Texas Instruments Incorporated ($TXN)

Why don't we get started. Good morning, almost afternoon, everyone. Thanks for coming. I'm Stacy Rasgon. I'm Bernstein's senior research analyst covering U.S. semiconductors and semi capital equipment. And it's my honor to welcome our guest here today, Haviv Ilan, the President and CEO of Texas Instruments. Before I start, I want to mention if you have questions you'd like to have asked during the presentation, you should have a link to the Pigeonhole form. I think there's a QR code that you can scan. You can submit your questions there. We'll have time for Q&A at the end. So Texas Instruments, TI. It used to be that they were sort of the boring semiconductor company. They're kind of proud of it. I think it's been a little less boring lately, both for TI as well as for the space overall. But TI over the last 4 or 5 years has embarked on a program of significant capacity expansion that at least temporarily sideline cash flow and return, and this is for a company who sort of pioneered the whole idea of 100% free cash flow return. But they always do think long term. And we're now at kind of the tail end of that investment strategy, the cash flow now seems ready to start coming through again. And now with a manufacturing footprint that they'll be left with that potentially leaves an increasingly advantage maybe in a world that is growing increasingly decoupled. I think more tactically, the post-COVID overhang and it was pretty long, seems to now be behind us. At a minimum, we've got an industrial rebound now that's driving upside. And now coupled with an AI environment that's gone mainstream and sort of dragging everything along with it, there's a data center story though that maybe becoming more of a primary growth driver for the company, which complements the traditional focus on industrial and auto. So I wouldn't say things are all that boring anymore. To tell us about it, it's my great pleasure to welcome Haviv. So thank you so much for being with us here today.

Thank you. Thank you, Stacy. Great to be here. Thanks for having us.

You bet. And maybe to start off, like just on that CapEx strategy because that really has sort of been the defining thing -- element of the company over the last like half, I can't believe it's been half a decade already, but it really has. But you're at the end of this like 5-year CapEx and investment cycle. And maybe just talk to us about how that's -- like what was the driver of it? What was the impetus? Like how has it gone versus your expectations. Where are we now? What do we expect going forward in terms of CapEx and cash flow and return and margins and all that? Like how is the company now, I guess, on the other side of it, different from where it was before you went in? And how does that advanc?

Sure. Sure. First, thanks for the introduction. You touched more or less the executive summary. I appreciate that. But in general, you're right, somewhere at the back end of 2020 and 2021, we got together and decided that we are going to set the company and prepare it for the next 10 and 15 years. This was not about the next cycle, but more of a longer-term thinking that secular growth in semiconductors will continue especially, as you mentioned, in industrial and automotive. We also thought that it's -- we need to be ahead because one of the areas that we've learned during the previous cycle that when you fall behind and can supply the parts or the sockets you've won that's just not a good -- that's not just a good place to be...

That kind of did bite you a little bit during COVID.

Our customers deserved that we can support them in every scenario. And then you want to model a set of scenarios and we said, "Hey, what could be secular growth in semis, does it accelerate or no?" The answer was yes. The second is do we only want to grow with the market or do we want to have an opportunity to gain some market share and also support our customers through the cycle, not only at the middle of the trend line, but also it peaks and of course, that's very, very important. And along the investment plan, there was all these geopolitical tensions that continue to rise, and we said, "Hey, having manufacturing in North America, but in general, a geopolitically dependable footprint that is broad is going to be very, very important." So that's what kind of derived our plan. As you mentioned it, it was a 6-year plan between '21 and '26. Yes, it is '26. It feels again a long time ago. Now in the last year ...

Looks like it went like that. So.

Some days are longer than others, but it does fly by, yes. So we are in the sixth year, and we went through a more than $20 billion investment cycle, and we always said we want to be positioned for every scenario. I mean a couple of years ago, as a cycle it deepened or the down cycle deepened and lengthened. To your point before, it looked like, hey, are we on right bet. But we always had the conviction that we are. And I think right now coming into 2026, after a year of growth in 2025, but maybe growing demand driven by industrial and data center, as you mentioned, we are very pleased to be where we are because we have the capacity. We have inventory that can support short-term demand from customers or surge of demand. And we also have the clean room or the footprint, especially if you are brick-and-mortar and clean room footprint to grow into even faster growth scenarios. So that's a very good position to be in because if you are falling behind in a situation like that, you have to wait 4 to 5 years if you haven't made the investments. So we are excited to where we are. I think, as you said, free cash flow should grow as CapEx goes down and demand goes up. And from here on, we should be more in a steady state of supporting our growth with investments as needed. So very happy to be in that phase of the investment cycle. Yes.

It might be helpful if you could outline for us exactly what capacity have you actually put in place over the last 6 years? And where does it sit? . I know there's Richardson and there's Lehi...

