Intel Corporation (INTC) Earnings Call Transcript & Summary

Everyone, I know people are still trickling in, but welcome to this afternoon session. Really delighted to have the team from Intel. I'm Vivek Arya. I cover semiconductor, semi-cap equipment at BofA Securities. I'm really delighted and honored to have Dave Zinsner, the Chief Financial Officer of Intel; and Sandra Rivera, who heads the Data Center and AI business. And I think David, it's your first conference.

It is, yes. I was supposed to do it before.

It's an honor. That's right, as part of Intel. And I know Sandra, it's your second one, so a little more pro at this. But really delighted to host you both.

And maybe, Dave, to start a few questions with you. I think you have now been at Intel for about 5 months or so. And I was hoping you could give us a report card of the good and bad and hopefully, not that many, at least. But just a general sense of what you have observed so far.

Yes, sure. Well, first of all, thanks for having us, Vivek. I appreciate it. And let me just give the preamble. We may make forward-looking statements. Investors are encouraged to consult the Form 10-Q and 10-K most recently filed for the risk factors that may materially affect those forward-looking statements. Yes. So been in the role 5 months. I have to say I'm really enjoying it, to be honest with you. Pat kind of, as he was recruiting me to come aboard, really was just in a really dynamic fashion talked about the transformation that's underway at Intel and the opportunity for me to be a part -- a small part of that, which is so compelling. And I've just been really happy so far with how engaged the team is in terms of driving that transformation. I'd say on the good side, clearly, on the process technology side, we are in a much better place, executing well against our milestones there. I think the operations team is world-class at Intel, and we've had a whole bunch of curveballs thrown at us over the 5 months I've been here even. In the operations team has ability to kind of react to all of those things and mitigate them has been pretty exceptional in my view. We have great customer relationships. Sandra and her team and her compatriots and the other divisions, I think, do a great job engaging with the customers. I think we get incredible amount of mind share, incredible amount of focus and help in terms of developing the next set of products that meet their demand and their requirements. On the bad -- I wouldn't say bad side but on the areas where we can certainly improve, the data analytics at Intel surprisingly are actually not as strong as I had seen in my prior companies. But I think there's definitely opportunity to improve that. And it may seem like a trivial thing but it's like flying a plane without some of the avionics instruments. It's hard to fly the plane without all of the equipment. And I think we can do a lot to help the businesses, like Sandra's business, do better across the board, whether it's R&D and development of products and early indicators there and how we're doing all the way up through how we are tracking in terms of backlog and sales and so forth. I mean we have that data but it's a big lift to pull it all together. Obviously, we're still a little bit uneven on the product side, some areas of excellent execution, some areas where we need to improve a bit. But we've -- I recognized when I got here that, that was a multiyear journey to get that to a better place. And I'm optimistic that we'll do that and exceed probably Pat's schedule in terms of when we get there. And then I would say the other area that -- didn't mean to go back, the other area that I would say is really strong at Intel is this like notion of bleeding blue, which Sandra's been here a long time so she knows it. The team here is really passionate about driving transfer rate, and I mean all the way from operators in the fab up through the executives. They're really rallied around the mission that Pat's put forth. And so I think we have everyone all in. We just got -- we just surveyed the team and all I saw were the initial like high-level results but they looked very strong. So clearly, his message is resonating with the team, and I think that's a good omen for us. Yes. So I'd say that's probably the high level. There's plenty of work for me to go do to make sure that we've got enough funding so that Sandra can deliver on her growth aspirations and the others can as well. So I'm excited about that opportunity.

Excellent. Maybe Sandra, on a similar kind of structure-type question. So you did make changes in terms of the internal/external realignment, right, of the Data Center business, put networking in a separate group. What was the thinking behind that, right? And what's kind of the core mission of the DC/AI business versus what DCG used to be in the past?

