Intel Corporation (INTC) Earnings Call Transcript & Summary

Hey there. Good morning. Good morning, everybody. I guess good afternoon to those in London. But my name is Matt Ramsay. I'm from the semi's research team here at Cowen, and welcome to our first virtual TMT conference, hopefully is useful for everybody, and we can get everything done without too many wrinkles. It's a little bit -- we're all learning on the fly here, I think, all the way around. So -- but really, really happy to have Trey Campbell here from Intel and have a little bit of a chat about things that are going on with the company and in the markets they serve. So thanks, Trey, for being here. I don't know if there's anything you want to say to kick us off, but thanks for doing it.

Well, hey, good morning and good afternoon, depending upon where you are. No. I'm excited to be here and good to catch up, Matt, and talk to everybody today.

Yes. So obviously, interesting times. Some of the industries that Intel traditionally and still dominates are behaving a bit differently than a lot of the other industries in technology where COVID is a big headwind economically and to a lot of different pieces of businesses. But Intel, the PC and the data center business, particularly the cloud part of the data center business, have obviously had some pretty big tailwinds from work from home and COVID trends. So maybe just starting off with DCG, Trey. I think there's a lot of moving pieces for the year, and I think folks are quite curious about those moving pieces. The cloud demand is strong. 5G infrastructure is strong, maybe some headwinds in government and enterprise. And if you could just sort of set us about how you guys are thinking about the year in DCG and potentially the puts and takes on the shape of the year, and then we can kind of take the conversation from there.

Yes. You did a pretty good capture. From what -- I'll give you a little bit of compare and contrast versus what we thought in January and then where we kind of ended up on the April call. But the main one, I think, that was different, if you go back to January, one of the things that we talked about was cloud. We thought it would be robust in Q1. You kind of had several quarters of pretty strong cloud build-out. This is pre-COVID. If you remember, coming back into kind of Q4 '18, early '19, we started to see cloud roll over a little bit after a number of quarters of really strong kind of build-out. And so second half of '19 and then through Q1, we had expected Q1 to be pretty strong. And we had also said starting in Q2 and then maybe transitioning into the second half of this year, we would have expected cloud to kind of go into a bit of a deceleration pattern where what we've typically seen historically is after a number of quarters of strong above trend line growth, you start to get some kind of mean reversion where just a lot of -- the majority of cloud service providers get optimization on some of those assets that they bought, start to catch up to some of those purchases, and then you get another kind of strong burst. And so everything that we have seen historically, we believe, which is that the path is up and to the right over a long-term basis at a really good growth rate, but we had expected it to kind of tune down a little bit starting in Q2. What we see now, and I'm kind of talking about end of April or 90 days later after our Q4 call is it was stronger than we thought. A lot of the trends that anecdotally people have seen and now you've seen all the cloud service providers report out, a lot of cloud demand in the COVID environment. Obviously, we can talk about PCs, but a lot of pull on people valuing their PC in a way they haven't in a long time in this kind of an environment. And so we, now, and we guided, we expect a strong Q2 in cloud. We've already had customers talking to us about Q3. So definitely, I think 90 days with more information, we think cloud demand is stronger than it was our expectation back in January. If you look at the comms business, that kind of end market segment, I don't think it's a lot different than what we thought back in January. We expected it to be a strong period where we're starting to see the 5G build out. We're growing our footprint out into the radio access network with a number of products, but one we've talked about, which is Snow Ridge. It's a 10-nanometer based SoC that does packet control plane in the base station. And so we're starting -- we have known designs at some key TEMs in that space, and so we expect that to start to ramp out. And so I think net-net, that's probably not a lot different. And then even though we saw a strong Q1 in the enterprise and government segment, our expectations are that there's going to be some headwind on that part of the business. It is pretty highly correlated with GDP. And if you see a 6, 7 point kind of inflection in GDP, we're going to see some impact on that part of the business. There are some verticals. I think where we sell a lot of gear into, they're going to be severely impacted, and you guys follow what those are. So that's kind of how we think about DCG. Maybe clicking down just a little bit into some of those other data-centric markets, the other places where we've already started to see some impact, and you saw that in Q1, the IoT business, pretty heavily leveraged into industrial, retail. Those kinds of markets have already started to see some drag from the -- from economics. And then Mobileye, obviously, is impacted by the reduction in global car sales. I think they're better situated than a lot of folks in that chain but still some headwind just as fewer cars are sold.