Yes. So from a footprint location, we made investments, mainly in North America of 300-millimeter wafer fabs that are not common, by the way, for our analog and embedded market. We are building very modern fab, fully automated. And the scale of TI allows us to do it. So we've built what we call the Richardson 2, that's our Fab 2. It's now almost in full production, almost fully utilized. We also decided to build a new site in Sherman, Texas. That's north of Dallas. And over there, we are planning a mega site, almost 4 fabs that altogether, we'll be able to support by -- per fab, about $10 billion of revenue.

Each?

Each. And also, by the way, the cost to build to one is also similar to that. So think about it at that rate. And we've built 2 of them. So 2 shell rooms are built. The first one is partially equipped, the Sherman one. It went into production. We had an inauguration date in Q4 of last year, ramping out very nicely, but we have a lot of clean room available over there that's going to give us the [indiscernible] we will need through this upside through this cycle that we are hopefully experiencing now. Lehi is really focused on 2 types of parts, mainly the embedded processing business is served by the Lehi Fab. That was an acquisition actually of a fab we bought from Micron.

That was the [ CrossPoint ] facility.

Correct. And what we decided to do is not only to retool the fab into an embedded processing type of a product, but also expand the site to the second fab, it's called Lehi 2 which is a much larger facility that can support even more revenue per the factory, maybe think about $15 billion of capacity over there. So Altogether, the plan is going well. Remember that the Lehi fab was not only to support growth, but also to support internalization, building actually built in the foundries, mainly in Taiwan into the U.S., very, very important for us to control our destiny, and we will grow into our fab. So if embedded used to be maybe 15% built internally, it will reverse. So by the end of the decade, we see embedded running more than 90% internally.

Where it is like right now, is it still 15%, 20% internal?

No. So right now, we are almost at 50-50 already because we are transferring our 65-nanometer embedded processing nodes into the factory. By the way, there is also some analog, high-speed analog mixed-signal solutions that are built over there. But you can think about it as 2/3, 1/3 embedded versus analog. And that's already moving. Our next step is to move up 45-nanometer now that is happening this year. So by the end of the year, embedded will be predominantly built inside our own factor.

45 do you have and below in your -- I can't imagine be that high, but maybe tighter than I think it is.

So 45-nanometer, we have a node that really for nonembedded flash technologies. So think about external memories for embedded, think about radar systems. That's the main thing that runs on that. But we are going a couple of steps forward. So right now, we are already sampling our test chips for 28-nanometer embedded flash systems. So we have several flavors, one more mix signal analog base and one more embedded processing, embedded flash-based. Our future MCUs, especially the larger one that has more memory footprint will be built there. And we can take it one more step forward into '22, that's more or less the plan of the sub. We are not planning to build a FinFET technology in Lehi. But that gives us, I would say, the excitement at least 10 or 15 [indiscernible] of runway to support our businesses. And in that sense, very unique. The fact that we have invested in our own technology that allows us right now to respond to developing -- to the developing situations across our markets, whether it's industrial, data center or hopefully automotive comes back soon as well, and we are feeling very comfortable about that.

You'd also closed some other fabs, right? There was also part of the internalization? .

Yes. So we decided through this down cycle to shut our last 6-inch fabs, one was in Dallas. One of them in same lane in Sherman...

That's done, right.

That's done. So we are now in the last -- also from an overhead perspective, Q2 will be the last quarter of any expense related to these fabs. We are setting them down. And that's good to be because 6-inch fab -- they rent for 50, 60 years, by the way, beautiful ROIC. But it's hard to maintain them. So when a tool goes down, it's really hard to fix it. So we took the opportunity to transfer all that goodness of long-lived parts into our 300-millimeter wafer fabs.

And so how does all this translate into gross margins then? Because, I mean, you used to -- clearly, when you first embarked on the 300-millimeter that was probably 15 years ago, like with the fab 1 and [indiscernible] and everything, I mean, you were running 4% of CapEx to revenue for 10 years or whatever. And I think gross margins peaked at 70%, close to 70%.

Yes.

And then we knew as you embarked on this investment cycle gross margins where they come, and you never hit that, and they came down, I can't remember where they bottomed like mid- to upper 50s. They're kind of creeping up now. I think -- I know you don't guide to gross margin, but the -- my math suggests the implied gross margin for the next quarter is right around 60, so at least maybe starting with a 6 handle. And a lot of the things that you're talking about, like, in theory, should be good for gross margins. We're through the depreciation slug, revenue and utilization, hopefully, are going up. You've moved stuff in, so like the cost structure gets better like where do I see those going? Like can that actually get back to where we used to run like back in the old days or -- this is a always the big question everybody always has for TI.

I understand. And we respect it. You know we run...

I know you don't run the company...

Yes, we'll get there. But let's answer your question because I think it's a fair one. First about Q2, I think you said it. I think -- yes, it's a good assumption. The more you wait the better it gets. And the reason is Stacy, that -- and that's just a math of depreciation and growth in internalization of wafers. Just invite you, if you want to see it in action, one way to do it. We now -- I think by segment, you can see the reporting of margin -- or gross margin by segment. Look at the embedded segment. So you will...

