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

Good morning, everyone. Welcome back to our session. I'm Vivek Arya. I cover semiconductor and semicap equipment at BofA. Really delighted you could make it to this session, and we are really delighted to have the team from Marvell, Jean Hu, Chief Financial Officer; and Ashish Saran, VP of Investor Relations. And my plan is to just have me going through a number of questions, but I will have some time towards the end for questions from the audience as well. But with that, welcome, Jean, Ashish, really appreciate you being here.

And I know you just reported a few days ago, right, very strong results and outlook, right, very different from just the external perception that investors have of this industry. So maybe, Jean, give us like a state of the union, what's going well? Where do you see as kind of headwinds that are small today, but investors should keep in mind, just the state of the union?

Yes. First, thank you for having us here. Really great to see everyone in-person Yes, we announced our earnings 2 weeks ago. We literally -- that's the Q1 for our fiscal '23. We really started our fiscal '23 with a very strong financial performance. The Q1 revenue is record revenue, growing 8% sequentially led by our data center business, right? So if you look at our performance, our data center now accounted for 44% of our revenue. And then enterprise networking, carrier infrastructure, auto industrial are also very strong, which accounted for 88% of our revenue. So if you think about Marvell, only 12% is a consumer-based business. We are most exposed to data center infrastructure, which is definitely the secular growth trend, right? In cloud data center, in 5G and automotive. I think that's why we have been performing really well because we have been investing in those secular growth markets over time. If you look at our Q2 guidance, we actually guided another quarterly significant up, right, 5% sequential increase, 41% year-over-year increase, which is entirely organic. We had Inphi last -- Q2 last year for the full quarter. Again, it's the data infrastructure market. It's the secular growth trend we have been investing. We now see the benefit. It's Marvell's unique new product cycles. Of course, our business model, we actually have more than 65% gross margin and the operating margin in Q2, at the middle point, will be 36.5%. So we're tracking to get to our target operating model.

Got it. And anything, Jean, on the other side from a headwind perspective, whether it is supply or other issues from a macro economy perspective that you think are concerns that we should watch out for?

Yes. Our team has done an outstanding job to improve supply. And if you look at our Q2 guidance at $1.5 billion revenue versus a year ago, which is $1 billion, so we increased our supply significantly. That being said, we continue to be significantly supply-constrained. Our demand continued to outpace the supply. We do think we have enough supply to maintain incremental revenue growth for the second half. But you're absolutely right, supply for Marvell, it's not only second of half. We do see it going into next year. We'll continue to face supply constraints across substrate wafers and other areas.

Absolutely. Jean, I wanted to talk through different parts of the business, but I actually wanted to start at the very high strategic level in that Marvell has put together a collection of very interesting assets, right? You had very strong IP from organic Marvell, then Cavium, then Inphi, right, then Avera, then Tanzanite and then Innovium. What's the big strategy? Like if I were to take you 3, 4, 5 years back and say, lay out the perfect -- how have you identified these targets? And how have you kind of put them together to make them part of a nice pull of the company?

Yes. I think the Marvell transformation story really started from when Matt Murphy, our CEO, joined the company 6 years -- almost 6 years ago. The first thing the company has done is choose the right market. So 6 years ago, Marvell is largely consumer semiconductor company, right? 40% of our revenue is HDD products and 85% of that HDD actually is consumer-based. The first thing we have done definitely is choose the right market and decide that the infrastructure, those are secular growth trends we see like 5 years ago. And also, they have a very long product cycle, very high barrier to entry. Once we chose the market, Matt and the team really did a lot of things. First is organic investment. We really invest our R&D significantly in organic investment to pivot our internal development to data infrastructure market. At the same time, we also were not free to do acquisitions, as you said, is we acquired Cavium very early on to either the compute technology. And then we acquired Avera to add ASIC capability, Aquantia and then we also add Inphi to really add the optic-electric, that kind of capability to build our portfolio. At the same time, we're also not afraid of divest, right? We sold our Wi-Fi business for $2 billion which we can use that cash to invest in the data infrastructure market. So the overall transformation really is surrounded to choose the right market, build the right portfolio. Today, if you look at the portfolio we have, we probably are one of the best positioned in data infrastructure market. That's the strategy, the execution. As you know, the team is very focused on consistent execution. So it's the whole package that allow us to drive today's performance.