Yes, it's an excellent question and we get that one a lot. When Pat came in, it was very clear that he wanted to increase the level of accountability in order to improve the execution and, frankly, to be more transparent with our investors. And as a result, he structured the organization with those points of accountability, responsible for the core Xeon CPU portfolio as well as the FPGA portfolio, the Altera acquisition that came in over 6 years ago now, and the AI accelerators. And I'm also responsible for the building out of the enterprise-wide AI strategy, both architecture and road map, partnered with Greg Lavender, our CTO and Head of Software because AI is as much a software problem or opportunity as it is hardware. So that's my remit. The way that I partner, not just with Nick, who is responsible for the networking portfolio business that's near and dear to my heart, having built into that business with my colleagues, but partner with him and in many ways, we sort of have a driver drafter. So I drive, obviously, the Xeon portfolio, the FPGA accelerator assets and then I have my internal customers as much as the external customers that build out of that IP in that portfolio. And then similarly, Nick is building out the networking portfolio, encryption, compression, network packet processing capability, the connectivity portfolio. Raja builds the graphics accelerated computing capability. And then, of course, Michelle is responsible for the client portfolio. So there's a lot of leveraging of the investments across the entirety of Intel, like more clarity in terms of who's accountable for which parts of the portfolio and how we serve not just the internal but the external customers. And this is what Pat is all about: accountability, focused execution, objectives, key results and driving to that product leadership that our customers actually value.

Do you think that creates any level of risk in that when we look at some of the companies, it's interesting that we have had the compute companies make acquisitions of networking assets, right, and then optical assets and right, so they're talking about a lot of integrated products. So do you think having this siloed approach creates more accountability? But do you think it prevents that kind of cooperation? Or are we reading too much into that?

I think you're probably reading more into that because in order to deliver a complete platform, you actually leverage assets from all over the company. Not just -- I mean, even if we're looking beyond the different IPs, if you think about, as Dave was talking about, our process technology, the manufacturing capability, the software portfolio, so much of the future for us is software-defined, hardware-enabled and very much leaning on Greg Lavender, his organization in the software horizontal capability that he's providing to all of us from a product perspective, again to unlock the value of the hardware. So we don't really look at these products in a vacuum, it's really a platform. And software becomes a unifying theme across the portfolio.

Makes sense. Maybe then Dave, stepping on to the demand side. As we were just talking before the session that there appears to be kind of this dichotomy between where investors and the market demand is going versus semiconductor companies, right, who are mostly sounding pretty strong about backlog and demand signals and so forth. What's on your dashboard from a demand perspective? What's going well? What's not going well?

Good question. I think on the macro side, clearly, it's weaker. And we, like everyone else, will be impacted by the macro events that are unfolding here more recently. That's clearly going to impact us, as it will virtually everybody else in not only the semiconductor industry but globally in terms of corporations. For us, the other thing was we had 3 kind of headwinds coming into the quarter, which we talked about. One was the match set issue where customers could not get enough components to build product. That was we expected to impact demand. The second was inventory. We expected customers to reduce their inventory levels, which would impact their demand on us. And then there was the China, Shanghai closure and we expected that to open up in early May. And it takes time to get back to normal but get back to normal on a relatively quick fashion. I think in all 3 cases, the circumstances at this point are much worse than what we had anticipated coming into the quarter. So that certainly is an impact to the business as well. I think when you look at it over the long term, though, it's -- semiconductor industries have cycles to them. Every company within the semiconductor industry has a slightly different cycle to them, and that's how we're used to managing through those things. Pat and I talked that we would manage the business, and we've managed our guardrails appropriately. And so we will do that regardless of the circumstances that unfold. But looking over the long term, I feel really good about our opportunity to hit the model within the time frame we talked about. I feel good about the growth rate moving to this 10% to 12% by '25 to '26. The margins to be back towards the higher end of our range of 54% to 58%. We expect to have good cash flow once we've made the necessary investments over the next 3 years. One of the more positive things that has also happened over the last 4 or 5 months is our capital offset, the smart capital strategy that we talked about at Investor Day is actually going pretty well. And I think by the time we talk in earnings, we'll be able to update and talk about a better number than we were originally anticipating even for '22. So I think that helps in terms of supporting the investments that we want to make for the -- to drive the long term. So overall, optimistic about the future. We're going to go through some choppiness for sure in the near term as everyone else will as well. And what we've got to do is kind of keep our heads down and drive the business, execute to the plan and things will have a good outcome for us.