You report operating margin by segment, not gross margin...

No, gross margins. Yes, that's the new rules. So you go to the 10-Q, you'll see there...

Really, I didn't even know that.

Okay, you see so. You always done something new. So you can look at it and see the progression of embedded because what you're seeing there are wafers that are coming into Lehi. The reason Lehi is now at 50% utilization is that transition, and it continues every quarter because now our 45-nanometer wafers are moving in. And you'll see gross margins right now, getting closer again to 50% on embedded, it used to be much lower. And you'll continue to see that moving. That's one example of how you move a foundry wafer into TI, that's going to be accretive to your gross margins. Same is going to be in Sherman because but over there, you have to grow revenue into that wafer capacity. And I think both are happening right now. So to your question, can it get back to the areas of before? And again, we don't think that way. The answer is yes, but that's not going to be satisfying for we care about free cash flow per share growth. So you can have very nice margin in high 60s and your free cash flow per share can do less than double-digit growth as we've done in the previous decade. So [indiscernible] how do we get back to the trend line of this about 10% CAGR of free cash flow per share. And that's how I talk to the Board, that's our commitment. And as you said, we'll take a big step forward in 2026. Why? Because we are seeing some revenue growth, okay? We are seeing internalization of Lehi, to your point. And we are starting to see the CapEx levels going down. So the math just worked together. And then as I said, the more you [indiscernible] it gets. So that's kind of the plan. I'm always cautious and we had an encounter last year where we're at the same point of time a year ago. And I want to see the demand continuing. But so far, the indications that the environment is a little different. This is a more broad environment of demand. I see it across regions, like we see it across markets to your point. Let's see how it develops. Of course, when we go out there in July, we will report further on the progress over there.

Let's maybe talk about some of those demand drivers, some of those end market. Maybe start with industrial. And so this historically was the biggest piece of your business. And this was probably the peakiest during COVID, given the shortages and everything and it fell off the most. I can't remember how much you were down...

Almost 50% down from peak on a quarterly basis.

So I guess where are we now? Where are you seeing regarding the industrial recovery? Which areas are stronger, weaker in where are we sitting today versus that prior because it's now... .

Yes, that's a great because I look at it all the time...

And you've been waiting, you draw -- you always draw this chart of like growth versus trend, and you've been making the point we've been below trend for, I don't know, multiple years, right? .

So the beautiful thing is that we are still below our peaks, okay? And we are talking about 4 years later. So let's go recap what happened first in Q1, and I think it's just good to set the stage. Industrial did grow sequentially 20%. So very strong growth of almost a comeback, right, year-over-year, close to 35%, somewhere between 30% and 35% but closer to 35%. But still Stacy, 15% below the peak.

1-5?

1-5, yes. Okay? Now it's catching up very quickly. So as I indicated in the call, the reason we see an above [indiscernible] in Q2 sequentially is driven by industrial and data center, which is a different story.

I'll get to the data center.

We'll get there. I'm sure. I appreciate the fact that you haven't stated there. That's great. But the fact that industrial is coming back is very visible for us. And by the way, some of it in sectors that are data center related, if you think about energy infrastructure as a sector, a big sector for TI, I think data center helps us there. That's measurement is a sector that is getting helped. So these sectors are growing fast but aero space and defense is coming in a big way. And we are starting to see this is my biggest excitement factory automation or industrial automation, robotics coming back. And that's an area where last year, it was waiting to come back. But I think then with all the anxiety about tariffs, I think our customers took a breather of making the CapEx decisions. And I think they are moving now, okay? So we are seeing this beautiful situation of industrial growing very fast on top of secular growth in data center. And if I may add, automotive is still hovering at nice levels, closer to the peak but has not grown back to where I think it should. So overall, I think we are in a good set up.

You sounded a little squishier on auto, I think on the earnings call...

No, because in automotive, if you look at the progression, in the last 3 years, peaked in 2023, automotive never dipped like industrial, hovering around the same level. And I think that's what happens with content growth. I mean there was inventory correction, but content growth continues to only keep it flattish. So not the industrial story. But I think that also continues to grow. There is secular growth in automotive continuing. And when that happens, our confidence level will be higher. So you're right, the automotive was growing about mid-single digits year-over-year in Q1. But very close to peak levels, which is a better story. Hopefully, that gives you a picture about the industrial.

It does. So to go back to Industrial. So any thoughts on how much of the strength we might be seeing might be customer restock given lean inventories versus actual like sort of fundamental underlying demand? And I guess would you be able to tell?

