Intel Corporation (NASDAQ:INTC) UBS 50th Global TMT Conference December 5, 2022 10:50 AM ET
Company Participants
David Zinsner – Executive VP & CFO
Conference Call Participants
Tim Arcuri – UBS
[Call Starts Abruptly]
While it is late relative to the original commit dates. Once we reset the commit dates, we were actually thinking it was probably something towards the end of this year that we would get out. The fact that they were able to pull it in, they debugged a lot more quickly over the course of the last few months versus what’s progressed from before that. So, we are actually in, I think, a pretty good position for Sapphire Rapids.
Now that does not mean we’ll gain share yet. It’s a good product in certain workloads. It’s not a good product in every workload. But I think it sets and establishes the execution that we expect to see when we bring out Emerald Rapids. And more importantly, in ’24, when we bring out a power-efficient product, Sierra Forest, along with Granite Rapids, I think we start to reestablish ourselves. We’re in a strong position on the data center side. We start to see the share erosion start to dissipate and actually turn around in the other direction.
And then — so that’s on the product side on the two major products. We also have a number of product areas that get a little less focused. Our NEX business, our network business, we continue to gain share in that business. That business is doing quite well. Our graphics business, while we’re relatively nascent at this point. We have started to introduce some products on the discrete graphics side, we see some good progress so far.
Obviously, we’ve got a lot of work to do in the product portfolio into a competitive position. But I think I like our progress so far in terms of how we’ve executed. And then lastly, of course, is the foundry business that gets a lot of attention. We’ve had very good progress on customer engagement. This will be, I think, a really proof point to where we think we are from a process technology perspective that we are starting to establish ourselves and get good customer traction on the foundry side.
And so, I think you’ll see us — we announced one customer earlier this year, MediaTek, I think you’ll start to see as we progress into ’23, more customer announcements that show that we have a real strength in the foundry space. Now transitioning to the other part of it, the last thing that I really wanted to execute on in addition to all those thing was the financial side.
We do think — I think we said on the call in October that we felt like we could reestablish ourselves as one of the best in terms of performance of the semiconductor space, and we associated that with a 60% gross margin, 40% operating margin business. Well, we’re not ready to call that yet the model. I think we have a really good line of sight to be able to do that. And our first step in that was to get this $3 billion of spending out in 2023.
We’ve got very good line of sight to the $1 billion reduction in cost. And keep in mind, that’s net of $2 billion of depreciation increase that we’ll see in cost of sales. So, really a $3 billion of cash flow improvement that we’ll get in ’23 on the cost of sales side.
And then $2 billion of OpEx improvement, taking our OpEx roughly from about $22 billion to about $20 billion in ’23, very good line of sight to that. And that’s just the down payment on a larger cost takeout that I think we can do. We said that in total, it should be more like $8 billion to $10 billion. First $3 billion, obviously, we get in ’23.
We have very good line of sight into about half of the remaining. And then I think as we see the results of separating the business into a foundry business and a product business on top of Intel. We’ll see a lot more accountability and transparency in both those organizations that I think will drive significant amount. I think $3 billion will turn out to be a relatively conservative number.
Question-and-Answer Session
Operator
Q – Tim Arcuri
Great. Since you brought up the cost cutting, I want to ask you about that. It’s been a roller coaster. Obviously, you were short capacity, and now you’ve had to cut utilization. And you have a 300 basis point headwind to gross margin from that in Q4.
David Zinsner
Yes.
Tim Arcuri
The question all the time is that you take out $3 billion this year, you take that 8 billion to 10 billion by 2025. How does that impact the process road map? I mean if you — if 2/3 of that $9 billion is cost, that’s basically $6 billion on a variable cost base of $23 million this year. So if I exclude the depreciation, the question I get is that’s a huge portion of your variable cost base. How do we then get convinced that, that doesn’t impact your process aspiration?
David Zinsner
Yes. So from an OpEx perspective, I know yours is more about cost, but let me just talk about it from an OpEx perspective. From an OpEx perspective, the one area I did not touch is TD is technology development. Now that they had a higher aspiration of how much they wanted to grow, but they don’t grow difficultly, but we made no reductions in technology development, and that’s because that has got to go flawlessly.