Is it -- so I know when you had the last call, the expectation was that second half, right, would grow and there would be seasonality, but it's whatever normal seasonality. But consumer headwinds have actually increased, to your point, since that time. So how does that impact Intel's second half?

So it's noisy for sure. I think we'll wait to give a firm number when we get all the data for this quarter and have a real -- we have a whole process by which we roll things all up for the second half and we'll have a much better feel. At this point, I would just say it's gotten a lot noisier than it was even a month ago, so. But on the flip side, we are paying attention to those things, and we're managing the spend levels and investments in a way that is appropriate to invest in the long term but also appropriate for the conditions.

Got it. And then maybe drilling down on the data center part, Sandra, which is one of the key parts of Intel? How do you see the demand signal shaping up there for your business?

Yes. So from the customer perspective, we typically do see that server and data center lag more of the client business. But server and data center will also be impacted in terms of demand that we see our customers, as they lean out their inventories, some of the match set issues that Dave talked about that they are still experiencing. And then waiting to see exactly how the enterprises consume where they may shift some of those workloads where the CSPs are still going to see growth, is that going to be muted or not. I will say that it's probably not one answer because it depends on the geography, Vivek. Like we know that in China, as an example, the government -- regulatory environment is defining or directing that some of those enterprise workloads can't shift to the cloud. And so we see the modernizing the enterprise on-prem infrastructure more so, right, than other parts of the world, whereas in the U.S. and in Europe, we see more customers doing -- the enterprise customer is doing that lift and shift to the cloud. I think what happens there, of course, is when they first land on the cloud, there's a lot of inefficiency because the software wasn't really written for a cloud-type infrastructure. So we have a great opportunity to go in there and do the workload optimization work that then unlocks more value even when they land on instances in the cloud. But I don't think that the server/data center will be immune from some of that choppiness. It's just that it typically will lag and then there will be, in different parts of the world, enterprise will still be very bullish on modernizing their IT infrastructure, a lot of the return to office, all of the conference rooms want to be, collab rooms, video room. And so you see the networking business strong. You see again some of those IT budgets still holding as the macro environment continues to play out. We may see a little bit more movement in terms of, again, different parts of the world, what's going to land in the cloud. Or are you going to see the same growth rates that they've had during COVID? Probably not. But overall, I see there's still build out, maybe a little bit more choppy in the near to medium term.

Got it. I had 1 or 2 questions about Sapphire Rapids. We had NVIDIA earlier in the day. And Jensen was actually very positive on Sapphire Rapids. He said, we are ramping with that as part of the -- I think, their ddx platform. So there's certainly a lot of interest and excitement in the ecosystem about that platform. What is the state of progress with that because there are often media headlines, so it's shifted out, it's not going to be ready? And then there are also concerns about just the availability and the price of DDR5, et cetera. So give us a sense for where Sapphire Rapids is, right? And how much can it be exiting this year as an example?

Yes. I was in Jensen's session. So I'm glad to hear that he is saying what we're hearing from customers, which is Sapphire is a highly innovative, highly-anticipated, highly in-demand product when you look at the innovations that we're driving forward, not just in the CPU generation but for the entire industry, the transition to DDR5; PCIe Gen5; the compute XpressLink CXL; high-speed interconnect in addition to all the integrated accelerators that we have in Sapphire for AI acceleration, network packet processing, crypto acceleration, database -- data analytics. I mean all of that is built into this platform. But as you can imagine, with innovation comes a level of complexity. And we see all the customers, OEMs, ODMs, the hyperscalers, strong demand in that platform validation with us, and we delivered the initial SKUs in Q1 that we had planned. At this point, we are building in more platform and product validation time. So we see Sapphire, the ramp being later in the year than what we had originally forecasted but the demand is still very high. And not all customers are going to move to Sapphire in 1 step, where you have high-performance computing workloads and platforms. I mean clearly, NVIDIA is both a competitor on one side but a very important customer for us as well because every one of those DGF platforms needs a head node and Sapphire is a beautiful product that complements their overall platform. So they see the TCO performance improvement, perk from TCO that they get from Sapphire and they will lean into those transitions and the DDR5 premium. But other customers will look for strong value, strong compute capability products that have already been qualified in terms of Ice Lake. And so Ice Lake continues to grow. We had record revenue in Q4. We had another record revenue in volume in Q1. And Ice Lake for -- certainly for 2022 will be the highest volume product as we ramp Sapphire later in this year and then, of course, throughout '23. And I had a show-and-tell, maybe one of my colleagues can grab it out of my bag because I can show you Sapphire, it's beautiful. It's in the outside zipper.

