Intel Corporation (INTC) Earnings Call Transcript & Summary

Good afternoon, everyone. Thanks for coming. I'm Chris Danely, your friendly neighborhood semiconductor analyst here at Citigroup. Next up a company that needs no introduction, Intel and a CFO that needs no introduction, Dave "The Hammer" Zinsner, is what we're calling you. In all sincerity, I've known Dave for quite some time. I literally think he's the best CFO we had in semis and I wrote as much when he first joined Intel. So before we get to me beating him up, I do have a couple of housekeeping items to talk about, I guess, this is some sort of a safe harbor statement. There we go. Before we begin, please note that today's discussion may contain forward-looking statements. Really? Great. I like forward-looking statements, that are subject to various risks and uncertainties and may reference non-GAAP financial measures. Please refer to Intel's most recent earnings release and annual report on Form 10-K and other filings with the SEC for more information on the risk factors that could result in actual results -- result in actual results, who wrote that? Result and actual results to differ materially and additional information on their, not our, non-GAAP financial measures, including reconciliations where appropriate to the corresponding GAAP financial measures.

All right. With that out of the way, let's get to the questions. So you guys had a very interesting conference call/print. You announced some changes. Maybe let's just hit the cost reduction plan, sort of broad picture, run through it again, where the cuts are going to come from? Could there be more? And then we'll -- I'm sure we'll go all over the place after that.

Right. Okay. So, yes, first of all, thanks for having me, Chris, appreciate it.

Thanks for coming, Dave.

Yes. So as we rolled out the new operating model where we divided the company into a foundry business and a products business and started to operationalize that but also manage the financials. One of the benefits to doing that was to begin to benchmark each of those entities to their corresponding competitors, to get a better sense of where we were overspending relative to other peers in the industry. And once we -- in the first quarter, we announced or showed our first quarter P&L in that format. And then I kicked off a process to start to do all of this benchmarking work. And internally and sometimes you hear Pat describe it this, we called it the clean sheet analysis. So it was basically saying, hey, we now have a fabless business at a particular size. What are the elements of how much finance do they need? How much HR, do they need? How much design engineering do they need? How much facilities expense should we expect? And we went through kind of line by line, through all of those. Now it was supposed to take the greater part of the year and we were going to finish it off by the end of the year. We'd have that now to work with and then we'd go forward into '25 and beyond, making all the necessary decisions to get to those benchmarks for both foundry and products. As it became clear that the back half of the year wasn't going to have this uplift that we were thinking which generally took the whole kind of 5-year horizon and shifted it downward, it became clear that, hey, we're going to need to action these and need to action them relatively quickly. And so we accelerated the end of it such that at least the analysis such that we would get there by the end of August, so -- or end of July, so basically the August earnings call that we'd be able to announce what we were going to find, all right and cut. And we looked at all 3 elements. We looked at cost of sales, we looked at operating expenses and we looked at capital expenditures. And in operating expenses, it was pretty easy. You kind of look at the quartiles of every competitor in terms of what they're spending and look at it relative to us. And we're fairly far out of the benchmarks. And so some of it was just, hey, finance needs to reduce their spending to be in the top quartile. HR needs to reduce their spending to be in the top quartile. Some of it was kind of overarching aspects or functions of the business. So we have repetitive processes in a number of different areas. Everybody wanted to hold their specific way of coming up with market share data. And so like relooking at all of those things to drive more efficiency. We have facilities, right? Pat mentioned this last week but we have facilities all over the world. Almost every one of them has a lab. We can't have 70 labs out there. I mean, that's just not efficient. So it's getting at all those things to drive more efficiency. And then there's just an aspect of it that is, we have a lot of layers within the organization and we need to have more of those layers, either actively in the design, actively in the production or actively in the sales of those products. The stuff that goes on behind the scenes, we need to have less of that, that activity kind of off the field versus on the field mix. And then lastly, we're just looking through the product portfolio and going kind of product by product and looking at areas where we don't see good profitability and we don't have good confidence around our ability to drive profitability in those individual product lines and make reductions in those areas. And what we came to is that, hey, we think we can get the OpEx down in light of all of that, down to kind of the $17.5 billion range for next year and then get it down even further in the following year. And I feel, basically, at this point, we're largely through a lot of it. We should be mostly through all of the cuts by the time we're announcing earnings this quarter. And there'll be some stuff that kind of straggles on but most of the decisions will have been actioned and communicated and we wanted to do this quickly. On the cost of sales side, it's kind of a similar thing. There's a function within cost of sales that is part of process technology development. And there are areas to be more efficient there and we will -- that's how we're going to get the $1 billion cut to next year's cost of sales, is through attacking that area because we're looking at process technology kind of holistically, some of it gets allocated to cost, some of it gets allocated to operating expenses. And then lastly, there's a CapEx part. And I look at CapEx in kind of 3 buckets. One is what do we need to invest in to keep the process technology moving forward. We're not going to do anything to affect that. Obviously, that's the [indiscernible].

