adminIntel Corporation (NASDAQ:INTC) Morgan Stanley Technology, Media & Telecom Conference March 6, 2024 1:15 PM ET
Company Participants
David Zinsner – Chief Financial Officer
Conference Call Participants
Joe Moore – Morgan Stanley
Joe Moore
Great, welcome back. I’m Joe Moore, Morgan Stanley Semiconductor Research. Very happy to have with us today the CFO of Intel, David Zinsner.
So quickly I want to read this edge, you guys have heard before. For important disclosures, please see Morgan Stanley Research Disclosure website at www.morgansanley.com plus research disclosures. If you have any questions, please reach out to your Morgan Stanley sales rep. That’s — I think this is my 19th fireside, so I’ve heard that maybe 19 times.
David Zinsner
All blurring together.
Question-and-Answer Session
Q – Joe Moore
Anyway, so maybe if we could just start, you know, you’ve been, you and Pat together for, you know, almost three years now. Can you kind of talk about Intel’s priorities? How you’re doing? Where you are? And we’ll go into detail on all of it. But…
David Zinsner
Yes, sounds good. You know, I actually probably should like say the disclosures too for Intel. You know, I make forward-looking statements, consult the risk factors in the 10-Q and 10-K, and I’ll mostly be talking about non-GAAP numbers. So there’s a reconciliation to GAAP on the investor relations website.
Yes, so actually, I’ve only been there for two years. Pat’s been there three, I came in a year after Pat, basically. And I think it’s — transformations are hard, obviously. They take a while, and there are always some unexpected twists and turns in the process. But I think — when I think about Intel, I think there are really three things we’re trying to do. Number one is, you know, get the core business into a good place. The second is to stand up the foundry business. And the third is to attack, you know, what is a kind of exciting and growing market in AI.
I think on the core side of the business, things are going extremely well. Most of that was around getting five nodes in four years, you know, we’re largely through Intel 4 now. Intel 3 is not too far away now at this point with Sierra Forest, which will qualify in the first-half of this year, and then we’ve got 20A and 18A coming and both of those look pretty good, so I think, you know, we’re in a good place on the process side and the product side is starting to, you know, execute well. I think the teams are starting to execute on the products. They’re beating their milestones in most cases as opposed to missing them significantly, which they had been doing in the past. So I think that part of it is going well.
And you kind of see that showing up and you know the market share data looks good on the client side in particular. You know, we’re still kind of working our way through on the data center side, but that has you know stabilized as of the fourth quarter and I think we’ll largely see that stable through this year. And then as newer products come out, I think we can make improvements on that. So core business, get that into a good place. I think we’re there at this point.
The second piece is standing up the foundry. That one is going to take a while, obviously. But I think a lot of the pieces of that puzzle are now starting to get assembled. You know, the process, five nodes in four years, aspect of this, like I said, is largely on track. Pat, I think, two weeks ago, whenever we had the Foundry Day, he mentioned that we have $15 billion of lifetime deal value. So deals are starting to get inked here. I think we have most of the top 20 customers on the foundry side running test chips with us. And so we look like we’re in a good place as it relates to getting a pipeline. And then we’re also making good progress on the packaging side.
So I think that part of the business is going well. It takes longer, obviously. It takes multiple quarters to get packaging wins, and that show up in revenue. It takes multiple years for you to get wait for wins, and that show up in revenue. So there’s a time lag. But all the things that we wanted to see in terms of milestones, I think we’ve largely executed upon.
The one that’s probably in the more nascent space is the AI part of the story. We have the Gaudi products that address the data center space. And the pipeline is increasing, but we’ve got to translate that to revenue. And we largely have not yet done that in a significant enough fashion. So more work to go on that side. But we do think that AI has a broader exposure to it, both in terms of the PC space and at the edge. And I think we have good positions in those areas and are probably more ahead relative to competitors in that area.
So I think I give myself or give ourselves, I guess, kind of a B in terms of the transformation. Not everything has gone well, but most of it is going largely according to plan. It’s just these kind of things just take time.
Joe Moore
Yes. You mentioned transformations are hard, and you’ve certainly done a lot here. Maybe if I could just ask about this broader ambition to do IDM and foundry together, not to rehash multiple year-old debates. But to me, you can add value through manufacturing the way you did 10 years ago, which is to make the best-in-class microprocessor transistor, and you made really good product.
