STMicroelectronics Aktie

Good morning, I'm Pavan Daswani, Citi's European tech analyst based in London. And we're extremely glad today to be joined by Lorenzo Grandi, CFO at STMicro. Lorenzo, thank you for joining us today.

You're welcome. Thank you to all of you.

Could you maybe start -- I've got a few questions prepared, and then I'll pass it on to questions from the audience. So could you maybe start with the near-term demand trends? At the recent results, you talked about strong bookings across all end markets, a book-to-bill of close to 2x and improving visibility. Could you talk about what you have seen over the last month or so since the Q2 results in terms of orders, lead times or inventory levels?

What can I say is that actually, the situation in respect to what we were seeing during Q2 and what we said at the earnings release of Q2 is unchanged means that the booking is still very strong. Our book-to-bill remained well above 1. This is true in general, I would say, across the various markets that we serve. So at the end, the situation remained quite healthy. We see a very positive trend, similar to what we had said exiting Q2. No changes -- significant changes in respect to that.

And any trends to flag in terms of end market demand?

Clearly, what we see is that there is a strong demand for the AI infrastructure. This is definitely one trend that is very visible. But I have also to say that when we look at our market like automotive and industrial, still there, we see very, very strong demand. Industrial, definitely now the lead time for our STMicro is increasing significantly for some of this product is even now reaching 50 weeks, so it's important lead time. And also in automotive, will be steadily the demand increasing and the book-to-bill also here well above 1.

And how does that 50 weeks compared to last year?

Last year was completely a different situation. Last year, if you remember, we were still in a situation in which we were having an excess of inventory and distribution, these kind of things. Today, inventory in distribution is super lean, especially for this kind of product.

Okay. So I want to come back on the automotive and industrial trends that you're talking about. But maybe let's start with the AI data center opportunity, which is clearly one of the biggest topics around ST today. You raised your data center revenue ambition to above $1 billion in 2026 and to well over $2 billion in 2027, which does leave a bit of room. What gives you confidence in guiding 2 years out? And what are the key capacity and supply constraints that controls how we should think about 2027?

No. Yes, clearly, let's say, we see a very strong demand. And as you know, as you have -- let's say, we now see our revenue for this year, let's say, above $1 billion in this application. And for next year, we see revenue well above $2 billion. So it means that we see really a strong demand in moving. Clearly, this -- how we serve, let's say, this application. We have served through -- there are three, let's say, the thermal flow, we have the power flow and the connectivity flow. When we look at these three flows, clearly, for us, the strong revenues are coming from the connectivity flow, where we have, let's say, the products that are the silicon photonics, we have the microcontroller. And we have, let's say, the BiCMOS EIC that are for the optical transceiver.

Yes, the demand is very strong. And indeed, what we expect, for instance, for next year, is that mostly of our revenue will be in MDRF for around the 80% of the billing. Only the 20% will be in APMS, where we serve the other two flows, let's say, substantially that are the thermal 1 and the power.

On the power, clearly, let's say, we are marginal today, but we are, let's say, creating our portfolio for the new architecture at 800 volt, and we do expect to start to have a meaningful revenue after 2028 for the power.

Okay. And when you think when that does kind of come in 2028 and when that starts to contribute more meaningful, how do we think about that opportunity beyond that in terms of how you're thinking about market share opportunity, et cetera?

At the end, let's say, what we may say is that clearly, for AI, we have a significant opportunity. We have the ability to have the portfolio that is serving the optical transceiver as we have all the ingredients, including our ability for the packaging, what will come, let's say, the NPO, the CPO, these kind of things. So we think that we are well positioned in this respect.

And then as I said, we have also the opportunity to grow where today, we are not so particularly strong. That is, let's say, when we talk about the power flow, the power delivered from the grid to the MPU, this is where we may have opportunity to grow even if not the short term. Short term will be mainly driven by the connectivity portion.

Yes. So on the revenues this year and next year, silicon photonics, how do you see the market share? I mean, right now, you've got a significant market share in that space. How do you see that evolving? How do you kind of maintain that market share?

Today, what can I say is that we have a significant number of customers. And these customers are asking us to long-term agreement. What does it mean -- long-term agreement means that they are locking, let's say, capacity volumes, pricing and also with cash advance for many of these agreements that we are signing. So I think that it's difficult to me to talk about a number in terms of market share, but I think that now with our positioning, our technology, our PIC100, we are really, let's say, playing a leading role in the silicon photonics.

So you've talked about long-term agreements. You've got your revenue targets this year and next year in silicon photonics. Do you feel your capacity is currently in the right place to meet that?

We have to invest. We have to invest in order to follow the demand. We have the infrastructure to do this because we have the 300-millimeter that is possible for us to increase in slicing and not create a big, let's say, infrastructure, but with this get away to create some capacity increase based on how we see, let's say, the demand evolving. The positive is also that investing in silicon photonics is a little bit -- is not an investment that is only specific for that, is fungible also for other, let's say, technologies like, for instance, the microcontroller. This give us maybe in future, if something change in terms of mix, the ability to use this investment for other products.

It's not something that if you invest in silicon photonics, then you are locked there. And whatever it happens, let's say, you cannot use this capacity for different products in your portfolio.

Yes, we will follow, let's say, we will invest in order to follow the demand.

And maybe that's a good transition into margins, which is probably the other key topic on top of investors' minds for ST. How should we think about the bridge from today's mid-30s gross margins moving back to that mid-40s target range that you have provided?

Well, the bridge is the following, let's say, take the margin where we stand today. Today, we are guiding 37% in this quarter, 37% gross margin. This 37% is impacted by one side, still by underloading charges for around 70 basis points. This progressively will disappear. Then we have an impact of around 50 basis points related to the fact that we have some extra cost due to this reshaping program, duplication of products, duplication of technologies, from one fab to the other fab. But then the big chunk of the improvement will come from the closing of 200-millimeter old fab and two -- 150-millimeter for silicon carbide fabs.

This will give more than 400 basis point positive impact on our gross margin. But then there will be the mix. I don't know, for instance, silicon photonics is accretive to our gross margin target. So these are the way that we may see, let's say, the evolution of our gross margin starting from now.

Yes. So manufacturing reshaping is clearly a very important driver of that margin improvement. What are the key milestones we should be watching over the next kind of 18 to 24 months to kind of gain confidence that things are on track to get that 400 basis points? And maybe if you could talk a bit about the timing of when you expect that?

Clearly, when we talk about the reshaping plan, we have to keep in mind that, let's say, the positive impact of this is a little bit a step function. So to gain in terms of improvement in your cost and your cost, you need to close the fab. And this is a step function, either the fab is open or closed. So when we will be in the position to close the fabs is when, let's say, all our customers will have qualified the products and move the products on the new fab, for instance, from the 200-millimeter to the 300-millimeter.

Today, the plan is in schedule, means that what is in our hands as a company, creating the capacity, moving the process, duplicating the product on the new fabs is done. Now we are sampling our customer and, let's say, waiting for their qualification. When this will happen, let's say, that we will be in the position to close the fabs. And so to see this step-up in our gross margin will be at the end of 2027 and maybe early 2028, depending on when we can really close the fab and to shut down the operation in the old fabs.

Okay. And I guess pricing is the other key driver in that margin bridge, which we haven't touched on in much detail as yet. So could you maybe give us some color on current pricing discussions? What product lines have you seen price hikes and what magnitude? And how should we think about the trajectory of ASPs for 2026 and 2027 as well?

In terms of pricing, clearly, coming back to the point that we were discussing before in which there is definitely an imbalance between your capacity and the demand. Today, we have a backlog and demand that is higher than our capacity, clearly is creating an environment in which you may have the opportunity to increase price. On the other side, you have also to keep in mind that on the input cost, you have a price increase.

How the company here is handling this is clearly, let's say, having a price increase in our top line for our customer that at least, at least, I would say, are netting the price increase in the input cost. In reality, we go a little bit beyond that. Let's say, we increase price more than what are the input cost, but not in a way that is too much opportunistic -- means that at the end, we will have some benefit, but will not be the situation similar to what, for instance, after COVID, in which there was a capacity reservation increased prices significantly.

So we are increasing price. We will have some benefit of this price increase that are more than offsetting our, let's say, input cost. So it means that at the end of this year, for sure, the price will not see declining price in 2026. Overall, in the average, we will be, let's say, substantially flattish, slightly increasing. Remind also that there is a portion of our portfolio, while it is, for instance, in the consumer distribution, small customer, medium customer, increasing price is not so complex, especially for products like our microcontroller, STM32, some analog, some kind of product like that. But there are engaged customer programs and big customers in which increasing price is a little bit more complex also because it's ruled by contracts that are somehow defining the price for this customer engage programs. This is the reason why, at the end, we will see substantially pricing, flattish, slightly increasing as a combination of increased price and maybe other business in which is more difficult to increase price.

So across product lines pricing, at least as much as the input cost increases.

This is the minimum. This is the minimum. Let's say, what we intend to do is really, let's say, to have a little bit of benefit of this price increase.

So you mentioned flattish in 2026. How do you think about that in 2027 as some of these annual contract renewals come up?

Clearly, we are not yet discussing pricing for next year. Clearly, what we have increased this year will stay next year. This is something that will not disappear. But I think that at the end of the next year, now, it's difficult to exactly say where we will position but will be a year in which clearly the price will be stable, slightly increasing -- That is different from the normal situation that we usually see in our sector in which pricing normally are going down, let's say. I think if the situation -- and today, there are all the ingredients to say that will remain as it is today, there will be still a significant gap between the demand and the capacity available.

Maybe it's a good time to pause and ask if any audience have any questions. Yes, in the front.

[indiscernible]

Sorry. Pricing? No, not really because today, there is, let's say, situation is such that at the end, let's say, our limitation is more in our ability to serve the customer than to try to use pricing, let's say, in order to gain, let's say, business. Today, as I said, we are facing backlog and demand that is well above our ability to serve the market. So reducing price and then not to ship is not really that we are thinking to do at this stage.

For your silicon photonics, the business, are you targeting a specific end application like NPO or CPO? Or is that something that's kind of determined based on customers that you're working with?

No. In respect to this evolution NPO, CPO, clearly, we have all the ingredients to follow. This will be driven by our customers. What we have in our hands, we have all the ingredients for this optical transceiver to be in the position not only with the products but also our ability for the packaging this kind of activity on which we are developing now this new application to follow the demand of the customer. So we feel ourselves well positioned in order, let's say, to continue to play a significant role in this activity. And this is what we plan and what we have embedded. We are confident in this respect.

I actually have a few questions. One, just I want to clarify something. When you talk about the gross margin improvement, you talked about multiple things, under absorption, but also closing down multiple factories, and you said sort of by end of next year, beginning of '28, that should be done. Is that going to be linear or just everything is going to remain in place? And then at the end of '27, everything shuts down? Or will there be things shutting down along the way? That's one question. And just if you could touch on trends within the satellite and mobility business, just what you see there.

Now in respect to the gross margin, we have to say the following. Clearly, let's say, we will see moving forward from the 37% of today, progressively some improvement in the gross margin. The message that I want to pass is that the biggest improvement, it comes from the reshaping plan. The biggest improvement, the 400-plus basis points will come, let's say, when we will close the fabs. So it means that at the end, there will be a certain point at the closing of the fab, when the fab will be shut down a step function in the gross margin. This does not mean that moving forward from the Q3, there will be some reduction in terms of underloading charges. There will be some reduction in terms of, let's say, extra cost because at the end, we will end this activity of duplication and so on. So means that mid next year, this extra cost will be very marginal.

There will be improvement in the mix. There will be improvement in the manufacturing efficiency. So we will see some progress over the time, let's say, in our gross margin improvement. But the biggest portion will come when we will be able to close the fabs. Here is where we expect significant change in our gross margin. And this will come when we close the fab. And when we close the fab, will be at the end of 2027, beginning of 2028. Of course, depending on the fact that our customer qualify the product because they need to qualify the product in the new fab. This is now our -- the underlying assumption is the qualification of the customer.

About satellite...

Sorry. On the satellite, this is a business that you know for us has been a significant business. Now today is a business that is reaching close to $1 billion. It's a business in which we will continue to grow in the next 3 years. Of course, now we have a position in terms of market share that is very high. Now with our customers, Starlink, SpaceX, we are now representing around 90% of market share. This is not sustainable. When the business is growing and this business is growing, clearly, in our industry, there will be second source, and these kind of things. Clearly, the good news that we see here is that this market is expanding with new entrants, in which we are well positioned to serve. And this is the reason why is that if one side with our historical customer, we may have, let's say, some decline in the market share even if compensated by the fact that customers is growing. On the other side, we will have, let's say, a new customer coming in. We are already shipping to new customers already today. So what we see for the next years is that our revenue will remain cumulative well above the $3 billion.

And Personal Electronics. Personal Electronic, clearly, this year, we have a seasonality that is different in respect to the past. We have already commented that. Why? Because at the end, there are some factors. One is the different way to introduce the product by our main customer that is not like in the past that everything was done in the fall. Now they think to split, let's say, the way to introduce product. And this, of course, is impacting somehow, let's say, our revenue.

And then there is this new foldable -- let's say, the new foldable, let's say cellular phone that they will introduce that is not have onboarding our image sensor. This is -- anyway, we see, let's say, Personal Electronics they are growing, continue to grow, is a lower pace in respect to other markets, but still growing between low to mid-single digit. So at the end, still contributing to the growth of the company.

And on the satellite business, the profitability? So on the AI, the Photonics business profitability is above the average gross margin. What about on the satellite business?

No, Satellite business is a business that, of course, it depends on the product because clearly, the profitability is different when we talk about user terminal, when we talk about satellite, when we talk about data [indiscernible] different kind of profitability. But then when we look at the average of this business, the average profitability, I would say that in line with our gross margin target. So at this stage, I would say that is accretive.

Any other questions? Maybe coming back to some of your comments earlier on automotive. What gives you confidence that the automotive recovery is more durable across demand, order rates and channel inventory?

Today, what we see in automotive is definitely demand that is strong. This is clear. We see also that this demand is not short term, but is covering more quarters. We see traction in silicon carbide. We see recovery in silicon carbide. This -- especially when we look at the European carmakers, we see, let's say, a positive trend for what concern this kind of -- we do not detect any inventory buildup. We see that as soon as we are delinquent in some of our products, especially in BCD, there is immediately escalation. So it means that, yes, clearly, there is no inventory in the channel.

Unfortunately, we had a little bit of hiccup during Q2 related to our Singapore fab. Immediately, this was escalated by our customer because they were expecting that -- Singapore is serving automotive for a significant portion. So yes, this is the reason why at this stage, we think that this trend is continuing. Since the beginning of the year, we have seen the forecast of our customer moving up. That this is normally not what it happened. Normally, they start the year with a certain level and then moving forward, let's say, going a little bit down. Now what we have seen during this year is continue, let's say, upgrade up of the expectation of the customers in terms of demand.

In terms of sales channel inventory?

Yes. We are in a situation that in average, overall, we are back to be in the normal situation. But for some of our products, like, for instance, our microcontroller is very, very lean. After many quarters, let's say, in which our revenues to the distributor were lower than the revenue of the distributors to their customers, and which the inventory has went down. Now when we look at this inventory compared to the demand, compare, let's say, to the final demand is really below what we consider a normal level of weeks that we should have in the inventory. Normal level of weeks is in the range of 12, 13 weeks. Now we have a situation in which we are much, much lower than that. I would say that at this stage, the inventory is more than under control in distribution.

Okay. And then maybe more broadly on silicon carbide. 2 years ago, investors were worried about EV demand. And today, the discussion seems to be shifting towards 800V adoption and accelerating silicon carbide penetration as well. So how do you see the silicon carbide opportunity more broadly evolving from here?

As I was saying before, we see now, let's say, demand increasing in silicon carbide. Today -- to be honest, today, in silicon carbide, the limitation factor still is our capacity because we are producing silicon carbide at 150-millimeter. Of course, we do not intend to invest in the 150-millimeter, but to move to the 200-millimeter, the silicon carbide. And here, we are back to the point of the qualification of the customer. So demand will increase. This year, demand is increasing. This year, revenue will increase double digit. The visibility that we have for next year in terms of silicon carbide for our revenues with the programs that we have in our hands is that we will be back at the level of $1 billion. The move of the silicon carbide to 200-millimeter should significantly improve, let's say, the profitability of this line. But then we are, let's say, introducing the new generation from Generation 4, Generation 5. So here, we see, let's say, a trend for silicon carbide that definitely is in the recovery pace.

We've got a few more minutes left. So any final questions from the audience.

So still on silicon carbide, can you share an update of the production ramp in the Catania facility for 200-millimeter? Are you still expecting profitability, no earlier than 2027, as you said...

In general, talking about the silicon carbide in Catania in the Power and Discrete. Taking about Catania, Catania is such as a facility that is quite effective in terms of cost. You have to -- not to forget that a significant portion of the investment in Catania has been subject to grants from the European Union. So it means that we have a really cost-competitive site over there in Catania. Now we need to move to the transfer. Clearly, today, it's not a secret that in our Power and Discrete, we are suffering the 6-inch production, the fact that we need to improve our mix in terms of technology and also in terms of mix of customers. We expect that moving to 200-millimeter in Catania, that, as I said, is a cost structure that is competitive.

Moving the 200-millimeter in China, Chongqing that, by definition, is quite competitive in terms of cost structure. This gives us the opportunity really to recover in terms of profitability in our Power and Discrete, where, of course, you can understand silicon carbide is important portion. Even if it will take time because we are back to the point, let's say, that it is not overnight that we can switch from one fab to the other fab. So -- but this is expecting that we will, let's say, to be back to profitability at the end of next year, early 2028, also in Power and Discrete.

Any other questions? Maybe just to finish things off, you talked about accelerating bookings, improving visibility. What are the key indicators that you're personally watching to determine whether this recovery is sustainable into 2027 and beyond?

