Texas Instruments Incorporated (NASDAQ:TXN) BofA Securities 2024 Global Technology Conference June 5, 2024 6:20 PM ET
Company Participants
Rafael Lizardi - CFO & SVP, Finance & Operations
Mike Beckman - IR
Conference Call Participants
Vivek Arya - Bank of America
Vivek Arya
Delighted and honored to have the team from Texas Instruments join us. Rafael Lizardi, the Chief Financial Officer; and Mike Beckman from the Investor Relations team. Very delighted that…
Rafael Lizardi
Good to be here. Thank you very much.
Question-and-Answer Session
Q - Vivek Arya
Thank you. What I'll do is, start with my questions, but please feel free to raise your hand if you had something that you would like to bring up. But maybe, Rafael, let me just kick it off at the high level perhaps give us the state of the union now we are almost kind of middle of the year. How has the demand and the inventory situation developed so far this year?
Rafael Lizardi
Yeah. Well, let me step back and just remind everyone of our plans, our investment plans that are -- that we put in place and where we are on those and why we're doing that. First, why we're putting these plans in place -- we put these plans in place and we're executing to them is one, the secular growth in semiconductors in analog and embedded and in particular in auto and industrial. And we all can see, I have a lot of examples of what's going in there and what's driving that with automation, electrification, safety that's driving that secular growth.
Second, our position in those markets is tremendous. So about three quarters of our revenue now is in auto and industrial. So we're playing in the best places. And then the last piece of that is this sector -- I'm sorry, the geopolitically dependable capacity that we're putting in place that our customers are clamoring for and that just puts us in a great place to have capacity at scale outside of China and Taiwan.
Vivek Arya
Okay. So maybe let's kind of pick up from that. From what you said, does it mean that the $5 billion per year CapEx plan that you have, is there any scenario in which it can change?
Rafael Lizardi
Let me first -- I'm going to answer that question directly. But first, let me walk you through what we're getting for that investment. I'm going to break it down in about four pieces, okay. First is RFAB2. RFAB2 we built that factory four or five years ago and we've been equipping it since then. We have about two years left of putting equipment in that factory. That's the factory that's going to be the workhorse of the next upcycle. Because that factory, if you go back to 2020, we actually let some money on the table.
We could have done more and if we had RFAB2 ready, we would have done more. So we don't want to miss out on that. So that's RFAB2. We're also RFAB2 is one where we're shutting down an old factory -- an older factory, 150 millimeter, 200 millimeter factory and we're moving those products to RFAB2. And as they move from 150 millimeter to 200 millimeter to 300 millimeters, they're a lot more cost-efficient, okay. So that's RFAB2.
RFAB1 is the factory that we bought from Micron in 2021 and we have been qualifying an equipment since they already had equipment, but we have we have to put additional equipment on that. Think of that factory as a transfer factory. We don't need growth to fill that factory. All we're doing is moving loadings that we've had and we continue to have to a large degree in TSMC and UMC and we're moving them internally.
When we do that, the trade-off, the cost trade-off is huge. We go from paying about $2,500 per wafer to essentially a $200 per wafer cost on a variable-cost basis because the factory is already there, right, the people are already there. That factory is in the middle of -- it has already qualified some processes. We still need to qualify additional processes. And then we have to go part by part and qualify this and send sample to customers. That's why it takes a while.
So let me tell you one more point on RFAB2 and Lehi 1. The ITC, the 25% ITC credit applies to those only expires in 2026. So that's another consideration that we have to take into account that we -- from that standpoint, we want to maximize the equipment that we put there through 2026 because anything that we put in '27 and beyond does not get a 25% credit. Okay. So that's for those two factors.
Now let me go to SM1 and 2, that's the new Sherman facility. At some point is going to have four fabs, one, two, three and four. For now, we're building two, one and two. One, it's going to be a full fab. In fact, it's almost done. It's going to start getting equipment in third quarter and then -- and by equipment, I mean a pilot line.