I didn’t think at that time you were particularly competitive with foundry because the wafer costs were very high. And then the company sort of tried to address that with a super ambitious 10-nanometer, and we know how that went. So like is the — there’s sort of that level of can I make the best microprocessors? And there’s another level, which is kind of compete with the best foundries in the world. Is there a risk that trying to do the second one kind of makes the first one harder?
David Zinsner
Yes. Well, I think there’s a confluence of things that I think drive us to want to be in the foundry business. Number one is it’s a lot more expensive to build a fab, an EUV fab, than it was to do a pre-EUV fab. And it requires a fair amount of wafer volume to get your return on one of those fabs. And I think it’s unlikely that we would have enough volume ourselves to be able to fill the fabs to the level that drives enough of an ROI. So part of it is just necessity. We need more volume, and that’s going to come — have to come from other customers. I think that’s number one.
Number two is things are coming at us in a way that’s actually fairly advantageous. The need now and the recognition of how important leading-edge process technology is to the world, I think, has been recognized now over the last couple of years. And so governments are now looking for opportunities to see production done outside of Southeast Asia. And so that also helps because now the other factor that kind of kept — inhibited you from being able to do it was it was fairly expensive to do one of these fabs in the U.S. or in Europe.
Now with the help and the assistance of governments, that can happen. We have the ability to drive a cost structure that would make sense. I think also when you look at our position now, we’re now in a space where we’re going to be competitive on a leading-edge process technology.
But having a multiple-year lead is probably not something we can sustain. And I think when you make that transition, you need to suddenly be thinking about, okay, you can’t just bring out the best transistor at any cost, and it all makes sense for us. Now we do actually have to drive real efficiency. And so part of the benefits of being in the foundry space is you get a lot of feedback from your customers that helps you make, yes, better transistors, but also better transistors at very effective cost.
The one thing I would just say is — and we get this question a lot is, okay, well, the profitability and the cost structure of the leading competitor in this space is significant. How do you get there? And I think it’s probably the wrong way to think about it because I don’t think we are looking to try to get to that level. We can be profitable, meaningfully profitable in the foundry space, be well underneath what the leading player in the space is, and still drive significant profitability for the overall Intel company because we get the margin stacking benefit in at least the part of the business that we sell into our own fabless portion of our business. So that’s really the way I kind of think about it.
And I think the last piece of this is kind of restructuring how we manage the business. So we look at the entire manufacturing and TD footprint as a separate P&L and kind of manage the company accordingly to that. And I think we’ll start to see a lot of the efficiencies that we think we can yield and get ourselves more competitive from a cost structure from managing the business in that way.
Joe Moore
And we’re I guess a month away from seeing what those numbers look like once you start to break out the profitability of the foundry organization separately. But it seems like the expectation is that you’re sort of a little subscale in that part of the business. And so the earnings almost kind of shift to the product groups a little bit because you’ll have foundry struggling a little bit profitability-wise, and you’ll have the — is that kind of the right way to…
David Zinsner
Yes. So one thing, on April 2, we’re going to do a webinar. Pat and I are going to do a webinar. We’re going to issue an 8-K that gives historical perspective on how these businesses look in the new way we’ll segment the business. And there’s a lot there to take, I think. So for investors, we thought it would be a good idea to kind of walk everyone through it and show what the opportunity is. So we’ll do that on April 2. And it will be me and Pat. And we’ll do some sort of webinar and then do some Q&A. And you probably have an opportunity to ask questions. I’m sure you’ll have it.
It’s the right way to look at it. I would tell you, the foundry — if our margins are in the kind of mid-40s space, call it roughly, and product, call it the fabless part of our business, the products are kind of north of 50, and we get margin stacking. Then it kind of tells you that the foundry part of the business is in a pretty challenged state in terms of profitability.
And partly, that’s scale. Partly, that’s — we have a big, relatively decent-sized footprint that’s more than we were set up to need. So we’re kind of underloaded in a sense, and that’s hurting us a bit. And — but a lot of it is just we have older wafers that don’t generate the revenue or the profitability that as we transition to leading-edge wafers that have good cost structure and also command better pricing, you’ll see a pretty big lift in terms of the profitability of the business. But the one other point, which I think kind of you’re making, the product business actually has good profitability.
Joe Moore
Yes.
David Zinsner
It is a good business.
Joe Moore
Better than we see at the overall EPS line, yes.