Got it. Yes. No. That's -- I think that's a good overview. I think, for us, trying to figure out as well, there's a couple of things going on at the same time. You just mentioned the market forces within DCG primarily, but there's also -- the competition has strengthened some and also you're in -- and Intel is in the process of ramping towards 10-nanometer Ice Lake launching late this year. I don't know, maybe you could give us an update there. Everything is still on track for some shipments in the fourth quarter. And when should we think about that portfolio around the 10-nanometer data center business being a meaningful part of that revenue?

Yes. I mean we're on track for Ice Lake, end of the year. If you think about 10 nanometer kind of broadly, one of the comments that we had on the call in April was that we're accelerating 10 nanometer, more units on 10 nanometer than we expected back in January. That's really not a comment on Ice Lake server, which comes at the end of the year. That's more a client comment. That's more base station kind of SoC comment, but we're shipping Ice Lake into notebooks and the base station SoCs now. And then as we get to the middle part of the year, we're excited about Tiger Lake. That was -- one of the discussion points we had on the call quite a bit was some of the gross margin effect of shipping Tiger Lake at the beginning of Q3 and then taking -- basically taking the reserves on that in Q2 ahead of the launch in Q3 for what we think is going to be a very, very nice holiday refresh with lots of design wins on the table. And then as we get into the end of the year in Q4, we'll have Ice Lake server. And so -- but there's 9 or 10 products coming out on 10 nanometer this year. I just hit really the flagship products that are going to be the ones that are the most meaningful, I think, in terms of the business. But yes, we'll have Ice Lake at the end of this year. So think of most of this year really being on 14 nanometer with Cascade Lake and Cascade Lake refresh, bringing in Ice Lake into the mix in DCG in the last part of the year. And then as you get into next year, you'll have Sapphire Rapids. And so that -- you have more of the product on 10 nanometer as we get into 2021.

Got it. It's interesting you bring up Sapphire Rapids. When I speak to investors about the competitive landscape for the product portfolio in DCG, there's a lot of focus in the investor conversation around Ice Lake, just because it's -- I think it's the first 10-nanometer launch, and that's been such a flag post to get to. But the work that I've done is that the real sort of next competitive bake-off, I will say, between you and your competition and data center is really around a jump to the tile with all the chiplet-based architecture with Sapphire Rapids and then the Granite Rapids product after that, that it may -- Jim Keller and his team maybe have a bit more influence on as they joined the company. So I mean are we -- are you guys thinking about Ice lake as sort of a seminal product and a big refresh for the business? Or is it sort of -- or is it right to characterize it more as a big achievement to get the data center footprint on to 10-nanometer and then we'll push forward to the next designs that are coming out of the pipeline after that?

No. I think we're excited about Ice Lake, and it's -- we've been -- obviously, it's taken us longer than we would have liked to get to 10 nanometer, both in client and in server. But starting 18 months, 24 months ago, we started making really good progress on 10 nanometer. And obviously, we want to get more of the product line over there. But I would characterize -- Cascade Lake is a great product. We've been -- for all of the comparisons to Rome on Cascade Lake, we've been doing pretty well in the marketplace. As evidenced by Q1 and the degree of -- we've certainly factored in a more competitive environment as comp has better products now and moving forward, we'd expect. But we've been competing vigorously. We brought Cascade Lake refresh out. That brought higher kind of capability from a transistor perspective. We pulled some higher core counts lower in the stack and gave a pretty meaningful kind of price to performance bump to a lot of our customers. And then as we get to Ice Lake, we'll bring in some other capabilities. That's the first product for us that takes the new step in PCIe. And we had transitioned that a little bit later, assuming the -- when the market and more of the meat of the market was ready for that transition, so there'll be more capability. Unfortunately, we haven't given a lot of details on Sapphire Rapids. As we get a little further down the road, we'll likely talk a little bit more about it, but we're excited about that product as well.

Got it. Got it. I think we -- since we're talking about 10 nanometer, and you mentioned the gross margin commentary from the call and the focus on some of the moving pieces with gross margin, it sounds like maybe a bit of a pull-forward and some strong ramp of 10-nanometer yields to get Tiger lake out for back to school in the back half of the year. And maybe you could just -- I think, for me, personally, it was a pretty big surprise when the numbers came out and the movements up and down in gross margin with the reserves. So I think many on the line here probably are familiar with that, but you can maybe walk us through it one more time. And as we get through the third quarter and some of those impacts reverse, I think the big question on a lot of folks' minds is with the 2021 outlook still at a 57% on gross margin, is that still the right way to think about? And how do we get from where we are now to maybe that number with some ups and downs in between?