So it's essentially -- think of it as -- excuse me, 5% capacity worth that we have to put in place one of a kind tools so that we can have a pilot line and start churning our products and then we send samples to customers so they can qualify them. That's SM1. SM2, think of a bare concrete shelf, okay, that is just a shelf. That is just so that is there ready for a potential -- if we need it, we can then move very quickly to put a clean room and then equip in that.
Then the last piece of that is LFAB2. LFAB2 is going to be a three-level fab, a pretty large fab in Lehiin Utah, right next to LFAB1, but it's going to behave like an extension of LFAB1, meaning that we don't have to qualify it separately. Once LFAB1 is qualified and a customer says, check, I'm going to take products from LFAB1, LFAB2 is just another floor of LFAB1, it's just an extension. We do our own -- install the tools, we're on simple qualification tool-by-tool and then we start ramping.
That we broke ground this year, but that can be done on a modular basis. Of course, equip the concrete you kind of put in place all at once, you don't have to -- but the clean room you can do more modularly, and of course, the equipment you can also do very incremental.
And then to answer your question, we're going to spend $5 billion this year and next year to do those things that I described. For 2026, can we accomplish what I just described with less than $5 billion possibly, in fact, very likely, but we're in the process of understanding that. We'll give you more information of that overtime.
Vivek Arya
Okay. What is changing that view versus three months ago?
Rafael Lizardi
Time has passed and we have more information and time to assess where we want to go. And at the end of the day, look at big picture, right. We spent about $2.5 billion in '21, $2.5 billion in '22, $5 billion in '23, we're going to spend $5 billion this year, $5 billion next year. And then the remaining $5 billion that we just talked about in '26. You add that up, that's $25 billion. So could we potentially shave a billion or two from that and still accomplish everything that we want to accomplish that's a possibility. But in the big scheme of things is $1 billion or $2 billion out of $25 billion investment cycle.
Vivek Arya
I see. And then, Rafael, even longer-term, I think the plan was that you would have a few years at this $5 billion, and then the exit would be 10% to 15% kind of CapEx intensity, right. So that's more kind of top-line base. But even that 10% to 15% is much higher than what TI used to have, right. Like in the past, you were kind of low mid-single digit type of CapEx. So why is it so much more inflationary to put this kind of capacity now versus before?
Rafael Lizardi
I tell you what, forget about that 10% to 15%, frankly, because that is -- that 10% to 15% was a function of revenue growth. So if revenue was headed towards a very high number, then of course, we would have to put additional CapEx in order to equip those factories. Just as I mentioned, SM2, we would then have to equip LFAB2. We would then have to equip, but that depends on revenue growth. So if revenue growth is not there, then we don't have to spend anywhere near that.
Vivek Arya
Okay. But I mean, doesn't 10% to 15% automatically take care of what the revenue level is?
Rafael Lizardi
Well, depends on the revenue. You tell me the revenue and I tell you what I'm saying it precisely, but depending on that revenue growth. So for example, the 15% corresponded to a 10% CAGR from 2022 at $20 billion, you do 10% CAGR out to the end of the decade, that would have corresponded to a 15% CapEx at that point. If instead of that, we're at a 7% CAGR, that would have corresponded to a 10% CapEx intensity. If you're lower than that, then it's less than that.
Vivek Arya
Okay. So let's say we get past this inventory correction, do you think that your addressable opportunity is growing at a 7% CAGR or a 10% CAGR?
Rafael Lizardi
Let me put it this way. You go back to the peak-to-peak analysis of the last two semiconductor cycles. So 2014 to 2018, 2018 to 2022, what happened in those years. From 2014 to 2018, of course, that's four years, we grew at 21%, the peak-to-peak. From 2018 to 2022, that's another four years, we grew at 27%.
If you pass forward, just use that same math and get to potentially what could be the next peak in 2026, you can get to roughly $24 billion to $26 billion of revenue. And you also should consider that in 2022, we were constrained as far as how much we could -- we could have shipped. So we could have shipped more than the $20 billion. While we're talking about this, let me also comment on the fall-through to gross margins.
