Intel Corporation (INTC) Earnings Call Transcript & Summary

All right. Good morning, everybody. I'm Ross Seymore, semiconductor analyst here at Deutsche Bank. We're very pleased to kick off our second keynote of the day. We have Sandra Rivera here from Intel. She's the Executive Vice President and the GM of the Data Center and AI group, DCAI. About 1/3 roughly of Intel's revenue, about a $20 billion annual revenue run rate. So a big company -- a big business within a big company. Before we get started with the Q&A, they want me to read a little bit of the disclosure, the legal disclaimer side of things. So today's discussion includes forward-looking statements, which are subject to risks and uncertainties. Please refer to Intel's SEC filings available at intc.com for more information on risk factors that could cause actual results to differ materially. So Sandra, thank you very much for coming out.

Thank you.

Why don't we kick off with some near-term discussion before we get to the much more important longer-term discussion? In that little disclaimer, we talked about near-term results potentially differing materially, and Intel had that situation creep up last quarter. Now since then, we've had plenty of other companies that have had material differences. So talk a little bit about what happened in the second quarter, what you're thinking for the second half? Just for DCAI overall, how much of the issues were macro versus company-specific, et cetera?

Yes. Okay. Well -- and I will speak to all that, Ross. And as you know, I will anchor all my statements on our Q2 earnings. So I'm not intending to update our guide that we provided, but I will obviously speak to what we shared then and some of what we're seeing now in terms of Q2 and the second half of this year. So we did provide a guide in terms of where we see the year finishing out just based on what we have been seeing from a macroeconomic perspective, which did include, of course, all of the issues around softening demand based on inflationary pressures looming potentially, the R-WORD out on the horizon, the situation in China in terms of supply chain constraints all over but just to match that issues that our customers continue to endure as they're trying to get their products to market. And so all of that, including what's happening, of course, in Europe with Russia and Ukraine did clearly impact our second quarter and is impacting our year. But I have to also acknowledge that we had execution issues ourselves, particularly as we were looking to drive our flagship next-generation Xeon CPU out into the market of Sapphire Rapids we uncovered an issue that had not been, of course, foreseen. And we had set -- we have set a very high bar in terms of the quality that we will -- that we've established to ship our products and decided that actually we need to do another small metal layer stepping in order to get that product out at the quality that our customers expect and that we expect of ourselves. And as a result, some of the execution challenges that we had in Q2 were based on the misses that we had in terms of expectations. And so that has pushed us later in the year for the ramp of Sapphire Rapids. So I would say that while there was quite a bit of macro, Ross, we have to own the issues that we have in terms of execution. And so we're taking material steps to address all of those issues in terms of our forward products, ensuring that we take those learnings so that we can be more predictable in our execution going forward.

On the macro side of things, again, we've had a couple of other companies. Just last week, some of the GPU or networking companies also have slightly choppier data center businesses results in guidance than were expected. So we chalk that all up to the macro side. So within macro, was there a specific region that was problematic for you, customer type kind of enterprise versus cloud? How do you split down were the source of weaknesses?

Yes. And I will -- I guess, I will make a comment as I answer that question, Ross, that we probably took a lot of grief for coming out and signaling that the market was softening and that there were a number of issues from a supply chain perspective because we were the first ones to come out. But then, as you saw, subsequent companies have also seen what we've seen, which is the market softening, a little bit more caution in terms of the OEMs and our customers from an inventory management perspective. And still, these massive issues that perhaps -- and I was just talking with one customer last week that they don't have 100 parts that they're chasing to ship that rack or that cluster, but they still have 2 parts, and they still can't ship that rack or deploy that cluster without those 2 parts. And so it is better, but it still persists. And so clearly, that has been an issue for us. From a regions perspective, we are the largest player in China. So we do ship quite a bit of product into China. There's a lot of innovation. It's a huge market. There's a lot of demand for compute. And as a result, we were perhaps been ordinately impacted as COVID impacted China and some of the policies around Internet companies there as well have had an impact on us. But overall, when we look at the market for the TAM opportunity for this year in data center and particularly in compute -- Xeon CPU compute, we see that the market is still going to grow. Roughly 4% is what we're looking at, but that is going to be growth from hyperscaler somewhere in that mid-30s where the enterprise and government is actually slowing down. So we see them dropping mid- to high teens. And so from that perspective, you can see that there is a little bit more caution on the enterprise side, a little bit more -- see how the economy plays out before they start spending as much of their budgets or even more constrained budgets than they've had. But even the cloud customers at this point are signaling some level of slowdown.

So last question on the near term. When you talked about the second half of this year, you said the second quarter would be the bottom for your business. And then you'd grow, I think, moderately or modestly into the second half. Talk a little bit about the assumptions that you have behind that growth. How much is Intel-specific versus some sort of macro view as well?