Got it. So you think the band is now complete? Or you think that there are other parts of the -- that you can have the nice -- right orchestra at this point or...

I think if you look at today, we are more focused on organic investment, right? So we have the right portfolio. We have tremendous design wins. Ashish can talk more about it. So the team is very focused on executing those design wins. So we have been adding engineering talent to ensure the execution to delever future revenue growth. There are always small technology acquisitions like Tanzanite, very small, but strategic important. So we'll continue to do the packing small acquisitions.

Right. I asked this because you have other larger peers in the industry who have chosen to say, look, no more semis, right? We will do something else. But -- and I just wondered if it's your size that maybe gives you that advantage where you can actually go through all the regulatory hoops, right, and consolidate more in the semiconductor industry versus, say, some of your peers?

I'll start. I think we're very focused. We chose the data infrastructure market. When you look at the semiconductor companies in that market, there are actually not many, right? We -- the beachfront properties we got like Cavium, Inphi, those are so critical. So the team really was aggressive we took out those acquisitions. But as of today, we -- there are small companies there, but we are very focused on the market we participate.

Yes. I think, Vivek, I think the different reasons for acquisitions, I think we've done them to drive growth, right? And it's all about building things around our existing platform. We've already done most of that. And I think the other thing we're doing is we have a lot of opportunities to take our existing platforms and get a lot of leverage out of them. Just a few examples is we talked about entering the CXL market, the active electrical cable market. They're using existing platforms. You don't need new investments. We've talked about our OCTEON compute, right, our DPU platform which started off years and years back in Cavium in the enterprise market, which is now shipping in enterprise, shipping in cloud, it's shipping in carrier in 5G base stations. And soon, we've got a design win in automotive with the same OCTEON platform. So I think there's a lot of leverage we can do. We don't need to necessarily acquire to drive the next phase of growth. I think we can invest organically in what we already have.

Got it. Now on this topic of supply constraints, what did Marvell do incrementally in the last year to help improve that? And where are the constraints today? And then obviously, part B of that is, what can you do over the next year to address them?

Yes. Our team has done a tremendous amount of work. If you look at the Q2 last fiscal year, our revenue was like over $1 billion. And we guided Q2 this year at $1.5 billion, at the middle point. And that supply improvement the team has drive over the year because for the last year to now, you really don't have a capacity guarantee, right? It was -- if you started the year earlier, it's all like you try to improve supply chain. We do work with our suppliers to build the long-term agreement, to secure future supplies. As we talk about this, we have tremendous design wins, the team is executing and most of our products are so sauce. So we have to secure the future supplies. The team did a lot of work to have a capacity agreement, long term, even leverage our balance sheet to make a deposit to assure the long-term capacity. In the near term, right, is we continue to see constraints in the trailing wafers technology. We definitely continue to see substrate -- complex substrates we largely use in networking products. Those are continued to be into next year. We don't see the -- for us, from our angle, we don't see the supply chain constraint to go away next year.

Got it. Jean, is it fair to say that supply constraints are injecting a level of discipline in the industry because people are so worried that there is an excess, right? That do you think that supply constraints as much as they are not helpful, that they may actually be working in a way to inject some discipline?