If you look at our fall-through gross margins in those peak-to-peak years, so 2014 to 2018, these are just in our financial statements. I'm not making any predictions. This is -- I'm just telling you what the actuals are. 2014 to 2018, that fall-through ex-depreciation was about 90%, was actually a little more than 90%.
And then you go 2018 to 2022, do the fall-through again ex-depreciation, it was also about 90%. So now, is it going to be 90% in the future, not necessarily. For example, part of the reason we moved so high was as we were doing less personal electronics and more auto and industrial. Now that we're 75% auto and industrial, there's less tailwind on that, but we should still do very well on that fall through.
Vivek Arya
Got it. I thought one of the reasons TI suggested kind of move towards the 10% growth model was that a lot more of the business is coming from auto and industrial. Like even in those years, your auto and industrial, right, definitely grew at a very nice kind of double-digit CAGR where some of the weakness and the volatility was on the consumer side, like for some of those years, a large smartphone customer was the largest customer for TI, right, not as much so, I would imagine now. So doesn't that mix-shift towards auto and industrial make it more likely that you can grow at that 10% base?
Rafael Lizardi
I tell you why…
Vivek Arya
Not talking about it the right way.
Rafael Lizardi
I prefer not to speculate on that, but what I'm confident -- I can confidently tell you is that if it grows 10%, we'll be ready. And that is what this plan gives us, okay. That is why it's so important that we complete these investments that we're making as we -- as I described RFAB2, Lehi 1, Lehi 2, SM1, the shell in SM2 because if in fact, we have a 10% CAGR or even a higher CAGR because of the factors that you just described, we don't want to leave any revenue on the table.
Vivek Arya
I see.
Rafael Lizardi
And we wouldn't. Just like I described, we would have tremendous optionality to put more equipment in SM1, more equipment in LFAB2, and even finish accelerate the clean room in SM2 and ramp that up as well.
Mike Beckman
And I'll just add that automotive and industrial, that is a factor in it. But that didn't happen by accident. That was a bias toward auto and industrial from an R&D perspective. And it's not just a one-year bias, it's been a longer-term bias. So as you pointed out, those are markets have grown at a faster rate than they overall. They're a higher percentage of our revenue today. But is that potentially something that's going to drive our growth faster over time, we want to be ready for it, and it's again, it’s not by accident.
Vivek Arya
All right. If more than half of the customers for automotive and industrial are outside the U.S., why do I need to have all of my capacity in the U.S.?
Rafael Lizardi
Because -- well, you don't need to have it all in the U.S., you want to have it where it's geopolitically dependable. But the U.S. is a great place to have fabs, okay. Now that we are at par with the CHIPS Act on the 25% ITC and we'll see what happens on the grants. We don't have anything to announce on that. But at some point, we should hear and we should have something.
But now that is at par on an even playing field, the U.S. is a great place to put fabs. Texas, in particular, we get tremendous advantages there. Of course, we have fabs there. It's easier to deploy additional fabs. Electricity cost is very favorable there. The access to talent is also really good. So I put Texas next to any country at this point in terms of building additional semiconductor fabs.
Vivek Arya
Got it. Absolutely. So I can understand that from a supply perspective, I'm trying to think that does it become -- is it a factor in terms of share gains? Because when we talk with a number of your peers, right, who have a hybrid strategy where they are sure many of them are still dependent on Taiwan, but TSMC is building out more fabs in Japan. So why isn't the capacity access they have in Japan give them that geo -- that dependable capacity at a much lower capital intensity than what TI is doing?
Rafael Lizardi
The difference is -- a good question. The difference is how many fabs is TSMC build in Japan, like one, right. How many are they building in Europe, like one. How many are they build in. So a customer that's looking for a check the box A, I have dependable capacity because I'm building -- I'm buying from ADI and they're sourcing out of -- they could source out of Japan, they could check a box on a PowerPoint slide.