As I said before, let's say, today, what we see in terms of demand, in terms of backlog is that it's covering more quarters. Today, what we see is the fact that it is extending not maybe 1 or 2 quarters but well inside 2027. When we look, for instance, to the silicon photonics, as I said before, we have long-term agreements covering more than 1 year, 2, 3 years. So at the end, this give us some comfort to say that, yes, this entering 2027 and moving forward, we should see, let's say, really, let's say, our top line growing at a fast pace.

Great. I think that's a very good place to end things. Thanks again, Lorenzo, for joining us, and thanks, everyone, for the questions.

Thank you. Thank you.

Ladies and gentlemen, welcome to the STMicroelectronics Second Quarter 2026 Earnings Release Conference Call and Live Webcast. I am Moira, the Chorus Call operator. [Operator Instructions] The conference must not be recorded for publication or broadcast.

At this time, it's my pleasure to hand over to Jerome Ramel, EVP, Corporate Development and Integrated External Communications. Please go ahead.

Thank you, Moira, and thank you, everyone, for joining our second quarter 2026 financial results call. Hosting the call today is Jean-Marc Chery, ST President and Chief Executive Officer. Joining Jean-Marc on the call are Lorenzo Grandi, President and CFO; Marco Cassis, President, Analog, Power and Discrete, MEMS and Sensors Group, and Head of STMicroelectronics Strategy, System Research and Applications and Innovation Office; and Remi El-Ouazzane, President of Microcontrollers, Digital ICs and RF Products Group.

This live webcast and presentation materials can be accessed on ST Investor Relations website. A replay will be available shortly after the conclusion of this call. This call will include forward-looking statements that involve risk factors that could cause ST results to differ materially from management expectations and plans. We encourage you to review the safe harbor statement contained in the press release that was issued with the results this morning and also in ST's most recent regulatory filings for a full description of these risk factors.

Also, to ensure all participants have an opportunity to ask questions during the Q&A session, please limit yourself to one question and a brief follow-up. Now I'd like to turn the call over to Jean-Marc Chery, ST President and CEO.

Thank you, Jerome. Good morning, everyone, and thank you for joining ST for our Q2 2026 earnings conference call. I will start with an overview of the second quarter, including business dynamics. And I will hand over to Lorenzo for the detailed financial overview. I will then comment on the outlook and conclude before answering your question.

So starting with Q2. Our second quarter net revenues of $3.49 billion came in above the midpoint of our business outlook range, driven by higher revenues in Communication Equipment, Computer Peripherals and Automotive. Gross margin was 34.8% and non-U.S. GAAP gross margin was 35.2%. Both were in line with the midpoint of our business outlook range. Non-U.S. GAAP diluted earnings per share was $0.31. During the second quarter, inventory in our balance sheet remained flattish. In distribution, inventory further decreased and is now below our standard target. We generated a positive $75 million free cash flow.

Let's now discuss our business dynamics during Q2. During the quarter, demand increased further with strong bookings and book-to-bill close to 2 overall. We were well above 1 in all end markets and significantly above 2 in Communication Equipment & Computer Peripherals, mostly driven by optical connectivity, including silicon photonics. We saw improved visibility and signs of tight supply in several product categories.

In Automotive, revenues came in better than expected, increasing 14% sequentially and 16% year-over-year. This growth was driven by our solid position on application-specific ICs and sensors for conventional applications, electrical powertrain and ADAS. Automotive design momentum continued to build across multiple OEM and Tier 1 ecosystems. We secured design wins across hybrid, electric and conventional vehicles, including applications in onboard chargers, powertrain and active suspensions.

These wins were across our application-specific ICs and sensors. Specifically, our smart power ICs wins include custom devices for airbags, electronic stability control and suspension applications based on our proprietary BCD technologies manufactured in our [indiscernible] 300-millimeter wafer fab. We have progressed well with the integration of NXP MEMS sensors business acquired in February. As we anticipated, the complementary technology and product portfolio is strengthening our Automotive Sensors business with awards at key players for active safety application and tire pressure monitoring.

Industrial improved 20% sequentially and 34% year-over-year. Importantly, inventory in distribution further decreased and is now below our standard target. This solid growth was driven by our general purpose microcontrollers and by analog with their wide ecosystems and by our application-specific analog products, complemented by power conversion products. We are strongly positioned to support the ongoing transformation of factory automation, robotics and power and energy infrastructure. Our portfolio is uniquely addressing the emerging needs of physical AI where intelligent sensing, real-time control and efficient power management are increasingly critical.

During the quarter, we saw design wins across industrial automation, power systems, building automation and home appliances. We continue to introduce new advanced sensor for this application. We launched a new series of industrial MEMS sensors with embedded AI tailored for the fast-growing industrial condition monitoring market. We also announced a new compact 3D LiDAR module, delivering AI-ready output data for low compute Edge AI systems running on microcontrollers and high-performance sensing in applications such as robotics, industrial automation, smart buildings, ER/VR and health care. This is in line with our strategy to move beyond stand-alone sensors and deliver integrated sensing systems that support real-world Edge AI.

Finally, we announced a further expansion of our collaboration with NVIDIA to accelerate physical AI. As a partner in NVIDIA Halos for robotics and end-to-end functional safety system for industrial and humanoid robots, ST is bringing its microcontrollers, sensors, motor control and security solutions to support Halos readiness across chips, evaluation kits, software and reference designs.

For Personal Electronics, second quarter revenues were up 3% sequentially and 20% year-over-year. This growth was driven by increased content by device in our engaged customer programs and better than normal seasonality. During the quarter, we introduced secure chips that help smartphone and personal electronics manufacturers prepare for quantum-ready security requirements. It combines post-quantum cryptography acceleration with NFC, secure element and eSIM functions on a single-die for use cases such as digital identity, payments and digital car keys.

We also introduced a new generation of ultra-low power global shutter image sensors that deliver high-quality, always-on vision to compact devices operating on batteries or harvested energy such as wearables, AR/VR and IXA/IXS, smart home appliances and medical devices. They are engineered to deliver rich visual context and AI-ready data under tight constraints on power, size and cost.

With the depth sensing technologies I mentioned earlier and this device for AI vision, ST delivers a complete perception stack for Edge AI to customers. For Communication Equipment & Computer Peripherals, second quarter revenue were above expectations, showing increase of 13% sequentially and 50% year-over-year. This growth was driven by our engaged customer programs with our custom design products, boosted by our microcontrollers for optical connectivity. Indeed, ST is a core enabler for the Cloud AI era. We see strong traction on optical connectivity driven by silicon photonics ICs, electronic ICs, microcontrollers.

For the power stage of Cloud AI, we are already successful with our microcontrollers and high-voltage power and analog products, and we are building a pipeline of design wins for low-voltage power and analog products. Therefore, we are raising our revenue ambition for data centers. We now expect revenue above $1 billion in 2026 and assuming the current dynamics continues and with the current engagements we have well above $2 billion in 2027.

During the quarter, we secured multiple design wins across a range of products from optical connectivity driven by silicon photonics ICs, electronic ICs and microcontrollers to silicon and silicon carbide-based power solutions. In addition, we see a growing number of nontraditional AI server companies, including players coming from industries such as solar power and battery storage, where we are leveraging strong relationships to support their expansion into this field.

In May, we held a webcast on the Low Earth Orbit satellite communication and new space opportunity for ST, highlighting how ST is positioned as a core semiconductor enabler across this new industry. We see a significant opportunity here with our addressable market expected to reach around $3 billion by 2030 or about 4x the 2025 level. ST expects to generate well above $3 billion in cumulative space revenue over the period 2026, 2028, mainly with our BiCMOS, FD-SOI and Panel Level Packaging technology.

Finally, in June, ST joined the EUR 115 million Series A financing of Quobly to accelerate the industrialization of its silicon-based quantum computers and bring its first commercial product to market by the end of 2026. For ST, the scale needed by high-performance computing customers can only be achieved if breakthrough quantum systems can be industrialized and integrated with semiconductor grade standards and backed by a robust ecosystem. We are leveraging years of shared expertise in FD-SOI and deep technological collaboration to accelerate the commercialization of Quobly's products through our 300-millimeter silicon fab environment.

In May, we held our 2026 ST Microelectronics Annual Shareholder Meeting, where all resolutions were approved. Following the AGM, ST Supervisory Board appointed Mr. Armando Varricchio as the Chairman and Mr. Nicolas Dufourcq as the V Chairman of the Supervisory Board. Now over to Lorenzo, who will present our key financial figures.

Thank you, Jean-Marc. Good morning, everyone. Let's start with a detailed review of the second quarter, starting with revenues on a year-over-year basis. By reportable segment, Analog Products, MEMS and Sensors grew 26%, mainly due to Imaging and MEMS and to a lesser extent, Analog. Power and Discrete products increased by 3.7%. Embedded Processing revenues were up 35.5%, mainly due to general purpose microcontroller and to a lesser extent, custom processing and connected security. RF & Optical grew 32%. By end market, Communications Equipment & Computer Peripherals grew 50%, Industrial 34%, Personal Electronics 20%, and Automotive 16%. Year-over-year sales to OEMs and distribution increased 23.3% and 33.1%, respectively.

On a sequential basis, by reportable segment, Analog Products, MEMS and Sensors increased by 8.2%, Power & Discrete by 19.2%, Embedded Processing by 17.7% and RF & Optical Communications by 8.6%. By end market, on a sequential basis, Industrial grew 20%, Automotive 14%, Communication Equipment & Computer Peripherals 13% and Personal Electronics 3%.

Turning now to profitability. Gross profit in the second quarter was $1.22 billion, increasing 31.1% on a year-over-year basis. Gross margin was 34.8%, increasing 130 basis points year-over-year, mainly due to lower unused capacity charges and better product mix. On a sequential basis, gross margin increased by 100 basis points. Non-U.S. GAAP gross margin was 35.2%. Q2 gross margin included about 60 basis points of negative impact resulting from nonrecurring cost related to our manufacturing reshaping program. The negative impact on gross margin just mentioned, nonrecurring cost is expected to remain at a similar level over the rest of the year.

Total net operating expenses, excluding restructuring amounted to $970 million in the second quarter. Non-U.S. GAAP OpEx stood at $960 million, in line with the expectation given in April. For the third quarter of 2026, we expect non-U.S. GAAP net OpEx to stand at about $980 million. The sequential increase is mainly due to start-up costs and employee share award expenses that are more than offsetting the positive vacation seasonality effect. Excluding these 2 headwinds, Q3 '26 non-U.S. GAAP net OpEx would have been at about $920 million. For full year 2026, we now expect non-U.S. GAAP net OpEx to be slightly above $3.8 billion taking into account increased employee share award expenses and the temporary impact of the start-up cost, reducing our other income and expenses positive line.

For full year 2026, non-U.S. GAAP net OpEx are expected to increase low double digit year-over-year. Excluding NXP MEMS business acquisition and the exchange rate impact, like-for-like net OpEx should be up high single digit year-over-year. Our cost saving plan is delivering its expected benefits. At the same time, we are accelerating our investment in new business opportunities. In the second quarter, we reported $187 million operating income, which included $58 million for impairment, restructuring charges and other related phase-out costs. These charges are related to the execution of the previously announced company-wide program to reshape our manufacturing footprint and resize our global cost base.

Q2 operating income also includes $24 million purchase price allocation effect from our acquisition of NXP MEMS sensor business. Excluding these items, Q2 non-U.S. GAAP operating income stood at $269 million and non-U.S. GAAP operating margin was 7.7% with Analog Products, MEMS sensor at 10.1% Power & Discrete at minus 21.4%, Embedded Processing at 19.7% and RF & Optical Communication at 21.2%.

Second quarter 2026 net income was $222 million compared to a net loss of $97 million in the year ago quarter. Diluted earnings per share were a positive $0.24 compared to a negative of $0.11 one year ago. Non-U.S. GAAP net income stood at $291 million and non-U.S. GAAP diluted earnings per share stood at $0.31. Net cash from operating activities totaled $502 million in the second quarter, including $44 million outflow related to restructuring compared to a net cash flow from operating activities of $354 million in the year ago quarter.

Net CapEx was $409 million in the second quarter compared to $465 million in the year ago quarter. We now expect 2026 net CapEx to be at the high end of our $2 billion to $2.2 billion range, reflecting accelerating investments in the selected growth drivers, including cloud optical interconnect. Free cash flow was positive at $75 million in the second quarter compared to a negative $152 million in Q2 2025. Inventory at the end of the quarter was $3.19 billion compared to $3.17 billion in Q1 '26 and $3.27 billion in Q2 2025. Days sales of inventory at the quarter end were 126 days, in line with our expectation compared to 140 days for the previous quarter and 166 days in the year ago quarter.

Cash dividends paid to stockholders in the second quarter of 2026 totaled $75 million. ST maintained its financial strength with a net financial position that remains solid at $2.01 billion as of June 27, 2026, reflecting total liquidity of $6.03 billion and total financial debt of $4.02 billion. During the quarter, ST issued a new $1.5 billion dual tranche senior unsecured convertible bond, Tranche A and Tranche B for $750 million each due to 2031 and 2033 and announced the early redemption of its $750 million convertible bond due in 2027.

Now back to Jean-Marc, who will comment on our outlook.

Thank you, Lorenzo. Now let's move to our business outlook for Q3 2026. So we are expecting Q3 revenues of $3.7 billion plus/minus 350 basis points at the midpoint of our Q3 '26 net revenues will increase 6.2% sequentially and by 16.2% year-over-year. We expect our gross margin to be about 37% plus/minus 200 basis points, including about 70 basis points of unused capacity charges. This business outlook does not include any impact for potential further change to global trade tariffs compared to the current situation.

To conclude, in Q2, we saw demand further accelerating, strong bookings in all end markets and improved visibility. In Q3, revenues are expected to continue to grow sequentially and year-over-year and gross margin to continue to improve. As previously anticipated, Personal Electronics seasonality this year is different compared to previous years. Revenue growth for Personal Electronics is expected to be below normal seasonality in Q3, moderating ST sequential growth in the third quarter. In Q4, we anticipate a revenue growth acceleration, mainly driven by our engaged customer programs in AI data centers and Low Earth Orbit satellite communication.

We expect Q4 revenues to be above $4 billion, representing a sequential improvement better than normal seasonality. This translates into H2 versus H1 growth above our normal 15% seasonality. ST growth driver remains solid. We continue to see strong demand in AI data centers, reflecting the success of our product and technology portfolio. We are raising our revenue ambition for data centers. We now expect revenues above $1 billion in 2026 and assuming the current dynamic continues and with the current engagement we have well above $2 billion in 2027. This confirms ST's strong position in the evolving AI data center. Thank you, and we are now ready to answer your questions.

[Operator Instructions] The first question comes from the line of Janardan Menon from Jefferies.

I was just looking into your second half guidance and trying to get a feel for your gross margin trend into Q4. So I know you don't want to guide on Q4, but you are talking about a bigger increase in revenue quarter-on-quarter. And we don't know how much it is, I agree, because you just said more than $4 billion. But I was wondering directionally, whether you can give us any qualitative comments on how your Q4 gross margin could proceed and how you see that evolving into 2027 as well?

Okay. Thank you for the question. I take the question, and Jean-Marc, about the gross margin, the evolution. Clearly, let's say, the midpoint of Q3 non-GAAP gross margin or gross margin for us in Q3 will be 37%. We have guided at this level of gross margin that is increasing, let's say, about 180 basis points compared to the one of Q2. Clearly, starting from this Q3 gross margin at 37%, we do expect, let's say, for Q4, a sequential improvement in our gross margin, considering that our revenue will increase significantly.

But you have to keep in mind that there is some headwinds as well. One is that, let's say, our level of unloading charges will not change significantly in Q4 due to the fact that we are starting some fab, particularly in China, let's say, which we will still have some negative impact on our, let's say, level of unloading. And clearly, there is also still, let's say, this cost related to the transfer of technology related to our reshaping program in our manufacturing infrastructure that will be still there, similar to the one that we have in the current quarter in Q3.

So yes, I confirm there will be improvement, but there will be also some, let's say, headwinds that are, let's say, impacting our gross margin. Overall, anyway, I confirm that in Q4, there will be an improvement in our gross margin, sequential improvement compared to the 37% of the Q3.

But given that your revenue jump is going to be bigger, can we assume that your gross margin jump will also be bigger? Or is that a speculation?

As I said, they will improve, let's say, compared to the 37%, but still, let's say, it will be impacted by some ingredients that is related to this level of unloading that will not decrease, let's say, while when you look, let's say, the dynamic between Q2 and Q3, unloading was, let's say, decreasing. This will not be similar what will happen in -- between Q3 and Q4, let's say.

And as well as, don't forget that when we look at the dynamic of our gross margin moving from Q3, we had also benefit from the FX, let's say, that was improving in respect while, let's say, in Q4, this effect will not be there. It will be neutral. So yes, I repeat that there will be an increase in our gross margin, but you have to also consider in your modeling that there are some headwinds, temporary headwinds, but there will be in Q4, let's say, that will be limiting somehow the improvement of our gross margin in Q4.

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

I guess I want to start with the data center number. Can you provide some more granularity on what's driving the big increase in 2027 versus the prior expectations? Like how much of this is optics versus power? It did sound like there was some positivity on the power side. And then I guess also is you mentioned you were capacity constrained before. Is that number greater than $2 billion, assuming you're still constrained as well?

So before I pass the question to Remi to go in further detail, it is clear that our growth in 2027 will be driven both by our specific engaged customer program on this field of activity. But clearly, the overall success on optical cable connectivity. Clearly, it will be the main driver of the significant growth we will do next year. Now I'll let Remi comment more in detail.

To complement what Jean-Marc said, we see clearly an acceleration in the adoption of 800 gig and 1.6 terabit per second pluggable optics. And those are actually generation and categories of transceivers where now we are seeing a triple effect and this triple effect is a fairly large market share when it comes to the microcontroller, taking care of the control plane, a growing share when it comes to the electronic IC driven by our BiCMOS technology.