And we looked at other areas, mostly around overhead, some of the product portfolio. We felt like was relatively ancillary to what we were trying to do and didn’t drive any meaningful growth. And so, we have kind of looked at areas where there was low ROI, low NPV, and we’ve kind of pruned the portfolio. And that will continue, by the way, as we progress into ’23.
On the cost side, I look at it as really efficiency. When you look at how we’re structured now, the product area can with the manufacturing organization around quite a bit. They could do as many hot lots as they want, they can do as many samples as they want. They can do as many stepping as they want. They can change their forecast pretty much every week if they wanted to do that.
And so there is a ton of inefficiency in the fab just by the way it’s been structured and what we’ve optimized to. But by the way, we made a ton of sense when doing a monopoly. It just doesn’t make sense anymore. And so one of I think the things that perhaps isn’t quite well understood about this creating two P&Ls and really managing the P&Ls in a different way is to drive this behavior and make it way more efficient.
The other thing that I feel was the right decision in a prior period in Intel’s existence, but is no longer the right approach is this notion of kind of coming up with process technology, giving it to the fabs, not letting them adjust it in any way, running it for as long as its lifecycle is and then converting over to the next process technology. And I’ve heard from even my compatriots and CFOs in the equipment side is we don’t have uptime to the level that we should.
We’re not getting the output that we should be getting — and that comes from just relentless focus on improvement in manufacturing that just wasn’t something that we did, which now partly as a result of now wanting to be in the foundry space where we extend these nodes a lot longer, we’ll just need to do. And so fab managers will be empowered to go out and drive an eke-out efficiencies.
It’s just we just let go, yield improvements, relentlessly looking at overhead, make sure that we properly allocate the right overhead for what the spend is. So, I think there’s a ton of just as we move progress, we are just going to get more wafers out than we would ordinarily have gotten just because we’re driving more efficiency.
Tim Arcuri
So the inefficiencies really were that high.
David Zinsner
They were, yes. I mean I’ve heard statistics of like uptime and certainties of equipment that were 20% for us that when you look at the best-in-class, it’s 80% uptime, that piece of equipment. So it’s meaningful. Now I understand it. This is just the way — Intel, the ROI just wasn’t there when you’re converting over these nodes so quickly. But now that we plan to extend them hopefully, 10, 15 years, it makes a ton of sense to go drive that efficiency.
Tim Arcuri
Got it. Can we — so you had talked about the separation of the foundry business, the separate P&L. Can you just talk about the sort of — what that actually does for the Company? Is it being done because customers want to have a separate sort of an optically separate foundry business? Is it really just that the product groups are being held more accountable from a cost perspective? I’m sort of impressed to think that there were that many inefficiencies that resulted from the product groups being able to do what surprise.
David Zinsner
I think, Tim, who runs the factory, must complain every other day about something that he’s been with around on. But the number one, originally, as I was thinking about this, first kind of thinking about how to approach this. I was really thinking about it in the form of standard cost because at Intel, do everything on a kind of on an actual cost basis, and there’s a lot of noise around actual costs, what you think your cost would be versus what it turns out be in month one versus month seven versus month 10, can be all over the map.
And it’s very hard for the business, a plan around pricing and volume and loading and all those things, if they don’t have a good sense of what their pricing curve is going to look or the cost curve is going to look like. So, I was originally thinking about it in the concept of that, hey, we need a more formalized senior cost approach to how we do it, which is basically how almost every other semiconductor operates, a semiconductor operates.
And then, as I was kind of unpacking all of it, I realized that we needed more transparency between the fab organization or call the foundry organization, which is both supporting internal customers and supporting external customer versus the product company business because of these inefficiencies that I saw, but more importantly, because — when you look at our margins and when you look at the margins of somebody that uses a foundry and then sells the product, the margin stacking just does not mathematically make sense.
There’s no way that — I mean clearly, there could be slight differences between different companies in terms of their scale and so forth, but not to this magnitude. That made no sense. And so, when you look at what we call TMG, which is the technology organization manufacturing and foundry together, as one and look at the P&L. And we haven’t finished it. Our hope is that we can provide something to investors by 2024.