But if something actually happens where the fabs are not-- the supply is not available coming out of China or Taiwan, all those customers are going to be asking for the same parts from the same fabs in Japan and Europe and other places. Now we don't have that problem and we're going to have those fabs in the United States, as I described earlier and all available with plenty of capacity.
Mike Beckman
At scale is essentially, what is required.
Vivek Arya
But have you -- Rafael actually seen that in practice, so I understand the concept, but if practically our customers saying, yes, I was planning to give a part to whoever else, but I would rather give it to TI because you have that fab capacity in the U.S.
Rafael Lizardi
It's an excellent question. CEOs of our customers will tell you that twice a day and three times on Sunday. But when it comes down to the designers actually putting in the part, you have to be competitive. You have to be competitive with price, you have to be competitive with the part that you're putting out, you have to have good deliveries inventory available. So it gets more tactical when it comes down to the actual designing and winning the sockets. But at the high level, the CEOs and the CP -- the Chief Purchasing Officers and their influencing on the -- on the designers, yes, we are seeing that.
Vivek Arya
I see.
Mike Beckman
With deadlines, dates, they want to get certain percentages too. I mean, they are actively trying to do this.
Rafael Lizardi
Which we're doing the same thing with our supply chain, okay. And we are largely diversified and dual triple source, but in places where we're not fully or not to the level that we won, we're doing the same thing. We're looking for other sources so that we have our own geopolitically dependable capacity when it comes to more compound, lead frames, wire, etc., for our various operations.
Vivek Arya
I see. At a recent conference, Haviv mentioned that there is a -- whether it's a plan, whether it's a trend line or something that can get Texas Instruments to potentially above this $12, right, per share and you manage your business for free cash flow per share. So, where does that $12 number come from? Is that a trend line, like what needs to happen for TI to hit those…
Rafael Lizardi
Okay. So go back to the peak-to-peak comment that I made earlier, okay. 2018 to 2022 or you can do 2014 to 2018, do the growth rate there. And by the way, it happens to be four years in between those, and four years from 2022 happens to be 2026. So it actually works out pretty well. Do that, do the fall-through at similar rates.
In fact, do it a little less, not don't use 90%, use 80% to 90%, pick 85%. Do that, model it away and it's not hard to get to a -- and the CapEx, as I said, I haven't given you a number, but it doesn't have to be $5 billion in CapEx. You do that math, you get to a very healthy free cash flow per share.
Vivek Arya
I see. Is that -- Is that how you are kind of putting all your CapEx and OpEx plans in place to get to that number conceptually at some ---
Rafael Lizardi
Conceptually. Yeah. And we want to be able to maximize the revenue opportunity. So as I -- if you account for the math that I just gave you, it doesn't even account for the fact that we -- our revenue was compressed in 2022, doesn't account for potential issues of geopolitical tensions that could give us even bigger opportunity. It doesn't account for the fact that auto and industrial could accelerate growth. This is why we want to be prepared to do even more than that.
Vivek Arya
I see. And does this change in CapEx plans, Rafael, does it change the depreciation schedule that you have given for the next several years?
Rafael Lizardi
At this point, I've only given depreciation for this year and for next year. I haven't given anything beyond that. So that is not changing because the CapEx for those years is not changed. But at some point, we'll give you '26 and beyond, or at least '26. We'll see how often we give it. And of course, that -- if we spend less CapEx, that's going to drive the increase in depreciation lower. And then we also have to take into account the grants whenever we hear about that and how that affects that. Clearly, it can only bring it down versus what it would have been.
Vivek Arya
Got it. And from what we have seen, right, many other companies who have gotten the CHIPS Act funding, it has been -- so of course, ITC you mentioned, right. So that is already part of the financials that you're reporting, right, so far. But what we have seen with others is about that 15%, right, or so broken out in terms of grants and loans. Is that sort of the structure that we should contemplate for TI?
