Texas Instruments Incorporated (TXN) Earnings Call Transcript & Summary

Good morning, and welcome to the Texas Instruments 2026 Capital Management Call. I'm Mike Beckman, Head of Investor Relations, and I'm joined by our Chairman, President and Chief Executive Officer, Haviv Ilan; and our Chief Financial Officer, Rafael Lizardi. This call is being broadcast live over the web and can be accessed through our website at ti.com/ir. In addition, today's call is being recorded and will be available via replay on our website, along with the complete presentation and prepared remarks for your convenience. This call will include forward-looking statements that involve risks and uncertainties that could cause TI's results to differ materially from management's current expectations. We encourage you to review the notice regarding forward-looking statements contained in our most recent earnings release as well as our most recent SEC filings for a more complete description. With that, let me turn it over to Haviv.

Thanks, Mike. Let me start by welcoming you to our 2026 Capital Management call. During today's presentation, we'll share more detail on our approach to capital allocation and our investments. We'll begin with a recap of our objective, strategy and business model that is built on our competitive advantages. We will review our scorecard for 2025 and update for 2026 as well as a historical view of our capital allocation. Then I will provide additional insight into our growth expectations, where we continue to see excellent opportunities across all of our end markets and especially in industrial, automotive and data center. Next, we will provide a brief update on our progress in strengthening our competitive advantages. And lastly, we will review our free cash flow per share performance and wrap up with a review of our cash returns. If you haven't already, I encourage you to read our investor overview, which provides insight into our business model and competitive advantages. It is available on our Investor Relations website at ti.com/ir. The following guiding principles will help frame our discussion today. At TI, we run the company with a mindset of being a long-term owner. We believe that growth of free cash flow per share is the primary driver of long-term value. Our ambitions and values are integral to how we build TI stronger. When we are successful in achieving these ambitions, our employees, our customers, communities and shareholders all win. Our strategy is comprised of a great business model, a disciplined approach to capital allocation and a focus on efficiency. Our business model is built around four sustainable competitive advantages: manufacturing and technology, a broad product portfolio, reach of our market channels and diverse and long-lived positions. After investments in the business to grow free cash flow for the long term, the remaining cash will be returned over time via dividends and share repurchases. With that as a framework, our objective is to maximize long-term growth of free cash flow per share. We believe this is the best metric to judge our performance and generates long-term value for the owners of the company. Our strategy to achieve this objective has three elements: first, a great business model that is focused on analog and embedded processing products and built around four sustainable competitive advantages, advantages that we continue to invest in and make even stronger. Second, discipline in allocating capital to the best opportunities. This spans how we select R&D projects, develop new capabilities, invest in manufacturing capacity or how we think about acquisitions and returning cash to our owners. And third, striving to constantly increase our efficiency, which is about achieving more output for every dollar of input. Our strategy is designed around four sustainable competitive advantages that, in combination, provide tangible benefits and are difficult to replicate. First, at the bottom of this slide, we start with the foundation of manufacturing and technology. This provides us with lower costs and greater control of our supply chain and provides our customers with geopolitically dependable capacity. Our second competitive advantage is a broad portfolio of analog and embedded processing products. These products provide us more opportunities per customer and more value for our investments. Third, the reach of our market channels, including our sales team and ti.com. This provides us access to more customers, projects, socket per project and insight into their needs. And lastly, we have diverse and long-lived positions, resulting in less single point dependency and longer returns on our investments. With that, I'll turn it over to Rafael, and he'll review our approach to capital management and the scorecard.

Thanks, Haviv. We have shared our capital management scorecard with you since 2013. You can see that the scorecard includes descriptions for our long-term objectives for each metric as well as the target range. The long-term objective provides insight into how we make decisions and run the business as opposed to only a number or a range. In 2025, we again met our objectives. Capital expenditures were about $4.6 billion and cash return was about $6.5 billion, which is a reflection of our continued commitment to returning all free cash flow via dividends and repurchases over time. We are pleased with the consistency of these results over time that have been enabled by our business model, discipline in allocating capital and constantly striving to increase our efficiency. For our 2026 scorecard, we have updated our CapEx targets. We expect CapEx in 2026 to be in the range of $2 billion to $3 billion. For 2027 and beyond, CapEx will continue to depend on revenue and expected growth. Our long-term objective remains the same: to support new technology development, revenue growth and extend our low-cost manufacturing advantage. We have also updated our inventory days target to 150 to 250 days, which allows us to meet our objectives of high levels of customer service through a range of market conditions by providing competitive and stable lead times while minimizing inventory obsolescence. Turning to cash management and debt. Our objectives are unchanged. We plan to fund the recently announced acquisition of Silicon Labs through a combination of cash on hand as well as debt financing that we will arrange prior to closing. Silicon Labs will enhance our leadership in embedded wireless connectivity solutions, and we expect it to close in the first half of 2027. Our commitment to return all of our free cash flow over time is unchanged, which includes a sustainable and growing dividend as well as repurchases when it is accretive to future free cash flow for long-term owners. In the 10-year period spanning 2016 to 2025, we have allocated about $109 billion of capital. Given that magnitude, you can appreciate why capital allocation is a job we take quite seriously and one that has a significant impact on owner returns. Our largest category of capital allocation, about half of the total, has been investments in critical areas that drive organic growth such as R&D, sales and marketing, capital expenditures and inventory. For reference, R&D and capital expenditures have accounted for the majority of our investments over this 10-year period. As we previously mentioned, we have long had a commitment to return all free cash flow to owners over time via dividends and repurchases. And finally, potential acquisitions are evaluated through two primary factors. There must be a strategic match, and they must meet certain financial objectives. These factors remain unchanged. For simplicity, we have not included changes in net debt, which over this 10-year period increased about $8.3 billion. Now I would like to turn it back over to Haviv to share additional insight into our growth expectations.