This will be a first for our conference, actually.

Yes, if you have it, yes, that would be great. Thank you. I just forgot to grab it. Yes. So this is Sapphire Rapids, which is a package, 56-core beautiful, high-performing built-ins, DDR5, PCIe Gen5, CXL, AI acceleration, data movement accelerators, network acceleration and encryption. And this is the thing that we are working so hard to get out the door.

Sandra, do you think this product can bring you one step closer to kind of reversing the share right ship story? Your competitor has an Analyst Day later this week. And I imagine they will talk about, right, their next-generation platform. So do you think Sapphire kind of helps you reverse that or will it really take the next 1 or 2 generations?

Sapphire will be leadership in the market, absolutely, but we would have liked more of that gap, more of that leadership window for our customers in terms of when we'd originally forecasted the product to be out and ramping in high volume. But because of the additional platform validation that we're doing, that window is a bit shorter. So it will be leadership, depends on where the competition lands. We know that Emerald Rapids, which follows Sapphire second half of next year, a nice performance bump in terms of the memory, the networking and the overall performance in the same platform, socket-compatible. So a much easier upgrade for our customers and, frankly, for them as well, an ability for them to get a bigger return on the investment in the platform and all those innovations. So Sapphire will be out this year, Emerald in '23. And then, of course, Granite and Sierra are firmly in '24. Both that P-core, performance core, and the E-core, efficiency core, that our customers are valuing also in the same platform to give them that flexibility and that ROI for the new platform between, again, the P-core swim and the E-core swim lane. So as we look out, what we said at Analyst Day still holds, which is we are going to be in the mid- to high digits in terms of growth in the near term. And then in the outer years of the window, the '24, '25, '26, we absolutely get back to that growing at market. And really, of course, we have a goal to exceed that, of course.

Dave, back to you. How -- so the overall mix is dynamic, right? Consumer is facing headwinds. The data center is coming out with these new platforms. How do you manage capacity with this kind of a backdrop? Is it that fungible? I imagine it's -- is it challenging? Is it easy? How do you manage that?

It's not easy. We do have a great operations team that I think really is a differentiator, quite honestly, in the marketplace. We're looking at volume every -- certainly every month, sometimes even weekly on certain situations, and adjusting our outlook. It tends to be a multiyear thing. You can obviously turn a fab back and forth in a month. So we have to look at it over a multiyear period and look at what we need in terms of quarters of each node. And we're constantly adjusting that and rebalancing that based on the outlook that everyone has. But I mean, you take a business like the DC/AI business and you have to believe it's going to be a high growing business, right? I mean it might be lower growth than the market early on because of our need to get back to dominance. But the -- over time, it's going to start growing, as Sandra said, at the market level so that we kind of build that into the assumptions over the long term.

Got it. Makes sense. And then one last thing on the China lockdowns that I should have asked in the prior section, which is the impact. You think that is now getting to be at a stable level? Or is that still -- I think you used the word noisy, so we are still not out of the...

I think that one is, clearly, we are starting to recover. It's going slower than we had originally anticipated but we are starting to see things unlock. It will still take some time. I doubt that's a significant headwind out into the second half. I doubt. I think more likely, the other areas are going to be more of the headwind that we'll face. But again, anything can happen, right? Another lockdown occurs somewhere else that could obviously impact the supply chain. I would go back to the operations, I've got to get a bonus from them. But the operations team has done a very good job flexing the situation to try to overcome some of the challenges. And it could have been a lot worse. And they've done, I think, a very good job overcoming a lot of the issues.