But in the just spreadsheet kind of exercise of this whole thing, I mean, this thing has a really rough P&L, Intel does not work. And so, I started — so Pat and I started to realize, hey, we actually need to create more accountability. They need to be able to measure themselves against other foundry competitors. And they need to be eking out performance. They just aren’t focused on — they’re focused on delivering and quality and those kind of things, which are the metrics that they were measured to in the past.
So, that’s really the primary reason. Now your other point is right. When you do this, it does have customers looking at this a little bit differently. Okay, I actually now see a clear bifurcation between the fab organization and the product organization. I feel more comfortable around my IP. And so, as we went out to customers within the days ahead of announcing this, just to get their take universally, it was pretty positive. Okay, this actually even makes me feel a little bit better. So, it was a fringe benefit, but it is a benefit.
Tim Arcuri
Yes. Got it. Obviously, a lot of people agree that there’s a lot of potential for things to go right from here. I kind of think of ’18 as being the most important full component for the Company over the next few years. But that process node is still a ways away. So, the thing I think many folks struggle with is, what are the mileposts that we can see tangibly over the next 12, 18 months on AT&A otherwise to justify people buying the stock? Is it selling up a big foundry customer? Is that the single biggest full complaint that we can see that’s tangible that we cannot sure?
David Zinsner
Yes, I mean, from our own perspective, we have a whole set of milestones about when things detain, when they tape out and support, that when we power up that are going to be our own internal milestones. And we do provide breadcrumbs around those to help investors understand how we are progressing. But I recognize from an investor perspective that it seems a little more amorphous, not like something tangible.
Clearly, from our perspective, if we hit all those things, I think we’re going to be successful at AT&A. But from an external perspective, for sure, if you see — if we get customers, which we expect to get customers, signed up on AT&A that’s the validation that we’re progressing. And so I think that you’re right, that will be real evidence. And our hope is that earlier in ’23, we’ll be able to announce some of those customers. The progress has been very good.
We’re engaged with seven of the 10 largest customers within the foundry space. And so, I think you’ll start to see some true points over the course of ’23 that give you confidence that not only do we know we’ve got it going from an internal perspective, but somebody has put us through the paces externally. And it’s interesting, we thought we were pretty good on Intel 16, but lo and behold, when you get another customer in there putting you through the paces, you learn a lot about things that you can do to improve, and we’ve already experienced it even in Intel 16.
Tim Arcuri
So do you think that once we get inside ’23 that we could realistically expect to see big foundry customer sign up for SG&A?
David Zinsner
Yes.
Tim Arcuri
Let’s — I’m giving a question from the audience, which is leading into the next topic that I want to talk about I have a [indiscernible]. The question really is around the rationale for building the factories in Ohio and the choice of Brookfield as a partner. And then that’s going to lead into a bunch of other questions that I got.
David Zinsner
Yes. Well, I mean, obviously, we looked for a location that we felt like had the right level of water, had the rates infrastructure, had a good set of universities around it that could — we could attract talent, how — was in a state where the government was highly supportive, educating and training the workforce, helping us do that because even the line workers have a level of capability that is not typical of just a generalized manufacturing facility. So those were all key components.
Obviously, incentives played a factor in it. And so that drove part of it. I think the level of engagement from Ohio was pretty unmatched relative to any other states that we were looking at for that greenfield fab. No. Obviously, we’re also expanding in Arizona. We have a facility that we’re going to expand into in Germany. I think we got the same kind of vibe from the German government in terms of their level of interest engagement and support. And so, that’s what drove most of the decision-making. Now on the Brookfield thing, was there a specific question on Brookfield or do you want to…
Tim Arcuri
No, just kind of talk about them as a part of how that all.
David Zinsner
Yes. So, the challenge with the fabs is there’s — you look at Arizona, to do those two months is going to be somewhere in the $30 billion range, maybe a TAM bit less. It’s significant amount of cash outlay and it takes quite a while before you start seeing cash inflows from producing wafers and so forth. And so, we recognize that discrepancy between money going out, money going in, and we needed a structure that improved that balance.