Core Capex, I guess, what you call that.

What's that, core?

CapEx.

Called core CapEx, maintenance CapEx, whatever. Then there's an amount that we are going to want to invest in to be a bit, to have some white space. And we've dialed back how much white space we think we need but we still need to have white space because the problem with white spaces or call it shelf space, is that if you don't have it and suddenly decide you need it, it's 5 years before you're going to get it now because it takes so long to build that out. So we want to make sure that we're still kind of leaning ahead on that space. And then the rest of it is capacity. And we can modulate capacity based on demand. And obviously, with the different -- the change in the demand profile, we saw an opportunity to reduce our spending. And I wouldn't say it was easy but there was a fair amount of low-hanging fruit in that capital spend to be able to adjust it. One of the things that we did that will benefit us in CapEx will also benefit us on the spending side in the P&L is, we're going to kind of skip over productizing [ 28 ]. That's another example of something that we'll do. So 18A which is our most advanced process now had its 1.0 PDK in July, looking really good. Pat mentioned last week that we'll have Panther Lake out in production on 18A by the end of the year. At this point now, having the 20A was a nice to have and relatively expensive. So we figure we'll spend or save about $0.5 billion just from that activity between some CapEx and some spending in the P&L to get us there. So no stone will be left unturned as we kind of work our way through hitting these targets that we talked about.

Okay. Just a few questions to dig into on that. I think Pat talked about this at one of our dastardly competitors' conferences last week. Would you say that -- so -- and I think you alluded to it, would you say that the main catalyst for the cost cuts was the second half sort of plan being lowered, or was there something else there? So I'm just wondering but turn on the light [ holder ].

I would call it, yes, I would call it, it was all -- we were going down the path of this no matter what. We understood we were overspending. It didn't make sense. When we provided our 60-40 target model by 2030, it implies 20% of revenue for OpEx and we're well in excess of that as we are -- as we look today. So we definitely needed to bring that down. What we did though is accelerate it. So rather than just meander it out over the course of the year and into next year, we decided we're going to commit to something and go action a lot of these things pretty quickly. And so that's kind of where we are.

Yes. And then, I guess, an operational question. So when you announced the cuts, thousands of people and this much money and stuff like that, how do you guys sort of come up with that number because it sounds like most of the cuts will be done by the end of the quarter. Do you just figure, okay, looks like we're going to get rid of some of these people or this much and that much and then we'll figure out the nitty gritty later or was pretty much most of the specific cuts already done in that number? I'm just kind of curious because this kind of...

Yes. I mean because we had done all this analysis, this blue sheet analysis, we kind of understood function by function, what the spend should be and what the people count should be in each one of those areas. And so -- and that's the majority of it. The area that we still needed to work through is kind of portfolio activities and that does affect specific groups of, or specific teams within the company. But we'll largely actually be through that by -- again, by the time we announce earnings in the third quarter.

And do you feel like the plan you have is fine for now? Or could there be additional cuts -- was this the first cut of the cuts? Or could there be anything else?