Yes, that's definitely a good point, right? If you think about supply constraint, how we force the companies to really work with your customers, right, to really understand what's the true demand if they're building the inventory. That really helps us a lot because it's in allocation. So the first thing you do is you don't just give them what they need. You really try to figure out what's the end-market demand, how you -- what's the attach rate. You have that very diligent process established between suppliers and the customers. That discipline within semiconductor companies today since the pandemic and the supply constraint is much more robust. You do see, like Matt, CEO, he works with the suppliers, he work with the customers. So we do have a much better visibility. I think that discipline will continue, right, is semiconductor companies, there are always cyclical dynamics, but the secular trend, how the company can work with the customer to understand what's the secular growth, right, versus the inventory bubble. The supply constraint definitely helps the company. Like for us, we see very strong demand, right, continue to outpace the supply. So it really helps us to really make sure we manage our business well.

I think, Vivek, the other thing for us is a pretty big chunk of our demand growth is really coming from our own product cycles. So I think it's more a reflection of the share gains we won, which didn't turn into revenue until really late last year because now our customers are shipping their new products, right? And we're seeing higher content, right? So I think part of why we are seeing, call it, elevated demand in dollar terms is because of, I think, what we've been able to do, not necessarily that the units have taken off massively. I think that's the other difference, I think, with us.

No, that's a good point. The inventory on balance sheet for, not just Marvell, right? We have seen this across the board for many semiconductor companies. Inventory is going up. And sometimes when investors look at that, they're like, how is that possible that these companies are all complaining about supply, yet inventory keeps on building up. Isn't that a yellow or red flag for the industry?

I think that overall, we do see the inventory going up across the supply chain, right, not just semiconductor company, OEMs. But I do think from our angle, we have significant revenue growth, right? And we look at the second half, we'll see continued sequential revenue growth. So we have to build the raw material and working process to prepare for that revenue ramp. There's definitely a complexity today, right? We have a very diversified product portfolio. So the shortage of component, sometimes it's very unpredictable. Suddenly, there's a factory closed down somewhere, then you have shortages. So you tend to hide a little bit more, right? You tend to try to build a little bit more to have a flexibility to support the revenue growth. That's the complexity of today's supply chain environment. I don't know what the other companies and how they think about it. But for us, it's definitely the case. You have to build a little bit more because you just don't know which component shortage will hit you during what time of the quarter. And one good example is our enterprise networking. We see strong demand and strong growth. So we actually guided for our Q1, we could grow double digits sequentially. But we couldn't get there because there are different component shortage. And then if you don't build enough inventory to prepare for that, it's going to -- your revenue growth is going to be impacted.

Got it. So how do you go through the process, Jean, of saying that, look, this is the right amount, like is it a number of days that, look, once it goes beyond a certain number of days, that's the time that you'll start putting on the brakes. Like how do you make sure that there is discipline into that specific part of...

It's not even at that high level. In today's inventory management, you literally go down the part level. The detailed process, you work with your business unit, working with your customer at the part level. For us, I think the benefit is actually we tend to have large customers, large OEMs in each profile and the market. We have top 3, top 4 customers. So we do have a very strong relationship with them. We literally look at each part. That's how disciplined our supply chain has been is you watch each part number, what you build, what inventory, at -- what is the wafer level or it's at the testing assembly line or sometimes finished good, you couldn't even ship at the end of the quarter because of the timing. So in today's environment, you literally have to manage it at the part level.

Yes, Vivek, I mean if you think about just an example, right? So it's really at the program level. There are run rate businesses where you manage things differently, probably a little bit tighter versus, for example, last year, we knew we're going to have a big Q4 step-up in our 5G business. We got it up 30% sequentially revenue growth. We prepared for that earlier, right? Because we know it's a new cycle. We're in a very similar situation right now where we have some very big ramps in front of us back half of this year, especially next year. For example, we have a big cloud optimized silicon design wins we've talked about, which will drive $400 million-plus of incremental revenue next year. We have to start to prepare for that now. You can't wait till the last minute. So I think the world of just in time is not quite there. So I'm not surprised we've seen in across the industry. And for us, these are sole-source designs, long life cycles. These products typically shift for 3 to 5 to 6 years. I think the risk for us is not having the product at the right time versus being concerned that the demand is only in a short window. It's quite -- and not for us. So I think for us, it actually makes more sense to give us more flexibility, and that's what we've been able to do.