Yes. First, you're right. We are trying to be very humble about we don't have that information. We just have anecdotes, right? I will say that I believe because we are 4 years further away from the peak. And because we are still 15% lower than the peak, I think it's very early in the -- hopefully, the recovery phase. And typically, in the early phase, customers are not building inventory. Typically, that happens towards the end of the cycle closer to the peak. Now the customers are sounding and I did visit in Asia, we visit in Europe actually this month. And it does sound that customers need more parts to really support ramps. We have seen also -- I've seen anecdotally, this is where the only way I can tell with data that there is -- I appreciate your 6 weeks lead time, but I need it now, meaning customers are not carrying inventory. But that is for 1 or 2 parts. That's not across the board. This is not close to what we've seen in the coverage cycle. But there is more -- there are more cases where TI need to solve a problem. A little more -- and yes, there is -- my impact was very quite for 3 years. So it's coming back right now. Let's see how it plays out, Stacy. I think [indiscernible] also picking up in the first half of last year, then it took that a little bit of a breather. I want to see it one more quarter in so far, so good.

In general, your lead times are -- do you effectively still have 100% availability of -- 100% of your parts are pretty close to it. .

Yes, we're saying 100%, then I get a call from a customer A you own me that part. But in general, yes, our lead times are stable. Most of our portfolio, by the way, the general purpose portfolio is 6 weeks lead time. The more application specific, that are more unique and have less diversity, we usually keep them at 12 or 18 weeks of lead time depends on the part, but mostly 12. The average lead time is at around 12 weeks, okay, across our revenue footprint. And that has been very stable. And for a good reason, we have built inventory. We have capacity. So we have done -- and I expect it to stay for the foreseeable future that way. Again, demand, I don't forecast right now. But as long as demand continues to be strong, even at that level, we should be fine. The only area where we've seen a little bit more work to do on our side is on the assembly and test side. So over there, the mix is changing sometimes because of demand signals that are coming to solve problems, as we just mentioned. And this is where you never know the mix upfront on this.

This is like so many different types of package.

Correct, so many SKUs. We have thousands of SKUs on the assembly and test and it's different bonders, different lead frames that you have to take care of. So we are putting a little bit of adjustment on our -- how we allocate CapEx in 2026. It's really towards more on the assembly and test side. We also see the OSAT, the outside semi test houses that we are still using, more compressed. So we are actually accelerating the internalization our manufacturing into TI, we have the clean room...

I can't remember what the target was for like front end, I think, was something 90% internal. Back end, was it similar? .

Similar. 2030, what I can comment about, it's probably going to happen quicker because we are seeing tightness externally. We just have to qualify more technologies internally, and that's what we are doing right -- we speak right now. .

Are you guys holding like more die bank now because of this. .

We do. We do hold more die bank. Again, it gives you that flexibility of what does the customer want. It's also a more financial, it's more a fair way to hold inventory.

Got it. Got it. Let's talk about data center. And so... .

28 minutes or 23 minutes. I'm proud. that's Amazing. Yes.

We're not half way through that, even.

That's amazing. It's good. Usually, it's the first question. so I appreciate it.

We got to build to it so let's talk about it. So you're reporting it as a segment now.

Correct.

As an end market. And your old segmentation, you have this like enterprise systems, which I think was primarily like more traditional servers and things like that. And it looked like to me you took some of it out of the -- what used to be in industrial and some of -- out of what used to be com, put it all together and that's data center. And it's growing, I think last quarter, it grew 90% year-over-year or something like that, right? I guess the first question -- and it was -- I can't remember, 9% of your revenue last year. So relatively small still, but growing at a very run rate. I guess maybe the first one is just what is in that segment? What are you serving? And how much of it would you classify as like actually like AI driven versus more traditional data center. And by the way, maybe those 2 things are the same thing now. Like I don't just give in the way the market's going. But like see if we can peel the onion back a little for us on what's exactly in that segment?

No, I think it's -- let me take a moment to just [indiscernible] the data center market for our type of chips because you guys spend so much time...

It's like zillion things. I'm assuming.

No, but you exactly, but guys logic and memory, you know the story over there, and that's not our area. We make what I like to call foundational chips that are analogue and embedded chips, there are many of them, okay? I think even at the street level, people like to talk about a few sockets, but there are so many of them, really thousands and sometimes you count on tens of thousands per reg, okay? So very broad opportunity in the sense of product diversity. Customer diversity is more narrow, but the product diversity is very high. Now in our case, we decided to define the end market as based on it because it was too -- I mean, first, it was very small at the beginning, I would say, go 5 years ago, when we made our plans, I did not envision data center running at -- now Q1, 12% of our revenue.

12% in Q1. Okay.

In Q1, 12%. Last year, it was 9%. Exactly as you said. And the definition I like to -- and the definition for us is whatever it's inside the walls of the data center, some people call the white zone unlike the energy infrastructure that sits in the gray zone where are dust is. So we're inside the...

Okay. So if you got like stuff in like a turbine that's -- you wouldn't count that. .

SSTs or some of the UPS system, they see it outside. So that's not in our number. That would be in industrial.

But also growing.