I think this puts us in a good place in terms of benchmarking. I think we will make the necessary decisions around businesses that we think we don't have high confidence around performing. So I would say never say never but largely, this should be it. The only thing is that there are some things that have a longer period of time between the time you make the decision and the time you actually can make the action or do the actions necessary to hit the target. And so there will be a few things that's -- kind of drag into '25. And that's why the savings that we talked about for '25, which is to get down to the $17.5 billion OpEx, we said, hey, by the end of the year, we'll actually be exiting at a lower rate than that. And so we could see a lower number for '26 and that's because some of the things kind of straggle into the '25 year.

And so we'll definitely see the -- does that mean we'll see the bulk of the impact in Q4 or like Q4 and then first half of next year as far as the bulk of the benefits.

Yes. I think by the time -- we actually -- there are some things that are around notification, so for -- by the time it actually gets action we'll see a little bit of savings in the fourth quarter and probably most of the savings you'll really see in '25. The other thing that will happen is, in the third quarter, obviously, some of this stuff is going to come with restructuring or onetime charges associated with it. That will mainly show up in the third quarter. I could see some of it straggling into the Q4 and maybe into the Q1. But for the most part, I'd say most of it will get kind of done in Q3. The one other kind of nuance to this is there are some onetime charges that we clearly can call restructuring and clearly pro forma out of the P&L. And we alluded to this in the earnings call, I'm not sure if you picked it up. But there are some of these charges that we have traditionally not pro forma-ed, even if they're big and onetime of nature. So there may be some things that show up in our non-GAAP numbers that will be kind of associated with some of these activities. And we'll identify them. We'll call them out to make sure that, that's clear. And then we'll make the modeling a little funky for the next couple of quarters.

Okay. And it seems like you broke out the cost cuts in 2, there's definitely going to be some OpEx. Sounds like it's a little more towards R&D than SG&A, certainly some product lines and CapEx. Would you say those are the 3 biggest? And then where do you expect to save more on? Would it be on OpEx or gross margin or CapEx? Or are they all sort of equal?

On the cuts?

Yes.

Yes. Mostly, I would say, hard to say, you've got to define what's the biggest. I mean we're bringing CapEx down by more than $10 billion next year. So and I guess...

I was thinking that would be the biggest...

It's the biggest reduction. OpEx is, depending on what your model suggested next year would be somewhere in the kind of $3.5 billion to $4.5 billion reduction off of whatever you thought next year's OpEx would be and then cost of sales is the $1 billion.

Yes. Okay, great. And then you guys talked about moving some manufacturing. I forget exactly which chip line to Ireland have resulted in a gross margin impact. Maybe just run us through that, sort of what happened? Where did the miscalculation occur? Is this something you typically do and it was earlier than expected? Or maybe just take us through that?

Yes. I think stepping back to the second quarter gross margins, there were 3 things that impacted gross margins. There was this shift, quick shift to Ireland. There was some -- well, kind of, reserves we took either for demand, or in some cases, we took reserves associated with businesses that were kind of end of life thing. And so all those things, we hadn't baked into the model for one reason or another and they all kind of hit us at once. And I would say, we generally have a few things that go this way for us. But usually, there are some things that offset it. And so you don't really notice that we just -- it was just an unlucky quarter for us where we had things that went against us and nothing really hit us to the positive. And so it kind of exacerbated this -- the gross margin situation for us. On the Ireland, specifically the Ireland move, typically what we have done, as we ramp up the volume to a pretty reasonable level in Oregon and then once it's kind of clicking along there, then we switch it over to a high-volume fab to really ramp it up to a production level. In this case, also trying to kind of be very smart around our CapEx. We had the opportunity to, hey, if we shift it faster, we can reduce the spend that we would ordinarily need to have in Oregon to expand that footprint. Now we have opened up space that they can just slot into for the advanced nodes. And so it was good in that regard. I think it saved us like $1 billion of cash flow to do it. And then the second benefit was we -- if you ramp the high-volume fab, the fab that you have identified as a high-volume fab quicker, you get to a better cost for those wafers quicker. Now you have to take some pain in the beginning. But over time, it's actually a good move. And so we went hard to shift it and we also pushed the team to sell Meteor Lake as hard as possible. And those 2 things combined drove a pretty significant gross margin pressure on us. If we missed it by I don't know, 300 or 400 [indiscernible] like 500 basis points, this was like 1/3 of it that we -- that drove the [indiscernible]. Now we kind of knew we were doing that in the quarter before. I would say it was a combination of -- or merely when you're pushing hard on something, it's usually not negative to you. But in this case, it was just because the -- it's the early ramp of the wafers that we just weren't kind of baking into the forecast. And then the second piece of it was with this new model that we have, where we have foundry rolling up a P&L and the products P&L, we kind of lost the storyline on this one. It was a good answer for products and we didn't kind of track how this would be actually negatively pressuring gross margin.