Yes. So let me start with kind of what we said in April on gross margin. And as you referenced, there was a pretty big gyration year-over-year and sequentially in Q2 for our guide, so down about 6 points sequentially, down about 6 points year-over-year. And the majority of that was really driven just by kind of some mechanics of how we account for new products when we qualify them and a conservative practice we have to basically reserve those until we reach a ship qualification and can get those shipping in high volume. And so from -- we do this for every product at the company and we have for the 20-plus years I've been here. It just so happens that it really only kind of flutters corporate gross margin when it's a client product that, one, has a lot of volume; two, ramps into kind of a holiday window, very steep window; and three is accentuated when it's at the beginning of a quarter, meaning, in this case, with Tiger Lake, we talked about having more than 50 designs already teed up for the holiday. So we're going to make a lot of this product. So we're going to make high-volume production throughout Q2. In all of that period, we'll reserve. So it's just a -- it's a conservative practice. We expect to ship the product. But there's a sliver of possibility in any product that you have an issue that you can't sell until we take the conservative stance to reserve it, which means you take a big charge in Q2. That effectively means, in Q3, you're going to get the revenue for those products and you're going to sell them out at kind of 0 cost. And so think of a 300 basis point plus kind of headwind in Q2 that becomes a tailwind in Q3 and then mostly normalizes out by Q4. And so -- well, it's a standard practice, and it's something, I think, if anything, I'd read it as confidence in the demand for the product heading into the holiday season. And then the other part of it was just, hey, we're shipping more of the base station SoC. We're shipping more of Ice Lake, which is the first-generation 10-nanometer notebook product. And all of those things are creating some headwind. Net-net for the year, we pulled the full year guide for all of the items because of the -- there's just some -- there's a larger-than-normal uncertainty in the second half of the year based on how COVID plays out in some of our markets. So we pulled the full year guide. But what we said is, based on Q1 and Q2, we're roughly in line with what we expected for gross margin for the year. But we've pulled the full year guide in the sense since we pulled this year, we're not really kind of going to walk into next year because we need to see what kind of this year holds, and then we have a better kind of vantage point to talk about what all the items are going to be next year.

Got it. No. That makes sense. A bit -- just to follow up a little bit on the variables that, I guess, George, had laid out in an Analyst Day, 18 months ago or whatever it was, around mix of 10-nanometer and 7-nanometer start-up costs. And none of those at a broad brush strokes, none of those variables going into 2021 would be materially different, absent COVID effects.

Yes. I mean the things that we talked about, and it's been about a year since we did the investor meeting kind of in May of '19. But where we talked at that point, when we talked about kind of gross margin in 2021, you hit it, which is, today, we're moving down the yield curve on 10 nanometer, but the comparison to 14 nanometer cost is pretty tough. One, we've been running 14 nanometer. If you remember, we kind of started it at the end of 2013. And so you've been running 14 nanometer for a long time. You have very high yields. You have a lot of the equipment set. We depreciate equipment over 5 years, so a lot of that equipment set is fully depreciated. And so when you look like-for-like 14 to 10 is challenging, even as you move down the 10-nanometer yield curve. And so that's where we kind of said, hey, as you look into next year, one of the factors is going to be -- you're going to have more of your products. We're still -- we'll have -- start to have more of the Xeon product. We've still not migrated desktop over to 10 nanometer, and part of that is you want to get to this -- the future kind of second version of 10 nanometer where you get more performance out of the transistors and higher clock frequencies. And so you'll have more of the 10-nanometer product. That will be a headwind to gross margin even as it's a tailwind to product performance and leadership capabilities. So I think you had that characterized correct.

Got it. Got it. One -- just jumping around a little bit in the topics. Given the amount of time we have or don't have, there's so many things to dig into, in addition to being forced to try to be a bit of a virologist and talk about COVID. All of us are getting our crash course in U.S.-China diplomacy as well being in the semiconductor industry. There's been a few different announcements that have happened over the last few weeks, TSM potentially coming to the U.S. There's always commentary behind the scenes and questions about what products Intel may or may not keep in-house versus eventually outsource. I don't know. I've been asked by a lot of investors to just, I don't know, come to you and ask if you could give an overview of what those 2 topics, really, how U.S.-China relations are impacting or not impacting Intel's business given you guys manufacture most of your stuff in the States. And secondly, how are you thinking about competition in the fab space from TSM? Does it change if they move to the U.S.? And how does it change your views on what needs to be done inside or outside of Intel?