Yes. So with path, and Dave talked about at earnings was that we see that second quarter -- third quarter, sort of 2Q, 3Q as being the bottom. And that's why they provided the guide, right, in terms of just because there's a lot of uncertainty, we really can't pinpoint it. We can provide a guide with the lower end of that guide being just -- well, the -- being rooted, I should say, in the fact that the macroeconomics may not play out as any hope in terms of a resurgence in getting back to addressing those macroeconomic issues or the supply chain constraints and particularly in China, getting somewhat out of that COVID downturn. But overall, I would say that right now, we don't have any reason to see that things are getting more optimistic, meaning that we still see those issues playing out. There's a lot of uncertainty. The market does seem to be a bit more cautious in terms of just waiting to see what will pan out. We do see, particularly on the client side, and you've seen some of the other companies come out where the consumer spending is down. There's more inflationary pressure there. And we only see that happening first, obviously, on the client business before you see it impact the server business. So still a lot of uncertainty, still a bit cautious in terms of do we see the market coming back in full force towards the end of this year. I would say on the good news side that there's still quite a bit of demand for compute. Data continues to grow 50% every single year. It's done that for many, many years, and it continues in the meta process and compress and encrypt and move and store all that data. Still is driving the need for more compute. And so we are still operating in a growth market. And for us, we have a lot of opportunities, particularly in the areas of high growth workflows like AI, like networking, like encryption, like storage, where we have strong leadership position. And so we are leaning into a lot of that opportunity in terms of where we're focusing our efforts.

For the health of the market, you mentioned about the TAM, I think, growing 4% or so, but it was kind of a tale of 2 cities between enterprise and government being down and the cloud side being up. Let's focus on the positive side of the cloud being up. How immune do you think that dynamic will be to some of the macro issues that is uncertain that you highlight?

Yes. I don't think that anyone will be completely immune, but you're talking about them is the growth of the market going to flatten out or will it decline, and we don't see that. I mean, the market will still continue to grow. As I said, the amount of data being created and the need to turn that data into valuable insights, which is what AI helps us do will continue going forward. It's just how muted or how much you bend the knee of the curve in terms of exponential growth. And that, that I believe, everyone is signaling that it's probably a bit more muted because, ultimately, the enterprise customers are -- and the consumer are the customers of the cloud. And so if you see a bit less spending there, you're going to see it reflected in the cloud as well. Now there's a huge opportunity there because a lot of customers, whether they're enterprise customers or government customers, they still are deploying cloud architectures, cloud business models, cloud deployment models. And so whether you're deploying in the public cloud or onprem, we do see the market moving to cloud architectures, and it really drives so much of our thinking on our road map of how we have products and swim lanes, particularly on the Xeon road map, that are more attuned to those cloud-native workloads, those micro-services workloads so that we can optimize the portfolio around where the market growth is going to be. So I don't think, Ross, about cloud is just a set of customers because cloud truly is an architecture and a deployment model that all customers are embracing. And even if you see some compression in the -- or more cautionary, I should say, spending on the enterprise side, they're still going to want to embrace the cloud model, even if it's onprem or on the edge or moving workloads into the public cloud.

The last question on the end market as a whole. The last time, from a semiconductor perspective, we went through these cloud digestion periods was late '18, early '19. Intel had that issue then and then came roaring back. How would you compare this uncertain period to that one? I guess macro is probably a broader issue now. It was a little more semi-specific back then. But how do you compare these 2 times?

Yes. It's a good corollary. I actually think that there's probably more uncertainty now than there was. Back then, the market is more complex. There's a broader range of workloads and use cases. There's more coverage across the globe in terms of just how customers even think about their infrastructure and innovation and where they're trying to drive time to value. So we do believe that we come out of this, and of course, we see the demand in compute continuing to grow throughout the decade. The semi industry itself is growing into a $1 trillion TAM opportunity by the end of the decade, as we've talked about, which is why we're building capacity and why we are investing in leading process technology. But I think in the near term, it will be choppy. And long term, I don't think the thesis changes that compute -- as the demand for compute continues to grow, Ross, and we have evidence of that just from the amount of data that's being created in the world that you want to turn into some kind of value.

So let's get into, for our next topic, your road map of products and how you're going to address some of these workloads. I think you talked at your recent analyst meeting about AI being like a $40 billion TAM by 2026. So huge market opportunity just in that slice of what you're doing. Let's start with Sapphire Rapids. So I know it's going to be a little late. You had the issue one more meta layer, et cetera. But talk about the road map that you have and how it addresses those opportunities, starting with Sapphire and then we can go into Sierra and Granite and all the other code names that maybe people don't know. Just talk about what that road map looks like and what the opportunity is for you to address.