Growing pains, probably.

Yes, a little bit on the kind of the new reporting structure. And it's different because this is -- there's a massive amount of elimination when you combine these 2 businesses to create the total company. So it makes it a little noisy. And we've worked out those kinks but we probably should have had better line of sight to those gross margins when we guided the second quarter back in whenever it was, April.

Yes. Do you think that this is something that you'll start to do on a more normal basis? Or is this more of a one-off as far as the early move because this is going to be a longer-term game and so.

It is a good question and we're looking at it. And I think you'll see that there will be opportunities for us to be more aggressive in ramping our wafers in the high-volume fab earlier, to get the learnings earlier and to keep the capacity, the precious capacity in Oregon available to us to ramp the newer nodes early on. So you're likely to see us play around with that a little bit. And it is the way generally the industry works.

Okay. Let's shift gears a little bit but stay on the CapEx topic, CHIPS Act. So you guys have your grant allegedly in hand? How does the sort of CapEx cut fit in with CHIPS Act? Is CHIPS Act going to change? Are you already seeing some money flow in? How long is this going to take? Maybe just sort of run us through the expected benefits and time line of the CHIPS Act.

Okay. Yes. So let's start. There's 2 parts of CHIPS. There's -- one is the grants and the other is the investment tax credit. Now let me go into the CHIPS part and that down. There are kind of 3 components to getting awarded money for CHIPS. First was the, there was an application process that led to a term sheet and that's what we announced with the $8.5 billion. There was then a process to get a long form, so kind of an expanded term sheet essentially, that's just kind of, folks -- certainly, you know what a term sheet looks like. It's usually got like 4 sentences for every topic. That's a major topic. So it's got to be expanded upon. So we went through that process and we are complete through that process. Now we go through kind of the documentation of the awards in a really detailed fashion that will feel like a definitive agreement almost to some extent. And we're not there yet. So there's been no money sent to us yet on the CHIPS grants and it won't be until we have that and then there'll be some kind of process by which we go then say, okay, we've accomplished a milestone. Now let's get the grant money. So the likelihood is, we don't see actual cash coming in the door until later this year, is my guess. And that's the process that has kind of been built in by the CHIPS' office. Everyone is going through that same processes.

And do you think it will get paid out over 1 year or 5 [indiscernible].

So it's milestone based. It's over a few years that it gets out and -- but it's like specific milestones you hit this, you hit a -- you finished off a certain construction or implementation of equipment or you sign up a customer or you get to a certain wafer level. So there's a whole bunch of milestones that we have to hit. And every time you hit one of those milestones, you get a certain amount of the grant money.

Okay. And then the investment tax credit.

Okay. So then on the investment tax credit side, that one is a little bit more straightforward. There's essentially 2 points in which you can claim a credit. One is when the fab is kind of ready for tooling. And the second is, when the tools are in and it's production ready, at which time you file it on your tax return and then it kind of nets out in your taxes subsequently. We have already, in some cases, started to account for some of the investment tax credit because we know we've already earned it. There was something like, I don't know, last year, it was like $0.5 billion of credits that we -- that showed up in the balance sheet. We still got to collect that, the actual cash on that but that's already in flight. And so those will extend for some time as we kind of build out the project. And it's simpler in the sense that you invest $1 and you know kind of what kind of grant you're going to get, it's $0.25 on the $1 and so you get $0.25 through this process that I just outlined.