Absolutely. So is it then -- it may seem like a self-serving statement for the industry that may be the market should just be prepared for semiconductor companies to just have higher inventory levels that, that by itself just because as the industry has become more fabless, it's not like you have to go fill a fab and suddenly you see a lot of inventory, and that can be a problem that as the industry has become more fabless, then we should just be prepared to see higher levels of inventory, but that's the new normal.

I think definitely, it changed, right? I remember before the pandemic, at Marvell, we were managing our days of inventory at 80 days, 85 days. Really tight because it was like all about the cash flow conversion and everything. But that's how we got into trouble, right? 85 days when the pandemic, the supply chain situation hit us we couldn't even handle it. But then over time, you definitely understand today's environment. If the supply chain environment loosing up, the company can manage it, right? I definitely think the days for -- to keep inventory at 80 days for us is history.

History. Indeed.

Yes.

Yes. I don't think it's just semis, Vivek. I think it's across the supply chain. I think it's -- you'll see this at another levels as well. Everybody has the same problem. I mean there's multiple examples, right? Where you just can't rely on the just-in-time system, where if you had sudden extra demand, you would say, yes, within the next 6 weeks, and you would get it. That simply is not the case anymore.

Got it. You mentioned trailing edge, right, not having enough capacity there as one of the sources of supply constraints. The external perspective is that, look, as demand for some consumer products slows down that's going to free up capacity. So part A of the question is, have you seen that slack start to help you, right, given your mix? And part B is, we are starting to see a number of people in the analog microcontroller industry really ramp up CapEx. When does that incremental capacity start to help you out? So kind of 2 parts, 2 trailing edge.

Yes. We definitely have not seen.

Have not seen?

Yes, we have not. And we definitely talk to our suppliers using the same information you have, right? So we basically tell them, hey, look, the PC market is slowing down, you should allocate more to us, but we have not gotten...

Is that surprising, Jean?

I think naturally, we were hoping the same thing. But on the other side, if you think about she has a very good thing to say, a point is about a lot of our customers they have a PC side, they also have a server side, right? It's -- they may not let go their capacity. They may just move the capacity within their demand. So not clear yet, and -- but we have not seen it.

I think, Vivek, I mean, our assumption is, if it happens, it's kind of an upside, but we've already taken care of our own capacity needs, which is we've been able to get more capacity as we did late last year into this year, we're driving 40% plus organic growth at this point. We expect to get more capacity every single quarter, right? So I think we've taken care of being able to drive sustainable revenue growth for a very long time period. To the extent there is changes in the PC supply chain and some of that starts to open up, that's a nice bluebird, and I think it's going to take some time. Just like we have typical NCNR non-reschedulable on our customers. I suspect the same things across the industry. So even if there is a softness in demand, if you order it, you're probably still going to get it, right? The question is maybe 6 months out, you get some flexibility. So yes, we've certainly told our suppliers, hey guys, we're happy to pick on anything where somebody else doesn't want it, but I don't think it's going to happen in the next 1 to 2 quarters, right? It's going to be probably beyond that time frame.

Got it. Absolutely. The one other interesting dynamic, I think, Ashish and Jean, you mentioned also is there's a lot more company-specific share gain aspect. But still, look, 40% plus growth is a big number and people are so worried. So if you look at the different parts of your business like data center or carrier or enterprise, where do you think trends are maybe kind of -- where is that demand above trend line today that you think becomes a little more normalized as you look at next year?