Yes. Let me start and let's -- and then interject, and we can unpack it wherever you want. And so I think, one, I'd start with China is a super important market for us. Like a lot of companies, China is the second-largest PC market for end consumption. It's a big consumer of a lot of our data-centric products, and so we'd like to see a world of free trade and a world where we can participate in all of those markets. And so we work hard in -- with both sides on that. We hope that the bilateral discussions that have been ongoing are fruitful and continue to generate a more open market. That said, we have a really good relationship in China. We've been there 35 years. We've got factories. The memory factory in Dalian. We've got an assembly plant in Chengdu. We've got lots of engineering teams in Shanghai and Beijing. And so we've continued through COVID. We continue to operate in Dalian and Chengdu. And we continue -- we've complied with all of the U.S. export restrictions and have continued to ship most products into China. There are certainly some places where -- with some supercomputing applications. We comply with if anything we know is going into a military application, those places where -- and there are certain known customers where we're restricted. We've continued to look at the things that came out recently from commerce. I think our first read of those would be that will probably not be a huge impact to us in terms of our shipments, but we're continuing to assess that and look at it. So yes, it's a huge market. We hope everybody -- we could kind of work through some of these things and continue to ship into that market. That said, we do have most of our manufacturing in the U.S. from a fabrication perspective. We also got fabrication plants in Ireland and in Israel, and so we've -- we are a very large U.S. manufacturer working closely. There have been some communications that we've had in terms of how we think about the capability of U.S. manufacturing in the future. Obviously, TSMC made their comments about potentially being in Arizona. So we're working through all those. Obviously, if there was going to be a world where there was a U.S. manufacturing kind of presence, it would certainly be more economically attractive to not have TSMC here, but I mean, it's one of those where we'll work through all of those options. I think you also said something on outsourcing. Was that -- is that right? I want to make sure...

Yes. It's kind of all tied together there a little bit in the back end, so yes.

Yes. I mean -- and we're -- we hit this before, but 20% of our volume historically is kind of out in the foundry network. Historically, that's really not been CPU products. That's been things that are more high mix, low volume. Or even if they are high volume, it's been the modems or some of the FPGA products. We've bought a lot of companies through time, and we continue to manufacture Mobileye outside. We have products from Habana that are going to be outside. There's a -- we don't force everything into a mold here, and I think the management team doesn't have allergies to having things outside. Historically, though, we've seen a stronger ROI from having some of these high-performance logic devices inside in our factories, and that's why we manufacture them there. So I think it's one where we're doing a lot of things to improve our capabilities over time. Some of these will resonate with you from your background. But we've historically done a lot of bespoke IP development. We probably didn't need to be. I mean we're taking valuable engineering time to go do things like USB and PCIe blocks that are commonly available out in the ecosystem. And so I think one of the things that we're on a journey is how do we use the ecosystem more openly, whether that be the EDA vendors, the standard IP vendors, the equipment vendors. How do we share in that ecosystem so that we can put our capabilities and really generate return on the things that we can customize and deliver a lot of value on and not burn a lot of calories on things that are already standard and available? The secondary or ancillary benefit of that, I think, is that if you decide that you want to outsource something, if you're on the same kind of ecosystem and capabilities, it's a lot easier to go make that decision than if everything you've done is bespoke. So I think we're going to do that to make ourselves a better IDM, but it creates more option value in the future, too.

I would imagine that, that path that you're talking about there as some of the products become more multi packaged or tile-type architectures, that would probably lend itself to that trend over time. A bit of a pick and choose as you put products together. Is that sort of the idea?

Yes. I mean I think it's one of the benefits, I think, that you get in that kind of a tiling kind of world. I mean the other one, I think, is just you -- one, we've seen this in the FPGAs already, where we create an FPGA fabric. Then we've got some of the packaging technologies like EMIB that we've deployed in FPGAs where then you can take on some of the tiles for the FPGA, and they can be on different process technologies or from different internal or external foundries. So you have definitely that capability as you move into more kind of tiled capabilities. The other one is that you lower the size of the -- it's different than having a 600-millimeter square monolithic block and that you can kind of have multiple smaller tiles which also have some benefit. And as you think about where you're putting those tiles in terms of leading edge and how you can get them potentially quicker in the leading edge if yields are lower, if it's a smaller die size, those are -- that's also a beneficial capability.