Rafael Lizardi
Well, what you have seen so far is less than 15% on average we've seen roughly…
Vivek Arya
10%, 15% right?
Rafael Lizardi
Yeah. But it depends -- it depends on the timing of projects. So, we'll have to see what the Department of Commerce offers and where we land on that.
Vivek Arya
I see. And when we look at that denominator of that CapEx, is that of all of RFAB2 and LFAB and all the Sherman fabs and LFAB2, like what is that denominator that Department of Commerce could be potentially looking at and say, okay, I need to fund 6% of that?
Rafael Lizardi
What I'd tell you in there is, I think their focus is projects that are happening now through 2030. So they're focused on the short-term. So for example, Sherman 3 and 4, they're not even -- they're probably not going to be in the picture, right, because we're not talking about building any of those anytime soon.
Vivek Arya
I see. So if I look at CapEx, let's say, $5 billion towards $4 billion, right, somewhere 20 to 25, and is that kind of the rough map between now and 2030?
Rafael Lizardi
We will comment when we get the information. You're trying to ask many different questions on that, but I don't have any information to give on the grants.
Vivek Arya
Okay. And just the last thing, let's say you get a grant, does that also then impact your depreciation schedule? Because then it will be net CapEx. So even what you have given for this -- next year, could be…
Rafael Lizardi
That's right -- I doubt that it would affect next year because it would most likely be longer-term than that, but it would affect for sure 2026 and beyond versus what it would have been. That doesn't mean it's going to be lower than 2025. It may just not increase as much as they would have otherwise.
Vivek Arya
Right. Because where I was going with that is because that formula does affect right as much as you manage the business for free cash flow, we looked at gross margin, right, for sure.
Rafael Lizardi
So on that point, let me just give you a quick update on that. At the last earnings call, I think I told you depreciation for this year between $1.5 billion and $1.8 billion, but we're going to be at the lower end of that. For next year, $2.0 billion to $2.5 billion, we're also going to be at the lower end of that.
Vivek Arya
Got it. Okay. So, whenever you hear about the grant, does that kind of move that a little bit or that's more a…
Rafael Lizardi
It should not move 20 -- definitely not '24 and unlikely that it moves 2025. It will likely move 2026, which I haven't given an update on.
Vivek Arya
Got it. So that only changes. Okay. The second thing I wanted to ask you is back to the demand side. So you did guide June to grow, right, sequentially…
Rafael Lizardi
At midpoint. At our midpoint.
Vivek Arya
At the midpoint. Would you consider that seasonal? Would you consider it not like should we think June is now a normal quarter or you still think you are battling inventory headwind that the end-market has not really stabilized? So I'm just trying to understand where we are in the spectrum moving to normal. Are we there?
Rafael Lizardi
Mike, do you want to take a shot at that? [Multiple Speakers] Yeah, go ahead.
Mike Beckman
So I think it's important to remember that this cycle has had this end markets moving out of face with each other dynamic. So it's probably hard to answer the question directly of whether seasonal in the current guide. But I think we're starting to see this first-in, first-out phenomenon kind of playing out, right, because you saw personal electronics was the first to correct. It started seeing some sequential growth in the middle of last year.
Industrial had some sectors that began correcting in late 2022. You had a set of sectors that kind of hung in okay in '23 and then late '23 joined and also began to correct and everything further corrected the fourth quarter. But then last quarter and first, we saw mixed results there where some of those earlier sectors flat, some grew, some still down on the later sectors to correct down double-digits. So more mixed results there. And again kind of following that first-in, first-out, it's been what seems to have been so far materializing.
Automotive was the last to begin correcting for us. I think fourth quarter was down mid-single-digits. First quarter down also mid-single-digits. Obviously didn't fall off a cliff. Yeah, we'll see where it lands and obviously, I think the guide speaks for itself. But yeah, I think that's what we're seeing so far. I don't know if you have anything to add, Rafael.
Rafael Lizardi
No.