Yes. Actually, they've also been really such a powerful asset even as customers had wanted to adjust their mix of products, again based on market conditions. That's true on the client business. It's certainly true in the server business where, well outside of normal lead times in Q4, we were able to just say yes to customers and shift the production line to make sure that hundreds of thousands of units, we're able to ship even though, again, it wasn't during normal lead times. We lead into that again in Q1 in terms of just saying yes to customers. So we're in really good shape from a manufacturing perspective. We're in great shape from a process technology perspective. One thing I didn't mention on Sapphire is it's on our 7-nanometer node. And so the process is quite healthy. In fact, Alder Lake, which is our client product ramped 15 million units. I think we announced at Q1 earnings, which makes that the fastest ramping -- one of the fastest-ramping client products in almost a decade. So the process is healthy. The capacity picture is good. But we've got some of these other issues that we're dealing with and customers on that issue still working through that.

All right. One thing on the foundry strategy. How much of that is kind of accretive to the model? And how much of it is a distraction to the core business?

So I think I'd step it up a level and say, first of all, there's a need out there for foundry that is more diversified than is -- that is a great reason for wanting to see a more distributed, more global supply chain. So there's a need out there. And I actually didn't mention it in the big surprises, but to me, that has been a big surprise, is how engaged we are with customers that I really wouldn't have expected we'd have that level of engagement in or with that really gives me like a lot of optimism around this business. We still got to execute on laying out the product offerings in that space, wafer offerings in that space. But I feel really confident we've got the demand in place. As it relates to margins, obviously, the margins of the foundry business are more like 50, low 50s percent versus, ideally, when we get the IDM business back to a better place, it's probably going to be approaching 60%, if not 60%. So it's dilutive in that regard, but it should come at a fairly high growth rate. And if you look at the valuation that we're going after, I mean, it's like a $500 billion, $600 billion valuation that we're going after, that's growing, that we can grow into that market, and there's demand. It just seems like too good an opportunity in terms of ROI for investors not to go after it. And then lastly, I would just say there's a great kind of synergy between the IDM business and what's required there. And without disrupting or, in any way, impacting other companies, IT and whatnot, but just in terms of the base investments that you have to make to drive both a foundry business and an IDM business, that we can get a lot of synergies there. So while the margins on the growth side might be dilutive, on the operating margin side, when you model it out at levels where they're really at their run rate, it's actually fairly accretive to the operating margins.

Got it. So one other thing, Dave, that if I go back to the Analyst Day, the assumption was the PC and the client business would kind of stay flattish, plus/minus, right, over the next several years. Let's say, hypothetically, that the PC market declines 5%, 6%, 7%, a year for the next few years just because we are at a cyclical high. How does that influence your foundry strategy? Does it create more incremental capacity that you can then allocate to something else? Like what is that interplay between the PC assumption, the actual being different than your assumption and your foundries?

I mean that's a great thing. We have a diversified model. We also have graphics and networking. So all of those things, we can play off each other based on which ones are driving higher demand. Some things will be lower than expected. Some things will be higher than expected. We'll be able to manage our volume within that envelope. And you're right, as more capacity is freed up for foundry, that's certainly a big opportunity there. Now having said that, my boss would kick me in the behind if I didn't say that we don't believe -- subscribe to that idea on the client side. We do think that this market has shifted up to a new level of demand because of all this work-from-home, education online and so forth. So we do see this as a more permanent model for the future. Now could it -- there could be machination, obviously, quarter-to-quarter or what have you, but we do believe that this is more like a well into the 300 million unit market as opposed to something that's going to gravitate back down to 250 million or so.