It's still growing, yes. That's one of the sectors in industrial that I mentioned before, is energy infrastructure. But go back to the walls. Now you talk about 3 sectors. And it's really -- if you think of the rec, Stacy, there is the compute tr . And this is why I don't say CPU -- traditional -- today, on a compute node, you can have a CPU and 2 accelerators, right? So to me, it's compute. And I'm not trying to break it between AI and non-AI. To me, it's all compute. There's networking that is larger than you would think. I'll actually give the numbers in a minute. And then there is what we call rec power and cooling. And the rec power is traditional PSUs. AC comes in, DC comes out, but there is also a change over there as you might know. But overall, the market, if I quantified last year, I call it in our area, $7.5 billion, less than 10% of our TAM, okay? That's...

That's now you're at. .

That's the market. We've done 1.5%, so it's about 20% share.

It feels low, like, just given all the...

That's the numbers we are -- we've done our math and it's not very far away from what we've seen. I call it this year, and you can check some of the analysts out there that we came in very similar numbers. I think it's 12.5% this year. So the growth rate is 65% or so, that's high. TI in Q1, just to complete the point, that grew 90% year-over-year. And the run rate is right now above $2 billion. So it's close to somewhere between $550 million and $600 million. So that's just to set a stage. Now why do we want to talk about it because of the growth rates. More power is added to data centers. And power density also growth and architectures are changing. There is a very important and, I think, unique growth opportunities for our type of chips, sitting around power, sitting around communication, clocking, sensing, cooling, protection, you name it, and we want to play there. Our play in data center is across the rack. I would say when the architecture change from 480-volt AC to 800 plus minus 400 DC, our play is going to be in the entire rec. And I would segment it to 3 sectors, as I said. So I think half of the -- it's more like in the compute side, power.

This is like analog and other stuff on...

Feeding the power to the consumers on the board. Think about the compute trays, okay? But not only power socket. There's also a lot of signal change, clocking, et cetera. Then there is networking about 35%, and the rest is data center power and cooling. So that's kind of percentage of footprint.

Repeat those percentages for me.

So 50% will be compute, compete trays, 35% we count -- our estimation for networking. And network includes also optical networking that optical modules, et cetera.

Only 15% on the power side?

Power delivery, we are talking about the AC to DC, okay? There is power also close to the GPU. That's part of the compute now, okay? So that's the way we think about it. Think about it the end equipment is the tray, okay? That's the way we think about. What's the content? And the beautiful thing is, it's very rich. There are so many sockets per board, power, signaging, clocking, protection, sensing. And we play in all of them, Stacy. So our growth even last year was in all these sockets that no one likes to talk about. It's like the broad portfolio that we have, and we serve it very, very well. I love this type of sockets. They don't generate a lot of attention. They are not attacked every other day. But we also want to play -- that was my comment in the last call and more application specific one. Some people will call it -- I like to go it multi-phase power, what feeds that. You have to take, I don't know, 2,000 amps into a GPU. You don't do it in one wire. This is a heavy socket with multiple -- voltage regulators are delivered in different phases of time. That some people call it the Stage 2, will play over there. And over time, as architecture is growing into, let's call it, 800 volts or first one is 400, we also want to play there. because we do have a GaN technology that we've invested in. And that's a great technical solution to convert energy convert power, call it from 800 volts all the way to [ 12 or 6 ]. You can do very nicely with GaN solutions. So that's the opportunity for TI. We can grow into more application-specific sockets, very competitive sockets. You have to win at multiple customers, but we're going to play there.

Is it fair to say that like on the application specifics of that, that is where you have to compete for this socket. But the other stuff you really don't have to compete as much. Like I said, it's more -- I don't -- I'm probably simplifying, but catalog... .

I think competition is tougher for the application specific, also because they're defined by the customers multi-- usually, they'll define the footprint. And if you want to serve the socket, you're not doing it alone. This is not only yours, right? So you have to compete with other players, plus the revenue concentration, we talked about it last year. Revenue concentrated, but socket is high. Every win or lose is a lot, okay? So obviously, there is more competition there. Now our supply footprint, the fact that we have such a broad portfolio serves very well the general-purpose sockets. And there are maybe on average $0.20 or $0.30 per socket, but they add up. That's how...

How do you Think about content like per rack for you guys.

Yes. So we -- you can do all kind of math, but I will say it's in the tens of thousands of dollars per rack.

Actual content or opportunity?

Our win rate last year -- our share last year was 20%, do we want to grow market share we do. So look, I call the TAM at about 65% growth. So far, 1 quarter, we grew 90%. Can we keep it up at that clip or can we outgrow the market? We'll have to wait a year and see. But that's always our intention to outgrow the market.

I mean auto and industrial are clearly your biggest segments... .

Correct, like 1/3, 1/3, and now we are talking about 12%.

Well, data center become -- does it overtake the other ones at some point.

I don't know if it overtakes the other one, but I think it overtakes consumer or PE over time just because of the clip of growth rate. If you are thinking about...