We, actually, if you think about our transformation journey, we're just started, right? And if you look at the data center, which is 44% of our business. Specifically, we're more focused on the cloud, right, the cloud, the data center, all the growth drivers we have, we still have a lot ahead of us, right? If you look at the electro-optical side of you, have 200-gig, 400-gig to 800-gig transition, there's still a wave for transition ahead of us. And then you have 400ZR, which adoption just started, we are shipping not only to one customer and the more customers compared to the first generation of inter-data center connectivity. And then on the storage side, we have significant design wins in the SSD DIY space. Those are still ramping. Then networking, the switch side, Innovium side, we're just getting started. It's supply-constrained. So for us, we do have a lot of growth drivers. We're at the beginning, right, then on the 5G side, the same thing. We do see our 5G businesses started to ramp like -- especially Q4, we grow significantly. But going forward, you do see, not only 5G adoption continues through U.S. and Europe and India, they still have not started yet. But more importantly, our content. We do see our 5-nanometer 5G solutions start to ramp later this year, which will add incremental additional comments to next year. So automotive, that's -- we're still at the beginning of our ramp. So for us, we're actually very excited about the opportunities, not only this year, but ahead of us in the future. And then we also on CXL, I know it's a little bit further way out, but it's a very large opportunity, too.

Yes, Vivek, I mean just going -- just kind of summarizing, right? So I think your question is, hey, you're growing 40%, which is 2x the top end of your long-term target. And I think the way to think about it is, is that 2x growth is not because the end market is growing 2x. The end market is growing faster than we expected, sure. But I think what we are seeing is the faster realization of basically content as well as share gains, right? And kind of the perfect example is enterprise networking, right? The end market is not growing 30%, 40%, 50%, 60% year-on-year. It's growing, sure. But I think what we are seeing is all those share gains starting to show up, right, our multi-gig Fis which have much higher ASP. So I think it's after realization of our own product cycles is the reason why we're growing significantly faster than what was kind of our upper end of our long-term target. You shouldn't assume it's because the end market is growing 2x faster. It's growing faster, but not 2x faster. I think that's the way to think about it.

Got it. Right. No, that's a good point. Because I imagine that -- the good thing is you're not exposed to the consumer, and that's where a lot of the headwinds are. So you have taken care of a big issue just by that strategy alone. But then you still have parts of the business, right, that are exposed to cyclical end markets, whether it's enterprise or automotive, right? So do you think that any one of these markets has the potential to slow down? Or do you think your share gains can kind of power through that next year?

I think the cyclical dynamics in semiconductor is always there, right? I think -- the way we think about it is if you look at the long-term growth trend, we'll always grow above the market, whatever the market. So that's what Ashish said is we do have our unique product cycles, whatever the market growth, the cyclical dynamics of semiconductor, we all know, right? But because our unique product cycle, the content gain, the share gain, will continue to grow above the market. Enterprise networking, in the long run, it's not going to grow as high, right? Right now, there is a modernization of enterprise. There are a lot of upgrade cycle going on. In our Analyst Day, we basically said enterprise networking in the longer term, we actually targeted the low single-digit growth from a market perspective. But for us, we do have all the content gain, share gain and also our unique product cycle, we continue to drive, which will continue to grow much faster than market.

Yes, we're very confident, Vivek, that we'll be able to continue to grow almost despite the end market. Having said that, the reality is all these end markets, we haven't seen any sign yet. People actually want more demand, more product from us every quarter. So our delinquency, how much you can supply, I mean that number actually went up. So while I know from a macro perspective, there's a view that, hey, next year is going to be looking very different. We haven't seen anybody take the foot off the gas right now. And part of it, I think, is because we never came even close to what they wanted us to deliver this year. I think that's part of it as well, right? Why we're not seeing it as much is because we didn't come even close. Back to your earlier comment that maybe the supply contains are acting as a governor on unsustainable growth, yes, I think that certainly applies to us. I mean we just came -- I mean, enterprise networking, but probably the biggest hole, we're the most behind because we just can't come even close to what they actually want. So even if the end market ends up moderating, let's say, some time out of the future, then you just feather into that lower growth rate, but we don't see this kind of big step off, at least not for now.