Got it. And just the last one there is we often see headlines that Intel is working closer with Asian foundry is that you might be the lead customer, right, for the next-gen process and so forth. I know Intel hasn't itself made that specific announcement per se. But how should investors reconcile Intel wanting to have its own foundry business at working versus actually outsourcing more of its product to an Asian foundry? How do we reconcile it?

No, it's a balance. I mean we kind of look down at the volume we need for the IDM business, most of which we want to manage with our own infrastructure. And we also want to grow this foundry business in a manner in which drives good return to shareholders. But we do want to keep some flexibility for one reason or other. Either the node doesn't make sense for us or we want to have some headroom that we're not putting the investment in that we can flex if we're uncertain on the demand side or what have you. And so having a foundry relationship allows us to create the maximum flexibility to take advantage of the assets that we haven't invested in but also give us headroom and flexibility when we need it to meet customers' requirements.

Yes. And I was going to say that the other thing that, that foundry model, both internal capability but external foundry and strong relationships there, they're important relationships for us as we continue to disaggregate our architecture, less monolithic, more tile architecture. Not everything benefits from being on leading node, as you well know. And so there's no reason to take on the time, the cost, the risk of porting all of those IPs over. And if we can continue to leverage foundries for N minus 1, N minus 2 in the FPGA portfolio way, way, long-tail nodes that do just fine, there's no reason that we wouldn't use that foundry. And increasingly, that chiplet tile architecture, the disaggregation, the modular building blocks just gave us more flexibility and optionality in terms of what we build in-house, and then where we partner with the foundry providers, which are, again, important in terms of the overall ecosystem to us as well.

Got it. Sandra, I had 2 questions for you on the AI business and just kind of the broader XPU strategy. So on the AI side, you have 1 competitor in the market, where the product side is, look, I'm going to invest in the GPU forward/backward. This is the product, right? And then there's another strategy which has a lot more choice, but sometimes choice is not always a good thing that I can do a lot with Xeon. I can do a lot with accelerators and there's also a GPU product. So how do you drive that right balance between focus and enough customer choice?

Well, we do see that the growth of compute is coming from, again, the need to compute and process all of the data and the data is heterogeneous, right? There's structured, there's unstructured. There's dense data, there's sparse data. And so AI, in particular, will play an important role. But it won't be one-size-fits-all. If you have a portfolio of applications that you're running in your environment, then having a CPU with integrated accelerators with the tool chain and an ecosystem that is ubiquitous is your best answer. If you are running deep learning training models, those small to medium complexity or classic machine learning, actually integrated acceleration on Xeon did quite nicely. If you're doing deep learning training, you have a couple of options. The GPU is primarily the engine that drives a lot of that today. We have a GPU option as well, which Raj is building in his accelerated graphics organization. And that does quite well if you have a combination of HPC, AI and graphics workloads. But if you're doing deep learning training only or 100% the large -- enjoy the time, then that's where an AI accelerator like a Gaudi, right, Gaudi1 that landed on, of course, being offered by AWS in their public cloud as a DL1 instance, where they're positioning it as 40% better price performance because, again, it's a dedicated accelerator, very, very tight in terms of the architecture, the design. It doesn't have to do a lot of different things. It has to do that one thing really, really well. And then, of course, on the inference side, that build-out of the edge, accelerated by 5G, whether you're in a centralized data center or you're distributed computing, that lends itself again, Xeon has a nice footprint there. The unifying theme for all of that really comes back to software, right? And so that heterogeneity and the fact that you want to be able to move workloads is not even something that's certainly a data scientist or a subject matter expert in biology or genomics or retail that they even really care about. We have to make those underlying architecture somewhat abstracted and we have to provide that software stack so that it's easier to consume. And the way that it becomes easy to consume, of course, is by landing in all of the standard frameworks and building out the models or being in the standard models that all of those developers will come looking for those innovations, and that will be in TensorFlow that will be in PyTorch. That will be really demonstrating performance against the natural language processing models, the computer vision models ResNet-50 showing performance. So for us, the investment that we're making in software is probably even more important than the underlying architecture. But it won't be one-size-fits-all. And you see the competition doing the same thing, which is they say, "Well, you know what, I don't think I'm going to be able to address all of that compute with just 1 architecture."