David Zinsner
And better than you see now because we take a whole bunch of costs from the factories and we just — right now, the way we’re allocating it is just spreading it through those facilities. So they’re kind of artificially lower than they really are. Once you kind of say, okay, look at them as a fabless company that’s buying wafers from a factory, in this case, the Intel factory, then you notice that, hey, this is the problem, it’s more closely aligned to what you see in a fabless company.
Joe Moore
So in a way, the worse IFS is, the better, except that it says there’s a lot of movement here…
David Zinsner
Exactly. But I would say, if you can’t — the glass half full kind of way to look at the foundry side is because you look at it and you see the challenge of the financials and you know how — the kind of position they’re in, in terms of the age of the technologies that are getting shipped in terms of wafers, the fact that the cost structure has never been a big focus of that part of the business. The opportunity to improve it is pretty significant. And if you just move that to something that is reasonable, I’m not talking about like leading-edge players’ profitability, which is a reasonable profitability, it’s a significant tailwind to the earnings of the company.
Joe Moore
Great. That makes sense. So maybe talk about five nodes in four years, which was, again, you were pretty far behind when that started. And so you gave us multiple milestones to assess. A lot of enthusiasm from you guys on the progress of that. And a lot of the productization is still to come, right? Because you’re sort of 2.5 years into that. And we’ll have seven and four that are out in products. So been a quick succession, we get three, 20A and then 18A. Can you just talk about where that confidence is coming from at how good the 18A process will be?
David Zinsner
Yes. I think when Pat — when I joined, was it, January of ’22, Pat mentioned this five nodes in four years. And I had just come from Micron, who had done like a node a year, and that was like breakneck speed. I thought five nodes in four years, oh my God, that is fast. I almost thought we will never get there, but it’s a good kind of aspirational goal.
But to the credit of the team, I think we’ve done a very good job executing on these processes. Partly, they got the benefit of because the leaders are out in front a bit, it allows — catching up is a little easier than blazing the trail. And so that has helped, I think, a bit in terms of getting us to where we are.
And then secondly, yes, we missed the EUV thing, and that set us back. But largely, this is a good — really good team, the TD team. So once we got readjusted to fixing that, their ability to execute is actually quite, quite good. Yes, as you mentioned, Intel 4 is now out with Meteor Lake, and that’s already ramping and shipping. So we’re — we can check the box really on that one. And Sierra Forest will qual by the end of the first half of this year. So — and that’s the first Intel 3 products that pretty much check the box on that one.
So what we have left is 20A and 18A, which are kind of like cousin nodes, I call it. So — and what’s been good about 18A, which has been different probably than any other process Intel probably ever done, is they’re — we’re getting a ton of feedback from customers outside of our own design teams on the node and helping us kind of improve upon it. So I think all things look really solid. They’ve hit or beaten every milestone that we put forward on 18A. So I suspect Clearwater Forest will be our first product. And I think we’ll meet or beat the time frame for that one. And then we’re on to the more leading-edge nodes that — and unveiled at Direct Connect the other day.
Joe Moore
Yes. And it’s also become increasingly evident that this backside power is a big deal, and it’s something that people want. Can you talk about what that may look like from a foundry standpoint in the sense of — we’ve talked to some of your big potential customers. And when we saw — for example, we saw a lot of this when people moved to Samsung. It’s that we give something small, just make sure they can execute on something. And then they give them something bigger in the next wave. Is that how you see this? In other words, people will give you something…
David Zinsner
Like customers will get a small volume?
Joe Moore
Yes, it can be a small project. And from that concept, to a big project.
David Zinsner
We probably won’t win anybody’s major volume in 18A. We’ll win some smaller SKUs, and that’s all we need, to be honest with you. That will be very significant to us, even though it seems maybe marginal in the marketplace, particularly if we can collect enough of these customers. Backside power, this combination of gate-all-around, what we call RibbonFET and the backside power, which we call PowerVia, both of those will be part of the 18A solution, which really makes it, I think, a really compelling offering. And what we’ve seen a lot is that — what does that mean? It’s not working?
Is this working? Yes. Okay. What we’ve seen a lot is that when it comes to like the high-performance compute part of the market, that’s really where we’re starting to see a lot of our uptake. The particular aspects of 18A with PowerVia and RibbonFET, combined with our just legacy of experience on high-performance compute, I think, makes us a really compelling partner for customers that are in that space and want to develop products.