And starting from next year, but really accelerating next year is actually our growing revenue in silicon photonics supporting photonics IC that is being part of those pluggable transceiver. Like we will explain, we have a lot of scalability in terms of capacity on that technology because of the structure we have in our Crolles factory. So we are not right now gated by capacity expansion to go and capture revenue at this stage.

Maybe a follow-up on that also. I totally appreciate what Lorenzo, you were highlighting from the manufacturing transitions and underloading charges. But on a like-for-like basis, should data center as it grows, be accretive to gross margins?

Yes. At the end, yes, this kind of, let's say, clearly, this has been already, let's say, moving from Q2 to Q3 and will be also contributing from Q3 to Q4, product mix is contributing in a positive way to our gross margin. Yes, I confirm.

The next question comes from the line of Jakob Bluestone from BNP Paribas.

So I had a question and a follow-up. On the pricing, could you maybe just give us a bit of an update what are you seeing in terms of pricing tailwinds? And when do you think that might impact revenues this year?

Yes. In terms of pricing, I would say that, of course, it's twofold. On one side, it's true that we see, let's say, in our input cost price increase. Clearly, there are different materials or maybe, contractor activity that are increasing prices. On the other side, I would say that there is the other side that is -- that we confirm that in this context, let's say, of higher input cost, yes, we're increasing the prices on selected products.

Clearly, this is an ongoing process that is, expanding in terms of, let's say, price increase. And I would say that at the end, what we see in our input cost is more than offset what we do on our top line. So I would say that at the end, at this stage, the 2 impacts are more or less offsetting each other.

And if I just ask a quick clarification. On your data center revenue guidance hike, was that increase driven by increased demand outlook or by a faster expansion of your capacity and supply?

It's both. Because clearly, above $1 billion 2026 revenue, the demand is well, well above, but thanks to our capability to grow in the various assembly and test manufacturing. So that's the reason why we have increased our indication for this data center business. And next year is the same. Next year, clearly, we will closing the gap between the demand and our capability to supply but it is really driven first by demand, then it is covered by engagement. Backlog for this year, 100% coverage. And next year, okay, engagement are covering our expectation. And definitively, our capability to grow on this advanced 300-millimeter technology is a very important competitive factor for us.

The next question comes from the line of Sandeep Deshpande from JPMorgan.

Could you talk about the revenue growth guidance into Q3 and then potentially into Q4 by your segments? Because you said on the -- earlier in the call that Personal Electronics was weaker in the third quarter. But can we look at how the growth was in the other segments? And based on what you are indicating for the fourth quarter at the moment of greater than $4 billion, at least directionally, how to see the different segments in terms of your end markets into the fourth quarter based on your order book today?

I will take the question. Well, clearly if I am starting by the key growth driver, it's clearly our verticals, Communication Equipment & Computer Peripheral. I can say that in Q3 on a year-over-year growth, this segment will grow very similarly what we have achieved in Q2. So means close to 60% growth. And definitively, where will have in Q4 a very strong acceleration, means, okay, we will be about 90% growth.

The second really positive growth vertical is Industrial. Industrial in Q2, we grew 32%. And step after step, Q3 and Q4, we will go close to 40% growth year-over-year in Q4. Well then, clearly, Automotive is performing above what we expect and what the market is expecting. You know that for semiconductor industry, the Automotive is expecting to grow about 13%, 14% year-over-year. This is what we will achieve on this segment.

Well, PE is definitively a different profile, let's say, in the year-over-year growth because it will be slightly negative, let's say, mid-single digit during Q3 and Q4 as we anticipated in our previous call, that will put this segment at the end of the year for the full year growing, let's say, from low to mid-single digit that we already anticipated in the previous call. So my takeaway is really very strong growth on Computer and Communication moving from 60% growth in Q2 to close 90% in Q4.

On Industrial after Q2 of 30% growth year-over-year, we will grow close to 40% in Q4. Automotive, let's say, low double digit as expected consistently with the market. And this year, okay, different profile for PE will be year-over-year negative on H2 after having been positive in H1. But on the full year, it is low to mid-single digit, which is consistent with the market of a smartphone that is decreasing because on the low end, let's say, device, there is an impact because of the memory price. So this is the profile of the revenue, Q3, Q4.

Any follow-up, Sandeep?

Question would be that you've seen this significant strength in Computer and Peripherals associated with the AI market. Is there not any flexibility in filling your capacity in the third quarter itself? Why does this have to wait until the fourth quarter given that you still remain underutilized in the third quarter?

Let's say, all the advanced technology that are enabling our growing Industrial market and Communication and Computer, we are in a permanent growth and exactly at what we expect. And here, the only flexibility we could have is to continue to accelerate permanently, which is already on really a competitive path. Then clearly, where we have still some, let's say, underloading and even sometimes, okay, in Q2, we face a slight delay in our capability to ramp up, it is on legacy -- pure legacy analog technologies because if you remember, Q1, we were totally underloaded.

So the Q2 ramp-up was challenging. So we delay a little bit this ramp-up that in a certain extent, has impacted also our capability on Q3 to fully fulfill the demand on legacy technology. So this is, let's say, pure temporary capability to ramp and some unloading charges specific to really legacy technology on analog that is, let's say, showing this figure.

The next question comes from the line of Domenico Ghilotti from Equita.

I have a follow-up on the gross margin. So in the past, you were guiding -- you were suggesting that at $4 billion sales per quarter, so the profitability gross margin would have been at least in the 40% range. Now you are suggesting that you will probably not be at that point in Q4. So I'm trying to understand how this -- so first of all, if you are still confident to get to the level of profitability you were suggesting and there are some specifics -- temporary effect on Q4 that we should take into account?

And a follow-up on the AI data center demand and your customer engage program. Can you give a sense of how concentrated is today the demand there and the engage program supporting your $2 billion revenues for 2027?

Maybe I take the first question about the gross margin at $4 billion in respect to our model to be above, let's say, the 40% gross margin. Here, I have to remind all of you that, yes, this is our model, but let's say, we always said that there are 2 conditions that we need to achieve in order, let's say, to be above 40% when the company will be at $4 billion. One is the revenue, of course, let's say, that this is what is happening. But the other point is that we complete our reshaping manufacturing program. Means that actually we have, let's say, done this transfer from the 200-millimeter to the 300-millimeter for the silicon closing 2 fabs and let's say, from the 150-millimeter to the 200-millimeter for the silicon carbide.

We are not yet there. We are, let's say, in the middle of this transformation. So clearly, here, we are not in the condition to have our manufacturing infrastructure at the right level of efficiency. The other way around, I would say that in this moment, in the quarters, Q3 and Q4, we have some kind of extra cost that are related to this transfer, the qualification of the products, the redo of the mask of the products, all these kind of things that are impacting our gross margin and that clearly for optimal efficiency due to this transition.

So at the end, we confirm our model to be, let's say, above 40% when the company is there. But once we have done the transformation when we have completed our programs that, as you know, it will be at the end of 2027, not before. So this is the reason why it is not enough to be at $4 billion to have gross margin at the right level of the model.

On the second question, maybe, Remi, you can comment, again, the key growth driver that will put us on a trajectory well above $2 billion next year.

Yes. We have discussed earlier about what we're doing on the optical front. And I insist on the fact that we are now at the intersection of 3 main vectors of growth, which is the oversized market share we have, 800 gig and 1.6 terabit per second in MCU, the growing adoption of our [indiscernible] BiCMOS process for electronic IC and the steep ramp-up we have on our photonics IC platform, our silicon photonics platform in 300 millimeters, which has proven to be really, really well adopted across the board by all the major actors. And I think there was also a question related to the concentration of that revenue. And we see an evolution of our revenue that is pretty consistent with market share distribution between hyperscalers, which makes us actually quite confident about the composition of our revenue.

The next question comes from the line of Didier Scemama from Bank of America.

My first question is on the outlook. I think, Jean-Marc, in your prepared comments, you mentioned that your book-to-bill was close to 2, I think, overall and above 2 in certain segments like Optical Interconnect in particular. So I'm just wondering, does that imply that your first quarter seasonality might be a bit better than normal? And I've got a follow-up.

Thank you. Yes, with this book-to-bill, what also is interesting is that out of the 100% of the booking we received in Q2, well above 50% were for next year. So it means the customer now they have understood that they have to provide us visibility more. And the good news is that now our total backlog is representing about an average of 4.5 to 5 quarters of Q2 average revenue, which is clearly an improvement step. Coming back to standard of visibility. So clearly, yes, Q1 is today on the dynamic to be well loaded and clearly continuously boosted by our revenue related to AI data center.

Okay. Great. And as a follow-up, I just wanted to make sure I got the right end of the stick on the financial model. So I think in the past, you were talking about 45% gross margin on $18 billion and then maybe 50% gross margin on $20 billion. Obviously, that's contingent on execution of the restructuring plan on manufacturing. So just wanted to make sure that this is still the case and whether you've got increased confidence that you can deliver these sort of numbers around 2028?

We confirm our confidence level to reach $18 billion in 2028. And clearly, announcing the dynamic and increasing our indication on data center, it is clearly one of key growth driver that will position our company on $18 billion by 2028. Then second, Lorenzo already commented that this business related to AI data center is accretive to our gross margin. So we will have this mixed effect model. Okay, I confirm that assuming we complete on time our reshaping program on manufacturing and that the FX will remain our model, we should be in position to reach the gross margin target consistently with our $18 billion model. More about $20 billion, okay, for the time being, let's reach together the $18 billion target, then we speak about the $20 billion.

The next question comes from the line of Stephane Houri from ODDO BHF.

So I would love to come back on maybe the satellite Low Earth Orbit opportunity. And if you can tell us what is the dynamic currently, notably with your main customer and the ramp of your second customer? And if you confirm the target of $1 billion for this year and if you have a view for next year already?

Thank you. I'll pass the question directly to Remi.

Stephane, the dynamic is quite healthy. Overall, like Jean-Marc explained at the beginning of the discussion, we stick to well above $3 billion over '26, '27, '28. You have noticed that and we spoke about that in the past that it's very much a launcher dependent business in the context of deploying satellite and then deliver the services which in itself is a fuel behind user terminal consumption. Clearly, things are progressing in the right direction [indiscernible]. Other constellation have faced a bit of a snag lately, and to be fixed in the coming quarters. But directionally, nothing has changed. We see actually a strong '26 and an even stronger '27.

Okay. And can you comment on the level of profitability on the gross margin on this business like you did for data center? Is it accretive?

Yes, clearly, let's say, is a business with a combination of different gross margin because clearly, there are different products, the one that are going in the satellite, the one that are going in the gateway, the one that are going, let's say, in the user terminal. Anyway, when we look, let's say, at the average of the gross margin of this business, yes, I confirm that also Low Earth Orbit satellite is contributing, let's say, to the improvement of our gross margin.

The next question comes from the line of Francois-Xavier Bouvignies from UBS.

My first question was on the capacity front. I mean we see an acceleration of growth from a cycle perspective, but also from AI data centers. And you are seeing some tightness, as you say in the release, also TI suggested as well some tightness. So I was wondering, how do you feel about your capacity in the next, let's say, 2 to 3 years? I mean, do you have, you think enough capacity to deliver the different growth scenarios? Or are you evaluating maybe some brownfield or greenfield expansion down the line so the capacity of ST in the next 2, 3 years would be helpful?

We have to see it under 2 angles. One angle is what is related now our capability to support AI data center and at this stage, we believe that we can sustain it and support it. Of course, under the assumption, we are well assessed the dynamic we have today, on optical cable and microcontroller. And clearly, the key success factor is Crolles. And Crolles will reach 15,000 wafer per week and will go above to support the dynamic of this business.

Clearly, we see some tightness is clearly what is related general purpose microcontroller. Why? Because first of all, we have 2 cumulated effect. There is first the enormous success of the microcontroller developed by Remi team for optical cable. And there is the solid recovery of the overall industrial market. And where, okay, I mentioned during my address that in distribution inventory now well below our standard and the POS dynamic is very strong. It is here that we are facing at this moment some tension on the supply and increasing lead time.

But here, basically, we have some key competitive advantage. First of all, we have our 300-millimeter fab, where clearly our microcontroller will be processed. The good news now is I got 300-millimeter, certainly will reach the full build-out pretty soon before 2028, and we'll be capable to support the growth of microcontroller as soon as we will have qualified as the 19-nanometer and 14-nanometer technology.

Then the second important lever is our China for China strategy that will start to pay back because if you remember, we have qualified 14-nanometer technology in China with our main partner that will enable us to support the growth in China. That is very demanding, both for Industrial in distribution, but industrial OEM and also optical cable. On the other, let's say, technologies and product, clearly, the challenge for us is our transition phase between respectively, 6-inch to 8-inch silicon carbide technology and 8-inch to 12-inch analog technology. And we are accelerating as fast as we can in synchronization with our customer that could create time to time, okay, some temporary tightness on the supply.

So this is okay, the overall picture. So 3 elements, more short term on microcontroller, but we have exactly in our hand, our capability to grow. All the new technology driven by AI data center, we have the path to grow. Some other tightness related to our reshaping but only temporary. Then after, we have some pocket of capacity limitation time to time with OSAT, but okay, we manage it.

And maybe my follow-up would be on silicon carbide actually. I mean, we see a lot of change in terms of silicon carbide demand, one driven by the Chinese EV carmakers adopting 800 volts supporting the silicon carbide growth, but also we see one of your main customers delivering strong deliveries. And on top of that, you have the 800 volts opportunities for silicon carbide. So I was wondering if you saw an improvement on the silicon carbide front? And if you could provide any maybe guidance on what to expect for that business, that would be great?

Thank you. So I will pass the question to Marco, who is managing the product line. Yes, okay, all the positive dynamic you described, we see it, but Marco will connect.

Yes. I confirm that we see the positive dynamics that you just highlighted. Actually, in Q2, we saw the revenue growth in the low teens year-over-year. So we are back to growth year-over-year and mid-30s in terms of quarter-over-quarter. This is also supported by strong bookings, a book-to-bill that is well above 1, which is resulting in a growing backlog. So the dynamics are confirmed and are factual is what we see.

So in this context, I can confirm that this year, we should grow the silicon carbide revenues double digit in '26 versus '25 based on already design wins and backlog, which is already visible. So the dynamics are positive. Of course, we are facing the transition between the 6 inches and 8 inches, which have sometimes created some tightness in terms of supply because the products need to be qualified by end customers. But I confirm that the dynamics are positive and are growing positive day by day.

Thank you, Francois. This is ending our call for this quarter. So thank you very much, everyone, for joining us. And we remain at your disposal should you need any follow-up questions. Thank you.

Thank you.

Thank you.

Ladies and gentlemen, the conference is now over. Thank you for choosing Chorus Call, and thank you for participating in the conference. You may now disconnect your lines. Goodbye.

Ladies and gentlemen, welcome to the ST The Leo Opportunity Conference Call and Live Webcast. I am Sandra, the Chorus Call operator. [Operator Instructions] The conference is being recorded. [Operator Instructions] At this time, it is my pleasure to hand over to Jerome Ramel, EVP, Corporate Development and Integrated External Communications. Please go ahead, sir.

Thank you. Thank you, everyone, for joining ST the EO Opportunity Conference Call. Hosting the call today is Remi El-Ouazzane, President, Microcontrollers, Digital ICs and RF product books. This live webcast and presentation materials can be accessed on the ST Investor Relations website. A replay will be available shortly after the conclusion of this call. This call will include forward-looking statements that involve risk factors that could cause ST results to differ materially from management expectations and plans. We encourage you to review the safe harbor statement contained in the press release.

[indiscernible] and also in [indiscernible] finding for a full description of these risk factors. Also to ensure all participants have an opportunity to ask questions during the Q&A session. Please limit yourself to 1 question and a brief follow-up. I'd like now to turn the call over to Remi El-Ouazzane.

Good morning, and good afternoon, everyone. It's great to be with you today. As a space electronics market is becoming an increasingly important topic. We're going to share ST perspective on the LEO opportunity, why it matters and how it is translating into growth for our company. For us and to use aphorism, it's been a 10-year overnight success. We look at the market momentum, the technology shifts driving this expansion and the pivotal role ST is playing and will play across the LEO ecosystem. So let's get started. I want to begin with the broader orbital infrastructure landscape. We can actually split it into 2 fairly distinct words.

The first is traditional space. This includes exploration missions as well as operation in geostationary and medium Earth orbit. This world remains very important for applications such as video broadcasting, radio broadcasting, mythology and GNSS. These are established mission-critical domains. They are typically based on large satellites have long-term development and life cycles and follow highly predictable service models. The second word is what we call new space. This is mainly centered on orbit, and it is expanding rapidly into broadband connectivity, direct-to-sale services and eventually orbital data centers. This segment is driven by smaller satellite, much faster deployment, lower latency and entirely new business models aimed at individual users and mass adoption. This shift is not only about Orbit altitude. It's really a change in the economics of space. It means a broader set of applications, much higher deployment volumes and much stronger semiconductor intensity.

On a live note and in reference to the date, we have selected for this event. Maybe some of you have supported a very special star. On the upper left part of this chart that is at the very far end of the Galaxy. And although we won't be discussing that type of object today, made the fourth be with us throughout this presentation. ST has been active in space for more than 45 years. Over that time, our technology has supported the full spectrum of space application. On the traditional space side, we are in [indiscernible] Arian launches and major exploration missions, including the Chinese [indiscernible] telescope.

As such, it should not be a surprise that ST was also involved in new space from the very beginning, enabling with our technology the first generation of large-scale LEO programs, both in space and on the ground. Now let me double-click on SC's LEO business. We have seen a very sharp increase in revenues from this business, reaching approximately $600 million in '25 up from about $175 million in 2021. That represents a compounded annual growth rate of about 36%. This is a strong and solid performance that it positions ST as the leading semiconductor players in the LEO market. In fact, John estimates our RF market share for LEO at about 90% in '25.