Thanks, Rafael. First, let me make a few comments about the overall market environment. This slide, which we have shared for the last few years, shows all semiconductor units shipped, excluding memory on a trailing 12-month basis over the past 30 years as reported by WSTS. You can see here that market recovery is continuing, though the slope of recovery is more modest when compared to previous upturns, with units shipped still below the historical trend line shown in gray. This historical trend line, which grows consistently over time, guides us as secular content growth continues and our confidence in the strategic opportunity remains high. During the last decade, we have worked hard to focus our product portfolio on analog and embedded products and strengthen our position in large growing markets. This includes investments in process technology, package technology and the expansion of our product portfolio. As we discussed in our earnings call a few weeks ago, we reorganized our end markets to include data center. We see opportunities in all of our markets. However, we place additional emphasis on industrial, automotive and data center. In combination, these growing markets made up around 75% of our revenue in 2025 and are gaining momentum. I would like to expand a bit on our broad portfolio of general purpose and application-specific analog and embedded processing products are well positioned to serve the industrial, automotive and data center markets. Starting with automotive, we are continuing to see growing opportunities across our automotive sectors where subsystems with higher content are becoming standard features in more vehicles. We are seeing content expansion across all vehicle types, battery electric, hybrid and internal combustion engines. As a result, our exposure in automotive is broad with growing opportunities across customers and geographies. Turning to industrial. Our positions are diverse and long-lived. We see content growing across many sectors as increased automation, more sensing requirements and increased energy efficiency are expected to continue for the foreseeable future. Our general purpose products and ASSPs are able to serve a broad array of sockets in the industrial end market with many designs lasting for decades. Lastly, I'd like to spend a moment describing our position in the growing data center end market. This end market is comprised of sectors found within the walls of the data center, which includes data center compute, data center networking as well as infrastructure related to rack power and thermal management. This end market is supported by a broad set of products from TI. For example, DC to DC voltage regulators, clocks, hot swap controllers, current sensors, interface products and point-of-load controllers are just a few of the many general purpose products and ASSPs that are enabling data center growth. These products are used throughout the data center, including the entire power tree from high-voltage AC to DC conversion to the intermediate bus, all the way down to the multiphase controllers and power stages. Our product will also play an important role in the long-term transition to 800-volt DC architectures where our GaN technology will drive higher power density per rack. To summarize, we're exposed to the best markets. Our revenue has grown from about 43% to about 75% in industrial, automotive and data center, which will be excellent market for our long-term growth. Second, we have a stronger product portfolio. The breadth of our analog and embedded processing products, which span both general purpose and application-specific, combined with our investments in process and package technologies have strengthened our portfolio offering. As a result, our exposure to large, fast-growing markets and our strong product portfolio positions us for growth. Next, I'd like to spend some time discussing our progress in strengthening our competitive advantages. To start, I'll update you on our manufacturing and technology competitive advantage. I mentioned earlier that for each of our competitive advantages, we work to ensure that they provide tangible benefits and are difficult to replicate. Our investments in manufacturing and technology helped to extend our cost advantage and give us greater control of our supply chain. Today, we will provide a recap of the progress towards our capacity road map that will support growth over the long term. Before we do that, I'd like to provide some insight into the benefits of owning and controlling our supply chain and the benefits of 300-millimeter. There are several benefits to owning and controlling our supply chain. First, these investments provide the capacity necessary to support growth. Second, we have more control of our supply chain as more than 90% of our wafers and assembly will be manufactured internally. Third, our process technologies, which are focused on 28-nanometer to 130-nanometer, are optimized for analog and embedded processing products. Lastly, we have a structural cost advantage because of our increasing 300-millimeter wafer fab footprint. All of these benefits allow us to deliver geopolitically dependable capacity with equipment and process technologies that last for decades. This example, which we have shared for many years, is an illustration of the cost benefit of internal 300-millimeter wafer production. In addition to the benefits of moving to 300-millimeter, we also continue to benefit from moving externally sourced products to our internal manufacturing, which are more cost effective. The combination of internal manufacturing along with 300-millimeter wafer production provides lower cost, and we will be increasing both our percentage of wafers produced internally and on 300-millimeter over the next several years. Now let me remind you of the phases we outlined for our 300-millimeter capacity investments. The focus of Phase 1 was to support 300-millimeter transfers and incremental growth by equipping RFAB2 and LFAB1. Phase 2 was focused on new fab preparation and primarily included long lead time work that spans several years before a fab can produce a single wafer. This is what allows us to be in a position of modular capacity expansion or Phase 3. In Phase 3, we will equip and ramp fabs according to customer demand without any requalifications. We have executed well on this road map, delivering on time, on budget with high levels of efficiency and have transitioned into Phase II across the majority of our manufacturing sites. This will allow us to scale CapEx according to demand and deliver free cash flow per share growth across a range of market conditions. Our 300-millimeter wafer fab manufacturing investment spans across three sites in Richardson, Lehi and Sherman. Here, you can see more information about each fab. In RFAB2, we completed transfers from our 150-millimeter fabs and RFAB2 is ramping towards full build-out, more than doubling the capacity of RFAB1. LFAB1 continues to ramp with new products in 45-nanometer to 65-nanometer process technologies as well as transfers from external foundries. In addition, qualification of our latest 28-nanometer process technology is underway. As a reminder, in the future, LFAB will play an important role in manufacturing products from Silicon Labs following closing of the acquisition. Lastly, the SM1 clean room is complete with production underway. It will ramp according to customer demand. The SM2 shell is also complete, which eliminates construction lead time for future expansion. This 300-millimeter footprint in combination with our existing 200-millimeter wafer fabs and back-end assembly and test facilities provides our customers with geopolitically dependable capacity. Our CapEx investments over the last several years have been important to position the company for growth. As we are nearing completion of our 300-millimeter capacity expansions, we are reducing our CapEx and expect to spend about $2 billion to $3 billion in 2026. As a reminder, these CapEx figures do not include CHIPS Act benefits. To date, we have received $630 million in direct funding, including $555 million this quarter based on completion of milestones related to our U.S.-based capacity expansions. In 2027 and beyond, CapEx will be determined based on revenue levels and expected revenue growth. Finally, at the bottom of the slide, we have highlighted several key metrics that this road map is already beginning to deliver, which will continue through the end of the decade. In 2025, we continue to make progress on growing the percentage of wafers and assembly manufactured internally. By 2030, we expect more than 95% of our wafers to be sourced internally with more than 80% on 300-millimeter. We also expect to assemble more than 90% internally. Overall, this will provide us with lower costs and greater control of our supply chain and provides our customers with geopolitically dependable capacity. Now I'd like to make a few comments about R&D. We allocate our R&D investments to growth opportunities to strengthen our technology and product portfolio while improving diversity and longevity. On this slide, we summarize the current direction of our R&D investments and our revenue breakdown by end market. For the revenue breakdown, we have provided data for 2013, 2024 and 2025, so you can get a sense of how the portfolio has changed over the longer term as well as compared to last year. We can find great investment opportunities in all of these markets. As shown in the second column, the direction of our R&D investments is consistent with prior years. Our investments in R&D are biased towards industrial, automotive and the data center end markets and continue to be up broadly. This reflects our belief that these end markets will be large and fast growing due to semiconductor content growth in industrial and automotive and the overall investments and growth in the data center market. Personal electronics and communications investments remained steady. Here, you can see the strategic progress we have made in industrial, automotive and data center. In 2025, those markets combined represented about 75% of TI's revenue compared to just 43% back in 2013. Success in these 3 markets require a long-term commitment and the willingness to invest broadly across sectors and product categories, both of which we have done and continue to do. I'll now turn it over to Rafael to discuss free cash flow per share growth and cash returns.