Got it. I guess we have a couple of minutes left here. And one of the things you and I have exchanged a couple of notes about and I think the Cowen team has spent a ton of effort in the last number of months digging into this new concept of computing at the edge. It's not really new per se, but it's certainly new in terms of a nascent market that -- at least in my team, I think that it's going to be a big deal over the next decade or so as intelligence gets pushed out to the edge. And I just wonder, a, do you guys agree with that holistically at Intel? And b, it seems like coming from the data center strength, the relationship with some of the comms service providers, there's been some challenges to the overall market share in DCG. And -- but it seems like the long term, over the next 5 to 10 years, it seems like the pieces that Intel has put together might be beneficial for the edge in the long term. I don't know if I'm characterizing that right or if you'd agree with that, but I just -- if there's any thoughts you might have there, we'd really appreciate it.

Yes. No. We did change -- exchange notes on this, and I think it's great that you guys are really kind of digging in on this. I think edge is one of those things that's talked about all the time, but it's also one of the most misunderstood topics on what it is and what the taxonomy or the definition is. For us, the -- it's about taking compute resources that were centralized in hubs and moving them out to point of service or closer to the origin of the data, and there are several reasons why customers want to do that. It's about application latency. It's about quality of service. You can get better TCO. You can comply with data sovereignty requirements. And hopefully, you can get better insight based on the data. And the other big driver is that it's just a huge -- it's going to be a huge market. Whether that's kind of looking at edge of the network or out to kind of edge premise where you're out in a factory or even out to a device like an autonomous car. And so we see a $65 billion market in the next 4 to 5 years. That's a TAM, but we also believe we're getting kind of on the order of $7 billion to $8 billion today in edge. If you look at kind of our IoT business, our Mobileye business and you have to deconstruct a little bit of our networking business and talk about where it is out at the edge versus out of the core. But I think it's a huge example. And in that market, I'm really not even capturing the autonomous driving kind of TAM. When we look out to 2030, we see markets that are $70 billion, $160 billion when you think about autonomous driving and AV and capabilities and then all the way up to kind of mobility as a service. So this is a huge opportunity, and so we're super excited about it, how our products play in that market.

No. That's helpful. I think we've focused a lot of our work on the edge cloud, so sort of the extension of the server market and the cloud server market, but there's obviously tentacles of that, that go into other big areas as well. I think last one, we got a couple of minutes, just normal financial question stuff that I've gotten pain to ask a few times. You guys got -- made some changes to the buyback structure and a few other things in relation to COVID. Any new or different things sort of variables there? I was a little bit surprised maybe, just given the financial health of the company that they made -- George and the team made that changed so quickly. Any considerations there or just sort of conservatism, and let's see where things play out given the dynamic market we were in at the time?

Yes. I mean I think it was prudent conservatism. I think you go into a niche situation with a global pandemic, where we don't know a lot, and you don't know about the breadth or depth of how long it's going to last, we certainly -- it wasn't for a worry about having free cash flow generation. We're going to generate a lot of free cash flow under a lot of different scenarios, but it certainly -- you can look at the different returns of companies who were highly levered versus not levered, right, when we went through this. I think for us, we invest a lot of dollars in R&D. We invest a lot of dollars on CapEx. We want to make sure that we have those pools of cash to invest under a variety of different corner case scenarios. and so we went out. We got over $10 billion in the debt market, just -- and that was really just to help us during this situation, have that kind of buffer. We're not in a business to hoard cash. That's not what we do. We -- I would say is paused the buyback. We didn't stop the buyback. We paused it in this period. We fully intend to resume the buyback. That's a key means of how we return cash to shareholders. And at the same time, we tried to emphasize and underline how important the dividend is and that we're fully committed to the dividend in this situation. So I think our capital allocation philosophy is intact, to invest in the business. Some of that comes through, some M&A capabilities we did. We had a really exciting, I think, acquisition with Moovit to help our autonomous driving business over the last several weeks. Then it's the dividend, then it's making sure that we return that buyback opportunistically when we feel like it's the right thing to do based on the stock price.

Got it. Well, we scratched the surface of a lot of interesting topics. There's so much stuff to dig into with a company as broad as Intel. But really, really appreciate your time, Trey. For the investors on the line, obviously, follow up with my team or with Trey's team as you see fit, but thanks, everybody. Have a good conference and a good week.

Yes. Everybody, stay safe out there. Good to talk with you.