Got it. Because especially in enterprise, right? Again, the outside view is that really large networking OEM just warned, took down their growth rate, right? They have been -- they are an important customer for you. Then we see all the headlines about layoffs or issues at major tech companies and so forth. And I think investors are justifiably concerned that at some point, this has got to hit enterprise demand. And that semiconductor companies because of these supply issues, will just see it later, but they will see it, but it's just going to be later. So how do you ensure that the quality of that demand that you've really kind of battle-tested it to make sure it's going to be there?

Yes. As I said earlier, in today's environment, the relationship between customers and the semiconductor company and the suppliers are very tight in the sense is you really -- because it's allocation, because each customers, you end up engage, not only at the CEO level, but also every level to really try to understand. I think our team is very mindful, right? We have a lot of varices in the semiconductor executives in our company. They are very aware of the history. So they are very mindful, very disciplined from just working with the customer, understand their demand. And then we can decide where to allocate our capacity, right? If you look at our revenue performance, it's supply-constrained. So each quarter, when we get into the quarter, like Ashish said, we know last Q4, we need a wireless ramp significantly. So we have to allocate capacity to wireless side. And then each quarter, we literally try to assess, make sure, right, is we have customers, they may have a line-down situation. And we're not going to allow somebody to build the inventory when other customers, major customers have a line-down situation. So that discipline certainly today, I do think the team is really try to understand and the market demand. The good news also is for us is our product, a lot of them actually, you can literally see what's the 5G launch and what our customers' market share in the market. And the same thing in the cloud data center, you can see attach rate. You can see how the product to the data-centric architecture, how they're rolling out their 200-gig, 400-gig, 800-gig. You actually have a lot of visibility there. So I think that's a good side of our business we -- frankly, we have a very small portion of our channel business. Even our channel business it tend to be demand of fulfillment, right? So we do have a good understanding of them.

You're not relying on replenishing that channel, too, right?

No, no. Not at all, right? The channel largely serve as demand fulfillment for some of Asian OEMs and so.

Got it. Jean, how do you think about free cash flow generation and conversion? So I understand the company is in a growth phase and growth phase is not where you try and optimize free cash flow generation. But at what point do you think your ability to generate 35%, 40% EBIT margin should be translating in an ability to generate 20%, 25% plus free cash flow margins.

Yes. I think in a normalized environment, you are very familiar with the business model, right? For us, our target operating margin is 38% to 40%. And we have a relatively low tax rate and we have a little bit interest expense. So typically, our operating income to net income conversion, it's probably 5 percentage points between different other things. So when you think about that, the 35% of net income margin, okay, what's the free cash flow conversion? For a company like us, if we're growing 40% like last -- this Q2 or last Q1, it's challenging now to build working capital. Right now, we do see the working capital investment. And in Q1, we also paid out like $85 million for capacity guarantee, right? So those are really what caused the low free cash flow generation in Q1. And going forward, once it's normalized, I do think the revenue growth and the incremental working capital to support the revenue growth probably will have some leakage from net income to free cash flow conversion. But I think we should be able, at the 85% to 90% free cash flow conversion to net income, right?

Got it. So 25% plus free cash flow margin to whatever.

Right, 25% to 30%.

30%. Right, or -- got it. absolutely. When do you think that happens? That happens when growth slows down? Like do you have -- is that...

I think once it's normalized, right? Right now, the supply chain environment is really challenging. So the days of inventory, as you can see, we get to like 140 days, right? So that definitely -- I don't think that's normalized inventory days will keep. So I do think going forward, frankly, our Q2 free cash flow should be very good because seasonally, you really built on the Q1 that you started to maximize the free cash flow.

Yes. Got it. Maybe let me pause there and see if there are any questions from the audience. Please just raise your hand if you have a question. All right. No worries. I can keep on going. So pricing as a dynamic, Jean, how long does that persist in the industry. Is that a cyclical thing? Is that -- there is kind of this tight supply, right, it's kind of almost like a [ flock of birds ] . I know it's not the right phrase to use at this time. But there's a state of confusion right now. So maybe customers are just willing to pay up a lot for supply assurance. But as that supply situation normalizes, do you think that removes a pricing dynamic from the equation?