We also recognize that if you are trying to raise cattle for these fabs, the existing pools of capital are relatively limited. I mean, there’s obviously availability, but there’s a limit on any one-time what you can raise. This was an opportunity to leverage somebody had a lot of cash that they wanted to put to work, a big slug, and we needed a big slug. And so, we worked with them over months to kind of optimize this to work both for Brookfield and for Intel. And we came up with a structure that essentially looks a lot like a JV.
We invest 51%, they invest 49%. We do it over time as we test to build out the factory. And then as the factory starts generating revenue income, the income kind of comes back to both parties. There’s a certain amount of upside that we think we can drive that we get to keep. But there’s a — they are protected on the downside in case we can’t load the factory as high as we want.
When you look at it from an internal rate of return basis, the rate was below kind of our weighted average cost of capital, but it was higher than kind of a borrowing rate. But when you factor in how much money we needed, probably that borrowing rate had upward pressure on it anyway. And given that rates have moved quite a bit, it’s now at a point where it’s pretty commensurate with debt, and it accesses a pool of money that we just don’t ordinarily have. We have bondholders. We have a certain set of banks. We get all of their limited partners providing equity to us.
And then they will back leverage it. But when they back leverage most of that leverage is to institutions that we don’t deal with. So, it was pretty attractive. Our goal will be to do more of these, but we’re going to be thoughtful around which ones make sense based on our confidence around when loadings are at a level that will justify some structure like this.
So, I would look for more of them, but I wouldn’t necessarily say every one of the fabs I just mentioned are going to be candidates. We’re going to have to be somewhat thoughtful about that.
Tim Arcuri
Got it. Yes, just on the point about third-party financing. Currently, you have three buckets. You have the CHIPS Act, you have tax credits and you have the Brookfield, right? My estimate it could be up to $10 billion per year, each of the next three to four years, I mean it’s a lot of money when you actually add up the whole thing. I get the argument from a lot of investors that say, well, if they’re that dependent on government money to be competitive, that I’m not going to pay a multiple for that. And I also hear that well, what if there’s regime change. And there’s not the type of commitment to subsidizing manufacturing onshore. How do you sort of think about that?
David Zinsner
Yes. I mean, first of all, hopefully, you’re not investing in any other semiconductor company that manufactures because I guarantee you every one of them gets grant money. That happens in Asia, believe me. So everyone is already getting grant money outside of the U.S. and Europe. As I look at the — how long can kind of count on this, the invest tax credit is basically four years.
The CHIPS money is projects, it’s a project base things that will extend beyond four to five years, but there’s a limit to the time frame of the project. And as we are looking at the cost structure of those fabs and whether we can make it work from a cost competitive perspective, we’re assuming this, but we’re not assuming anything beyond this. So our assumptions are built around what’s been — what we know for sure.
I’m hopeful that actually they use because I think in order to be competitive with Asia, it will need to continue. But I’m not too worried about it right now. I think that what we have gotten through assuming that we get the right level of CHIPS grant money that we think we will, assuming we get the rate level of CHIPS grant in Europe that we will together with the investment tax credit, I think we are in a position where we do have a competitive cost structure relative to Asia and that I think we can be pretty effective, as a foundry supplier and as a manufacturer of wafers in the U.S. and Europe.
Tim Arcuri
Just sort of dovetailing on that, can you talk about the dividend? I get a lot of people asking that, wondering how sacred that really is. And I — in some ways, I think I would have maybe wanted you to just pull off the band-aid and just cut the dividend rather than acting all these other sources of — so how safe [indiscernible] to the dividend and sort of how was the calculus that actually went into the potential to cut the dividend versus access to these third-party?
David Zinsner
Yes. I mean, I think, obviously, there’s a Board decision ultimately. So, it’s being a little out of turn probably. But I would say that our thinking, given the amount of investment that we’ve got to make is that growing the dividend, obviously, it doesn’t make a ton of sense.