Thanks, Haviv. First, it is helpful to consider how our operating cash flows are enabling our long-term investments. Specifically, operating cash flow in 2025 was $7.2 billion, an increase of about 13% from last year as we began to see recovery across our end markets. At the same time, CapEx was $4.6 billion or 26% of revenue. As shared previously, in 2026, we are reducing CapEx as we near completion of our capacity expansions. We have often said that the best measure to judge a company's performance over time is the growth of free cash flow per share as that is what drives long-term value for our owners. Here, we are showing our 2004 to 2022 free cash flow per share trend line continuing at the same rate through the end of the decade. In 2025, free cash flow was $3.23 per share, an increase of 97% from 2024. As you can see, free cash flow per share is trending up and beginning to approach the trend line in 2026 as growth returns and CapEx begins to moderate. We are on track to deliver more than $8 per share of free cash flow in 2026. After 2026, free cash flow per share growth will be driven by revenue growth and our CapEx strategy. This underscores the strength of our business model, including the scalability of CapEx with modular capacity. This will allow us to deliver free cash flow per share growth aligned with the long-term trend line. And finally, long-term free cash flow per share growth will continue to guide our capital allocation decisions. As mentioned before, our long-term objective is to provide a sustainable and growing dividend to appeal to a broad set of owners. For 22 consecutive years, we have steadily increased our dividend, including a 4% increase in 4Q 2025. These increases represent 8% for five-year and 15% for 10-year compound annual growth rates. As of February 20, 2026, the dividend yield was 2.58%. Our objective in repurchasing shares is the accretive capture of future free cash flow for long-term owners. While the ultimate assessment of return on investment of these purchases depends on the future cash flow stream, the track record of this approach is encouraging. We have reduced shares outstanding 47% since 2004. We ended 2025 with about $20 billion in open authorizations, having bought back about $1.5 billion worth of stock in 2025, a 59% increase from the prior year. With respect to cash returns, in 2025, we returned $7.13 per share. Over the last 10 years, we have returned a total of 130% of free cash flow. Returns have grown at 13% since 2004. The strength of our balance sheet will allow us to maintain our commitment and track record of returning all free cash flow over time. It may be helpful to frame our performance versus others in the S&P 500. Our free cash flow generation puts TI in the 52nd percentile and is a reflection of our decisions to invest to make the company stronger for the long term. Underlying this is our operating cash generation as a percent of revenue, where we rank in the 87th percentile. Our cash returns puts us in the 94th percentile and return on invested capital in the 70th percentile when compared to the S&P 500. We believe our performance versus the S&P 500 is a reflection of our focus on growing free cash flow per share over the long term and the three elements of our strategy. First, a great business model that is built on our four competitive advantages, advantages in which we are continuing to invest and make even stronger. Second, discipline in how we allocate our resources, focusing on the best product opportunities as well as areas that strengthen and leverage our competitive advantages. And third, striving to constantly increase our efficiency, which is about achieving more output for every dollar of input. We believe if we can continue to do these three things well, we should be able to grow free cash flow per share for a long time into the future. With that, let me turn it back over to Haviv.

