, and this is a number of key design wins. These are all basically in development, right. And these will start to tape out to drive revenue, call it, a year from now, right? So I would say that's a big part of those design wins, a number of those tape-outs, right? On top of that, you obviously got the processor platform, which has many variants, right? We've got the 5G products we discussed a few minutes back as the first set. But then remember, the same platform goes into multiple end markets, right? So those are also taping out Ethernet switches, PHYs, SSD controllers. It's a pretty broad set of products. But I would say compute and networking would probably be the 2 areas. And compute, of course, in this sense would apply even to the service provider market. Those are the key drivers, I would say on 5-nanometer. Storage would fall a little bit later.

Perfect. Let's talk about your optical connectivity and your data center interconnect business, right? Much of this came from the Inphi acquisition. And so prior to the Inphi acquisition, consensus estimates for this year was for Inphi to drive $900 million in revenues. But if I did the math correctly, given the parameters you gave on the last earnings call, I believe that the Inphi business is on track to do about $250 million plus this quarter. That's annualizing $1 billion in revenues per year. Number 1, is that correct? Number 2, what are the biggest drivers of the outperformance for the optical interconnect business? And maybe the other question alongside that is -- for Chris is, what's your assessment of how the Inphi team pre and post acquisition manage their supply chain?

So maybe, Chris, why don't you lead with the supply chain and then I'll answer kind of why we're seeing better demand. So lets go on.

Yes, sure. So there's sort of -- there are multiple parts of the Inphi supply chain, okay? They have [Indiscernible] just like Marvell does, okay? And then they also have broadband analog products, which are very different foundry, very different sort of supply chain, and then there's also the optics, right, which -- with the silicon photonics and all of those pieces that are actually really leading edge and cutting. So there's areas where I would say that Inphi is doing things a lot like Marvell and it fit sort of right into our portfolio. And then there's an area where Inphi was really doing things Marvell has never done if, in fact, nobody's really ever done before. And they're really inventing sort of a new set of supply chain and how to ramp silicon photonics-based modules. And ultimately, I mean, you asked for my assessment, my assessment is that they did very, very well. Now being part of Marvell helps a lot, okay? Because we're larger, we have a broader supply chain. We're more strategic. They did very, very well for their scale. And what I would say is that inside of Marvell, we're doing even better. And they're doing even better. I mean, we kept that entire team and they're part of Marvell and they fit into our organization and are helping Marvell to continue to ramp our supply and our products. So ultimately, I would say that they've done very well. But of course, it's challenging across all of these different pieces out there and we're still working on it.

Yes, Harlan, I think your maths kind of in the right cord. I think if I remember correctly, you're right, the consensus numbers for this year or for last year was supposed to be around $900 million plus. It's Inphi is comfortably ahead of that, right? I mean -- And kind of similar to Marvell, right? I would say even our back half numbers are now running better than what we all expected, and it's a combination of very strong demand. So I would say in Inphi's case, as you know, the majority of the revenue is really exposed to cloud and no surprise by that market certainly has some strong growth. But I would say, on top of that, I think Inphi in fact, your question is executed extremely well. I think it was a seamless integration once we acquired them. I think they really enable the entire market, right? I think there are some customers we serve directly. There's some customers we serve through our partners. And I think -- and by the way, we sell TIAs and drivers to a broad number of customers, including some who actually end up competing in the end market level. I think this is the flexibility of the business model and the idea that, look, we want to enable the entire market, realizing can do it on yourself. I think we've been such a good partner that everyone felt comfortable in continuing to work with Inphi, and we've really maintained very strong share, quite frankly. So strong demand, excellent execution. And we still have -- I mean, I would say there's still a long way to go, right? I think in the cloud market, there's only really a couple of customers who really converted to PAM4 within the data center, 200, 400 gig. I think there's a whole host of customers which are going to start to convert in the next 2 years, right? And then you've got the entire -- between data center market, which is ZR, a 400ZR market is just starting. We have already started chipping products. No surprise, we were first to market. But that's going to be a much bigger market than it was at 100 gig, right? So I think there's a lot more legs on Inphi in the next few years.

So on that note, on the 200- and 400-gig optical upgrade cycle, it's pretty amazing because Marvell, Inphi and based on my calculation, you guys still own 80% market share, right? You started with Google. You started -- you ramped into Amazon. Right now, you're in the midst of upgrading Facebook as they start their transition to 200-gig. I believe Microsoft upgrade cycle is right around the corner. I guess first question is, how has the Marvell, Inphi team been able to keep 80% dominant market share on this fast-growing market, number 1. Number 2 is, when do you guys expect the China cloud titans and the Tier 2 U.S. and global cloud guys to start their upgrade cycle?