Got it. And one other one, Sandra, for you is ARM as a potential threat in the server market. I know ARM has tried many times before and not succeeded. Do you think it's different this time, that now that 1 established player is putting a lot of their weight and their established kind of software and developer resources behind it, there are some other start-ups, right, et cetera. Do you think that there is more renewed enthusiasm behind the ARM ecosystem? Or it's still too early to see them being a meaningful player over the next few years?

Well, I think that the share that ARM occupies in the server market today is still quite small but we watch it very closely. And for sure, the CSPs are looking at what can ARM deliver that they may not want or need on the type of performance core, the Rapids portfolio, Sapphire, Emerald, Granite that we would have. But that is exactly where we've targeted the efficiency core and the Sierra Forest swim lane because a lot of those cloud customers are looking at efficiency core type of workloads, really don't want all of those additional features. They just want high-density throughput, single-threaded performance and they want lots and lots of cores for some of the workloads, right? For other workloads, they care a lot about the performance per CPU and all of the capability that you get in the performance core. So it isn't -- it's going to be largely driven by the workloads, and we see customers wanting both types of CPUs around the portfolio. And of course, the ubiquity, the consistency of the software, the tool chain, the contract that we have with the customers really benefit our customers in terms of just knowing that the software out there will run first and best on Intel architecture. But there is one thing that we keep an eye on because some customers, they want the maximum performance per core count. They actually don't want more cores when you're looking at on-prem licensing models, right? So when you talk about VMware licenses or in-memory database licenses, they actually want -- I want the best performance, but I actually want to cap the number of cores. So again, the positioning for both our performance core and efficient core, we're hearing from customers that they like having that optionality and the consistency of the architecture. And just as we've introduced in '24, landing on the same platform, leveraging Intel 3 process technology, which again, is very healthy. We're very happy with that decision. I'd originally targeted a Granite for Intel 4, that Intel 3 is a slight derivative. Intel 4 is with the client Meteor Lake product, but I get more performance of the higher -- a more dense library from Intel 3, and it's a derivative. It's not a big process change and so I get to leverage that process entitlement in that product.

Got it. In the minute-plus that we have, Dave, any final comments, closing thoughts?

So maybe I'll start. Sandra can certainly had -- the one thing I think that is hard to capture for investors, many times overlooked but it's almost always the most important thing, and I saw it at my prior company. And I tried to tell them, hey, there's a cultural difference here of like really sweating the details, having a view of where you want to go, setting objectives around that each year and cascading that down, so everyone knows their role and then tracking them to that. That was really what Sanjay at Micron is amazing at. And I would just tell you that transformation is also happening at Intel. Surprisingly, that muscle, for whatever reason, atrophied. And we didn't really do that. And so really, it hasn't been until just recently, ever since Pat joined, that they really started this. And really, the only permanent thing you can do in a company is really change culture, right? Because everything else changes. Technology is moving at light speed all the time. And so the thing that's permanent is the cultural change you make. And I would say this is the cultural change, like I saw at my last shop, of like everybody now understanding their role, understanding what milestones need to take place over the course of 1 year or 2 years, 5 years, what have you, and execute. We don't execute all of them, I mean, with their misses, of course, but there's accountability and there's course correction and we go at it harder again the next time. So that would be the only thing that I would just leave investors with the mindset of is that's what's going on at Intel, and probably is the most underappreciated, I think, aspect of the transformation.

Yes. People and culture, I mean it is such a differentiated competitive advantage for the long term. And I will say, having been at Intel 21 years, we have an irrational love for Intel. And even as Dave points out, we also -- we just love our customers. I mean it's really -- they're so innovative. They're so -- just the trust, the collaboration, the ability to work through the good times and the tough times, we're leaning into customer obsession, disciplined execution, getting back to that level of predictable cadence that they expect from us. And we really want to delight them and to grow with them going forward.

Absolutely. Terrific. Thank you so much, Sandra. Thank you so much, Dave. Really appreciate your time today.

Thank you.

Take care.