Thanks, Rafael. Let me wrap up my prepared remarks with a few summary comments. As engineers, it's a privilege to get to pursue our passion of creating a better world by making electronics more affordable through semiconductors. We were fortunate that our founders had the foresight to know that passion alone was not enough. Building a great company required a special culture to thrive for the long term, and we continue to build this culture stronger every day. We will remain focused on the belief that long-term growth of free cash flow per share is the ultimate measure to generate value. We will invest to strengthen our competitive advantages, be disciplined in capital allocation and stay diligent in our pursuit of efficiencies. You can count on us to stay true to our ambitions to think like owners for the long term, adapt and succeed in a world that's ever changing and behave in a way that makes us and our stakeholders proud. When we are successful, our employees, customers, communities and shareholders all win. Thank you. With that, I'll turn it over to Mike.

Thanks, Haviv. Operator, you can now open the line for questions. In order to provide as many of you as possible an opportunity to ask your questions, please limit yourself to a single question. After our response, we will provide you with an opportunity for an additional follow-up. Operator?

[Operator Instructions] Our first question comes from the line of James Schneider with Goldman Sachs.

I was wondering if you could maybe talk a little bit about how you're expecting 2026 to shape up from both a free cash flow and inventory perspective beyond what you commented on the slides. I think you talked about at least $8 in free cash flow in 2026 at the current revenue consensus. I think your slide sort of shows a wider range extending to $11 or $12 at the top end. Maybe just kind of frame how you're thinking about sort of the low and high end of that and your confidence in achieving over $8 this year.

Thanks, Jim. I'll start, and Rafael, maybe can provide a little bit more color. First, the -- at a very high level, it depends on revenue. And we are not going to call the revenue for the year. We are one quarter at a time. We gave the guidance for Q1. And I think in Q1, the year-over-year growth is about 10% at the midpoint. And of course, we need to let it play out. We are just out of Chinese New Year, and we'll have more visibility as we go into March. Having said that, as we put it on the slide, our confidence is high. The company needs to grow, obviously, to reach $8 of free cash flow per share in 2026, but it doesn't need to grow a lot. Once you get into this kind of mid- to single-digit growth to maybe 10%, we should be in that range. But we also are prepared for other scenarios. If the market wants to really grow fast in 2026, and we'll have to see how it develops, we can react to that and then free cash flow per share will be higher. On the CapEx side, and again, this is where Rafael can provide more color. As long as the growth is similar to last year, I would say it should be more on the lower side, maybe close to $2 billion for 2026. But if we see high growth developing during the year, we would, of course, want to be ready, and we will take it to the higher level, maybe closer to the $3 billion. So that's why you see a range over there. Rafael, anything else to add?

Just beyond that, that $8 that we talked about in the prepared remarks that is based on revenue on consensus for the year. So you can do the math if you want to model higher or lower, just use 75% to 85% fall-through on revenue to do that, and then you can model that away. You mentioned -- Haviv already addressed CapEx. You also mentioned inventory. We're pleased with our inventory position, and we'll just continue to adjust that depending on expectations for demand.

James, do you have a follow-up?

Yes. Just a clarification on Rafael, what you just said. Just in terms of the inventory range now being 250 days at the high end. I believe last quarter, you sort of indicated that you wanted to take factory loadings down a little bit. I think your inventory is around 208 days, if I recall correctly. So I'm kind of curious, under what circumstances would you want to run sort of at the higher end of the range if you thought 208 days was a little bit too much.

Yes. Just to clarify, that comment that you attributed to me, that was two quarters ago. That was the October call. And in the last call, we said we're comfortable with our inventory position, and we'll continue to operate there. And to try to answer your question, the days, that's probably more of a function of when you're in -- where you are in the cycle, whether you're in the upturn, the downturn, the peak or the trough, and that kind of drives your inventory days more than anything else.

All right. Jim, thanks for the question. We'll move on to our next caller, please.

Our next question comes from the line of Harlan Sur with JPMorgan.

Thanks for hosting this call. Haviv, with the renewed focus on embedded over the past sort of five to six years, you obviously added to that embedded capability with the recent acquisition of Silicon Labs. But as you've expanded your mass market strategy embedded, what's been TI's recent track record of attaching analog and power management products to new embedded opportunities and vice versa, right? Any way to quantify analog and power dollar content attached to new embedded opportunities?

Yes. I will provide, Harlan, a similar answer to the one I provided during the call we had earlier this month related to the Silicon Labs acquisition. It's always good to start from the embedded processing socket, whether it's a wireless connectivity solution, whether it's an MCU, a low-power MCU or a very advanced high-performing MCU all the way to low-power processors and DSPs, right? So, as I mentioned, customers always like to start with the center of the board. And in many cases of our applications across the analog and embedded market, that is the embedded processor. And we are seeing definitely with the growing portfolio of our embedded processing parts, and we are just releasing to market every year more and more parts. I think we are now at a 6x rate versus six years ago. We are seeing more opportunities to attach the periphery, mainly analog parts, whether it's power or signal chain around that, what I call Alpha socket, right? So we are seeing this opportunity. We are seeing this on our assigned accounts, where usually the first discussion with the customer is around that embedded processing socket. But also on ti.com, we are seeing more and more looks into our system solutions that are displayed over there. We also have some -- a whole product portfolio that provides software, application code, sometimes even the whole module or board that we provide to our customers, and we are seeing an uptick there. I will also say to me, it's only the beginning, right? Our embedded processing strategy has pivoted 6 years ago, but we have so much work to do. I mean the backlog here, the creative backlog we have in front of us is very, very broad. And we are getting our machine to become more and more efficient to go after it.

Do you have a follow-up, Harlan? We'll move on to our next caller, please.