Vivek Arya
Okay. So automotive, do you think that is in a better state than industrial? Like if you had to think about where TI gets back to seasonal trends in the back half. Is it more likely to happen with industrial or more likely with automotive?
Rafael Lizardi
Well, what I would tell you, just as Haviv said last week at the Bernstein Conference, we expect a shallow dip in automotive.
Mike Beckman
And I think we want to be ready for whatever eventuality comes out of any of those markets from an inventory perspective. But obviously, automotive has its own unique growth trends underneath it. The secular growth there has obviously been very strong last several years. The EV transition is a piece of that. I also think there's an inventory component and that supply chain evolved a bit over the last several years. So how all that plays into what this could look like as Rafael said, probably has a little more of a shallow cycle to it, but we'll have to see.
Vivek Arya
Sorry, not to parse words. So have you seen a shallow dip in auto already or is that shallow dip yet to come?
Rafael Lizardi
Well, you've seen some of it in our first quarter results and our guidance, I think our midpoint embeds that. I mean auto and industrial are 75% of our revenue. So our midpoint reflects -- what happens to our midpoint is showing in our…
Vivek Arya
But are they both increasing?
Rafael Lizardi
Well, we're not going to comment on second quarter. I'm just saying our guidance and if you look specifically, our midpoint and second quarter embeds is consistent with what I just said.
Vivek Arya
Okay. Next thing, Rafael, there's a lot of debate about what does all this buildup of capacity in China mean, right, for -- from a competitive perspective for you. So first of all, how much is China domestic kind of demand as a percentage of your sales?
Rafael Lizardi
Yeah.
Vivek Arya
And have you seen any design-outs for TI based on the capacity that's being built up?
Rafael Lizardi
Yeah. No, good question. So let's talk about that. So roughly speaking, and we published this in our Ks and our Qs, right, about 20% of our revenue comes from companies that are headquartered in China, okay. In fact, as of the last quarter was 17%...
Mike Beckman
17%.
Rafael Lizardi
Okay. So that means that 83% of our revenue is obviously non-China, so United States and Europe primarily. And those customers, let me start with those customers care deeply about having geopolitically dependable capacity, okay. And it's largely -- they are largely unaffected by whatever SMIC and HH over there decide to spend to build foundry, okay. Now on the 17% that is in China, we compete there very well, okay. And just because a foundry throws money at something doesn't mean they're going to beat us, okay. We have our competitive advantages to rely on and it starts with our manufacturing and technology and we just went through our footprint and our 300-millimeter footprint and cost leadership there.
But it's not just about cost and the footprint, it's also about the broad portfolio and the sales channels and the longevity, but let's talk about the portfolio, right. We have already over 100,000 parts and counting, right. So every year, we're turning out more and better parts and family of parts. When you're competing particularly in industrial and automotive, you need a -- families of parts. You don't just need one or two parts.
Personal electronics, frankly, it is easier to come out with one part that is perfect for this particular application and you optimize it just for that, for a phone that's going to sell 100 million units. Okay, that's hard. We can compete there too, but it's easier for an entrant, a newcomer to go in, partner with a foundry, and go after that. Industrial and automotive, a lot harder. You need families of parts and you need the reliability, you need the inventory position, the capacity.
And then on top of that, many of those customers in China also export. They're not just selling into China. They're export. So they care about the geopolitical ramifications of their -- of their choices. Finally, let me give you one more. It's not an all-or-nothing decision. Many of these customers have multiple boards that they have, some boards with TI, some boards with the local and they can flex I'm talking in China and they can flex how much they have depending on the situation, depending order, shipping, etc.
Vivek Arya
Got it. But when we look at how much China was as a percentage of sales for a lot of the fab equipment, right, tool companies, it was over 40%. So what capacity are they building? Like is it all consumer, like is it all -- like there is zero overlap with what TI is doing?
Rafael Lizardi
I'm sure there's no zero overlap. I mean, is it an ideal situation of course not. Okay. So it is a risk, but it's a risk that I think we can manage and we are managing for the reasons that I -- that I described earlier.