Joe Moore
Great. Maybe we could shift and talk a little bit about the nearer term 2024. In Q1, you described kind of a seasonal environment for your core business with some idiosyncratic factors. Is that sort of how you see the year playing out? Is there a seasonality in your core business, and those idiosyncratic factors kind of aren’t an issue beyond Q1? We heard from Mobileye this morning, pretty confident in the rebound in Q2. So how about that?
David Zinsner
Yes. So maybe unpack it, I’ll just like remind everybody what we said on the call. So essentially, the real kind of core, core part of our business in CPUs is largely going to be seasonal in terms of its transition from the fourth quarter to the first quarter. But we’re sub-seasonal when you look at us in terms of our guide because kind of these adjacent markets, Mobileye, the foundry business, kind of transitioning away from traditional packaging into advanced packaging. There’s a little bit of an air pocket there. Altera business, kind of later to see the inventory correction, and so that impacts that business.
So there’s kind of like this kind of air pocket trough that’s hitting us in the first quarter. And some of those businesses may take longer, and they may kind of stretch into the second quarter before they really start to see a rebound. We’ve said that we would be up — when we talked on the earnings call, we’d be up sequentially, and we’d be up year-over-year in the second quarter. But still, it’s not off to the races.
I think in the back half of the year, we have a really good — we’re set up really well because, one, all of that inventory stuff we know only have — it takes a couple of quarters to work through. And so you start to see a recovery in the back half of the year from those adjacent markets that are having the inventory correction. But on top of that, our core businesses, I think, have some really good tailwinds to them that really probably start hitting in the back half of the year. And you kind of hear it from customers who have talked in earnings and so forth.
On the client side, we do suspect that the AIPC will be — will drive momentum as we progress through the year. And I think, in particular, it’s going to be more of a back half play. We know Windows 10 is going into service. There will be some refresh activity that’s likely to be part of the second-half. So client looks like it’s going to have actually — yes, it will have the seasonal benefits it normally has, but it’s also going to have a little bit of a cyclical benefit because of those things happening. And then on the data center side, as we know and I think others have talked about as well, the units may be relatively flattish on a year-over-year basis. But the core count is growing significantly.
And the ASPs for core are more stable than they’ve been. They have been like declining at a rate that — where you haven’t seen quite the growth rate, but we suspect that, that’s going to change. And so we will have core count growth through the year that you’ll start to see momentum. And then while like I said, we’re not where we want to be yet on the AI front, AI will be a contributing factor to growth as we progress through the year. And so that will be another tailwind.
So all that said, I think first half, yes, be better than the first half of last year. But the second half of the year feels like it’s going to be really strong sequentially on a half-over-half basis. And of course, we did say on the earnings call that we thought it would be up year-over-year as well. So I think we’re set up well for a very strong back half.
Joe Moore
Yes. Okay. That’s very helpful. Maybe just a couple of questions on that. On the data center side, last year, my perception is that AI crowded out a lot of the discretionary spend. So obviously, we bought servers when we need servers, but a lot of the sort of 3-year replacement cycles got stretched and things like that. Do you see that changing over the course of this year?
David Zinsner
Yes. I think there’s an inevitability that you have to like start doing some refreshing in the CPU side. In addition, AI needs CPUs as well on the head nodes. So I think that the likelihood is we’ll see a lift from AI even in the CPU space as we progress through the year. And then on top of that, I do suspect that we’ll see strength. And some of it is the strength that you’re talking about, your wallet shifting around a little bit. But a lot of it is just — they’re just bringing out more advanced higher-core count solutions, and that’s going to be a strength for us.
Joe Moore
And then market share in that data center segment, your unquestioned leadership is still a year away, but you’re getting closer on the product side. Can you talk about how you see market share there?
David Zinsner
Yes. I think we have stabilized. Quarter-to-quarter, it might be — have a little bit of volatility. But I think in general, we’re kind of bouncing around at this level for market share until we get our products that are more competitive out. Sierra comes out, like I said, before the first half of this year is over. That’s an important one because it’s our first kind of efficient core product and addresses a place in the market that we were more challenged in terms of what customers are looking for.
And then Granite will come shortly after that, and that is a really great product for us, high-performance product. And then we’ll have a — like I said, the follow-on first 18A product, Clearwater Forest coming out next year. So we’ve got this kind of steady drum beat of improving product portfolio that I think puts us in an increasingly better position competitively.