20%?

PE is now 20 -- low 20s, I would say, but it doesn't grow faster, Stacy. Again, you've probably talked with people who are saying there's going to be new personal electronics, I don't know wearables or whatever. I have not seen that market develop yet. But that would also change. But right now, this market has been growing at the low single digits. So when you grow a market at the 60s, you can come -- it can become our third market or third largest market very quickly. Yes.

Maybe to touch on PE, just -- and I get it, it's not a driver and it's not the big.. But I'm a little worried about PE just into the back half on MIM pricing and everything else. What are you seeing there?

No, I think that's also part of the reason I want to see 1 more quarter before I call the second half. To me...

I don't think you should call the second half. I think it's better. .

Call it the third quarter, right, to give you a third quarter forecast, right? So I talk about it right now. But to me, our second quarter, I mentioned 8% sequential growth, nice year-over-year growth. Let's say it play out. But as I said, so far, we are on plan. If had to some to report, I will let you know. To me, the PE is one of the question marks because I think memory is a constraint. I think they would want to build more end equipment, but they can't. So how -- and it's still 21%, 22% of our business. So that could change the number. So we want to see it play out.

Are there memory impacts on anything else besides PE?

I specifically have not seen it. But I know our customers are busy around that, and they have to -- I heard it's supply limitations. Also, the costs are higher. But luckily, I guess, we don't build memory anymore. And we are focused on our stuff.

That's right. How are costs in general trending for you guys? I mean we're clearly in another inflationary environment. You got other like -- I mean maybe if you're in-sourcing, you're a little more protective from foundry price but foundry pricing going up and memory and other things and there's been -- I even know it's a room, right? I mean you guys have been taking up price along with many of your other peers in this environment. I guess how do we think about -- maybe the right question is we think about your pricing actions both on the wake of cost increasing as well as the potential like to capture more value for yourselves? .

I think you set it up correctly. There is -- we are experiencing inflation, especially when you think about energy in Asia, for example, where we have all of our assembly and test -- cost of resin. So you think about what's coming inside the package and the fillers or the mold compound. And that's all higher Cost of metals are higher. Yes, that's part of what we're seeing. But pricing environment is better than last year. So if you think about the COVID cycle where prices went up all the way into 2022, we have seen 3 years of our model coming through like '23, '24, '25 this low single-digit price effect, if you will, like-for-like didn't materialize in the last 3 years. But as I commented on the call, in the first half of this year, we expect pricing to be flat, which is good news for us because usually when you start the year, prices usually because of price negotiations does fall down a couple of percentage points. So that is flattish. But yes, I -- we have started a discussion with our customers to talk about costs and what the supply/demand situation. And I think the second half of the year, prices could be a little higher. And we will be a follower here, Stacy. We are not trying to set the market price, but we are watching the market environment. And right now, you're right, the market prices are going high.

I Just thought it's interesting because I look at the equipment guys like lagging edge equipment demand has not been great. I know we're seeing maybe a recovery now in some of the end demand for some of these, right, but the end demand in some of the trailing nodes has not been great. And like, usually, you see price increases when supply in these markets is really tight. I'm not exactly convinced that in some of the markets that you play -- that supply is like incredibly tight, and we have been seeing pricing coming. So is that just a function of like we saw during COVID, it is inflationary and you do have the ability to at least pass those costs along without the customers...

I think it's a fair question. But again, this is why we have to be cautious because it's only a couple of quarters in. But you're seeing what's happening in industrial. You've seen the data center market becoming. And I just came back from Taiwan, I was there last week, really focused on more data center customers and the supply chain, if you will, is heavily there. And I think we are starting to see areas of supply and demand mismatch even in our area. So especially on the analog side, Stacy, we do see that. So the discussion I had with customers last week was all about, hey, make sure you continue early times as they are, make sure that you can upside if we need to. And I always tell that we have parts. We have capacity. We have inventory, bring it on. But it's not -- I think we are a little bit different than average there. I have seen -- and you can see it also in lead times. All lead times you've asked before, are stable. It's out there that lead times have been increasing, especially on the analog side, sometimes through a year, okay? So I think TI is uniquely positioned. And we are not surprised. I think we have been disciplined in adding capacity through the down cycle, we are unique there, okay? And it sets us up, you always said to every scenario. The scenario is still getting built as we speak. And you and I will watch it together. But I think if it wants to continue to be strong in the second half, we will be ready. If it wants to continue into '27, we'll be ready. So that's where we are, and I love where we are because if you fall behind, and we saw it in the previous cycle, and he said, "Okay, I want to chase it now. You [indiscernible] 4 years from the minute you move there to substantial output from your fab, it's 4 years. So you're not talking about anymore. So in that sense, TI doesn't have that lead time issue, okay? We are very well prepared.

Got it. Talk to me about the slab deal. You guys sat out of M&A.. While the industry was consolidating for a long time and I don't know what it was. It wasn't -- there wasn't a strategic fit or the returns on the valuations weren't there. And why now?