Vivek Arya
Makes sense. Finally, on pricing, I think you have been kind of more measured in your approach saying that industry pricing over time could get back to kind of the historical trends of low-single digit. I think many of your peers have kind of maintained a flattish profile. So when you say low-single digit decline, is that because that is what you are seeing, or is that just because of the conservatism in how you see?
Mike Beckman
Frankly, no, that's what we're saying. That's what we're saying.
Vivek Arya
You are actually saying prices go down?
Rafael Lizardi
Yeah. We're saying that.
Mike Beckman
And that is in-- as you design in new devices to the customer as they open up their board and you have opportunity. The pricing discussion has moved back into that engagement with customers. There was a period there during '21, '22 where that wasn't the -- that wasn't in the conversation very much. Obviously, that's come back very much like it was prior to 2020. And so it -- our base case assumption is you start to see that. I would also add that we exercised a lot of restraint I think in pricing in '21 and '22.
Obviously, we did price to the market and as market moved up, we moved with it, but I think there was more restraint from us. So I think over time, like I said, our base-case assumption is low-single digit decline and that's just based off similar trends that we've seen in engagement with customers that we saw prior to the last cycle.
Rafael Lizardi
Right. So some clarification. This is not the price going back to what it was before, the rate of decline going back to what it was before the pandemic, okay. So prices ran up. Now they're declining 4% or 5% a year or so. That's one clarification.
Mike Beckman
Probably two-ish, three-ish.
Rafael Lizardi
Yeah. The other clarification, it depends on the end market, it depends on the region, it depends on customers. So there's some end-markets where you're not seeing much of that. And it also depends more than half of our revenue, I don't know, 60%, 70% is on multi-source parts. So those are more price-sensitive. The parts that are more ASSP, more unique, they're less price-sensitive, okay. So it happens. But in aggregate, 2% to 3%, 4% declines on an annual basis is what we are seeing so far and what we expect for the foreseeable future.
Vivek Arya
Do you think the fact that you're building a lot more capacity makes you a little more open to having the pricing discussion?
Rafael Lizardi
It gives us the ability to compete, but that's not -- it doesn't drive us to that.
Vivek Arya
Okay.
Rafael Lizardi
And we price to market, if our cost was 2x, 3x what it is today, there'd be some portions of the market where we wouldn't even be playing.
Vivek Arya
So are you then surprised when you hear your peers on public conference calls say that there is no change in pricing while you are giving a price discount?
Rafael Lizardi
Yes and no. I think I'm a little surprised, but there's some -- they don't compete -- the America – U.S. and European competitors or at least some of them don't compete to the extent we do in multi-source. Now they like to say that we only do multi-source or even commodity, that's not true. We also compete in their backyard. But in addition to competing with the high-end, innovation, the ASSPs, we also compete the multi-source space.
Vivek Arya
Got it. Okay, perfect. And before we close, Rafael, just any update on your kind of return of cash plans that as the free cash flow starts to improve from here, you think we should start to see buybacks start to resume, right. What changes?
Rafael Lizardi
So I don't have anything to announce on that front. But high-level, I'll tell you -- let me answer it this way. The reason why you have seen our buybacks decrease over the last couple of years versus what we've done before is because primarily of this CapEx investment cycle. All else being equal, that's the most important capital allocation, more than buybacks.
So that's why you have seen the buybacks decrease. You've also seen us take on debt, increase the cash on the balance sheet. So as the CapEx is behind us or mostly behind us, right, we're almost past the midpoint at this point. In fact, we are past the midpoint. Then it's likely that you see a shift on the capital allocation.
Vivek Arya
Excellent. On that optimistic note, thank you so much, Rafael.
Rafael Lizardi
All right.
Vivek Arya
Thank you, Mike.
Mike Beckman
Thank you.
Vivek Arya
Thank you for your time. Thanks, everyone.