Joe Moore
Great. That’s helpful. And then on the AI side, you said you’re not where you want to be. I mean you have talked about, at one point, $1 billion of backlog, and then I think that doubled. Is that still the ballpark of what you’re kind of thinking about doing? And when you say you’re not where you want to be, what gets you where you want to be?
David Zinsner
Yes. I mean, first of all, it’s pipeline that we talked about. I think we said at the earnings call, it was like a $2 billion-plus pipeline. You got to then translate that into revenue, and some of the pipeline falls off when you kind of get to the revenue. So we need to build that pipeline up more significant than it is today. I think it’s, in some ways, a matter of just kind of grinding it out a little bit. Gaudi is a really good product in terms of performance. And there’s been a lot of papers and analysis around that, that shows that Gaudi is a really good product when it is matched up against other hardware.
What we need to do is mature the software stack. And that’s what we’re in the process of doing and helping customers stand up the solution. So I think it’s just time, just continually building the pipeline. And I think AI will be an important contributor to the business. In addition, as things move from really big model training to smaller models, particularly as that moves to the on-prem because some of these models are not going to the cloud because the data is proprietary, and they don’t want to move into the cloud, then we’re more in a sweet spot for us, I think. That’s where I think Gaudi and even our CPUs that address AI are better solutions in that space.
And we have an ecosystem around that space that positions us better, I think. And so as things — kind of buying behavior kind of transfers away from larger models, I think we have a real opportunity to drive a better position in the space. And then lastly, I would just say we’re also coming at these competitors as potential customers on the foundry side. And that’s another opportunity for us to ride the AI wave is to be the preferred foundry supplier to customers making AI chips.
Joe Moore
And there does seem to be a lot of AI opportunity to the traditional CPU business. I mean you talked about at one point, most of the inference was happening on CPUs. And that’s changing at the margin because some of the complexity of transformers is moving to specialty silicon, but there’s still an awful lot, I think, that you can do to accelerate inference in those markets.
David Zinsner
Yes, for sure. And like I said, it’s — I think when I even look at Intel, we’re just now starting to look at all the data we have and starting to think about how we can kind of take advantage of that to do better in manufacturing, to do better in terms of our decision-making, to do better or have more of our programming automated through AI. All of that stuff is in kind of its early stages.
And I imagine that’s true across most of corporations that are looking to take advantage of AI. And so we’ll start to see more and more products rolled out that are either in the data center or even at the PC level. And that’s, I think, where we have a tremendous opportunity.
Joe Moore
Great. So I did want to focus a little bit also on capital spending and cash generation. First, on the CapEx side, can you remind us kind of how — I know you have a way of thinking about capital intensity minus offsets. Can you kind of walk us through some of that? And then there was a change to Ohio that I guess the time frame on Ohio got maybe pushed back. Does that have any effect on your capital intensity?
David Zinsner
Yes. So largely, what — in the period of kind of catch up, we think that the net CapEx intensity of the business should be roughly in the mid-30s as a percent of revenue. We were higher in ’23. We’ll be lower in ’24. The blend of that will be kind of in the mid-30s, and it’s just a function of — it’s hard to call the timing on these offsets. They come in kind of lumpy fashions. And so we had less of it show up in ’23. We’re going to have a lot more of it show up in ’24.
So — but I think still, the thesis was right, mid-30s as a percent of revenue as you’re starting to catch up. And you got to remember, in Intel’s case, we were not — we did not have like a shell ahead strategy, which is tough because if you don’t have a shell ahead strategy and you suddenly decide you need a fab, you now have 4 years to get a fab built.
So what we want to do is be ahead of that. So we had to, A, build up the fabs that we needed, but also get shells built ahead of that so that we could equip it as needed. It will then probably settle in the kind of mid-20s as a percent of revenue on net CapEx intensity. Clearly, the ’23, ’24 time frame, I think we’re mid-30s. We haven’t quite built out the plan for ’25 yet. I’m not sure whether it’s still in the catch-up mode or it’s roughly down into the mid-20s. But within a year or 2, I think we’ve settled in the mid-20s as a percent of revenue.
We’re going to do really well on the offset side for sure, which just gives us an opportunity to put a few more dollars at work just to invest in the factories because we thought we’d be in the kind of 20% to 30% offset range, which generally, when I give a range, it means I think it’s the mid.