Yes. So we're very excited about the road map that we have and a lot of the work that we've been doing over the past year. And I've been in this current role for a little over a year now, Ross. And we spent -- certainly, I spent the first 6 months really listening intently to customers and getting feedback on what they want to see from Intel's products in our road map. And they consistently ask us for road map simplification and, of course, their need to address these emerging workloads around micro services and cloud native that demand that's high throughput efficiency from many, many cores. And so from that perspective, when we came out at Analyst Day in February, we unveiled the road map that we've been working on, but we did simplify it. We refactored it so that we have our performance core products, which will -- which you would know as the Ice Lake generation, the Sapphire generation, the Emerald Rapids generation and then Granite Rapids after that. So we feel very good about having a predictable cadence on what was -- what has historically been the performance core product, where we introduced Ice Lake in '21, Sapphire here in '22, Emerald Rapids firmly in place for second half of '23 and then Granite in '24. But we also then introduced to the world something we've been working on for quite some time, together with our customers, which is the efficient for swim lane, again, high density, lots of cores, high-throughput type of workloads. And the Efficient-core products are ones that we will bring to the market. Actually, we have been working with co-definitional partners on that product, large hyperscale customers. And that comes to market in '24 as well. But that allows us to have an offering that is targeted at a different set of workloads, where, particularly if you're working at such large scale that you're always looking for how you optimize your TCO. So from a TCO perspective, for those high-throughput workloads, the Efficient-core swim lane with CR4 being the flagship product that we'll launch in '24, is addressing that workload. So the road map was simplified in terms of the number of SKUs that we took off and the number of new products that we were bringing to market. We have established a predictable -- a much more predictable cadence that our customers can plan their businesses around with as Ice Lake, '21; Sapphire Rapids, '22; Emerald Rapids, '23; and Granite Rapids, '24. And then we introduced the Efficient-core swim lane CR 4 is coming out in '24. And we've gotten very positive feedback from our customers for listening, for simplifying and for giving them an opportunity for a better ROI. And just 2 additional points there is that Emerald Rapids is socket-compatible to Sapphire Rapids, dropped in the very same platforms. So all of the investment our customers have had in the Eagle Stream platform, the platform that we're introducing to the market. They get to do a platform refresh with more performance and new features with Emerald Rapids with, again, minimal validation so that they can get to market quickly. And then similarly, for the Granite Rapids in CR4 since '24, that will be in the same platform for that customers, again, can leverage their investment and then we can validate and get to market more quickly.

So with Sapphire, you mentioned about being late, you want to have the quality up there, there was a security issue, et cetera. That being pushed out a bit doesn't impact the timing of Emerald?

No, because -- let me explain a little bit more with Sapphire Rapid. So Sapphire Rapids is a very, very innovative platform. We are introducing the first -- we're the first CPL introducers PCIe Gen5. We're the first to introduce CXO. We're the first to incorporate DDR5 memories. We have a number of integrated accelerators into the platform, AI acceleration, network acceleration, encryption, compression acceleration, data movement acceleration. So there is a lot of innovation packed into Sapphire Rapids. The challenge that, that presents to us, of course, is that with all of that innovation comes the level of complexity that has to be then validated, not just at the IT level and SSC level and the C-level, board level, but with our customers at the server and at the rack level. And so that level of validation and the time to validate all of the new innovations in the platform has taken us a lot longer. So for sure, Sapphire Rapids is a product that all of our customers want. And if you talk to our customers, you hear them say that we are ready. As soon as Intel releases a product, we intend to ramp. We're going to support our customers with all those launches because they value all those innovations. But for us, we introduced the level of complexity that created this need for validation that extended longer than we wanted and when, where we introduced some quality issues along that journey that we know we needed to address before we got it out. But all of the learnings that we've had on Sapphire Rapids, which, of course, any silicon development cycle is a long multiyear cycle, we have channeled into our future products, which are maybe simple things but impactful things, where you shift left in terms of your IP validation before you take in your product, which is what we did with Emerald Rapids. All the new features fully validated before we take it. We have a lot more test coverage on the post -- validation side. We have built out huge clusters so that we can actually validate at scale, and we've invited customers to fully validate with us. And as a result, we know that Emerald is actually executing beautifully, high confidence of where we see that coming out because all of that risk has been pulled out through our new methodology out of the product development cycle and so high confidence there. And then similarly, with Granite Rapids, a lot of those learnings have already been put into that product to level cycle because, again, it's multiyear development cycles. And so my confidence is actually increasing not just on the product development side but, of course, the opportunity to leverage the process technology advancements that we're making and the -- getting back to leadership in terms of the process node technology that our technology development organization is driving.