We are just in the early innings of this market. Later, I will explain our ST's position to sustain and extend this growth with our differentiated technology, our manufacturing strength and the breadth of our portfolio. To understand why LEO is getting so fast, we need to look at the technology and the economics behind the disruption. The changes have made large-scale LEO constellations possible.

First, the higher launch base of reusable rockets has allowed thousands of satellites to be put into a bit at a much more affordable cost. Let me share a few numbers. The cost of launch per kilogram decreased from about $10,000 on Falcon 1 in the mid-2000 to $2,000 in the following decade with Falcon 9 and is expected to be divided by a further factor of 10 on starship. Second, satellite themselves have changed. They have now become lighter, more stabilized, enabling upscale and faster manufacturing cannot be launched in groups for example, 25 per launch. They are increasingly embedding innovation such as software-defined digital payloads and inter-satellite optical lens.

Third and last, the user terminal has evolved dramatically. It has moved from a traditional parabolic antenna to electronically steered antennas, leveraging beam forming which can follow moving satellites and hand over from one to the next. These antennas are not produced very efficiently in very high volumes where this was originally monthly at the beginning, a military technology.

Together, those 3 ships have created the new new economy. And that economy is large, it's scalable and it's highly semiconductor intensive. So the question is how big can this become? According to Gartner, LEO services spending is expected to reach close to $15 billion globally in 26 this year, and it will continue scaling higher through 3 major services classes. The first and currently the most important is broadband service. Satellite broadband is connecting the unconnected and reducing the digital divide. It gives rural or poorly connected community access to high-speed, low-latency program. Think about this. While there is roughly 1.5 billion broadband users worldwide,

Today, almost 3 million people remain on the wrong side of the digital divide. It also provides connectivity and redundancy to corporates and governments and improve resiliency in the event of natural disaster. And because LEO constellation can cover the globe, they are particularly well suited for mobility applications, such as in-flight connectivity, maritime connectivity and even over time, automotive use cases.

The second growth engine, which is emerging now is direct to sale. This connects phones or IoT devices directly to LEO satellite that effectively act like a cell tower space. It opens the door to roaming led connectivity, expanded IoT services like asset tracking and helps eliminate mobile coverage holes. And finally, the next frontier is the orbital data center. This is still an emerging category today, but it points to a future where LEO is not just about connectivity. It could become a platform boot scale in orbit, benefiting from the nearly limitless availability of solar energy. This idea of orbital data center becomes more relevant as launch economics improve. As I previously mentioned, the cost per kilogram to orbit has fallen dramatically and will continue to do so, likely eventually down to below $100 per kilogram at some point.

That trend makes more ambitious innovate compute concepts increasingly realistic over time. [indiscernible] for instance, there are 5 for 1 million satellites for this purpose. And as Elon Musk [indiscernible], launching 1 million tons per year of satellite generating 100-kilowatt of compute per ton could eventually add 100 gigawatts of AS compute capacity annually with limited operational or maintenance needs. [indiscernible] with Project Sunrise also filed for NCC approval to support data centers in space and Google with its [indiscernible] project is gauging very similar initiatives.

So in short LEO is evolving from a connectivity platform into a much wider ecosystem with multiple growth vectors, both emerging in parallel. Now continuing on the service expansion, we just discussed, these opportunities have been amplified by rapid proliferation of LEO constellation. And something important has changed. The earlier space model was largely financed by public investment.

[indiscernible] is increasingly driven by massive private capital, which is accelerating constellation deployment supporting new business models and increasing the pace of launches. And because of the first examples of usage of today's LEO constellation, Governments have understood the geopolitical impact of these services. And as a result, Leo is increasingly becoming a global race.

Constellation are being deferred across geographies by a mix of established players and new entrants each targeting different parts of the LEO value chain from broadband to direct-to-sale to orbital data centers ambitions. Now let's try to translate this expansion into numbers. Over the next 5 years, the scale of the system will change dramatically. [indiscernible] cast deployed space is expected to grow by roughly 10x by 2030, reflecting the impact of the latest generation of launches and the rapid expansion of Constellation. Interestingly, the capacity deployed space by '28 may represent a similar order of magnitude than the total international Internet bandwidth on earth we would have guessed that 10 years ago.

At the same time, the backhaul infrastructure will also expand significantly with the number of gateways increasing by 1.6x. And this infrastructure build-out will enable a steep rise in usage. We are still at the numbers of new subscribers could exceed $200 million by 2030 from about $10 million today, driven by the proliferation and the improved quality of services as well as the arrival of new entrants from all geographies.

LEO is no longer just a niche connectivity technology. It's becoming a large-scale communication and eventually compute platform. It's difficult for us to go through a Leo presentation without talking about SpaceX. ST and SpaceX, it's a decade-long partnership built on co-development across key technologies for satellite and user terminals. The collaboration has produced billions of codesign products due to millions of selling new terminals and over 10,000 starting satellites. And products have been critical in helping SpaceX scale production through codesigning key chips, engineering services and high-volume manufacturing ST has demonstrated the exceptional value of its innovation and manufacturing capabilities. This collaboration has relied on differential ST technology and on the use of multiple products across the sterling constellation.

I will come back later to the specific technology and products building blocks behind this success. But for now, the key message is simple. The collaboration continues with a focus on ramping up rating us ongoing designs, excuse me, and architecting together the next-generation satellites and user terminal. And this partnership has yielded impressive numbers, as you can see on that chart. This chart illustrates the scale Starlink NST average together.

To date, ST has delivered more than 7.5 billion is 7.5 followed by [ 9 0 ] and the trajectory continues to accelerate as deployment expands. Now I know it's difficult to visualize what that really means. So let me give you a more tangible image. These close to 20,000 square meters of active silicon ship deliver, both in BiCMOS and [indiscernible] can cover up to 4 American football fields. -- or if you are a basketball fan, 48 basketball courts. This is a strong example of ST's ability to support high-volume space-grade program at industrial scale. It also shows our STC at the heart of stalling growth and execution. I touched base on this earlier when I spoke about capacity.

Space capacity is a key to LEO service expansion, especially broadband service expansion. As we discussed earlier, capacity has increased significantly over time from the first generation of stalling satellite to the latest V3 architecture. Across successive generation, Downlink capacity per satellite has increased from 10 gigabits per second to terabit per second level, roughly a 40x improvement. This step-up in performance reflects the evolution of the satellite platform itself. It's more integrated, it's more capable, and it's designed to support much higher throughput.

In other words, each new generation of satellite has not only improved technical performance, but it has essentially expanded the economic potential of the network. Parallel to this evolution, the ST bill of material was multiplied by 8x between the 1.5 and the VI satellite, now reaching several tens of thousands of dollars per se. That is important because it shows our STs not only prevent but increasingly embedded in the value creation of the platform.

As importantly, let me now turn to the ground LEO infrastructure and specifically the user terminals. Here, we have witnessed a very visible evolution in antenna design. Over time, due to terminal that becomes smaller, more integrated and more affordable, all of which we are critical to making this market scalable with our differentiated and mass-produced technology, supported starting value creation through bill of material cost optimization and product innovation.

We have now reached down to only several tens of per user terminal, which makes the user terminal much closer to a consumer like device making LEO broadband services bound to exploring. In practice, this has helped transform satellite broadband from an expensive and niche offering into a mass market product and now addressing more than 155 countries and markets.

Lowering the terminal cost has been a key accelerant of Leo broadband service adoption which I will now attempt to scope for you. Based on our current view, SAM for the broadband electronics was around $650 million in 2025, growing to roughly EUR 2 billion in 2028 and reaching close to EUR 3 billion by 2030. This reflects a rapid market growth and importantly, this still excludes additional upside from the orbital data center application.

As mentioned earlier, the market is becoming global with new constellation from new geography emerging. While starting still represents the LEO's share of the business today, it is not the only opportunity. On top of Project Kuiper, now rebranded Amazon LEO and European projects, I will also mention 2 Chinese projects, which are today at early stages but with strong growth ambition. The first one is China satellite network also called Starnet was planned 13,130 LEO satellites by 2035 and with 170 satellites in orbit so far.

The second Chinese project is space sales satellite, also called [indiscernible] which has planned 15,000 BO satellites by the early 2030s, with 130 satellites in orbit so far. I also want to point to the recently announced Blue Origin TeraWave project for a space-based network optimized for enterprise data centers and government customers. When it comes to as we have already been qualified at another major player, and we continue to build momentum with new constellation at the deploy.

Of course, the market growing at a 35% CAGR naturally creates appetite, a new supplier will try to gain share, which we have factored in our numbers. But our outlook remains very positive. The market is expanding fast. Our customers base is broadening. NSC, as you will see soon as a technology, the products and the manufacturing infrastructure to capture that growth. Talking about products, you can see on this chart, our comprehensive is our product offering. To keep it simple, we are present across the 3 layers of the ecosystem, be it satellite, gateways and user terminals. Our portfolio spans RF and digital ads, radiation capable MCUs and a wide range of products in power, analog mens, secure element, filters and more. This depth and breadth give ST a unique positioning in Leo.

It means we can address a very wide set of application needs across the value chain while also [indiscernible] our offer to the specific performance and cost requirements of each layer. Let's put some numbers around this now so that you can get a sense of the dollar amount of electronics and serve into this market. We provide that to several tens of thousands of dollars bill of material per satellite, a few hundreds of dollars per gateway and several tens of dollars per user terminal and to appreciate the order of magnitude, let me remind you what we discussed earlier.

Thousands of satellites are produced every year, thousands of gateway are being deployed and millions becoming tens of millions of user terminals are being shipped every year. As such, that does represent a very meaningful business opportunity for ST. But to understand why we are so well positioned in we need to look back at the technologies behind the offering. The first pillar is FD-SOI. This is a clear enabler for ASIC and microcontroller. It combines high performance, energy efficiency and structured immunity to radiation, we are also providing robust embedded memory. We are able to supply this unique technology from our 300-millimeter fab in France and also with our foundry partner.

The second larger pillar is a winning combination of the content and back-end technology we offer for the user's terminal front-end module. It's based for the front end side on [indiscernible] process, which is the right technology for flag, phased array and tenants, especially for low-noise amplifiers and power amplifiers delivers strong our performance in key satellite communication band with very low noise figures enabling to scale up the numbers of users served simultaneously with high linearity, reducing the numbers of antenna elements itself allowing user terminal cost reduction, hence, service demonetization and with the right balance of performance and cost. This technology is served from both our 20 and 300-millimeter fab in France. Another key enabler is on the back end of panel level packaging also called PLP.

In LEO, the scale requirement is massive. So packaging at scale, meaning several million units produced play becomes a real differentiator. PLPs high-volume production, strong RF and thermal performance and miniaturization. PLP is truly a [indiscernible] first. On top of the technology themselves, ST manufacturing independence also provides a market with a unique kind of geopolitical stability, if you wish, through a highly resilient supply chain setup.

Now it's one thing for us to sell claims, and it's even better to bullet up its customers' testimonies. And you can see here some quotes from starting executives, recognizing the value of our previously described technology. FD-SOI, including the recently introduced 18-nanometer nodes, for high-performance microcontrollers used installing mini laser system for inter-satellite optical links. CMOS the backbone of Leo user chemical front-end module. And on the back end, the PLP panel level packaging technology I was describing a minute ago. In Starling, is more than a supplier.

We are a key technology enabler across the platform, helping starting improved performance, cost and scalability. So bringing together our proven technology of scale of execution and the rapid expansion of the lab opportunity, ST has a very clear ambition. We aim to generate well above EUR 3 billion in cumulative revenue in the space market over the next 3 years from 2026 to 2028 combining LEO and traditional space.

On the LEO side, these numbers include only the broadband and direct-to-sale opportunities. Additional opportunities will continue to emerge including LEO data centers, which are not included in this ambition, it could further expand the market over time. We do consider the outlook for SC as quite promising. We have demonstrated strong execution track record, track record, excuse me. We have market leadership, and we see additional growth later. Overall, we are and we intend to remain the leading semiconductor provider in the new space.

To wrap things up, let's agree that a major disruption is happening in space, and LEO growth is fast accelerating. There are 4 things that I would like you to take away. First, manufacturing independence and our unique technology and manufacturing capabilities have put us at the heart of the LEO ecosystem. Like I said, like I said at the beginning of this presentation, a 10-year overnight success. Second, we have a differentiated portfolio, spanning every layer of LEO.

Third, the LEO [indiscernible] expected to reach around $3 billion by 2030, roughly 4x versus 2025. And while ST is targetd well above $3 billion of cumulative space revenue from 2026 to 2028. And last, the opportunity is still expanding. Orbital data centers will be a further growth driver, even though they are not yet embedded in our revenue ambition. In short, we expect space to be major growth drivers for ST in the coming years. I hope you for this presentation interesting. Thank you for listening, and we are now ready to answer any questions.

[Operator Instructions] Our first question comes from Stephane Houri from ODDO BHF.

I have one question and a follow-up. So the question is I'm trying to reconcile all the numbers that you have given, the sum of $3 billion by 2030. And also $3 billion plus space cumulative revenue by 2028. But at the same time, you you seem to say that you would reach $1 billion already this year. So what kind of trajectory revenue trajectory can we expect from 2026 to 2028 if if you can simplify it for me. And the follow-up would be also to have a view on what kind of market share do you think you will have by 2028 or 2030, if you refer.

Yes. Thank you, Stephane, for the questions. I will clarify the numbers indeed. Over the next 3 years, '26, '27, '28, our ambition is to deliver cumulative revenues of well above $3 billion for space as a whole. You're right, we have indicated in 2026 that we may already be close to $1 billion. The growth rate beyond 2026 will essentially depend on many factors, and I will get back on that in a minute, Stephane. As we said, well above $3 billion, I guess, at least scope for growth and opportunity. We factored in, in our assumptions, some market share loss for our main customer. We do not believe that actually our current market share that is exceeding 90% is something that we can keep as is.

The number that we are planting right now do not include any potential upside coming from data center in space in the medium term. ultimately to say that actually our number, our growth rate beyond '26 and the close to $1 billion that I just mentioned, are going to be dependent on the pace of existing and new constellation. And if you need to go atomic as to what is going to be the first level critical path to that, and I would say that actually rocket launcher is at the top of the list. And clearly, we are keeping a close eye and a savvy eye on the success of Starship and ugly, respectively, of SpaceX and Blue Origin. Those are actually real game changer, especially Star chip that would be the largest market build able to carry roughly 5 to 7x more than what you can do with Falcon 9 at 150 tons, and that will make a huge difference in terms of capacity.

The next question comes from Sandeep Deshpande from JPMorgan.

My question is you -- on one of the slides you.

[Technical Difficulty] In terms of what keeps your supplier in this market, what is the majority? Is there 1 kind of chip? Is it the RF chips, which are the majority of your revenue? Or is it the MPU where in terms of the satellite mark -- and in terms of where the growth is coming, is it coming from those tips? Or is it some from newer chips that you are supplying into this market? And a follow-up to that would be, where is [indiscernible] comparing for you at this point?

Could you ask -- could you restate the last one, Sandeep?

The last question was, where are you seeing competitive threats at this point in this market?

Okay. Got it. The class of products that we ship into space differ between satellite and user terminal. Clearly, we are not going to give any breakdown, but [indiscernible] terminal are the largest contributors to the revenue when it comes to what we do. And when it comes to the user terminal, the -- they are being essentially dimension by what we call 10 modules. So what is a front-end module, it's actually something that does 4 things: Signal amplification, transmit and receive. Hence, what I mentioned earlier in terms of power amplifier and loose amplifier switching because you need to go and move from talking to listening and filtering to remove electric noise and also the face shifting that is happening when you're creating actually [indiscernible]. They're actually the dimensioning element in the cost of the user terminal and the best technology on the planet to go and build frontier models [indiscernible] CMOS. And as such, I would say, anybody that wants to compete against ST and BiCMOS [indiscernible] as a potential competitors. Our BiCMOS platform, I have to say, is super established in the context of both either its cost optimized patterns for Ku-band or in the context of its performance for KA then the different bands of the LEO market. So we clearly -- this market, like I said, will have competition. But we feel quite strong as well about the front and process technology differentiation we have in the context of our [indiscernible]

So I missed middle question, Sandeep but we'll have to go and I'm not sure I capture that one, but the connection was not great.

The next question comes from Gianmarco Bonacina from Banca Akros.

A couple of questions for me. The first one is about the China opportunity. I think you mentioned a couple of Chinese companies that want to be active in terms of the satellite market. Do you think that you can become a vendor given the current geopolitical situation? Or are you already let's say, working with them? And do you think this is a viable opportunity for you? And the second one just about the market share, if I understood correctly, so you will grow from the around $1 billion to clearly above $1 billion. So the market share in 2028 will be well above 50%. How about competition? So how do you think -- what kind of level do you think you can keep going forward, maybe looking at 2030. So is competition very intense? Or do you think that you can keep this, say, above normal market share in this market?

Okay. Look, I start with the chart question. We are unapologetically European. So we end up being actually U.S. and China compatible and the China compatibility will start and finish at user terminal because of export control, we cannot be in the central technology happening in China. So we are focusing our entire energy on user terminal. The entire logic of what I've explained in terms of user terminals or I mentioned and where the performance of the network comes from relating back to buy CMOS, this exact same principle applies to also to Chinese user terminals, regardless if they are actually doing analog or sobering. I have -- we have not yet disclosed anything specifically happening in China, but we are actually quite engaged in user terminal development happening there. And it's a market for us that we are looking into, even though like I've explained, we are in the very, very early innings because their satellite footprint remains quite small, and if I can put it at this stage.