And our goal is that to get to this 20% free cash flow as a percent of revenue as our optimal model. And once back in that place, of course, you might think differently about dividend growth. But until then, this is probably at the level we want. I think it is important. I think the Board feels it’s important. It gives investors a yield when we’re in an investment phase with the Company.
And so that we felt like that was important, and we felt, given our balance sheet, remember, we’re an eight-plus rated credit, it’s not like we’re putting pressure, undue pressure on the balance sheet or taking undue risk.
That said, I guess the only other copy I could say is that, we do — our best ROI is to invest in those in four years is to get our product portfolio to the right place is to leverage our manufacturing infrastructure for foundries. Those are the higher ROI opportunities. So, if something got dramatically different in terms of our cash flow, of course, that might change the calculus on the dividend, but that’s not the case today.
Tim Arcuri
Got it. Can we talk just about the shape of next year and maybe also how to think about CapEx next year? I think at another conference last week, you suggested that March is likely to be, I think you had no better than seasonal. Seasonal is down 5% to 7%, roughly. And if anything, it sounds like it’s a little worse the bias with each of the downside versus that. Is there any way to sort of characterize the macro as we kind of head into Q1?
David Zinsner
Yes. So, I’ll call out this by saying our visibility isn’t pristine in this case, it’s somewhat unclear how the macro is going to behave. But the way we kind of looked at it was just taking a little bit of a peak into Q1, and that’s all it is at this point is a peak.
Our estimation is that inventory digestion that’s been going on with our customers, probably doesn’t finish at the end of this year, it continues into next year. That, coupled with macro headwinds in basically every geography out there, didn’t give us a ton of confidence that there would be some reason for us to be better than seasonal.
And so, we felt like sending a signal to investors to make sure they under said, hey, we won’t be probably out of the woods in terms of inventory digestion at the end of Q4. So, we would model something more no better than seasonal.
And — and as you pointed out, Tim, I mean, depending on the analysis you do, if you go back three years, it’s actually more like low single digit declines. If you go out more like five, seven years, you get more to this 5%, 7%, which I think is the right starting point for seasonality because the last couple of years have been so noisy with COVID and so forth, it’s kind of hard to get good data.
And so, that’s kind of the near term. Obviously, it’s macro-driven and as we progress out of this at some point cycles that are down, turn around the cycles that are up, and we’d expect to be able to take advantage of that.
Tim Arcuri
Great. Following on that, I want to ask about CapEx. And net CapEx this year is sort of mid-30% as a percent of revenue. But that includes a pretty big $4 million offset here in Q4. Is the right way to think about your net CapEx next year that it would be a similar mid-30% of revenue?
David Zinsner
Yes. I think what we — the way we want to operate is to generally in our investment period, which is ’22, ’23, ’24 operate in this kind of mid-30s as a percent of revenue. Obviously, we had a different view. Obviously, we had a different view on revenue back in the Investor Day.
I think one thing you can take away from this is, we’re operating within our guardrails. So if the top line changes, then the net CapEx intensity come for the dollar number changes to make sure the percentage stays the same. And so that’s our intent. And after we get beyond that, my expectation is that our net CapEx intensity probably comes down to something more like mid-20s as a percent of revenue.
Tim Arcuri
Got it. And then on gross margin, can you talk about some of the put takes as we look into ’23. 45% in Q4 is — includes about 300 basis points, I think, of under utilization charges. So, is it fair to say about 48% as being the right baseline as you kind of head into next year — I think maybe help us think about some of the puts and takes.
David Zinsner
Yes, I think so. I mean we probably will be underloaded in the first quarter. I don’t know yet what the rest of the year would like, but clearly, they’ll that will be a headwind until it’s not. And as you point out, once we can load the fabs up to capacity that should take off a 300 basis point headwind for us.
We’re also getting this $1 billion of cost savings that I already talked about for next year. That should also help. Keep in mind also that pricing — we made some pricing moves in the fourth quarter that will certain points in time that should start to show up more meaningfully in ’23.
ASPs are stronger in Sapphire Rapids that obviously helps as well. So, there’s some definite tailwinds to our gross margins. Of course, it’s more could it be more dominated by the cycle probably for the next few quarters. But I think that as things start to improve, we’ll start to see some benefit.