Yes. So first, you said set out in making a decision, but we don't set out on making..

No, I understand you evaluated all -- I get it.

We have a quarterly process. Okay. So that's ongoing. Again, see that. We've done a big one in the early 2000s. We've done a big one in the early 2010 Yes. And we did another 1. So on average it's 10 to 15 years. So we are on the cadence, right? Now why this one? And that's, I think, let me set up a little bit of embedded because it's an embedded-centric acquisition. And this is when there is a change. I would say 10 years ago, we would not consider it because we are not sure about our embedded business. And then we will very -- I mean, Rich was very open about it. We come from -- you look in 15 years ago. custom business, more than 50%, big logic DSPs, very digital, mainly built externally. That's the embedded business we are building today. So 5, 6, 7 years ago, we said, okay, let's retool our embedded business towards our competitive advantages. Let's bring it in. Let's create a broad portfolio, meaning less -- big processors and more MCUs with some analog periphery, some application specific MCU for power conversion, for motor drive, connectivity solutions, radar systems. These are the type -- which are also part of the DSP investment that we have. So these are the kind of investments we are making right now in embedded. So a number of parts we build every year is higher. We are going to build them all internally. And embedded, it's going to be more than 90% internal by the end of the decade. Now when you come into Si Lab. This is where we had an opportunity to give ourselves a step function in our portfolio because we have some connectivity parts, but they are mainly serving very well the automotive market. This is where we are winning and expanding. Our industrial portfolio is slowly growing, and that's a onetime. We saw a onetime chance to really increase our portfolio in a step function. If you think about our competitive advantages, Stacy, it's a beautiful -- there are 4 elements, but they bring in 2, we add 2. So we talk about manufacturing and technology, building it internally in our technology. We talk about the broad product portfolio. We're talking about a very strong channel position or advantage, both ti.com but also our very large, largest in the industry of team and the position of diversity and longevity. What do they bring in? They bring in a broad portfolio, a broad industrial portfolio and a good position in terms of diversity and longevity. 90% of their business is industrial. You look at the revenue by socket, very, very broad. I think you can see it on our website. So they bring in 2 elements. You add to that. The fact that you can bring it inside our lead fab and build it internally and control your destiny in terms of technology. and manufacturing. And also the field and the ti.com that can really sell beyond that connectivity chip, they have a very narrow portfolio in that sense from a technology perspective. You get into -- the Excel sheet works, okay? So and that doesn't happen every day. Many times we look stuff, it looks good strategically, but we can't make sense of the price. And look, if you ask me, would I do this deal today? It will be harder. Look at what happened to the market prices of assets in the last 3 or 4 months. So in that sense, I think the stars aligned and we made the acquisition, very pleased about it, very pleased with the progress towards the -- completing the deal in the next, I would say, 3 to 4 quarters by the first half of next year. And I think it will be a great addition to our embedded business. So that's a story over there. You want to do this deal 10 years ago, you don't have Lehi. You don't have conviction that you are going to be successful with our portfolio. Our confidence level is higher, and that's kind of the way we'll think about it moving forward.

Got it. So you would have had confidence in the business and you wouldn't...

in general, the business...

you wouldn't have anywhere to put it even if you...

And the synergies will not come in because these guys are using mainly TSMC and...

How long will that take you to internalize their products? And how -- and how difficult is that to do. You have to...

So luckily, we've been doing this ourselves many times, including right now as we speak bringing stuff from the foundries into Lehi on our portfolio. But they've done a great job, and this is very rare in a company on an embedded field, very, very well-organized platform -- operation i mean. Very well invested -- it's only somewhere between 10 and 15 different die, which is a...

It's how many different products...

2,000 products, okay? So they have done a very good job on doing this platform approach that you can build 10 to 15 dies and then approach many, many products out there. So that would take not too long, okay? It will be mostly completed by the end of the decade. We will start...

at the end of decade.

Yes, it's going to take 2 to 3 years. But we're going to start immediately after close. We are -- and that's the way we've modeled the synergies. Now internally, I want the team always to move faster. But as you saw, the $450 million of synergies that we described, they are mainly supported through the COGS, and it's almost fully in by the end of the decade.

Got it. So that also means that it's harder to buy somebody that sells 10,000 different products.

There you go. And most companies, especially on the embedded side, the breadth of the portfolio actually well highIt's how to do. It's how to do. It's how to bring in the synergies when you buy a fabless company. Now in the case of National, they had to own fabs. So it's a different story. But this is why these things are not trivial. They are not easy to do this, but you see us doing it rarely. But I can tell you nothing has changed on the strategy. We'll continue to look at assets. And I think this one was a unique one.