Yes. So with SM1 cleanroom complete, Qs complete, early ramp this year, SM2 shell complete, LFAB2 clean room continued build-out, you're obviously executing to a number of CHIPS Act milestones, right? You've got the CHIPS Act ITC, which is 35%, still have $1.6 billion in direct funding or grants to capture. I think last year, you guys captured about $335 million of CHIPS Act incentives. If you hit your build-out and ramp targets for this year, how much in CHIPS Act incentives do you target to capture in calendar '26?

We don't have a forecast per se for '26, but let me first kind of review the actuals, what I can tell you. In '25, we received cash benefits of about $670 million. That is primarily ITC. There was some direct funding there, about $75 million of direct funding and direct funding meaning the grant. Moving forward, we expect -- we continue to expect ITC now at 35% because last year and the years prior to that was 25%. So now it's 35%. And the remaining up to $1.6 billion in direct funding, including what we released in the most recent 10-K that we just received $555 million this quarter, in first quarter from the grants. So, beyond that, what I could tell you is going to depend on CapEx, and U.S. CapEx. So the ITC is 35% of U.S. CapEx, and then the direct funding, the remaining of the $1.6 billion minus the $75 million and $555 million that we already received.

Thanks for the question, Harlan. We'll move on to our next caller, please.

Our next question comes from the line of Stacy Rasgon with Bernstein Research.

Haviv, you talked about thresholds or criteria for acquisitions, strategic as well as financial. I was wondering if you could go into a little bit into those financial conditions. I guess, how do they compare now versus what you used to talk about those in the past? I have to assume that the returns that you're willing to live with today are lower than they were before. And I guess how does SLAB like fit into those financial criteria? Like what is it about SLAB that actually meets those?

Yes. First, just I would say that the threshold has not changed. I think the position of the company has changed, Stacy. And we always said from our point of view, acquisitions will be centered around analog and mixed signal. I would put Si Labs in the mixed-signal bucket. And first, it has to make sense strategically. And as we discussed earlier this month, we love the portfolio that we saw from Si Labs. We love the position they have. Both market position, 85% in industrial across many, many end equipment. And at a very nice rate of growth of 15% at the market that I believe is still in early phases of adopting this technology. There is a secular growth around wireless connectivity as people are adding more and more information into their system. And the best way to deliver that information on already installed infrastructure is wirelessly, as you know. We also love the assets in terms of the technology. I mean, the engineers, the offering across hardware, software, application code tools, which is something that we wanted always to augment in our embedded processing. Now this is where Rafael keeps us very honest on every acquisition we look at, it has to make sense financially, and it was hard to make the numbers work five years ago when we had to keep these wafers running out of the foundries in Taiwan. So, in that sense, that's the main change that happened in the last few years. I have been, again, looking at Silicon Labs for many years. but it didn't make sense before. And now when you add OpEx efficiencies to the OpEx efficiencies, the COGS efficiencies, the number just add up. And I'll let Rafael comment a little bit more on that.

Yes. I'll just add that we have always looked at an acquisition on the financial side from a lens of return on cost of capital and that it's accretive to cost of capital. So, within three to five years, and that is the case here. And return accretive to cost of capital, very simply, our weighted average cost of capital is about 10%. If the acquisition is $100, can we get $10 of free cash flow on a consistent basis starting on year three, four, five. And that is the case here with our expectations that we will get at and above and then above our cost of capital within that time frame.

Do you have a follow-up, Stacy? Stacy, do you have a follow-up?

Apologies. Yes. I wanted to just clarify just so I know -- have a good feeling where the loadings are. So, RFAB2, I guess, is now in full production given the 150-millimeter transition. It sounds like LFAB1 is in full production and Sherman 1, the shell is done and you've got pilot production and everything else is shells, I guess, LFAB2 and Sherman 2 and everything else. Can you just clarify, do I just have that right?

Not exactly. Maybe at full production to the installed 2, okay? If you go to RFAB2, we've installed, I think, more than 75% of it, close to 80% or 85%, but we still have a little bit of clean room over there, Stacy. And this is where we are in Phase III between RFAB2 and then SM1. We have the clean room, and we will install tools as demand picks up, okay? So that's on the RFAB2 case. On LFAB1, we are still in the process. We have a big chunk of our revenue is in 45-nanometer, okay? So over there, there is actually more clean room available for us. And I think you know our 45-nanometer footprint is mainly on automotive, ADAS, high functional safety type of product. It just takes longer. That work is moving well and is scheduled to complete by the end of the year. So that allows us to continue to fill the clean room in LFAB1 according to demand. And just when we do that, and it depends, of course, on demand, LFAB2, shell is going to be ready towards the end of this year, beginning of next. And then we can build into it without further qualification. So that would put Lehi in Phase 3 as well. Hopefully, that helps.

Thanks, Stacy. We'll move on to our next caller, please.

Our next question comes from the line of Vivek Arya with Bank of America.

So, let's say, if you grow sales 10% this year and 10% in '27, what is conceptually the CapEx for '27?

Yes. I think I've answered it for 2026, Vivek, and it's going to be very close to that $2 billion number, as I mentioned, simply because we have always some -- we don't run the fabs at 100%, and I mentioned some of the clean room available. So -- that would be the lower number. Again, if you look at 10% this year and even 10% next year, that's not a huge load on our company in terms of capacity expansion. Some of it maybe we will have to go into the back-end factories where we are more tight because the lead times on the back end are always lower to get tools. But in terms of the fabs, I think it's still going to be probably at similar levels in 2027 and maybe even lower. I think we'd have to leave that for next year's discussion.

Do you have a follow-up, Vivek?