Yes. I think if you remember, Loi had really done a great job at our Investor Day and really laying out how Inphi was able to create this kind of industry-leading set of products. I would really credit -- look, this is a market where you have to have the right technology, deliver at the right time, at the right power, at the right price. I think Inphi and now with Marvell, we've done that consistently, right? So I think our ability to really maintain leadership in this market is because we've invented some of these technologies and they're consistently driving performance. And I think we're -- back to Chris' earlier comment, I think you've got a great partnership with customers. So I think it's a combination of excellent technology leadership, combined with really understanding what each customers want. As you know in this market, right, while you look at this cloud market as one single market, it really isn't. It's really individual customers, and they all want different things, right. So I think the Inphi team has done a great job of really working with each individual customer, understanding what exactly they want and implementing those things uniquely within these solutions, right? So I think that's one thing, which sometimes I think the market doesn't appreciate us. There's a fairly high degree of customization that takes place even in these optical products. While they look the same from the outside, it 200, 200-gig, not quite. I think that's been a big part of it. So I think that's the reason how we've been able to do that. Then I would say the -- Yes, I think the U.S. cloud titans have driven, obviously, the first phase of adoption. But as you mentioned, there's a few more customers in the U.S. still starting to adopt. And then I would say some of the large customers in Asia are still -- they typically, are a couple of years, I would say, between 1 and 2 years kind of behind where you start to see some of the big U.S. guys adopt. So I would say that's still a couple of years out. So there's still a long, long runway for PAM technology within the data center.

With the new product cycle that Marvell, Inphi is the data center interconnect, where you mentioned it briefly, the 400 -- data center interconnect to the new 400ZR standard. I believe you're ramping your 400ZR COLORZ II solution into Microsoft. I think the next customer to start to ramp, we believe, is Amazon, maybe first half of this year. First question is, do you guys see much competition here? I mean, I know Acacia and Cisco have been making noise, but we've not heard that they're shipping yet. But first of all, what is the market opportunity? And number two, do you guys see much in the way of competition?

Yes. I would say -- I think the really exciting part of the 400ZR market is it's a significantly larger market, right? I think 100 was really a great partnership between 1 cloud customer and in fact, to really create this kind of new product, which is simply didn't exist, right? And that was great, and it did really well, and we've got a fantastic relationship with that customer. And to your point, we're starting to leave its 400ZR with the same customer. But the opportunity is significantly larger. It's across all, not just the cloud customers, but also in the service provider market, right? So our view is, yes, I would say you should certainly expect other players to be in the market, I think that's healthy, 400ZR is an industry standard. So I think it's actually very healthy. That our hope, quite frankly, is that there is going to be more than one -- it's going to be more than just Marvell in this market, which I think would be good for broad customer adoption, right? It's 1 thing to get 1 cloud customer who are comfortable having sole sourced. But when you're talking about deploying this in telecom networks around the world as well as market for cloud customers, you really want an ecosystem. So I would expect, quite frankly, that we obviously have a huge first-mover advantage in terms of having invented 100 and now driving 400, but I would certainly expect to see more players in the market. But as I mentioned, we're obviously first in the market. We're very happy with our leadership position. And I think you'll certainly see us involved with more customers as time goes on. So very excited about this opportunity. I think it's already starting to happen. I think that's the key point.

Yes. Let's talk about automotive. Significant opportunity for the team. You drove close to $150 million annualized revenue run rate in your automotive business in the October quarter. Here, the team is leveraging its strong position in Ethernet networking. This in and of itself is a $1 billion-plus market opportunity. You recently unveiled your OCTEON-based compute platform for automotive along with your first semi-custom design win for compute. What is it that the auto OEMs are not able to achieve with off-the-shelf automotive compute solutions, whether it's from incumbents like Renesas, NVIDIA, Intel, NXP. Like what is it that they're missing that you can -- what is it that customers can't get that you can provide to be able to fill this gap for your customers?

Chris, you want us to lead us through a hole or rightly announcement and then what we are seeing from customers? That would be right.

Yes. So we announced last month, our Brightlane brand, which is a new brand that will encompass all of our automotive products. And we've been saying for 4 years now, we got into this market and we started with Ethernet. But our view was that sort of the cars of the future in the automotive industry of the future was really the companies that were going to be the most successful in the next 10, 20 years, we're going to be data-centric, IP-centric type of companies. And Ethernet is really just the first start, right, replacing all the legacy links with Ethernet was a start, and that's obviously what's been driving our business today, but we've always seen a vision for our entire portfolio, including storage and networking and compute and security all within the automotive framework. And so that's been a very consistent theme that we've been seeing. And what I would say in terms of what's not achievable off the shelf is, ultimately, automotive companies are realizing that they really need to own the platform and design a unique platform, right, because that's really going to be one of the primary drivers of the user experience going forward in terms of what types of vehicles that you're -- It's not just going to be about -- in the past, a lot of times, it was about horsepower and torque and 0 to 60 and a lot of these types of things. But once there's an electric motor, a lot of that stuff becomes more similar amongst vehicles. And then the experience within the car is much more about the software and the platform and the ecosystem. And so they're spending a lot of time investing in leadership platforms. I think you've seen, obviously, Tesla has done that on their own and built their own silicon. And I think a lot of other companies are saying, well, we can build our own silicon, we can go off the shelf or we can go somewhere in between. Some -- it's a pretty nice place to be, because you don't have to start from scratch, right, developing everything from scratch for companies that haven't really been in the silicon business. But we can still optimize it and customize it for our precise needs in our precise platform and then have our own software platform sitting on top of it versus just leveraging a sort of open ecosystem and having everything be the same.