Okay. That's the first question. In terms of market share, looking at that for the coming 5 years till 2030. Yes, for sure, we expect to go in a sustained pretty oversized market share, clearly, lower than the 90% that we're talking about right now. but still quite high in the context of what we're doing with Starlink, also in the context of our positioning, especially when it comes to front-end modules in the constellation. And there will be competition for sure. And the competition, like I said, to find to find -- I'll let you make your work on finding names of competition, but the pedigree of a competitor needs to be a company that is likely that is actually likely doing by CMOS as huge assembly capabilities and is able to master PLP. If you find the cost of company, they are likely to become the competitors of ours.

Next question comes from Didier Scemama from Bank of America.

I just wanted to come back to the question that Sandeep asked. Just wanted to understand -- if you could give us a high-level breakdown of revenues, '25 and whatever you think it could be in '28 between user terminals, gateways and the satellite itself, just high level so that we've got an idea on how to model this business and I really appreciate for all the details you've given. The second part and the follow-up would be, I think from your slide, you mentioned that there was 9 million subscribers to satellite communications. But -- there's been about 20 million user terminal chips being sold, and I think expectations for 30 million from your slide. Can you explain the difference? Are those chips also going into the gateways or any other market that would explain the delta between these 2 numbers?

Thanks for the question, [indiscernible]. The first question, I will not disclose precise numbers. Directionally, I will repeat what I just said earlier, user terminal are the largest contributors to revenue, followed by satellites, while gateways are less materiality when it comes to our revenue at this stage. To date, what you need to understand in terms of numbers of chips sold compare in contrast to numbers of subscribers. I think this was your question, Didier. It's actually a good question. which is a function of the amount of antenna elements in a user terminal. And that amount of interline elements, which what I talked about those content modules that actually manage and can elements to do what I said it was doing in terms of CX informing that actually varies greatly and has been very -- has been moving greatly between different countries of terminals. It's actually -- you need to think of user terminals not as something that is software-defined, the capacity, the bandwidth you will get at all is a function of the capacity of your terminal, the numbers of antenna elements it has -- and that number to make it sure that number can vary anywhere between, I would say, the 100 to more than 1,000. So based on the mix of user terminals and so on and so forth, this is where the close to 10 million subs ended up actually leading to more than 7.5 billion ICs. So you can make the math backwards a little bit [indiscernible]

Okay. Okay. So there are basically multiple antennas per user terminals.

Anywhere from more than one of us to more than 1,000.

The next question comes from [indiscernible] from TD Cowen.

Two from me. So one, and I'm sorry for continuing to ask about the share narrative. It's just we don't get many verticals where companies have 90% share very often. But as we think about the share moving lower over the next several years, how much of that is just naturally from SpaceX having less overall dominant position in the market share as others ramp versus dual sourcing on stock that you're currently engaged in, including at SpaceX. And then for my follow-up, you made a comment before that you think competitors will primarily be IDM. Can you walk through the logic here? I mean there are third-party foundries that offer FD-SOI and buy CMOS, for instance, is it because you anticipate these large satellite customers needing some element of customization or something else that's driving that the logic about it needing to be from IDMs.

So sure. Thanks. Those are 2 great questions. I'll start with the share question. I will not answer it directly. But if I look at the total SAM over the next 5 years, it's actually the life is from a linear standpoint, I mentioned by SpaceX. So as such, I would say that actually our members are more actually are more actually influenced by our market share in SpaceX that they are influenced by the success of names LEO 1 wave or a Chinese [indiscernible]. That's the first answer to your question. That's the answer to your first question, forgive me. The second question, to wire idea, I think it's a great question for -- and there is 2 reasons to that. But 1 is more important than the other. The first is what you've mentioned. -- often for the front-end modules, those are not actually ASSPs, they end up actually custom products. Think of an ASC that has to be debated often, you need to have design resources to go do that, which makes a pure platform a bit more complex. But I tell you, actually, it is a [indiscernible] issue. The first order issue is actually, especially when you build those front-end modules, the back end, end up being as important as the home. I will not share the exact details in terms of cost contribution for a module, but the competitiveness you have to build actually the back end is critical. And if you allow me, I would take a little of 2 on this because it's important for you to understand something. We are actually -- we have developed this technology called panel level packaging. And unlike traditional packaging, we're using circular silicon wafers, actually PLP used large rectangular form factor panels, if you wish, and final label packaging world they are quite large, and by no of millimeter. And we have also developed a specific direct copper interconnect, which allows actually to eliminate electrical resistance and [indiscernible] and making it actually super efficient from a power efficiency and server management. This BAP technology alone without the be will not be as compelling. The reverse is also true. The by small without the PMP technology will not be as company. And that's the reason why George was saying, "Look, when you stack up all those margins and you look at the technology end-to-end, I do believe that it's a bit complex to not be an idea in that market.

The next question comes from Jakob Bluestone from BNP Paribas.

You kind of gave a bit of color around the content per unit for satellites being in the tens of thousands of gateways being in the hundreds of dollars the user terminals being in the tens of dollars. And I was just wondering, can you maybe give us your thoughts on how do you see those numbers evolving? Do you think they'll go up or down over time? And any thoughts you can share with us on pricing trends given that may be a little bit of pressure on market share?

[Audio Gap] See quite sizable contribution from others or still very concentrated client base in 2027, 2028?

Product line cycles, I would say, are for user terminal Ale and roughly a month in terms of generation. And you -- and I'm very happy to share that number with you because I've shared some of the economics. And as such, you can now understand why only people with manufacturing excellence are able to go and evolve in type of life cycle because you're doing thermical ramp-up and vertical run down. And obviously, we have engaged customer programs that have told us how to do that at full yield and full quality. There is no coincidence eventually as to why we are where we are. our goal on the customer base will be okay in the next 5 years. I don't see more than 10 players. So [indiscernible] as broad as it gets because you understand the dynamics, right, for you to be relevant. It's completely dimensioned by launches and satellites. So there is only so many companies and so much money able to do that. So I think in terms of scale that will park ST economics, I don't see more than 10 players. And even that I have catched a number a little bit.

We take the follow-up question of Didier Scemama from Bank of America.

Yes. So it's not directly related to LEO satellite, but I wondered, is that number that you've given that just about $1 billion for this year. Does that include also military applications? Because I would imagine is also supplying into the [indiscernible] and some other military equipment? Or is that included in there? And is that substantial? And I've got a quick follow-up.

Not really, it's not and the answer in general outside of this would be its corporately marginal.

And since its May 4. I wondered what would be your silicon content to the [indiscernible].

That is a great question, is will require you to become Jedi foe me to share the answer.

I think this is ending our call for this presentation. So thank you, everyone, for joining us. If you have further questions, feel free to reach out the IR team, and I wish you a happy day. Thank you.

The conference is now over. Thank you for joining Chorus Call, and thank you for participating in the conference. You may now disconnect your lines. Goodbye.

Ladies and gentlemen, welcome to the STMicroelectronics First Quarter 2026 Earnings Release Conference Call and Live Webcast. I am Moira, the Chorus Call operator.

[Operator Instructions] The conference is being recorded. The conference must not be recorded for publication or broadcast. [Operator Instructions]

At this time, it's my pleasure to hand over to Jerome Ramel, EVP, Corporate Development and Integrated External Communications. Please go ahead.

Thank you, Moira. Thank you, everyone, for joining our first quarter 2026 financial results call. Hosting the call today is Jean-Marc Chery, ST President and Chief Executive Officer. Joining Jean-Marc on the call are Lorenzo Grandi, President and CFO; and Marco Cassis, President, Analog, Power and Discrete, MEMS and Sensor Group, and Head of STMicroelectronics Strategy, System Research and Applications and Innovation Office.

This live webcast and presentation materials can be accessed on ST Investor Relations website. A replay will be available shortly after the conclusion of this call.

This call will include forward-looking statements that involve risk factors that could cause ST results to differ materially from management expectations and plans. We encourage you to review the safe harbor statement contained in the press release that was issued with the results this morning and also in ST's most recent regulatory filings for a full description of these risk factors. Also, to ensure all participants have an opportunity to ask questions during the Q&A session, please limit yourself to one question and a brief follow-up.

Now I'd like to turn the call over to Jean-Marc Chery, ST President and CEO.

Thank you, Jerome. Good morning, everyone, and thank you for joining ST for our Q1 2026 earnings conference call. I will start with an overview of the first quarter, including business dynamics, and I will hand over to Lorenzo for the detailed financial overview. I will then comment on the outlook and conclude before answering your questions.

So starting with Q1. Our first quarter net revenues were $3.1 billion, including about $40 million revenues associated with NXP's MEMS sensor business, which we acquired during the quarter. Excluding this contribution on a sequential basis, net revenues were above the midpoint of our business outlook range, driven mainly by higher revenues in our engaged customer programs in Personal Electronics and in Communication Equipment and Computer Peripherals.

Gross margin was 33.8% or 34.1%, excluding the impact of the purchase price allocation, so-called PPA, following our acquisition of NXP's MEMS sensor business. Excluding impairment, restructuring charges and other related phase-out costs and purchase price allocation, PPA effects from our acquisition of NXP's MEMS sensor business, non-U.S. GAAP diluted earnings per share was $0.30.

During the first quarter, inventory in our balance sheet increased slightly, and we continue to work down inventories in distribution. They are now normalized. We generated a negative $720 million free cash flow, including $895 million cash out related to the payment of our acquisition of NXP MEMS sensor business.

Let's now discuss our business dynamics during Q1. Well, first, we had a strong booking momentum during Q1 with book-to-bill well above 1 across all end markets and regions. In Automotive, during the quarter, revenue declined 10% sequentially. Year-over-year, revenues increased 15%, marking the return to year-over-year growth. Automotive design momentum progressed with various OEM and Tier 1 ecosystems. We had design wins across electric, hybrid and traditional vehicles, spanning onboard chargers, DC-DC converters, powertrain active suspension and vehicle control electronics. Key products include power semiconductors, smart power devices, automotive microcontrollers, analog devices and sensors.

In February, we completed the acquisition of NXP's MEMS sensor business. The acquired technology and product portfolio are highly complementary to STs and strengthen our automotive sensor business. We are progressing as planned with the integration into our portfolio and operational flows. Industrial decreased by 1% sequentially and improved 26% year-over-year. Importantly, inventories in distribution further decreased and are now normalized.

In Industrial, our broad portfolio of microcontrollers, sensing, analog and power devices is strongly aligned with industrial transformation trends and the evolving needs of physical AI. During the quarter, we saw design wins across industrial automation and robotics, building automation, power systems, health care and home appliances. We announced our collaboration with NVIDIA to integrate ST sensors, microcontrollers and motor control solutions with NVIDIA Robotics ecosystem. This aims to help developers design, train and deploy humanoid robots and other physical AI systems with higher efficiency, reliability and scalability.

We are also proud to have been ranked the #1 vendor worldwide for general purpose microcontrollers for the fifth consecutive year based on research by Omdia. During March, we announced that the first batch of STM32 wafers fully produced in China for ST by our partner, Huahong, has been delivered to customers in China. This was a major step forward in ST China for China supply chain strategy.

For Personal Electronics, first quarter revenues were down 14% sequentially, reflecting the seasonality of our engaged customer programs and up 21% year-over-year, reflecting increasing content. During the quarter, we reinforced our position in mobile platforms and connected consumer devices, supported by both engaged programs and a broad open market portfolio spanning sensors, secure solutions and power management. We announced support for motion sensing and secure wireless technology on Qualcomm Technologies' newly launched Personal AI platform based on ST smart sensor and secure NFC controllers.

For Communications Equipment and Computer Peripherals, first quarter revenues were above our expectations, up 3% sequentially and 41% year-over-year. We continue to reinforce our position as a supplier of critical semiconductors that power cool and connect AI data centers from the grid to the core and from the core to the user. ST is now strategically positioned to capture upside from new AI-driven program, leveraging specialized technologies to enable the evolving AI infrastructure.

We confirm our data centers revenue expectation to be nicely above USD 500 million for 2026 and well above $1 billion for 2027. In a major development, we expanded our strategic engagement with Amazon Web Services through a multiyear multibillion U.S. dollar commercial engagement to enable new high-performance compute infrastructure for cloud and AI data centers. This engagement covers a broad range of semiconductor solutions, leveraging ST portfolio of proprietary technologies.

During the quarter, we secured multiple design wins for silicon and silicon carbide-based power solutions. These supports the drive for higher power density and increased energy efficiency for next-generation AI compute and data center architectures. We announced the expansion of our 800-volt DC AI data center power conversion portfolio with new 12-volt and 6-volt architectures in collaboration with NVIDIA. With this, ST now provides a complete portfolio for the 800-volt VDC power distribution inside gigawatt scale compute infrastructure, leveraging ST power, analog and mixed signal and microcontroller products.

We also announced the start of high-volume production for our silicon photonics-based photonics ICs 100 -- PIC100 platform used by hyperscalers for optical interconnect for data centers and AI clusters. The technology enables higher bandwidth, low latency and greater energy efficiency. As I mentioned last quarter, the momentum in optical interconnect technologies is also driving demand growth for our high-performance microcontrollers in pluggable optics. We are also seeing initial demand for our secure element in data server power supply units to support authentication and detect data manipulation attacks.

Our low-earth-orbit satellite business based mainly on our BiCMOS and panel-level packaging technologies strongly progressed during the quarter. We were selected to develop a power amplifier controller for direct-to-cell satellites based on our proprietary BCD technology by our main low earth orbit customer, and we continued to ramp shipments to our second largest customer.

For sustainability, we issued our annual integrated report during the quarter. This report integrates our sustainability statement detailing our performance in 2025. We made further progress and remain on track for our commitment to becoming carbon neutral by '27 on Scopes 1 and 2 and on product transportation, business travel and employee commuting for Scope 3. We also target the sourcing of 100% renewable electricity by 2027 and achieve 86% in 2025.

Now over to Lorenzo, who will present our key financial figures.

Thank you, Jean-Marc. Good morning, everyone. Let's start with a detailed review of the first quarter, starting with revenues on a year-over-year basis. By reportable segment. Analog products, MEMS and Sensors grew 23.2%, mainly due to Imaging and MEMS and to a lesser extent, Analog. Power and Discrete products decreased 1.8%. Embedded Processing revenues were up 31.3% due to general purpose MCU and to a lesser extent, custom processing and RF and optical communication grew 33.9%.

By end market, Communication Equipment and Computer Peripherals grew 41%; Industrial 26%, Personal Electronic, 21% and Automotive 15%. Year-over-year, sales to OEMs and distribution increased 24.5% and 19.2%, respectively. On a sequential basis, Analog product, MEMS and Sensor decreased by 9.1%, Power and Discrete by 5.4%, Embedded Processing by 4% and RF & optical communication by 9%. by end market, on a sequential basis, Communication Equipment and Computer Peripheral was up 3%, while the other end markets declined. Industrial was down 1%, Automotive, 10%; and Personal Electronic, 14%.

Turning now to profitability. Gross profit in the first quarter was $1.05 billion, increasing 24.3% on a year-over-year basis. Gross margin was 33.8%, increasing 40 basis points year-over-year, mainly due to lower unused capacity charges and better product mix. On a sequential basis, gross margin decreased by 140 basis points.

Gross profit included $11 million purchase price allocation PPA effects from our acquisition of NXP's MEMS sensor business. Non-U.S. GAAP gross margin, excluding this item, was 34.1%, excluding the impact of NXP's MEMS sensor business and related PPA effects, gross margin stood at 33.9%, 20 basis points better than the midpoint of ST guidance, which did not include any impact related to our acquisition of NXP's MEMS sensor business.

Q1 gross margin included about 50 basis points of negative impact resulting from a nonrecurring cost related to our manufacturing reshaping programs. The negative impact on gross margin from the just mentioned nonrecurring cost is expected to remain at similar level over the rest of the year.

Total net operating expenses, excluding restructuring, amounted to $904 million in the first quarter. Excluding the purchase price allocation PPA effects from our acquisition of NXP's MEMS sensor business, non-U.S. GAAP OpEx stood at $885 million. Non-U.S. GAAP net OpEx included OpEx related to the acquired NXP MEMS sensor business and a one-off impact related to a settlement with a supplier. Excluding these 2 items, non-U.S. GAAP net OpEx was broadly in line with the expectations given in January, which did not include any impact related to our acquisition.

For the second quarter of 2026, we expect non-U.S. GAAP net OpEx to stand between $950 million and $960 million. The sequential increase is mainly due to calendar base effect, start-up costs and 1 incremental month of OpEx related to the acquired NXP's MEMS sensor business. Excluding these items, Q2 '26 non-U.S. GAAP net OpEx would slightly decrease sequentially.

In light of our acquisition of NXP's MEMS sensor business and the new AI revenues opportunity, let me give you some more color on the 2026 OpEx. For full year 2026, we now expect like-for-like net OpEx to be up mid- to high single digit year-over-year versus our previous expectation for a low single-digit increase as we are accelerating our investment in new business opportunities, including NXP's MEMS sensor business acquisition and the exchange rate impact, net OpEx should be up low double digit year-over-year.

In the first quarter, we reported $70 million of operating income, which includes $71 million for impairment, restructuring charges and other related phase-out costs. These charges are related to the execution of the previously announced company-wide program to reshape our manufacturing footprint and resize our global cost base. Q1 operating income also included $30 million purchase price allocation effects from our acquisition of NXP's MEMS sensor business.

Excluding these items, Q1 non-U.S. GAAP operating income stood at $171 million and non-U.S. GAAP operating margin was 5.5% with Analog product, MEMS and Sensors at 12.2%, Power and Discrete negative 21.5%, Embedded Processing at 16.9% and RF & Optical Communication at 14.9%.

First quarter 2026 net income was $37 million compared to a net income of $56 million in the year ago quarter. Diluted earnings per share were $0.04 compared to $0.06 1 year ago. Non-U.S. GAAP net income stood at $122 million and non-U.S. GAAP diluted earnings per share stood at $0.13. Net cash from operating activities totaled $534 million in the first quarter compared to $574 million in the year ago quarter.