The other thing I didn’t mention back to your question before around confidence around the gross margin longer term. Beyond just this 300 basis points headwind, we’re also probably incurring about a 200 basis points or so headwind 4 to 5 nodes per year, just so many so much start-up costs jammed into current many years.
So, as we get out of the 5 nodes in four years, we start to go towards a normalized rate. That should also help by a couple of hundred basis points. I won’t have it in ’23. That’s not going to get a tailwind for ’23. But out in the longer term, ’26, ’27, we’ll be able to do that.
Tim Arcuri
Got it. If I look at book value currently in the mid-20s, so it’s about a 100 million.
David Zinsner
Book value.
Tim Arcuri
Book value. And I get the sense that there are — I get many calls from people who are sort of taking in on the stock, looking at book value. And I think people are looking at the replacement cost of your manufacturing assets and wondering whether you could, in this environment, you could replace your manufacturing assets for $100 million. So how do you think about it? Is there some way to think about the replacement cost for your existing assets?
David Zinsner
Yes. Did this a bunch at Micron because we have a way more value business. And it’s hard to come up with a great number to be on doing replacement value. I would just say, given that we’re depreciating these things pretty quickly. Equipment appreciate in five years, you can pretty much expect that the replacement was pretty meaningful because these things in year six aren’t worth of zero, right, obviously. And so, it’s at least 20% or 30% higher than the net book value when you look at it from a replacement value perspective.
I thought the other thing where you’re going, Tim was when you look at that as that’s a lot of value that’s kind of most your entire market cap. And then, we just got an unlock the value more recently with Mobileye. And you can look at the valuation has been pretty healthy. So as we look at it, I think one thing that maybe Pat hasn’t gotten enough credit for externally is how pragmatic he is and how focus is on unlocking shareholder value through whatever means is necessary.
And I think when you look at the Mobileye transaction, and I’m not saying there will be transactions like that, but there will be value unlock for sure. And so — and Pat, I think, focuses on that almost as much as he focuses on all the execution-oriented aspects of as well as well. So, we will continue to look for ways where we feel like the market hasn’t quite picked up the right value for the business and find ways to make that happen to enrich the shareholders.
Tim Arcuri
Well, the foundry business as you create a separate P&L, that’s certainly another asset that you could do something with, some of that you do with Mobileye for sure. Last question, we have about two more minutes left. But Dave, you were talking about aspiring to be a 60% gross margin business longer term. TSMC is 55% gross margin, but it was more like 50 even a year ago before they put through all these price increases. Can you justify maybe how you get back to 60% unless we’re talking about your product business, leveraging edge manufacturing, whether it’s yours or TSMCs, how do you bridge sort of where you are now to get to 60%?
David Zinsner
Right. And keep in mind, I don’t think we’re ready to call that the model, but we have an eye towards 60% gross margin. I would say it’s a number of different factors. Obviously, being meaningfully behind on the process technology side has hurt us a lot, not only just in terms of the costs associated with catching up, but also just our ability to have products that are highly differentiated that can command a stronger price.
We’re starting to make real progress now, but there’s way more work to be done there to get products that generate what we view as a fair price. And I think that will obviously help. As I talked about all these cost improvements, I think, significantly help. Some of the markets that we are entering into, like the graphic space, like the high-performance part of the data center space, they command margins that are well north of 60%. So that should be — we should be able to mix up in that regard.
And I think the area that — back to your original point, the area that hasn’t generated a level of margin that would support 60% in total is potentially in the foundry business. And of course, will be coming from behind relative to the leader will have all work out for us. But assuming we get to process technology leadership and assuming we were right that customers will need as we want another provider in the leading-edge foundry space, I think the margins should be commensurate with where you think of in terms of more leadership margins.
I think when you mix those together, I think 60% should be pretty simple, particularly when you don’t have a margin stacking challenges that our competitors have.
Tim Arcuri
We’re out of time but thank you, Dave.
David Zinsner
Thank you, Tim. Appreciate it. Thanks, everyone.
Be the first to comment