Okay. Does the restructuring change anything about the CHIPS Act as far as it concerns money, grants et cetera, et cetera. Okay.

Yes. It shouldn't.

And let's talk about the big news so far this year is, splitting of Intel into foundry and design. Look, what can we expect the company to look like in, let's say, 3 to 5 years? I mean, you guys have actually given like a 6-year time line. But we see 2 completely separate companies, separate stocks. Or will this be sort of 1 entity with 2 parts? Or what's the vision that you...

Well, I'll tell you what I can predict. What I can't predict is a lot of those things. But what I can predict is, we are going to create more separation between these 2 businesses. It's important for customers to see that separation. And I think it's -- it makes the whole system better, right? I mean the whole reason to split these off is, the foundry business starts to become very focused on cost and very focused on spending and very focused on making sure the utilization is at a maximum. That is not the way Intel ran the business. They didn't care necessarily about cost. It was all about getting the last wafer they could possibly get out. And if you were -- you had too much capacity, who cares, so be it. So it's creating that mindset within the business that's important. Same thing is true on the fabless side, their margins were mostly affected by loadings in the fab and that's all they ended up caring about. In reality, they should have been caring about how much silicon they're using, what their test times are, are they using the most effective packaging. Now they're focused on all those things. So we will get benefit. Part of what allows us to talk about the reductions we're talking about in terms of OpEx and the improvements in gross margins to ultimately get to the 60% gross margin by 2030 is because of this model that we're driving the business to. And we're going to put separate systems in place. I'm in the process of implementing ERP systems that upgrade significantly these old, funky ERP system, the ERP system we had. We will have a separate ERP system for products and a separate ERP system for foundry. And there will be lots of kind of firewalls in place to make sure that our foundry partners with products don't feel like information is getting shared back to foundry and that the IP that other companies are providing or leveraging within our foundry to develop their products aren't impacted by, or -- leading into the products company. And that's, that is our stated plan and it has been from the very beginning that we talked about this operating model. Where it goes from there is difficult to say. I'd just say our goal here is to drive maximum shareholder value. And that's the way we think about this every day. And I think you can expect us to continue to think about it in that light.

When we had you here in this same seat last year, you talked about the actual active benchmarking your foundry manufacturing versus other leading edge foundries, was already having tangible benefits. Can you just share with us now that we're 1 year past it, what has been like the better-than-expected part of the separation, besides obviously, waking up the TMG or engineering, the -- this is what TSMC does and we should probably figure it out.

Yes. I think we had a goal to -- I gave out like a stretch goal to -- so I'm tracking all the savings that we get from this new model and I have a team that looks at it on a regular basis and then reports it out. And we said we want -- I can't remember what the exact number last year but it was minimal. It was, I think, something like $200 million. We wanted to save $200 million in cash flow from the model already by the time we even implemented it. And this year, I said, okay, well, they got to $200 million, let's make it a big number. So I said, "I want $1 billion out of cash flow." Now it doesn't -- some of that got reinvested and so forth. But let's get $1 billion of cash flow improvement just from operating in this new model. And they are so close to actually hitting this target now.

And that was a stretched target.

It was a stretched target. And it's just, quite honestly, one of them was this kind of shift from Oregon to Ireland, was, one of the things they thought through as now just being their own business and thinking about their own profitability and how they could make their cash flows look better. They made that call. And so I think there -- and I can't remember exactly all of them but it all added up to about $1 billion of savings that we credited the organization with based on developing the new model. On the product side, we saw it kind of right way. I mean, the amount of focus on test times, which had -- I mean, I don't think anybody even talked about test times, much before that has been significant. It's all their focus. And in some ways, it's in this interesting dynamic or all of a sudden, the products business is like, I got to cut my test times down. And that actually is revenue and then so there's a little bit of go back and forth by both of those businesses because they're now kind of somewhat self-optimizing a little bit but it's the healthy thing to do over time. The one other thing that happened almost immediately was the products business would always just push a whole bunch of what we call internally [ hot bots ] but it was essentially all these lots of wafers going through that suboptimize the flow of the fab. And...