Harlan, it's the same playbook, if you think about it, what we did with OCTEON starting off essentially in kind of core networking markets a long time back. And then kind of expanding that into base stations over time. And then we've taken the same platform right, and it's now become a big part of our cloud customized silicon opportunities. It's kind of the same playbook going into the automotive market, really if you think about it, because back to your question in terms of who we compete with, right? I think if you want a semi-custom solution, but you want someone who can give you access to compute IP, right, storage IP, networking, right? Security, this -- I mean that when diagram actually shrinks fairly quickly. There's very few companies at the intersect which can do all of that and also know how to do automotive, right? That's a big thing. I think people shouldn't take the automotive capability for granted. It took us a long time to really invest in those maps and really build up that capability. And today, we're shipping at -- I think it's like less than 1 ppm, which is an amazing achievement and we able to give them all the semi custom works. I think we've got a very unique position in the market. And yes, I'm super excited. This is a much larger opportunity. So the $1 billion you talked about is really our Ethernet opportunity, and that trains that the station, that's well on its way. We've articulated at least a 50% market share. We expect to achieve in that market over time. So that revenue is going very rapidly. But this is a significantly larger multibillion dollar opportunity in compute over time. And we've already won the first design that's already in development with an active discussion with a number of additional customers, right, and to see where they want to go next.

Have you been able to leverage your significant design win pipeline on the Ethernet to get to the table and have these discussions on the compute side?

Absolutely. In fact, if you think about it, just like in traditional networks, right, you don't design compute independent on networking, right? So it's very similar in cars as well, right? As time's gone on, they're realizing you have the zonal architectures with compute clusters, right? That design is done in collaboration with what we want to do in terms of switching bandwidth, right? We can't do these things independently. So actually, to your point, it's a very, very powerful part of the ecosystem is the fact that we are the ones that developing that connectivity ecosystem around, and compute is done in parallel with that. So you're 100% right. It's actually been a big beneficiary, it's a fact that we actually have already -- we are the connectivity partner of choice. We set up 7 of the top 10 companies in the world have already started to use Ethernet solution for mobile. So you can imagine we've got a very strong pipeline and a very strong view into their long-term architectures.

Last question on the financial and capital return side. So once you delever, you're targeting 2x gross leverage target, you'll return greater than 50% of your cash flow. And I get this from investors all the time that we and investors are surprised that the free cash flow return is not something on the order of greater than 80%, right? Because the Marvell team has built strong revenue scale, right? The core business, Cavium, Inphi, Innovium. You've got a complete portfolio of IP in all of your target markets. So why the conservatism on the capital return target?

Yes. I would say, think about that capital return target as we set up a minimum, not a maximum, right? So I think we articulated, hey, it's at least 50%. It doesn't mean it will only be 50%, right? And if you remember that our history, if you look back in time, when we've done some large buybacks, right, even in some cases exceeded 100%, right, in that particular time frame. So think about that as a minimum target, right? Not saying you wouldn't do more. And you obviously want to leave some flexibility on the table, right, depending on market conditions, depending on what else we see. But to your point, I would agree that, as Matt articulated at our Investor Day, look, we've done the changes we needed to do. As you pointed out, we basically got all the key IP now. And quite frankly, there isn't that much out there. And I mean, if you look at it, right, I mean, for us to acquire another company, first, there's not that much we need. But who else can buy, which is going to drive the kind of growth we are driving ourselves today, right? There's not that many people out there driving 30% plus growth in the infrastructure market. So I think we're very happy with our asset base. And yes, I think you should expect us to start to deploy more of our free cash flow towards shareholder return. So look at the 50% as early as the minimum number as the way I would interpret it.

Perfect. Well, -- We're just about out of time. Chris, Ashish, thank you very much for joining us here today. Best of luck on the new year. It's setting up to be a pretty strong year for the Marvell team from our perspective. So looking forward to that, and thank you very much.

No, thank you, Harlan. Yes, we're very excited about what's in front of us for the next number of years, not just next year.

Thank you.