Yes. And second question is, what is your inventory optimization strategy? Is the goal to keep inventory kind of at this constant, I think, $4.8 billion or so and then the days are -- whatever the days are. And then as demand increases, you push up utilization, but you still kind of keep at this $4.8 billion. You raised the target number, so I understand the goal is to be prepared for any kind of growth. But still $4.8 billion is a lot of inventory to have. So, I'm curious, are you targeting days? Or are you trying to make sure that you always have about this $4.8 billion of inventory on the balance sheet?

Let me start. I think Rafael touched upon it, and just from a high level, and Rafael can be maybe more precise than me. But in general, we think about it as days rather than revenue first. And the second thing I would say you see a range. And again, nothing -- no cycle is a perfect sign wave going into -- up into the right, right, Vivek. But the way I think about it high level, you build inventory on the down cycle to a higher level of days just because you want to prepare for the future. And then as demand ramps, now it depends on the slope, you deplete it. You deplete because there is cycle time on the fab that is like three to six, sometimes nine months to get the chip out. So the inventory serves for the surge of demand. As I reminded so far, the -- if you go back to the trough in 2024, it's not been a very strong recovery. It's moderate, right? So right now, we are not challenged, and you still see the days high, but who knows what 2026 needs to do. So this is why you see the 150 to 250, and I'll let Rafael provide more color on the mechanics.

Yes. I'll just add a couple of things. Remember, the key is our inventory objectives, which are to maintain high levels of customer service, keep lead times stable and minimize obsolescence. Keep in mind that we're a different company than we were a few years ago. We have more industrial, automotive and data center. We have more direct business, and we have now more internal loadings, right? So all of that supports higher levels of inventory, everything else being equal. So we're very comfortable with the levels of inventory that we have now.

Thanks for the question, Vivek. We'll move on to our next caller, please.

Our next question comes from the line of Joshua Buchalter with TD Cowen.

Appreciate you guys hosting the call. I wanted to start with a big picture one. So you've posted the metrics on Slide 10 a few years in a row now. Can you maybe just walk us through why you think this recovery has been more gradual than others? And what sort of signals that you guys are looking for and we should be looking for of when we would get that sharper recovery where shipments can actually start hitting that trend line?

Yes. I think the why is related to what happened before the recovery, right? Think about the cycle of COVID, probably the most pronounced or a very strong rate of demand growth in the -- coming out of COVID with a trough in 2019 and then a boom in '21, '22. Then a very -- customers got anxious, built a lot of inventory and I guess, more than we even thought originally, although they all swore that it's not double ordered and they did it for real products. We saw a very large and prolonged inventory correction. That's just the size of it is historic. The second thing is the asynchronous nature. We all talked about it. It's a very unique cycle because markets behave differently simply because the way we consumed electronics during COVID, that's kind of dispersed the behavior. I think that's what makes everything longer and more moderate. And where we are right now, high level, we are, in essence, industrial automotive, data center company. That's our focus is 75% of revenue. As we said, although automotive was late to the cycle, there is so much content growth in automotive. the peak to trough was single digits, if you will, high single digits. And we already hit the level of the previous peak back in, I think, Q3 of 2023 -- sorry, 2025. And it's just doing well. Of course, we'll have to see how 2026 behaves, but we are continuing -- as I said in my prepared remarks, we are continuing to see generation to generation, just more features per vehicle. It doesn't matter what the powertrain is. We're just seeing that secular growth continues, and I think it still has a lot of runway in front of us. That's on the automotive side. Industrial is a big unknown. It's a big unknown because it's still with all the -- it grew nicely in 2025, I think double-digit growth, about 12%. It grew nicely in Q4, close to 20%, but I think it has a lot of room to go, okay? We are still trending about 25% from the peak. And in our case, other than -- we have about 10 sectors other than aerospace and defense, they are all like 30%, 35% away from the peak. So, to me, there is -- that's the big opportunity, if you will, in 2026. And I think we said it also in January. We are seeing some of it coming in, right? Our booking accelerated in Q4. So far, Q1 trended as expected on the industrial side. So I think there is a lot of opportunity here because, again, secular growth in industrial is continuing and maybe in early innings. The last one, I would say, data center is unique in the sense of it's just growing, right? I don't think we need to think about the cycle here. This is a secular growth because of the investment in data centers. As long as CapEx continues to grow and customers care about just the energy and energy efficiency and energy density, you will start -- you will continue to see a big opportunity there. We left Q4 with data center running at about $450 million, about 10% of our revenue, but the pace is very high. So I think this continues into the foreseeable future as long as AI and what it can bring in terms of efficiency to the economy continues, I see that as that trend continues. Hopefully, that helps. Mike, anything else on this one or...

Josh, do you have a follow-up?

Yes, please. Thank you for all the color there, Haviv. As we get out of Phase 2, I was hoping you could maybe comment on what's a reasonable floor for CapEx. So if I look back from 2015 to 2020, your average CapEx per year was around below $750 million. I mean, obviously, costs have gone up. But if we got into an environment where growth was slower than you guys anticipated and you didn't need to put in incremental capacity for Phase 3, what's a reasonable floor for how low CapEx could get in that environment as we think about a floor for free cash flow?

Yes. Again, first, Rafael, I think, made some important points about how much we've internalized our capacity, right? This is not only on the front end, also on the back end. And I call it maintenance CapEx. But even when you don't grow your overall revenue, there is always a mix change. And there is always one package you need more and the other package, you have what you need. So that is why I call it, there's always going to be some sort of a floor of CapEx. I kind of said I think 4% to 5% is a good number of revenue, but you don't need to grow just because you have to maintain your tools. You continue of course to go after packages and specific wafers that you are a little short about. There is always hotspots in our operations. And in general, we do -- we are going to do more than 90% internally. I think 95% or more on the wafers and probably a similar number on the assembly and test. So this is why you will see us running at probably that "floor". Rafael, anything else to add here?