If it didn't hurt their P&L, why would they care?

Yes, so why do they care? So they just did it. And it was such a fractional thing to them, ultimately and it got spread over all the businesses that any business never felt like their specific action affecting their specific P&L. As soon as it became a charge, like, hey, if you're going to do a [ hot bot ], I'm going to charge you. It went down like 90%.

Amazing how that happens.

Certainly didn't need [ hot bots ], as you mentioned.

So just to touch on the foundry business, a big push for you guys. When can we expect, I guess, material foundry revenue for products that are outside Intel?

Yes. Yes. Okay. So to start with -- and this was probably the biggest surprise, so the positive that we got with the foundry business is that our packaging is actually very competitive, quite interesting to customers and we're getting a lot of interest there. And so we had really kind of thought, okay, foundry is going to be kind of like 0. And then in 2027, it's going to be something because we were thinking it was all going to be about wafers. As it turns out, that's not going to be the case. We are going to win packaging customers. We'll have packaging revenue.in the back half of this year. We had a little bit in 2Q. It will be bigger next year. And so this will be a nice kind of growth business for us over the course of the next few years as we wait for the wafer revenue really. And the beauty of it beyond just getting revenue early is that it gives us an opportunity to kind of on ramp customers. They get to test us out in packaging and then we can pull them in on the wafer side over time. And so I think it's just a great opportunity for us to be successful on the foundry side. The wafer side, it's going to take some time. We have 12 RFQs right now that we're dealing with on 18A. I would imagine a lot of them will sort themselves out by the end of this year and probably maybe some revenue in '26 but probably '27 is the year where we see some meaningful revenue from that set of customers. And just maybe one other thing, sorry to jump on you. But we just -- in July, we talked about this on the earnings call but just to reiterate it, July 22 or so, we launched the 1.0 PDK for 18A. So what that does is, it says, okay, are your yields at an acceptable level? Or is your performance at a respectable level and kind of benchmarks that relative to where it should be in terms of its maturity. And we hit that point on July 22. And so then that allows the ecosystem partners to start building out the ecosystem to support those. And now we have -- now you start to see the RPUs come in, okay, now that you got something, this is real. We can start using our own silicon on the process, let's evaluate it. Will we win all of them? Probably not but I think we will win a meaningful amount of them. And we don't need a whole lot of wafers, quite honestly, to make this work for us from an ROI perspective.

Leads to my next series of questions. So you expect to have a material packaging or advanced packaging revenue next year on the foundry side, so that will impact it? And begs the question, what will the margins and like the margin profile look like of that business? Will it be accretive, dilutive? We don't know yet.

Accretive to where it is now because right now those margins are negative. But I would say right now, the margins we're getting on these -- this business is roughly what we hope would be the average of the overall business. And we said, we're roughly trying to get this business to be a 40-30 kind of business, 40% gross margins, 30% operating margins over time and that's kind of a goal by 2030. I'd say this is like supports that.

And so how will the margins go up? Is it just more volume, better yields? Any other [indiscernible]

What's keeping the margins under pressure right now for them is, we've got a whole lot of start-up costs because we're running so many nodes through it one time. It's really putting pressure on the margins. And then our cost structure in pre-EUV wafers is not great. And the price is not that great. And when you look at the progression from those pre-EUV nodes up through 18A, it's like a 3x increase in pricing relative to cost. And so that alone drives the gross margin significantly. So you take those 2 things that has a dramatic impact on the gross margins. I think over time though, of course, scale is going to matter. I mean, it needs to have some scale to ultimately get to the 40%, 30%. But a lot of it is really within our control.