Got it. Talk to me about China. So clearly, you guys have made a bet in the U.S. and I get it. And I understand the whole concept of geographically or geopolitically attractive capacity. At the same time, I know it's an investment controversy. I don't know how real it is, but it's an investment controversy about your position in China, and I was joking [indiscernible] has Texas right in the name. And most players in China, there has to be some sort of a China for China strategy, whether it's working with local partners or making it directly in the region or whatever. How do we think about TI's competitiveness over time in China relative to some of the local players given the trends that we're seeing on the geopolitical spectrum right now.

Yes. Let me break it. I think it's an important topic. Let me break it out to 2 parts. The first one you mentioned is kind of bet in the U.S. maybe on manufacturing or new manufacturing footprint. Not only market only, the market wise we are a global company, okay? So and the fact that we are investing in the U.S., it doesn't mean that we don't know how to solve China from China. We have a factory in China. We have a fab. We are still -- we have a big assembly test out. So we can support channel and manufacturing lines. I think the challenge in China is actually how do you compete in the market. And especially when you are not a Chinese supplier, I think that's more interesting. And first, we want to play there. Why? China is, what, 20% of all GDP more or less...

Most of your revenue, I guess, by headquarters...

Our revenue by headquarter is in last year was about 20%. Okay? More importantly, China has some very important customers, let's say, automotive that are technology leaders. You even go into data centers, optical links. China is a big player there in terms of headquarters company. And we also see areas in industrial where China robotics, for example, China is a big player. So you don't want to exclude yourself from China. You want to compete. It is harder. It is hard because of what you said. There is a growing industry of very hungry supplier base that is always expanding. I have my own China index. I review with my team every quarter, a set of 25 competitors, 25 that together add up to about half of TI. But they are doing well. I mean, they are competing, but TI can compete. And the reason we can compete is that we have some competitive advantages that are very, very important to our China customers. The portfolio -- the portfolio is very attractive. If I want to solve a problem on the board, do I bring 25 suppliers, some of them are start-up companies that are just trying to breathe versus an established player like TI that has good solution. Now the portfolio is not enough. Cost competitiveness is key. But can TI play that game? Yes. We are vertically integrated. As much as the foundries in China are serving these local competitors. They still don't work for free. They also some -- if you look at the utilization rates over there, they're also not -- they are not underutilized, let's say that. So we can compete by supply and also cost structure, very strong channel or testing that is entrenched over there and build relationship with customers for years and a customer base, Stacy, I mentioned data center, but that's not the case that is continuously wanting to diversify their markets wanting to go into non-China markets. It's actually where they make most of their profit. So that game has played well for TI. And the reason I say played well, just go put 2025. We report our China business. We grew close to 25%. The index I just mentioned grew slower, okay? And we are watching this every quarter. Q1 is now being reported. We are holding our share, okay? So you can argue that last year, we gained share in China in an impossible environment. This year, I think our odds are even better because of the supply-demand mismatch, usually everything starts in China. That's not an outlier this year. So over there, our capacity, our footprint, our cost competitiveness and our broad portfolio and channel advantages are playing on. And the risk I have about China, don't fall asleep, don't be complacent. Don't say these guys are commodity players, spec them, but also don't be scared, fight the fight. So we take the fight to Shenzhen to Shanghai. Our customers over there appreciate us. And by the way, every time there is a tie, we lose a socket. So we always have to be a little bit better than the local competition. Hopefully, that covers China.

No, that does. It does. [indiscernible] round.

Whatever it means, I don't know.

We've got some -- a few audience questions.

Okay. .

Robotics, humanoid [ general ] drones or autonomous vehicles as a driver of growth for you going forward? When do you expect it to show up meaningfully in the top line?

Look, humanoid content is astonishingly high for TI, not only the TAM. So we are talking about content for TI higher than a car, higher than a automotive, think about $1,000 humanoid. That's a content that -- now you tell me the number of humanoids that are going to be built. It depends who you listen to. But I think we are starting to see it, and I'm excited about it. I don't think it as soon as people say. but I can see why humanoids like robotics will be one day walking around our factory, especially assembly and test and providing value. So I'm excited about that, yes.

Got it. We've got 1 minute left. So I'm going to finish this up the way I always do with everybody. We've got a whole roof of full room folks here. Why should they buy TI stock? .

Yes. So again, I think part of your introduction helped me, but Stacy, as you discussed, we have been preparing for an opportunity for a long time. It's been a long journey. I even kind of wake myself up, boy, it's been 6 years. But we have done the hard work for being prepared. Now tell me what the scenario will be. I think we are seeing one more evidence that this is going to be a good time to be in semis. There is a secular growth in the data center market that is really not negligible anymore. TI is in a great position. There is industrial coming back in a nice way. And I think automotive is around the corner. So you put all these 3 together, you can envision a very strong demand environment. We are well positioned to support it. We have the inventory. We have the capacity, we can grow into it, allowing hopefully, share gains for the company. So thank you, Stacy. I appreciate being here. And thanks for the time.

I appreciate having here. Thank you so much. .

Thank you.