But in this case, the way things are tracking, the way things are playing out from a chips perspective, both in the U.S. and Europe, what we think we will get in terms of AMIC or the investment tax credit, what we will do in terms of leveraging partners like Brookfield, we think we’re actually at the higher end of that range at least over the next few years. So that means roughly, we’re kind of approaching 30% offsets, which I think is a tremendous opportunity for us.
Joe Moore
Yes. And you just mentioned the parts that we know about. There’s also stuff that we don’t quite know about yet. I mean Gina Raimondo was at your Direct Connect event, was very positive on the government’s relationship with Intel and the sort of focus that they have. So it seems like the grant money could — certainly, Intel should be a focus of where the grant money is delivered.
David Zinsner
Yes. I think we are — I think I want to say she said that we’re the national champion. So I think that, that’s a good adjective to describe us. We’re definitely committed to fulfilling the objectives of what the U.S. government wants for domesticating advanced semiconductor manufacturing. So all of that’s going, I think, well. We still have to get kind of settled on the CHIPS grant. I think we have a good sense now on what the investment tax credit will yield.
And so I think relatively shortly, you’ll be hearing an announcement from — either from us or from the CPO office on that grant and what that looks like. Things have gone — I think the partnership has gone extremely well with the government in terms of having this established in a way that fulfills everyone’s objectives.
Joe Moore
Yes. Yes, they’re actually here tomorrow, Todd Fisher, who runs the…
David Zinsner
Good.
Joe Moore
So maybe you could talk about cash generation for the year. I think you talked about a free cash flow-neutral kind of year. That’s not entirely intuitive to me, given everything else that you’ve said, but the cash of the CapEx is kind of tricky to figure out too because of the offset timing. And just kind of talk to us about how you think about free cash flow in 2024.
David Zinsner
Yes. Well, keep in mind, I’m saying this kind of mid-30s as a percent of revenue over the course of ’23 and ’24.
Joe Moore
Right. So it’s a little lower.
David Zinsner
’23 was much more elevated, so ’24 is going to be much lower. And so I think if you do the math, you’ll kind of — you’ll get there probably. And you’re right. Our goal is to be kind of roughly cash flow neutral for this year. And I would say this was kind of the year that we were roughly targeting to be cash flow neutral even back at Investor Day. And not surprisingly, this was not the revenue number we had predicted would happen.
Joe Moore
The PC market — yes, smaller.
David Zinsner
And so I think it does show that the company does a fairly good job of kind of taking the inputs of what’s going on with the business and then controlling the things that we can control to make sure that we fulfill that commitment. After that, the goal is to ultimately, obviously, to get cash flow to 20% of revenue. So our goal throughout the next several years is to kind of chip away at getting to that goal. Some of it is, obviously, how we manage CapEx, and CapEx can be modulated. We can still make the investments necessary to get to leading edge but modulate the capacity in a way that adjust the CapEx but still fulfills our objectives in terms of getting the leading-edge technology out. So that’s part of it.
Obviously, the offsets, we do better on the offsets. That also helps. But there are other things we’re driving. We’re driving working capital improvements. I think last year, we think we did at least $2 billion of improvement on cash flow just on working capital improvements. So cash flow for operations did better, I think than we expected, I think than what most people expected. Anything we can do in terms of being more efficient, obviously, helps as well. I think we cut $3 billion of spending out last year. We’re going to chip away at trying to drive efficiencies this year.
Ultimately, we want to be pulling out another $5 billion to $7 billion of spending efficiencies to help improve the cash flow as well. So I think investors should just kind of take away from this, it’s a huge focus for us. We know it influences the value of the company, the valuation. But also, it’s important for us in terms of our funding objectives. So we’re — I’ve got everybody in the company super focused on this as a metric to drive.
Joe Moore
Great. Well, you’ve made a lot of progress for sure. Let me pause there and see if we have questions from the audience. Here, Charlie?
Unidentified Analyst
Thanks, Joe. Hi, Dave. Thank you for your sharing. So I’m Joseph’s colleague, cover Asia semi. So I have a question about your foundry outsourcing. So you seem to show some interest to your outsourcing to TSMC, but TSMC also supply to your competitors, right, like AMD, business CPUs, et cetera. So what kind of commitments you can give to TSMC for this foundry vendor to build capacity for you? And also kind of a higher-level question, how do we think about the outsourcing strategy versus what you just mentioned, right, domestic manufacturing for the U.S.?