Yes. Is the advanced packaging you're offering, is that Foveros? Or is that something else?

It's Foveros. It's EMIB, both.

Okay. Great. So getting back to, I guess, your core CPU manufacturing. It seems like 5 nodes in 4 years, all on track, looking pretty good, hit all the milestones, products out.

Yes. I thought -- I think Pat probably would have liked to have beaten the 5 nodes in 4 years. We're going to hit it right around the 4-year mark for 18A. I think [indiscernible] others, he would have liked to see that come in earlier. But we hit what we committed and I think that's pretty significant, given how 5 nodes in 4 years, when I got here, sounded like such a crazy idea to begin with. We will have Panther Lake, which is our first 18A product in production by the end of the year, end of this year. All the...

Like, out on the shelves?

Just ramping in production in the factory. All the early silicon that we've been testing, both in Clearwater Forest and in Panther Lake, those products look really good. So I'm pretty happy with where we are in terms of the process, both from a foundry perspective but also from a products perspective because it's important that those products were going to be successful in the market and looks good. Sierra Forest, Granite Rapids, we should -- we said we would have it qualified in the back half of the year, looking good. I think that is a meaningful step for us competitively. Clearwater Forest will be on 18A, that will be next year coming out. And then we have Diamond Rapids also out, as the next generation.

Yes. Great. Have to check all those boxes. So manufacturing is on track. We've gone out and said, at some point in the first half of next year, you guys could catch up to the competitor. Do you share that assessment? And when will we start to see some tangible market share gains?

Yes. So on the client side, and feel free to say no, Dave, you're crazy -- on the client side, I would just say we're in a great place in terms of market share. And with Lunar Lake that we just launched, formally launched yesterday, I mean this has 120 platform TOPS. They've run it on -- continuously run it. And with applications running and it had 20 hours of battery life. This is going to be a terrific part. I think it answers any of the outstanding competitive questions around kind of ARM-based solutions in the client space. So probably not for me to make market share predictions but I would just say we feel like we're in a great competitive place in client and that was the only lingering question and I think we've answered it. And after Lunar Lake, there's Arrow Lake and Panther Lake and there's going to be a steady kind of success [indiscernible] set of products coming out. So that's in a good place. Where we still haven't completely gotten the business to a good place is in the data center side of CPU. I think Granite is a meaningful step forward for us in terms of making us competitive. Diamond would definitely put us in a good place competitively. And so it's just important that we kind of work our way through the road map to get us to where we want to be in terms of competitive position. We'll see.

Speaking of client, what's your take on the overall PC market? Is it still healthy? Is it plateauing? We've had other companies talk about that -- we saw a very strong PC market in the first half. And now there's -- any headwinds, flattening out, slowing down? Some folks are saying it's fine. What's your take on the PC market?

I would call it fine, all though we had thought we were going to see a pretty meaningful recovery as part of our -- part of our back half recovery thesis was that client would have a step up in the back half of the year. The AI part of -- the AI PC part of it has done well but it's mostly kind of cannibalized the non-AI PCs. And so that business is up a little bit from last year in terms of TAM, up a little bit on a units basis versus last year but it's not a significant improvement. And the way we're thinking about like funding the business, we're going to kind of be pretty conservative around our views of this -- of the client space, relatively conservative growth for next year and the following year. But it does have an opportunity to outperform. There is the Windows 10 and the service that should drive refresh. There is -- the AIDC might create an innovation cycle that could drive a more meaningful refresh. And there is a whole bunch of old PCs out there that were procured in COVID and before that do need to be refreshed over time. But I think for us, we're just trying to be conservative around our modeling there. And we'll be pleased for the upside if we're wrong.

Yes. And so I think you guys talked about a little bit of an inventory build in the PC channel in Q2 that's going to get worked down in Q3 and then it will take about a quarter than Q4 should [indiscernible].

Yes. Exactly.

Great. All right. I think that's all we have time for. All right. Thanks, Dave. Thanks, everyone.