Net CapEx was $362 million in the first quarter compared to $530 million in the year ago quarter. Free cash flow was negative at $723 million in the first quarter compared to a positive $30 million in Q1 2025. Q1 '26 free cash flow includes $895 million cash out related to the payment for the acquisition of NXT's MEMS sensor business.

Inventory at the end of this quarter was $3.17 billion compared to $3.14 billion in Q4 2025 and $3.01 billion in Q1 2025. Days sales of inventory at quarter end were 140 days, in line with our expectation compared to 130 days of the previous quarter and 167 days in the year ago quarter. Cash dividend paid to stakeholders in the first quarter of 2026 totaled $71 million. ST maintained its financial strength with a net financial position that remained solid at $2 billion as of March 28, 2026, reflecting total liquidity of $4.57 billion and total financial debt of $2.57 billion.

Now back to Jean-Marc, who will comment on our outlook.

Thank you, Lorenzo. Now let's move to our business outlook for Q2 2026. We are expecting Q2 2026 revenues at $3.45 billion, plus/minus 350 basis points. At the midpoint, our Q2 2026 net revenues will increase 11.6% sequentially and by 24.9% year-over-year. We expect our gross margin to be about 34.8%, plus/minus 200 basis points, including about 100 basis points of unused capacity charges. Non-U.S. GAAP gross margin is expected to be about 35.2%. This business outlook does not include any impact for potential further change to global trade tariffs compared to the current situation.

To conclude, in the first quarter, despite the macroeconomic uncertainty, we saw improving demand with strong booking and normalized inventory in distribution. In the second quarter, we expect revenues well above average seasonality as well as an increased gross margin. We have a clear path to improve gross margin while staying at the forefront of innovation. We expect 2026 revenues to show double-digit growth beyond our addressable market dynamics and our already engaged customer programs. This growth will be driven by new AI programs for which we leverage our specialized technologies to enable the evolving AI infrastructure.

Before handing over to Jerome, I am pleased to announce that as we did in March for Cloud AI and intelligent sensing, on May 4, we will host a dedicated call on ST's low-earth-orbit satellites, explaining how we are going to achieve our ambition of well above $3 billion cumulative revenues over the period '26 to '28 for this opportunity. You will receive the invitation today.

Thank you, and we are now ready to answer your questions.

[Operator Instructions] The first question comes from the line of Joshua Buchalter from TD Securities.

Congrats on the very solid results. So you have a lot of idiosyncratic growth drivers hitting this year across data center, silicon photonics, LEO satellite and then your largest customers, normal seasonal ramp. Can you sort of help us with the shape of the year and how we should expect them to layer into the model? Like should we expect 3Q and 4Q this year to also be above seasonal because of these company-specific growth drivers?

I am taking the question. Well, of course, I will not guide on '26, but maybe we can share a few elements. First of all, okay, the strong booking of Q1 has shown absolutely no pull-in order. It is, let's say, a well-balanced loading of the 2026 quarter-to-quarter. So the billable on '26 from the booking we received in Q1 represent approximately 85% to 90% of the booking we received. So this is positive to make us confident that in H2, we could achieve the usual seasonality H2 versus H1. Then what will be again positive on the year 2026, looking at the current dynamic in terms of growth.

In automotive, we confirm that '26 will be a growth for ADAS for sensor, of course, and also with the boost of the acquisition of MEMS from NXP and for silicon carbide. In Industrial, we will see a solid and strong growth on general purpose microcontroller. In Personal Electronics, okay, as we have already seen in Q1, our engaged customer programs in Sensor and Analog will be, let's say, a contributor of the growth but not a big one in H2 because a change of profile in the introduction of the new device.

Well, in data center, it is clear that here we are seeing a really strong growth in terms of demand, acceleration, including cloud optical interconnect, both for our PIC100 for our BiCMOS but I repeat for our general purpose microcontroller and analog and power discrete as well. So we confirm the revenue nicely well above USD 500 million. But the only negative aspect of the revenue in '26 is capacity reservation fees that will decrease, okay, $140 million compared last year.

So this is how we see the year 2026. So I repeat, backlog now well loaded, great confidence level to have H2 versus H1 at the usual seasonality on top of ADAS, SiC, sensor, general purpose micro, clearly, AI infrastructure and low-earth-orbit satellite will be very strong contributor to the performance of ST in 2026.

Really appreciate it. I was hoping you could comment on the pricing backdrop. I mean, one of your large competitors last night said it was coming in a little bit better than they originally planned and now expect flat pricing. Have you seen changes in the pricing environment over the last 3 months? And sort of what are your expectations on pricing for the year?

So, here, I'll let Lorenzo comment.

Yes. Thank you for the question. If you remember last quarter, we were talking about pricing decline on low to mid-single-digit expectation. But clearly, there is some evolution in respect to this expectation. I would say that in Q1, our price decline was as expected, low single digit. What today we see, we see an environment in which actually there is some selected price increase that also we expect. So at this point, I would say that in terms of pricing, our expectation is to have a very low single-digit, let's say, price decline. So it means that actually, in terms of pricing, we see a better situation in respect to what was a few months ago.

The next question comes from Francois Bouvignies from UBS.

Maybe just a follow-up on the pricing. I mean we have seen some announcements that you will increase your pricing in April and you are not the only one. So can you just give us an idea of how much of your revenues would be impacted by the margins? And also, Lorenzo, what about the gross margin with this pricing increase? I mean, should we -- I would imagine it takes a bit of time to fuel into your P&L. So when should we expect some gross margin impact from this gross margin increase that we see in the press? That's my first question on pricing, gross margin.

No. Clearly, let's say, when we look at the price environment, I would say that at this stage, yes, there is some selective price increase. It's not a price increase for what concern us across all, let's say, customer and products. Anyway, what I can say is that when we look at the dynamic, of course, of the -- dynamic of our gross margin moving from Q1 to Q2, and we may say that pricing is quite neutral in respect, let's say, to this dynamic, means that at the end, it's not a boost but it's not even a detractor. It will remain substantially flattish when we look at the evolution of the gross margin. That is not what is the normal trend when we look, let's say, the seasonality between these 2 quarters.

For sure, as a positive when we look at our gross margin, there is the mix. Mix is continuing to be, let's say, positive on our gross margin evolution. But clearly, there is also lower unused capacity charges. Our fabs that are better loaded capacity charges is declining moving from Q1 to Q2.

But there are still some negative. The negative is mainly related to our manufacturing efficiency. Why? Because there is some temporary suboptimal efficiency in the context of our reshaping plan. We are moving technologies, products from 200-millimeter fab to 300-millimeter from the 150-millimeter of silicon carbide to 200-millimeter. And we are really in the middle of this kind of programs that, for sure, let's say, are somehow impairing a little bit the efficiency of our fabs. And this, I would say, is the main detractor when we look at the evolution over the -- on a sequential basis of our gross margin. Pricing, as I said, is really neutral at this stage.

Maybe one for Jean-Marc. I mean, if we look at your customer programs, if I exclude the Personal Electronics, so if I take silicon carbide, photonics and satellites, so your big programs. Should we expect your revenues to grow quarter-on-quarter from here, like the fundamental that is increasing gradually. So no seasonality, I would imagine. So you should be able to see a growth across the board here quarter-on-quarter for the year. Is that the right assumption excluding Personal Electronics?

Of course, excluding Personal Electronics, this is what we expect.

Next question comes from Janardan Menon from Jefferies.

Just a follow-up on gross margin, Lorenzo. Looking into the second half, what would you see as the various puts and takes on that gross margin evolution? Your top line is growing perhaps much faster than what we had thought a few months ago. So would it be that, that utilization and underloading charges will get used up faster? And what is -- there's normally a lag between the revenue trend and the gross margin. So just if you could give us any commentary on how -- not in terms of actual numbers but just the puts and takes perhaps of the second half? And how do you feel about your sort of your model of getting to 45% given the kind of strength that you're seeing in end markets and the favorable product mix that you're seeing right now?

No. What I can say about the gross margin is definitely that the gross margin, let's say, this year will improve in respect to what has been in last year, definitely and will improve sequentially when we look Q1, Q2, Q3 and Q4. This is something that definitely we expect. This is what we expect to continue to see a sequential increase and sequential improvement over Q3 and Q4. What are, let's say, the driver we expect? Clearly, as you said, the unused capacity charges will improve, thanks to the fact that we will have higher revenues, even if I confirm that will not completely disappear. We will still have some areas in which especially related to the legacy technology that we will still have a little bit of unused capacity charge but much lower than what we saw, let's say, last year definitely.

There will be progressively some manufacturing efficiency improvement. Even if I repeat what I said before, we are not yet optimized because, let's say, we are in this transition. We will start to see this benefit of the transition mainly in 2027 more and then in 2026. But for sure, there will be some improvement moving forward from Q2.

Mix will be another positive impact. We will continue positive impact on mix. But clearly, we know that capacity reservation fees now are out. There will not be, I mean, much lower, let's say, there will not be a significant variation moving from Q2, Q3 and Q4, but are much lower in respect to what it was, let's say, last year.

As I said, there is this cost related to this transformation of our manufacturing infrastructure. Maybe what we will have, let's say, in the second half is a little bit higher input cost for our manufacturing considering, let's say, the overall situation. But definitely, I confirm that starting from our, let's say, 35.2% Gross margin in the second quarter, we will continue to see progressively improvement in Q3 and Q4.

And maybe just a quick follow-up. On your Q2 outlook of 11.6%, is there already a very significant contribution from the optical connectivity on the data center, which is driving that upside? Or is the Q2 more driven by a pickup in industrial, general purpose microcontrollers, et cetera, and the optical kicks in more meaningfully into the second half of the year?

The optical are starting to contribute. In fact, since Q1 is mainly through the high-performance microcontroller. But the main part of the optical overall with photonics by BiCMOS will be in H2 but microcontroller are already participating to the growth.

The next question comes from Gianmarco Bonacina from Banca Akros.

I have a question more for the midterm. You gave some figures for your, say, AI revenues for next year, above $1 billion. I just wanted to understand in terms of the commercial activity, we read -- we commented the big contract with AWS. So are you working on a commercial basis just to get potentially the revenues with other hyperscaler? And how confident you are that, let's say, the opportunity you realized with AWS can be also generated with other hyperscaler maybe, I mean, in the midterm, not just in the near term?

If we speak about midterm, our strategy on hyperscaler are the following. Basically, if you break down this, let's say, infrastructure in 3 main application domain. What we call the network flow. This is where we have spoken about, let's say, the optical cable and near technology, let's say, evolution with packaged co optic or near-packaged optics clearly here, one of the main driver will be AWS but clearly, ST is positioned to provide -- to be a provider of product and solution for optical cable far beyond, only AWS. This is the point number one.

And then the second big domain is clearly well known is what we call the power flow. So it means the capability for electronics to enable the supply of the processor from 20,000 volt to 0.8 volt. And here, ST is engaged now with a large product portfolio from, let's say, SPS, low-voltage MOSFET, microcontroller, driver, sensing and so on and so forth. And this will come far beyond AWS. Of course, AWS okay, will use this component but we will provide and we will compete far beyond AWS.

Then last but not the least is all the infrastructure around the thermal cooling of this infrastructure, and we are already there with our power solution, microcontroller and analog. So clearly, AWS will be a fantastic driver for ST for the growth of the revenue during the next 3, 5 years. But our ambition is well above, thanks to our product portfolio. And I repeat ST is a unique company capable to provide on this infrastructure from photonic solution, MEMS solution that will come pretty soon, microcontroller definitively, power switches, power drivers, controllers and including other sensor. So this unique position, clearly position ST in the future, to be an important contributor in terms of supply to this business line.

Okay. Just a quick follow-up for Lorenzo, if I can. The change in the guidance in the OpEx, just to understand correctly. So you are talking about your clean OpEx excluding PPA and restructuring.

Yes. Yes. Of course, we exclude PPA and restructuring. And as I was saying before, at the end, yes, apart of the fact that we have the addition of NXP that when we were talking previously was not taken into consideration. But I have to say that thanks to the fact that we see significant, let's say, opportunity in terms of revenues, we have some programs accelerating in terms of, let's say, development and bringing a little bit more level of expenses. I have to say that in any case, when we look at our net OpEx, the expense to sales ratio 2025 compared to 2026, let's say, in 2026, the expense to sales ratio will materially decline with respect to the previous year.

The next question comes from Andrew Gardiner from Citi.

Just a couple of, I suppose, follow-ups to some of the topics that have already been discussed. First, on the AI side, Jean-Marc, you've -- I think it's a reiteration of what you were saying last month in terms of the "nicely above $500 million of revenue for this year" and "well above $1 billion for next year." Just things are moving very quickly in this part of the market to put it mildly. What is the potential for upside there? And I suppose, more importantly, for you, where are you seeing capacity constraints at the moment that may indeed limit the level of upside relative to the demand that you're seeing?

And then a quick one for you, Lorenzo, just again on the OpEx. You said a low double-digit gain '26 on '25 on one of the items that you were looking at. Could you just provide us the baseline of that? I missed that when you were saying it in the prepared comments.

Never mind. It is clear that we are on some part of the technology and components that are enabling the solution we provide to customers, we are in ramp-up mode. Clearly on photonics and associated technology, we are in a ramp-up mode. Overall, what I can confirm today that the unconstrained demand we have today for '26 and '27 is well above the nicely above $500 million and well above $1 billion. And our ambition is to fulfill this unconstrained demand but the company first has to ramp up, okay, the capacity already installed in the second half of the year to implement additional capacity. And our ambition is to fulfill as much as we can the unconstrained demand of customers.

I will provide more color in July, clearly during our next meeting. But I really confirm that '26 will show a significant breakthrough in our revenue linked to AI data center.

For what concern OpEx, I confirm that net OpEx sales ratio will decrease in 2026 compared to 2025. What we expect, now we expect when we say OpEx like-for-like means, let's say, the same FX and same perimeter means not including the NXP acquisition to be up mid- to single digit, let's say, in 2026 compared to 2025. You have to consider that half of this increase is related to the start-up cost that we have, let's say, in the fab 300-millimeter and, let's say, 200-millimeter for silicon carbide that is, let's say, related to our transfer from 200 to 300, 150 to 200 for silicon carbide. So it means that this is something that is not structural is coming this year but will not stay forever.

If we include the NXP MEMS business acquisition, and also include the impact of the exchange rate, excluding the restructuring, we should be up low double digit versus 2025. This is assuming an exchange rate effective in the range of [ 1.15, 1.16 ] and is, of course, including, let's say, the operation of NXP MEMS business that we can estimate in the range of $50 million additional expenses for us in this year.

The next question comes from Sebastien Sztabowicz from Kepler Cheuvreux.

On the transformation program, where are you standing right now in terms of capacity build and so on? And when do you expect to have the full synergies benefit? Is it for '27 or more for 2028? And the second question is on silicon carbide and your JV with Sanan in China. Where are you in the ramp-up mode with the JV? And when do you expect the first volume to start to ramp up meaningfully in China?

For the transformation program, clearly now we are in the middle of the execution. Clearly, we have to respect the customer qualification time, when for analog technology, we move from 200-millimeter to 300-millimeter. This is -- they have a good incentive to do it because clearly, our capacity potential increase is related to Agrate in 300-millimeter. We expect that the benefits of Agrate at full speed will be more in the end of '27 and entering in '28, not related to the fact that we don't go at the right speed in terms of qualification internal, but more related to the customer normal constraint they have to qualify their own application.

On silicon carbide, it's a bit similar, in fact, okay, because here, we are moving from 6-inch to 8-inch, and this is mandatory to do it here is the same. We are not limited by our own capability, both in Catania and in Sanan in Chongqing. The limitation is more related to the qualification time of our customer. And you know that we are engaged in a very, very famous platform with an important player in Europe, which currently has a great success for this new platform in electrical car. And here, of course, we cannot take any risk and it takes time before to move to 8-inch. So for sure, are the same. The benefits will be more end of '27 and entering in '28, and in Sanan, we expect to start the production and to load this nice infrastructure starting the end of 2026.

The next question comes from Sandeep Deshpande from JPMorgan.

My question is regarding the acquisition of the NXP sensors business. How -- I mean, how did that business grow in the past? And how will that contribute to growth in the current year? And then my follow-up question is regarding the gross margin of the company. You've said that your underutilization charges do not fully go away this year. But should we assume that in '27, the underutilization charges go away and with the mix shifting more to the AI products as well as some of the satellite products, et cetera, that there could be a much bigger move in the gross margin in full year '27?

Thank you, Sandeep. So Marco Cassis will take the first question on NXP -- former NXP MEMS. And Lorenzo, of course, the second question.

Yes. On NXP, the combination of the capabilities of the 2 companies is translating in acceleration, of course, related to a market, which is automotive and clearly is moving at the speed of the automotive but it's an acceleration of opportunities of design-in and design win because we are putting together the best of the 2 worlds, which is a very strong positioning of NXP MEMS in accelerometers, where they do use -- sorry for a little bit of technical but monosilicon crystal, which are extremely good in terms of temperature performance for automotive and our capabilities on the 6-axis.

So we do see that we are going to grow with NXP at faster speed than what is typically the market growth in safety application. So it's going to be a contribution of the growth of the overall MEMS business. I hope I'm answering to your question, Sandeep.

So how much was the growth in the past couple of years in that business?

Well, it was in the range around low single-digit growth, which is the typical growth of safety application in automotive.

And you expect that to accelerate is what you're saying?

I am expecting this one to accelerate, yes.

Understood.

Okay. Maybe I take the one of the gross margin, let's say, confirm what I was saying before, the gross margin will improve starting from our 35.2% this quarter of Q2 after, let's say, quarter after quarter this year driven by the seasonality of the revenues, the continued reduction of the unused capacity. As I said before, let's say, still there will be some but reducing over the second part of the year and then the continued improvement of the mix. Clearly, let's say, this is our trend to the path above 40%. We said that when the company, let's say, will be with revenues above $4 billion quarterly revenues, let's say, we expect to have our gross margin at 40%. After that, our reshaping plan will be completed. So this is going in this direction, let's say.