David Zinsner
Yes. So we’re a big customer of TSMC. Obviously, I think that’s not surprising, and we’ll continue to be a customer of TSMC. That’s clearly the goal. I think probably, we are a little bit heavier than we want to be in terms of external wafer manufacturing versus internal, but we’re always going to use external foundries for wafers. In fact, we have this thing that we rolled out we call Smart Capital. Pat’s good at like catchy names, this is one of them, using finance and creating a catchy name a lot. But anyway, one of the big pillars of Smart Capital is to use foundries that we can’t build out the capacity in total for everything we need.
We need to make sure that others are doing it. We’ve got a great partnership with TSMC. Yes, we’re a competitor, and we’ll fight that out. I’m not sure we’re a major competitor to theirs. We’re kind of the second source or the one that can supply in markets outside of Taiwan. And that’s really kind of our goal as it relates to TSMC.
But we’re a partner of theirs. We are a customer of theirs. We actually sell stuff to them from our IMS business, our mask writing business. So it’s a great relationship. I know Pat and C.C. meet on a regular basis. They have a great relationship. So I’m not particularly worried about it. What was the second part of the question? I forget now. Yes, sorry. Whenever you ask a 2-part question to me, it’s always a failure because I can never remember the second one.
Unidentified Analyst
Yes. So second part was actually about how do you reconcile your outsourcing versus you’re getting support from the U.S. government, right, for domestic manufacturing? But I think you kind of covered that.
David Zinsner
Yes, I think I…
Unidentified Analyst
Yes. So more specific is about your commitments to TSMC because TSMC, at their earning call, they are sort of concerned about your kind of internal kind of technology development as well. But taking this chance, I actually have another small question about your Habana business. Question is the shipment to China, right? Because the performance seems to be great. How about the performance density? Does that comply with the U.S. export control? Just want to see whether there’s a kind of upside from your China business.
David Zinsner
Yes. I mean, obviously, China is an important market. It’s an important market for us broadly across all of our solutions, so we focus a lot on it. Obviously, there are certain restrictions that we have that BIS requires from us. Some of it’s related to technologies like that. Some of it is more broadly related to certain customers.
And we — obviously, we comply with all of those. And — but yet, we still need to make sure that we keep that market for us because it generates at least one-fourth of our revenue. And we’ll continue to focus on making sure we bring out products that obviously are compliant with what export licenses dictate, but also meet what customers in China need from Intel.
Joe Moore
Question over here.
Unidentified Analyst
Just wanted to ask a quick question about the Intel foundry service. And really as it relates to ARM v9 and your potential manufacturing of that, so ARM pushing into compute subsystems quite aggressively now. And I just wondered if you reserve the right of refusal to make a chipset on behalf of a customer that’s using v9 that would go directly against some of your own silicon?
David Zinsner
The way we’re going to run — without directly talking about it because some of that stuff is confidential, I would just say the way we’re going to run the company is foundry needs to go out and win business. And ARM, being an important relationship, an important ecosystem partner in the foundry space, will be a significant partner for us. And they will do whatever is necessary to meet whatever customers need from them. That is counter to the product business in some respects. But the product business is getting a different order, which is go out there and win every compute socket you can possibly win. And in some ways, it’s a conflict, but that’s the way we’re going to run the company.
Joe Moore
I think there’s time for one more question in the front.
Unidentified Analyst
How quickly can you ramp up in the foundry services your trailing edge or non-bleeding edge business to sort of start getting cash flow from these unutilized facilities?
David Zinsner
Yes. I mean we — so Intel 16, we do have customers on — a customer on that. We also have a customer — maybe multiple customers on Intel 3. So we are doing that. It does take time, too. It is like a year or two, in some cases, to ramp those customers. It also takes a little bit of time because those particular processes weren’t developed to be foundry processes, whereas 18A, I think we really came at it with that mindset. So we have to do a little bit of like lifting with customers that takes a little bit longer.
So it’s — onboarding a customer takes a bit of time, which kind of slows down the process. But it’s absolutely part of the approach. We also — in order to kind of accelerate it a little bit further, we’ve also partnered with certain legacy foundry customers to help accelerate it. We did that with Tower. We’re doing that also with UMC. And those opportunities also, I think, we can move more quickly in terms of generating revenue and cash flow to help drive a better financial profile and obviously be able to invest in the leading edge.
Joe Moore
Great.
David Zinsner
Thank you.
Joe Moore
Well, that takes us to the end of our time. Thanks, Dave. Thanks very much. Appreciate it.
Read the full article here