No, I agree.

All right. Thanks.

Thank you, Josh. We'll move on to our next caller, please.

Our next question comes from the line of Tore Svanberg with Stifel.

Thank you for hosting the call. Maybe to follow up on the last question instead of perhaps dialing into a specific number, I think in the past, you've talked about maybe CapEx being around 3% to 5% of total revenues. I assume maintenance CapEx is kind of what that means, right, in relation to your revenue. So is that kind of the ballpark number we should think about, about 3% to 5% of revenues per year, obviously, starting in '27?

No, I think just to clarify, again, when I talked about, let's call it, 4% of revenue is when you don't need to support your growth, right? When we need to support the growth, I will let Rafael -- I think there is almost an equation that we can give you when you think about long-term growth and what CapEx needs to do. Rafael will comment on that. But when you just have to run your factories, because we are now more -- running more of our capacity internally, we kind of modeled it internally at 4%. It could be, as you said, maybe a little higher, maybe a little lower. We'll have to see. Just because you have to maintain customer support, there is always areas where customers need more of certain parts, certain technologies, certain package types, modules, whatever you can think about the different permutations of our portfolio. And we will have to serve it. And of course, you have to maintain your factories to operate at a high level and some of it is just replacement and maintenance. And Rafael, can you talk about what -- how to model long-term growth on the related CapEx?

Yes. So, as we've said before, over time, as we run in Phase 3, the modularity phase, what -- the way you should model our CapEx is taking our growth rate, long-term growth rate over years and multiply that times 1.2 to get your CapEx as a percent of revenue. So, for example, 10% growth rate would equate with 12% CapEx as a percent of revenue. And now keep in mind that, that 12% in that example, that is gross CapEx before ITC benefits and before DFA.

And just to remind everyone, this is not a math, right? There is always lumpy, but it gives you -- if you put it over the long term, that's what you will get, Josh, okay?

Thanks, Tore. Do you have a follow-up?

Yes. No, that was very helpful. And as my follow-up, could you just square the circle with your segments? I think you pulled some percentages out of communications to sort of adjust for the enterprise system versus data center. Was that kind of more communications exclusively related to data center infrastructure?

Yes, absolutely. And there is a Slide 11 that shows that for you, so you can actually see the -- from two, okay, slide. But in general, when you think about comms equipment, it was all about optical communication within the walls of the data center. People can choose how to define it. We define data center that everything within the building, okay? The same we do for automotive inside the car, there is communication technology also going as we know, right? But the way we define the new market is the clarity is whatever goes inside the data center, okay? So comms, it used to be 5%, now it's 3%. That's a move on the optical communication technology moving into data center. You see industrial "losing a point". This is because of power delivery. The power supplies that we used to have under industrial in power delivery are now very unique for the data center. and they're not general purpose PSUs, and this is why it's smart to move it inside there. And the last one was enterprise, right? Most of the data center revenue was already in enterprise. Most of it was what we call enterprise compute. But now you see data center at 9% of revenue simply because of the contribution from comms and industrial. Hopefully, that makes sense.

I would just add that there was a small portion that moved out of enterprise. There was multifunction printer, enterprise machine projectors that moved into personal electronics, which is why you see it went up a little bit. Tore, thank you for the questions. We move on to next caller, please.

Our next question comes from the line of Matthew Prisco with Cantor Fitzgerald.

I guess to open, I would like to start on your thoughts on evolving customer relationships, potentially shifting to more system-level solutions versus historically more discrete kind of as you talked about with your ti.com and what you're seeing. TI, obviously, great product breadth. But how do you think about the ability to provide those more full stack solutions, including the software and tools where you can work with customers to actually solve a problem? And are there any changes in R&D focus that we should be thinking about around this dynamic?

We will always -- there is no change here because we've been always selling a complete solution. I think the evolvement of our portfolio is very helpful. We've talked about it around the embedded processing question before. It's always good to start from the big socket, right? If you have a processor, if you have a DSP, if you have a wireless connectivity solution, if you have an MCU, that's always a good start. But I also would caution all of us that you need to have the best parts. Everybody can talk about solutions, but customers will take the best part for the socket. Even if you have the complete solution, they are not looking for -- to cut their R&D. They want to add value, and they will always select the best socket. Even in general purpose parts, if you don't have the right interface part at the lower size at the lowest power at a very affordable cost, customers will take you out of the solution. So I always caution my team that we cannot just rely on system selling. We do have to refresh our general purpose parts. Of course, they usually get refreshed not as often, but you want to have the best in terms of size, power, cost. And the same on the application specific. The fact that you have a complete board doesn't mean that you get selected. The best part still wins. So that's the way we think about it almost in three vectors: customer, part, system. And that's the way we go after business in TI. That's been the case for more than 10 or 15 years now.

Do you have a follow-up, Matt?

Yes. Last quarter, you talked about depreciation being higher year-over-year. It would be great to hear if you have any quantification you can offer there. And then as we just think a little bit longer term, I mean, I think you have a roll-off of the RFAB2 and LFAB1 equipment, the initial installs there. But at the same time, you're bringing on new capacity. So does depreciation -- does that become a tailwind to margins at some point in the near term as we kind of see those dynamics play out?