Yes. I think today, the input cost increase, definitely, that's what caused the -- semiconductor company have to increase the price to just offset the input cost increase. I think the inflation environment, if the inflation environment is coming down, right, I think definitely will help the whole industry. Right now, it's purely inflation, the input cost increase is probably because of supply constrained. So we do see that semiconductor companies to offset the input cost increase. It's really -- it's dynamic. I think the industry has to adapt to this new environment. We'll see for the supply chain constraint can ease off or not.

I imagine -- so if you look at the kind of contractual arrangements you're planning for next year, do you think '23 is another year where the pricing dynamic works in the favor of the semiconductor industry? Or are you already starting to see signs of normalization as you look out to '23? Or do you think it's maybe '24 or '25 maybe before we get there?

I think from our angle, we definitely don't see '23 -- calendar year '23 will continue to be supply-constrained, definitely.

So that -- okay. So that dynamic is maintained, right?

Yes. From our angle, I don't know about others.

Yes. Okay. And then on OpEx intensity, I think Marvell has done a good job of constraining OpEx, working through with these NRE-type projects. Is that a big part of your OpEx, these NREs? And can they grow in line with sales? So we don't get to a point where suddenly NREs disappear and you have this big OpEx gap to fill.

Actually, it has been growing. I think the most important thing is all those NRE comes with design wins. The pace and the momentum of our design wins have been tremendous. We talked about it during our earnings call, we continue to win more and more designs. It's actually quite exciting. It requires execution, but the NRE has been growing significantly each year because the design win has been growing significantly. I think this cloud-customized concept or customization concept actually go through all our end markets. And what you are seeing is all our customers, they want to differentiate their solutions from their competitors. They want to customize their solution. And Marvell is best positioned, right? Not only we have the IP portfolio, we also have the ASIC team who knows the business model, how to do this kind of thing. So we tend to get more share of the design wins if it's customized which typically comes to be the customer funding. That's why I do think we'll continue to get more and more NRE, which help us to leverage more of our R&D investments.

All right. One other thing I forgot to ask about pricing, but I'll ask now. You see the leading-edge foundries still raising prices because we see it in the headlines that this or the other leading-edge, I won't name, right, is still continuing to raise prices even going out to next year. You think foundry price inflation is with us even for another year?

Definitely, for us, we're supply-constrained. The foundry inflation definitely is going to be there next year.

Like 5%, 10%. 10%, 20%, 30%, 40%, any range?

I don't have a...

Last one there...

Yes, yes. I don't have that kind of detailed comment.

Okay. But you still see them raising, right? And I asked that because I think you've done a good job in migrating customers to at least a next-gen road map, right? With 5-nanometer. So how much of the 5-nanometer investment is behind you, when do you start benefiting from those projects?

Yes. I mean we've got a number of tape-outs already completed. So we'll have 5-nanometer revenue coming in this year, and it starts to ramp up. Absolutely, yes. yes. And then it becomes a much bigger number next year. I think back to your question, Vivek, I think it's tied to more general inflation, right? I think part of the reason why people are expecting that expense next year in general to be higher. Forget even the supply constraint for a second is just because inflation is higher fundamentally, which applies to every part of the economy. So it's not a big shocker right, that you're expecting prices to increase. I think it's unlikely inflation goes from the high single digits today, right, at the worldwide level, back down to the low single digits. It's going to take some time. So if you were to say anticipating this period of higher inflation and just factoring it in, I think that's more to do with it. It's probably less to do with the fact that there's still, at some point, constrains resolve themselves, right? I think that happens, just not near term. But yes, on 5-nanometer, we do expect to start seeing production revenue this year to ramp much harder next year.

All right. Very good. Thank you so much, Jean. Thank you, Ashish, for sharing your perspectives. Thank you for joining.

Thank you.

Thank you.