So what I can say today is that clearly, let's say, in our gross margin, there is still some negative impact on this reshaping plan, temporary negative impact due to the activity that we are doing that we will progressively go down and transform, let's say, positive impact when we will start to have, let's say, the benefit of these programs. So yes, I confirm that at the end, let's say, there will be -- you will see a progressive improvement in our gross margin moving in Q3 and Q4 and then, of course, in 2027.

Thank you, Sandeep. We have time for a very last question.

The last question for today is from Lee Simpson from Morgan Stanley.

Great. Maybe just a couple of questions, if I could, around data center power and then on the photonics side. Just on the data center power, it did look as though you were saying you've seen some design wins. It looked as though with silicon, silicon carbide, most all of it first stage. I just wondered if you could give us a sense for the engagements you're seeing around gallium nitride, where regionally that may emerge? And then maybe just on the voltage regulation side on the second stage, anything really happening there, certainly as we look out to '27?

So Marco will answer the detail. Interestingly, for all of you, guys, maybe what I can tell you that the nicely above USD 500 million in '26 will be spread approximately between 40% related to analog and power and 60% related to microcontroller and radio frequency optical cable. Just for you to have the span of our revenue for 2026.

And I'll let Marco to answer the detail.

Yes. For what is related to power compared to our positioning 1 year ago, we put a major effort in expanding the portfolio to be sure that we can cover basically from grid up to driving the GPUs. And this goes through the full portfolio of ST, which is silicon-based, silicon carbide with different voltages and new packages that we are introducing where we are not present. And of course, the GaN, which is an important for the 800 volt where we are in sockets that I think will come to life during this year and next year. So the position overall in terms of portfolio is now much stronger than it was, and this will translate in revenues during '26 but mainly during '27. And this goes across the different ecosystem of suppliers, which means power supply makers based in many cases in Taiwan and of course, the ecosystem that we have in U.S.

So overall, the trend is going through the full portfolio of ST, again, with a portfolio that has been expanded and now is rich and covering all the stages of the power conversion.

That's very clear. Maybe if I sort of move it on to the photonics side. It always seems ST is extremely good at getting a big lead customer, pipe cleaning a new market opportunity and creating advantages, if you like, in technology, leveraging some of the IP in-house. But that transition to a standard product in the market for us always feels like the real ROI where margins can be accretive. Are we seeing when we look at the PIC and some of the engagements you have in the market, the possibility that this PIC100 becomes a standard product in the market?

Standard product, okay, I will not classify it as a standard product, okay? Maybe application standard specific, maybe yes. But one thing, I prefer to share with you again to show how ST is and will be a reference on silicon photonics. First of all, we are the unique company capable to provide silicon photonics technology on 12-inch. So we have the capability to increase our capacity in both in Crolles and possibly later on in Agrate. So for sure, ST, will compete on this market, largely to become a pure standard, you will have many innovation coming in the optical cable and optical solution. Again, the near-packaged optic, the co-package optics, all this will come maybe faster than expected. And silicon photon is a key enabler of all these technologies.

Okay. Thank you. Thank you, everyone. This is the end of this call. So thank you for joining us today, and we remain at your disposal if you have any follow-up questions. So sorry for the one who we couldn't squeeze into the question. So thank you very much. Have a good day.

Thank you.

Thank you. Bye-bye.

Ladies and gentlemen, the conference is now over. Thank you for choosing Chorus Call, and thank you for participating in the conference. You may now disconnect your lines. Goodbye.

Ladies and gentlemen, welcome to the ST Cloud AI Update Conference Call and Live Webcast. I am Maura, the Chorus Call operator. [Operator Instructions] The conference has been recorded. [Operator Instructions]

At this time, it's my pleasure to hand over to Jerome Ramel, EVP, Corporate Development and Integrated External Communications. Please go ahead.

Thank you, Maura. Thank you, everyone, for joining the ST Cloud AI Update Conference Call. Hosting the call today is Remi El-Ouazzane, President, Microcontrollers, Digital ICs and RF Product Groups. This live webcast and presentation materials can be accessed on ST Investor Relations website. A replay will be available shortly after the conclusion of this call.

This call will include forward-looking statements that involve risk factors that could cause ST results to differ materially from management expectations and plans. We encourage you to review the safe harbor statement contained in the presentation and also in ST's most recent regulatory filing for a full description of these risk factors. Also to ensure all participants have an opportunity to ask questions during the Q&A session, please limit yourself to 1 question and a brief follow-up.

I'd like now to turn the call over to Remi El-Ouazzane.

Good morning, and good afternoon, everybody. Today, I'm going to share with you some of our insights into tomorrow's advanced data centers and AI clusters, and how our company, ST, is positioned to contribute to them. I'm going to focus, both on the optical interconnect and power technologies.

In this presentation, you will see a lot of compute power and interconnect numbers, but it's good to remind ourselves that this is not just about making ChatGPT faster, high-performance computing as a potential to revolutionize society by accelerating advancements across various fields, be it in health care, in climate science or in agriculture, and this is really what it's all about.

I will skip the forward-looking information and go straight into the first half of this presentation. A handful of global hyperscalers drive the explosive AI data center growth, and they are actually projected to invest more than $700 billion of CapEx in '26 and in excess of $1 billion in 2030. Servers are creating a structural, and I may say once-in-a-lifetime growth opportunity, and you will see why ST is positioned to sharply increase its content, and as such, its revenue in data centers. Indeed, it's a combination of a lot of technological differentiation and well-positioned products that is going to allow us to go and engage on that market and support its growth.

Before we get going into some amount of details, we wanted to go and put some numbers behind the opportunity. And I think at this stage, there is no better way but to talk about our SAM per gigawatt of infrastructure. We see around $230 million. That number is approximately in nature. Should you write to $200 million or $250 million, I don't think it will change much. What matters here is the breadth of our offering this $200 million of SAM is backed up by about 400 products we have that are tuned to address the AI data center business. We can address every critical step that is actually relevant to us as an IDM that is power conversion, high-speed connectivity, control, monitoring and obviously, security. Each new AI campus built at this scale or each step up in rack power and rack bandwidth will automatically translate into more potential dollar opportunities for our company.

Let me deep dive into the first part of my presentation that is all related to power. As you know, the rapid growth of AI workload is driving unprecedented power demands in modern data centers. As AI models become increasingly more complex, computationally intensive, the infrastructure supporting these workloads must evolve to deliver amazingly high power densities.

We are coming from a world where traditional 54-volt power distribution system are absolutely reaching their physical and electrical limits. This legacy systems struggle to efficiently supply the massive power requirements of emerging megawatt-scale AI compute racks. And there are three main challenges that they're facing. One is a growing XPU compute, those chips are getting hungrier and hungrier from 500k watt -- from 500 watts to 1 kilowatt to 2 kilowatts to 3 kilowatts and more, delivering enough current at low voltages requires a massive amount of copper.

Size and space is the second problem. The thicker the cables and the more conversion stage we have, essentially there's less room you have for actual GPU content in a rack. This actually goes against compute density, which itself goes against high TCOs -- higher TCOs and lower ROI. The last piece is thermal management. Every time you convert power, you lose efficiency, and efficiency in this world translate into heat, which themselves will require actually more power to cool down. It's a very vicious circle. So that's why to address those challenges, the industry, as you know, is actually moving to more or less 1 megawatt per rack.

Why? Simply because by increasing the voltage to 800-volt in that case, the current required for the same power level is reduced, which in turn decreases resistive losses, for those who remember the physic classes and reduces the amount of copper cabling needed. This transition not only enhance energy efficiency, but will actually simplify the overall power infrastructure enabling more compact, more scalable data center designs that can actually meet the growing demand of AI workload, while actually reducing operational costs and quite importantly, environmental impact.

Now to unlock 1 mega plus density, ST is putting a combination of its most advanced technology, be it silicon carbide and gallium nitride as well as our smart power processing like BCD using galvanic isolation. It's actually not really an upgrade. It is truly a revolution and for ST, in particular, ST, it's a massive inflection point, giving us a chance to become an active player in this power conversion market.

If you go to the next slide. So what is the offering that we are supplying for this new paradigm? At the grid level, we leverage our leading position in wide bandgap material in the power supply unit to provide 800 volt AC to DC power shelf application using silicon carbide and GaN and setting the efficiency and power density baseline. Please do note that, that stage is actually sitting on a side car together with the BBU and also actually with circuitry that is allowing the rack to go and modulate itself as a function of GPU workload.

When you got this 800 DC, and that is actually being sent to the rack, we provide compact solution within this intermediate bus conversion with high power density, leveraging notably our GaN FET, and allowing us actually to go from 800 volt down to 54 volt.

Now I'd like to remind you that this is not new news. We actually announced that last October, together with NVIDIA. And we've announced that actually our team really designed two parts, hot swap protection circuit, which is actually a 1,200-volt silicon carbide. And the power convertor, the power converter in itself convert the 800-volt throughout the rack into this 54-volt needed by each server. And to provide this capability, like we've announced with NVIDIA a few months ago, we are providing actually a smartphone size footprint that uses actually 650-volt gallium nitride transistor in a stack half-bridge configuration.

On the last slide, which is actually what you see on the extreme right on that chart, we are developing silicon-based digital multi-phase controllers and smart power stage down to sub-1 volt of 08 compliant with market-leading XPUs, but in that case, we can actually cover the complete data center power chain of three stages, embedding a full set of power technology into integrated solutions that support customers from the grid all the way to the core. And I insist that actually this transition from 54- to 800-volt is actually creating a very nice inflection point for us actually to go and shine. This is it for the power section of my presentation.

I'm going to move now to the connectivity piece. Connectivity is the other critical lever to make AI factories truly scalable. As you know, in an AI data center, the interconnect is a high-speed data link, that allows thousands of GPU to, if you wish, talk to each other. It's efficiently directly drive energy consumed per bit and job completion time. It's actually a major cost and performance bottleneck. To build an AI infrastructure, you have to think on 3 dimensions: scale across, scale-up, scale-out.

Scale across allows to connect different data centers over long distances. And you can actually make the statement that this has been opticalized, if I can use such verb. Scale-out is actually becoming fully opticalized, and it's actually all about connecting thousands of those racks together to work as one giant cluster. The next big battle which is actually an order of magnitude more connection is scale-up, and it's all about making a single rack more powerful by packing more GPUs.

Today, scale-up is all about copper, but scale-up is actually we will talk about that migrating step-by-step to optical. All of these three directions, scale across, scale-out, and scale-up, the trend is the same. It's all about using light of fiber instead of copper to move massive amount of data in a very power-efficient manner and between AI server. This is where cloud optical interconnect and our Scale-X approach, as you see, the title become a key enabler for scalable AI infrastructure.

Now let me show you what are the needed semiconductor pieces to address these trends because it is not a coincidence that actually we are becoming a very large player in this market. On both ends, maybe I should start with a recap on the way these things work. At both end of each and every fiber in data center, there is a transceiver that converts a signal on light into an electrical signal and vice versa. And most of the time is a pluggable object, which one would plug and onto switches or a server, allowing to build a flexible interconnect network. On the left-hand side, you can see that actually such transceiver is made of three important semiconductor components.

To be truthful, I conveniently removed one, which is actually the DSP in charge of modulating the signal and other functions, but we do not actually produce such product. This is actually the property what Broadcom or Marvell essentially are doing. We have actually focused on the pieces, which are at the core of this optical engine and that actually where we're very engaged on. First is an MCU that controls the transceiver operation. Second is EIC, stands for electronic IC that is essentially driving the optical source in that case, a laser, but it is also actually amplifying signals in reception RX mode. And last is a photonics IC, often called a PIC, during the actual conversion of light to electronic and vice versa.

If you actually zoom in on its NPO and CPO devices or construct, if you wish, NPO standing for near package optics and CPO standing for co-packaged optics, and it's all about devices or constructs that do bring the fiber directly to be packaged together with the GPU or your switch unit. It's a matter of fact for NPO and CPO. The exact same building blocks are being used, again, albeit in a different configuration and with different performance targets.

So now, you can understand where we come from. ST provide to this system, all the related needed silicon technology, Silicon Photonics for PIC, BiCMOS from EIC, and I will tell you why BiCMOS is, by far, the best technology and tailored STM32 MCUs.

Now let me show you more details on the ST technology behind that in the next slide. Silicon Photonics for PIC is designed to support 200 gigabit per second per lane and to scale to 800 gigabit per second and 1.6 terabyte per second optical interconnect. This is actually a PIC100 process, and this platform is manufactured in our Crolles 300-millimeter facility. You may wonder what's a big deal with your 300-millimeter. Using 300-millimeter and in that case for PIC100, a 14-nanometer class lithography give you 3 big huge advantages. First, better critical dimension control, which is actually the smallest element in your silicon design and a uniformity that 200-millimeter wafers cannot touch.

Second, is a much higher yield and more predictable device performance, a key to CMOS digital yield levels. And last, and in this world of hyper demand 300-millimeter give you more dies per wafer, supporting the volume required by the hyperscalers and lowering cost per function.

The second part of the shuttle is BiCMOS. So why BiCMOS, and you will have to trust me on this, and we can go into details, is just actually the figure of EIC, which is actually the frequency MAX of BiCMOS process is critical in an electronic IC device for high-throughput transceiver in the context of data centers and AI infrastructure.

And here, we also have, by far, also on a 300-millimeter basis, we do have the best process on the planet. And last but not least, you need control. You need to be able to control this transceiver, and this is actually done with the microcontroller. You could ask me why a microcontroller, it's just to happen that actually the NVM provides super-fast or superior-latency in terms of loop closure and actually is favored as such, microcontrollers to be actually the engine control for pluggable or NPO or CPO.

And here, obviously, the STM32 MCU, the world-leading MCU platform provide actually all the benefits to this market as well. So now you understand that we provide the possibility to integrate all of those things in advanced packaging technology so that actually can create an optical engine form. From process technologies to PIC, EIC, MCUs, to an optical engine, assembled by us end-to-end. As a matter of fact, I will challenge you to find another company able to do that today. So it's actually this combination of products and technology that give us a strong, scalable position across pluggable, NPO and CPO solutions.

So you may have seen this morning that we've announced something, and what we announced this morning was twofold. First, we have announced that we are entering high-volume production for PIC100 in 300 millimeter for leading hyperscalers. Again, it's a combination of our technology platform, which is very unique with a backside implementation that is providing best-in-class performance for silicon nitride waveguide and build on a superior scale of our 300-millimeter manufacturing line. The combination of both give us a unique competitive advantage to support the AI infrastructure Super Cyber.

We have planned capacity expansion that will more than quadruple our output by '27 versus today's basis, and we have further expansion plan in '28. This acceleration of capacity is not done in a vacuum. It's fully backed by long-term capacity reservation commitments from customers.

The other thing that we have announced as well, which is worth noticing is that actually we are preparing the next step of the roadmap of our PIC100 process, which we call PIC100 TSV, through-silicon via. It's actually going to allow future generation of near-packaged and co-packaged optics solution that requires this level of integration to bring optics closer to the compute for AI, not scale across, not scale-up, but for AI scale-up.

It would be difficult to not talk in this presentation today about something that we have announced a few weeks ago. We have recently expanded strategic collaboration with AWS through a multiyear, multibillion dollar commercial engagement, serving several product category. It's a major milestone to position us in the AI revolution. And this collaboration will establish ST as a strategic suppliers of advanced semiconductor technology and products that AWS will integrate in its compute infrastructure, enabling them, obviously, to provide better high-performance compute instances, reduced operational cost and obviously, the ability to scale compute-intensive workloads more effectively.

ST will supply among other things, specialized capabilities across high-bandwidth connectivity, advanced microcontrollers for intelligent infrastructure management as well as analog and power ICs that deliver the energy efficiency required for hyperscalers, data center operations. It's because of this increasing demand in the AI data center. Our ability to provide the right products and technology portfolio. And last but not least, multiple deals we have closed over the past few months or are about to close that we've recently increased our revenue expectations, like Jean-Marc explained last week in San Francisco.

In data centers, including cloud optical interconnect and power and analog for AI server with all those dynamics, we now believe we can be nicely above $500 million of revenue in 2026 and already well above $1 billion in 2027.

A few words to conclude. ST optical technology are critical for AI infrastructure for pluggable transceiver, which amount for most of the market today, but also for the future fast-ramping near-packaged optics opportunities. Second, ST has also the power and a lot of technology, wide bandgap and silicon-based, galvanic isolation and smart power to address the 800-volt inflection point that is actually currently ongoing.

We are, as usual, supporting collaboration with a broad ecosystem, including research labs, ODM, module vendors and as you've seen, the largest of all the hyperscalers.

I repeat with the current market dynamics, we believe we can now achieve nicely above $500 million of revenue in '26 and well above $1 billion in '27.

Thank you very much, and we are now ready to answer your questions.

[Operator Instructions] The first question comes from the line of Joshua Buchalter from TD Cowen.

I appreciate you guys hosting the presentation. I was hoping maybe you could provide some more granularity on the composition of the $500 million of business this year and greater than $1 billion next year. Any details you can give us on the split between Silicon Photonics, and power, and then even within power across high-voltage -- high medium voltage versus Stage 2?

Not really. In that case, sorry, we were, as we just announced this morning, we are ramping in production on Silicon Photonics, which will start to go and ramp in revenue this year. We are in volume shipment for power devices in the AC/DC conversion and in the battery backup units. The 800-volt DC architecture is pretty new, and it's not involved in volume production yet. As you know, we've built reference design and proof-of-concept board that we launched last year in October, together with our colleagues at NVIDIA.