Yes. So I'll -- let me restate what I said at the last earnings call, it's the same. No change in that guidance, but it's -- our actual for 2025 came in at $1.9 billion, about as expected. For 2026, we expect $2.2 billion to $2.4 billion. That range. That range is lower than what we had told you before, but it's what I mentioned in January. And then for 2027, we expect upward pressure on that number, but at a lower rate of increase. So $1.9 billion going up to $2.2 billion to $2.4 billion, take that midpoint, that $2.3 billion, that's a $400 million increase. We expect another increase, but at a lower rate than that going into 2027. And beyond that, it's going to depend on CapEx, right? So it depends the more CapEx we spend, then there will be more depreciation. But of course, we will spend that CapEx only to support longer-term revenue growth.

Matt, thanks for the questions. We'll move on to our last caller.

Our last question comes from the line of William Stein with Truist Securities.

First question is about the M&A strategy. Of course, that's a very important use of CapEx that we've seen come up recently with the Silicon Labs deal. The question I'd like to ask about it, someone asked earlier about the financial hurdles and how they might have changed, and I think you said they haven't. But in terms of the strategic hurdles, I'd really like to understand how potential deals enter the funnel because it's my understanding that this deal came about after another company bid on Silicon Labs and that triggered TI's involvement, which I found sort of odd considering the very long duration between this deal and the prior one. Maybe talk about how TI strategically thinks about deals and the process through which they get into your funnel of potential deals? And then I have a follow-up, if I can.

Yes. So, again, I think I mentioned before, and I'll let Silicon Labs disclose what exactly happened in this case. I don't want to comment about that. I would just say that I've been looking at this asset for a while, and the interest was growing over time as we brought our own wireless connectivity and MCUs internally into Lehi and really like the results. And the asset, as we said, is a very high quality. Now regarding timing, you can never be -- it's never perfect, right? So it's something that you have to look at each opportunity case by case. And you if it makes sense strategically and to the owners of the company in terms of the financial impact. So, both criteria were met, and this is why we decided to make the deal.

Will, do you have a follow-up?

Yes, please. On another topic, you talk about investments in R&D, and that's very helpful. But one area that we haven't heard very much about is how TI could use artificial intelligence to either accelerate revenue growth or reduce costs. For example, I could imagine in product development where analog is known to be sort of talent constrained. Can you perhaps highlight for us how you're using AI internally at TI to improve your business?

Yes. I think -- again, I think we mentioned it maybe in previous calls, but I will just give you the three big ones that we -- I'm running at my level. Of course, there are many, many other smaller projects in the company. But I think the most interesting one for us is the top line growth, right? We have a very broad portfolio, and we call directly on less than 2,000 customers. So we have about 200,000 other customers that we never see. And this is where think about the technology that we have, the data, think about ti.com, the amount of just actions and people coming on our website and looking for parts to make sense of all of it, you really need these AI tools. I think it has been done in many, many other industries before semis, but I think we have the scale to exactly do that. That's part of the reason we've offered all these e-commerce capabilities in the last several years. It gives us more, I call it, golden data that you can act upon. So we are already running projects here, and we are seeing a clear ROI, if you will, on this investment on product recommenders and how you can do system selling for the customers you don't call upon. Now you can argue that over time, they can also do a better job than your sales team. But so far, it's not there, okay? So far, when we call on a customer, we do a great job. I would like the agents, if you will, to get to at least to show an improvement, and I think we are seeing that. So early phases of that, that's top line growth. The second one is related to CapEx efficiency. Look, there is so much data collected in our fabs and ATs that we couldn't act upon many years ago, just too much information. And we are seeing already how you route the fab, how you do your start plans, how you maintain machines. Let's just quantify it. If you spend $5 billion a year on CapEx and now you get a throughput improvement of 10%, that's $500 million a year in terms of CapEx cost reduction, right? So we are seeing that these projects are coming to fruition. I think we also alluded to it in one of our calls. We used to talk about 1.5x ratio between revenue growth and CapEx in terms of percentage of revenue. We are now talking about 1.2x. Rafael mentioned it before. That's really because of the efficiency we are gaining from the modern tools and the data that you can feed them. And I will say also that we are still seeing opportunities here. We are in the beginning of our learning cycle. We are seeing results, but more to come. The last one is related to OpEx. To me, the way I look at it, very high level. Look, there is a certain level of OpEx and R&D you can afford. So our creative backlog is always large and interesting. So if I can get more output from the same, call it, R&D, that's very, very interesting, right? So we are looking at that. We're already seeing it in software design, a little bit on the RTL, but still not on the analog side. On the analog side, we haven't seen great throughput improvement, but I think we are in early phases. I don't think the AI agent replaces a design engineer in the foreseeable future, but can we get more throughput from our dollar invested related to our efficiency, if you will, part of our strategy. That's something we always strive to achieve. I think there is something there. On the SG&A side, I think we can -- we have a great service. Can we do it for less money? That's always an interesting area for us. So we are in early phases there. And I think that's also something that could be interesting. So think about top line growth, CapEx efficiency and OpEx. These are the three big projects that we are looking at the company right now.

Thanks for the questions, Will. So, before we wrap up, Haviv, do you want to make a few comments?

Yes. To finish the call, I want to thank all of you for taking time today to go through our capital management update. Let me emphasize a few points. First, we remain focused on consistent execution of how we manage capital. Our disciplined allocation of R&D is delivering growth from the best general purpose and application-specific products in analog and embedded processing. Our manufacturing strategy is a unique advantage and will continue to benefit TI for the long term. And lastly, we remain committed to returning all free cash flow over time to our owners. Mike?

Thank you all for joining us today. A replay of this call will be available on our website as well as the slides that were used on this call. Have a great day.