For local conversion, we are in production with key customers on the 50-volt to 12-volt. While we have validated samples for the part of the 12-volt to portions, it's not been yet put into production. So that give you a bit of lay of the land, but we have decided actually not to break it down at this stage.

Okay. Understood. And then if I could follow up, maybe you could speak to which sockets you think are most applicable to silicon carbide and GaN in particular in the medium and high voltage, how meaningful can that be? And how important is having compound semis to winning business across the power tree?

That's a great question. Clearly, if you look at the -- this new architecture from the high-voltage grid to 800 volt DC, you will see actually a fast adoption of SSTs, solid-state transformers, where actually silicon carbide will be critical. If you look at the 800 volt to 54 DC gallium nitride, I think, will play a critical role as a wide bandgap. GaN has unique properties that lead to a low output capacitance and lower on state resistance, which make it actually an excellent material when dealing with such a high frequency operation.

The moment actually you land in the 12 volt down to 0.8, I think that you are in the land of silicon at this stage, most of the time, if not all the time, and we are moving away from wide bandgap material. We have not broken down the $230 per gigawatt of data centers. You know this slide that I've shown earlier, it's on purpose. We are actually looking at it holistically, and we have decided at this stage to not break it down.

Next question comes from the line of Sandeep Deshpande from JPMorgan.

My question is on -- first on the power side. There have been other players in the power market for the data center for many years before ST. What is the technology that ST is offering that differentiates itself and thus helps it to break into more market share in the power market today? And why, because ST was not a player in the power market before these recent announcements in any significant way?

Yes. Thank you, Sandeep. To this first part of the question, I -- you're absolutely right. I think that today, if you look at the 54 volt architecture, we have been actually marginal as the players. But this transition that is happening is giving us actually an opportunity to come in and if you allow me, I will leave it what I said earlier, because actually have come. I may not have come fully clear. First of -- the -- it's all about the amount of power density, you can pack in cubic -- in volume, if you prefer. And in that case, I believe, actually, we have something to put on the table. First, there is a hot swap protection circuit that is going to be critical as part of the 800-volt to 54-volt solution. And here, of 12-volt silicon carbide devices and on BCD, bipolar, CMOS, DMOS controllers with galvanic isolation are honestly the perfect fit for the technology. And we do believe actually we have a differentiation here, which has been proven to be the case.

On the power converter, there are two elements. There is primary side, there's secondary side. So the primary side is all of what we do on gallium nitride that I've explained earlier. On the secondary side, there is a lower voltage gallium nitride transistor and lower voltage gate drivers that we master and actually our STM32 G4 microcontrollers that we can actually package together to provide actually the appropriate solution for the secondary side.

We -- those two stages what I've explained, the hot swap part and the power conversion part backed up by our silicon carbide and gallium nitride and MCU technology, honestly, we are quite competitive. It's just so happened that before they were not really needed because actually, we are getting into the server rack at 54 volt, right? Because actually those racks were consuming far less power. But this opportunity that is actually popping up is creating an inflection point that -- like I said, I insist on the word inflection point that give us another opportunity to play in and to compete. I think...

And in that -- sorry, and just one quick follow-up. On this $230 million market that you're talking about in 1 gigawatt in the data center. How much of the SAM is ST actually addressing today in terms of market share, based on your AWS win and any other wins you've already had?

We have not documented that. We have focused on our SAM and -- sorry, you were asking me -- let me rephrase because I may have misunderstood your question, Sandeep. Are you asking me what is the SAM of the 1-megawatt rack?

No, I'm asking you in that $230 million of that SAM. Your SAM is $230 million in the 1 gigawatt rack, right? That is what you are saying?

Yes. Go ahead.

And in that $230 million, how much are you already addressing? And because there's some share you have and some share you still have to take, right?

No, I understand actually. You could assume actually by this year in terms of product availability. By this year, we can cover the entire $230 million in terms of product availability.

The next question comes from the line of Adithya Metuku from HSBC.

Can you hear me?

Yes.

Just two questions, please. Firstly, just a clarification. When you have these AICs and PICs, do they tend to come from the same vendor in a transceiver or in a CPO. And also, could you talk a bit about what market share you intend to have, maybe with a medium-term view or a long-term view in PICs with your technology and with PICs as well with [indiscernible] you have?

Okay. That's a very good question. There is technically -- technologically, no forcing function that forces you to use the same vendor for the microcontrollers, the EIC and the PIC. What we see more often assuming no supply constraints whatsoever is more a best-of-breed approach. And so you have to compete towards the best microcontrollers, the best photonics ICs and the best electronic ICs, I want to make that point first and foremost.

Obviously, that dynamic is a bit changing for the next 2 to 3 years because for those technologies, we can be guaranteed that actually there will be a lot of supply challenges in the industry. And obviously, that gives us, in exchange, an ability to provide to our customers a better service when we are controlling actually the three pieces, if I can put it this way. In terms of market share for PIC, which was your other question, and deniably, we want to become the market leader. So I think market leadership started 30%.

Got it. And in the EICs?

Actually, on the EIC, this mission has been accomplished. We are the market leader today.

And what share do you have, if I may ask?

I still have the same definition, I just gave you.

The next question comes from the line of Gianmarco Bonacina from Banca Akros.

Just a follow-up on the addressable market. I understand you gave some per gigawatt, but what's your expectation on the total addressable market in billions, if you can give a range on this for 40,000 in '27 and also for outer years? And the second question is given that the bulk of the growth would come beyond 2027, can you also share if you expect to grow faster than the addressable market, which I think in the press release this morning, you indicated some mid-teens beyond 2027?

I'm going to need you to reask your first question to make sure I got it, because it's a spin of Sandeep's, and I have failed in the first time, so I don't want to fail the second time. Could you please reask it again?

No, I just wanted to -- yes, a follow-up on the SAM, not per gigawatt, but in total billions. So what's your internal expectation for the SAM in billion dollars in total for ST in 2027 and maybe 2030?

Now it's a great question. Actually, to tell you the truth, it's on purpose we've not done it. And I have to tell you, it's, I'm going to give you some assignment, which is actually to go and modelize by yourself, the amount of gigawatt of AI data centers being deployed because the reason why we did it like this in all transparency with you is that actually, it is changing constantly. And as such, actually, the dollar value also is constantly changing.

So we have decided that actually it was better to normalize by gigawatt, and let's you modelize your way based on your own assumption because I have to admit that actually it is even of sale when we look at our forecast right now on an every 3-month basis, they changed so materially that I feel that it's -- we may be underestimating the total amount of gigawatts of data centers being deployed as we speak. So forgive me for not directly replying, but at least, I wanted you to understand my rationale for why we did it that way, okay?

Okay.

And you had a second question...

In terms of the growth beyond 2027, given that it seems that you are still, let's say, scaling up your product line, that is it fair to assume that beyond 2027, you will still grow, let's say, nonlinearly above the expected market growth because you are still say, probably not at your normal market share in 2027 or not, maybe, but, yes.

No, I think it's a fair assumption, especially. It's actually a fair assumption on multiple fonts. In Photonics IC because actually, we are exactly doing exactly what you said in terms of catching up to where we want to be from a market share standpoint, but we have also -- I must admit a lot of near package optics engagements, which are going to be such an accelerant for us beyond '27 that we are counting on. And also, when I look at my colleague, Macro with what he's doing on power and analog, I think actually Marco is spending a lot of time on this new 800-volt architecture. And I expect that 800-volt architecture to start to ramp in '27, and that's such that we should also see the benefits of that beyond '27. So your point is correct for those two reasons, I would say.

The next question comes from the line of Amelia Banks from Bank of America.

I was wondering if you could provide some more granularity around how concentrated the long-term capacity reservations are. Are they with a few hyperscalers? Or is it more diversified than that?

It's actually more diversified thing, Amelia. And I think you were asking this in the context of what we've announced on Silicon Photonics. The value chain is a very interesting value chain because actually, you have chip-level actors, you have module level actors, and you have hyperscalers. And we have actually, with those capacity reservation agreements, we are scanning the entire value chain.

Okay. Great, brilliant. Yes. And then just on a follow-up. Just in terms of the -- again, with the Silicon Photonic announcement, you're stating that the capacity will more than quadruple by 2027. I was just wondering if you could clarify how much of that is relating to sort of front-end versus back-end manufacturing.

Today, these announcement is 100% front-end. I may add one on top of that Amelia to go and help you this because it's something that I did not mention in my presentation, but it's something very important for you to understand. As you know, or may not know, the way we build our Crolles 300 fab is actually by gateways, which means that we can add tranches of capacity as we go. That's a unique advantage. Why? Not only because of the modularity because when we add capacity, our customers do not need to requalify. It's actually copy exact of the previous gateway.

So we can add capacity without adding to in -- without adding our customers to requalify. I want to insist that actually this is a value proposition today on photonics. I believe that only ST is able to provide, which is not the case of competitors of ours. And that makes actually the life of our customers way easier. And one of the reason why beyond the 300 millimeter asset, one of the reason also they have a huge interest in what we're doing. I'm closing the parenthesis, Amelia, but I wanted you to have this information as well.

The next question comes from the line of Stephane Houri from ODDO BHF.

Actually, I wanted to have a bit more visibility on the CapEx that has to be involved for this kind of growth. So you've been talking a little bit about Crolles 300, but looking at your CapEx budget, if the growth continues, as you expect, what kind of addition to the CapEx you have to do, and is only Crolles 300 involved in this CapEx expansion?

Good question, Stephane. I would walk backwards. And right now, yes, only Crolles 300 is actually the location where we have actually concentrated our Silicon Photonics activity for one of the reasons that I've described before because of this ability to go and systematically expand without needing our customers to get requalified. Can it be the only answer? Could actually we see eventually Silicon Photonics in Agrate 12 inch. Absolutely, yes. There is nothing preventing us from doing so.

In terms of polarizing the CapEx, we will not do it, but what I can tell you, though, to help and go in our direction is that actually to support the growth that we are seeing now. We came in October 2025 with a CapEx plan with, I believe, what is $2 billion to $2.2 billion of CapEx in '26. And so on and so forth for '27 and '28, which we have not shared. What we are obliged to do at this stage is, and we are now in March '26, what we are obliged to do at this stage is to accelerate part of our CapEx sticking toward $2 billion to $2.2 billion envelope, but remixing it because we need to go and accelerate Silicon Photonics. That is something that we are "because it's for a good reason, for us to do." But I have no absolute CapEx number dollar to share with you.

Okay. And if I can have a quick follow-up. Actually, when you speak with people like Infineon, they described for the Power AI business of a market of $10 billion by 2030, and they think they can have a market share of at least 30% to 40%. Your AI opportunity is, let's say, more diversified with Silicon Photonics, Power AI et cetera. Can you maybe give us some visibility on by market? What kind of size do you see, and what kind of market share?

This is what Gianmarco was also asking me earlier. I've been shying away from throwing those numbers because what happened, we share numbers, then actually, there are actually modernization being done, and then actually you go in the market share, the market share can seem really high. I think this is actually what happened to our colleagues in Infineon recently. And actually, you go and question whether or not those numbers are stable.

But actually, they may be stable or they may not be stable because this market is evolving at such a pace. So we've shied away from that. We'll let you go and make your own model in terms of overall AI data centers deployment, but we provide you this $230 million, all inclusive of what we do on power and cloud-optical interconnect, which give us a proxy, but right now, that's what we feel the most comfortable to share with together with, obviously, the numbers that we have shared in terms of being nicely about $500 million of revenue in '26, and well above $1 billion in '27, all of that being backed up by the deals that you know or the deal with analysts that you know and many more deals that you don't know yet.

[Operator Instructions] We have a follow-up question from Adithya Metuku from HSBC.

Yes. Just two more questions. Firstly, Remi, I just wondered if you could talk a bit about the scale-up opportunity. Some of your peers in the laser market have talked about that being a much bigger opportunity like by a factor of magnitude. So I just wondered if you could give us some sense of how much your SAM would expand from this $250 million or $230 million you've talked about, if you include the scale-up opportunity and when could this ramp happen? Is it a 2027 story? Is it a 2028 story? You have your peer Broadcom saying one thing, Credo saying one thing, NVIDIA saying another thing. So what is your view here? And I've got a follow-up.

Yes. I -- it did not escape me that there is a lot of spin right now on copper versus non-copper in terms of what is going to be the winning recipe. I will tell you the truth, we are everywhere in some shape or form. Obviously, the dollar content for us when it comes to Silicon Photonics is way higher. But I will try to remain objective.

I think that the depth of copper from a scale-up standpoint is only a question of when, not a question of if. It's due to gazillions of parameters, but think of poor consumption in terms of -- poor consumption per bit and do think as well in terms of the ability to go and pack way more compute density in a rack.

So I think it's a question of when, not a question of if. And you will see that similarly to EML for connectivity, it will be -- people will try to push and move to 448 gigabit per second service. Actually, some are doing that, but it's only bidirectional. There would be a lot of tricks to try to survive, but it's going to happen.

In terms of opportunity size, it's pretty simple. You multiply by 2. That's more or less what's going. You take pluggable, you multiply by 2 that actually give you the near packaged optics and co-packaged optics market size? And that's -- this is as big as it will become.

NPO versus CPO, Near Packages Optics versus co-packaged optics. Actually, I'm a believer for the next 5 years that the bulk of the business will be on near packaged optics just because of RAS; reliability, accessibility, serviceability, in the data center. And the fact that actually -- it gives hyperscalers way more flexibility on RAS than actually what they can do with CPO.

So NPO will be the winner for the 5 years to come. When will you see NPO was the second question. This, I will tell you very firmly that you will start to see NPO from second half '27 for AI clusters. That is going to happen. Is it going to be a hard switch 0 to 100? The answer is no, because you will see actually new architecture being introduced, both with copper technology and also with optical technology before actually they move fully to optical technology.

When could you think that a rack for scale-up will be 100% optical base? And at what point there would be no more copper. Okay, pick your points between 2029 and 2030. Let's take 2030 to be on the safe side. So starting to run similar '27, 100% coverage in 2030 is my opinion.

Got it. That's very clear. Maybe just to clarify, you said the opportunity would be 2x from -- is that twice the $230 million you gave? So if I include scale-up then basically, it should be $460 million. Is that the -- have I understood correctly?

That's a good question, but it's actually a very clever one, which I cannot answer because I didn't give you the breakdown of $230 million. But when this happened, I think the $230 million will grow, for sure, because if we look at the $230 million we gave you, it's only based on pluggable optics. So that's a good clarification question you just asked, I think. We have not included the growth of NPO. So this will come on top of the $230 million, but now I'm stuck. I cannot give you any numbers because then you can -- so I will have to find a way next time to answer this question more elegantly.

Okay. Got it. And then just secondly, on the bottlenecks and networking today or optical networking today, where are they from what you see today, where are the bottlenecks?

I think, there is a bottleneck of today, bottleneck of tomorrow. I think if you ask anybody today where are at the bottlenecks, they may tell you laser. I think that if you ask people what could be bottlenecks tomorrow, they will tell you photonics. Why, because lasers are lasers, are lasers, are lasers, they are used anywhere. So you may need more of them because it's EML, and it's one laser per lane.

You may be less of them because it's Silicon Photonics and it's 1 laser for 2 or 4 or 8 lanes, the lasers you need. But what's happening is that the transition to 1.6T is completely accelerating and 1.6T at 80% will be photonics. So now if there is an explosion in terms of gigawatt deployment at 1.6T, the pressure on photonics will be very high, which kind of explained why we're doing what we're doing and why we are thinking of doing even more, because of that very context.

Got it. Very clear. And when you say tomorrow, is that 2027 or 2028?

Yes. That ZIP code. I will not be able to give you something more accurate than what you just said.

There are no more questions at this time. I would now like to turn the conference back over to Jerome Ramel for any closing remarks.

Yes. Thank you, Maura. So I think this is ending our call. So thank you all very much for being with us. We remain at your disposal for any follow-up questions. So we look forward to asking you on March 16 for our conference call, ST Intelligent Sensing Enabling Critical AI with Marco Cassis. Have a nice day. Thanks.

Ladies and gentlemen, the conference is now over. Thank you for choosing chorus call, and thank you for participating in the conference. You may now disconnect your lines. Goodbye.

Okay. Good afternoon, everyone. Jean-Marc, welcome to San Francisco. We've got Jean-Marc Chery, he's CEO of STMicro.

Maybe just to kick off, you can talk about the momentum going into first half of the year. You gave a pretty good guide to Q1, talking about above seasonal aspect there, $3.04 billion. But as we've seen in a few weeks of this quarter already go by, are we tracking to those levels? And what other dynamics do we think we should be aware of for the first half?

I confirm that for the first half, the dynamic of booking we receive will enable us to do better than the usual seasonality in Q1. So we confirm our guidance. We have seen some distributor in Q1 that for the first year since many years, Q1 in Asia will be better than Q4, which is a pretty great sign because in Q1, you have the Chinese New Year. So to have Q1 better than Q4 is a signal that -- showing that something is happening. And in Q2, we confirm as well that to be at, let's say, seasonality, so plus 3% is at least what we can deliver. And the overall coverage to achieve this H1 that will be a significant growth compared to last year. Our backlog coverage is between 85% to 98%. So you see the confidence level today. Overall, for ST to grow significantly year-over-year and to do better than the usual seasonality is there.

Perfect. Super. So I mean if we think about now the puts and takes as we go through the rest of the year and how the phasing works, H2 on H1, maybe referencing things like mix and utilization, pricing and even the under-absorption on FX, how does that all impact margins as we go through this year?

To speak about H2, I think we have to go by segment. Well, on automotive first, for ST, the positive point is that we are now without any other inventory, so our major customer, OEM, the Tier 1, they are clean over inventory. The only headwinds we still have in 2026 is capacity reservation fee that will disappear basically over the year. So it means looking at the seasonality, we have usually on automotive, we can expect to have H2 growing. So contributing. I repeat, to the usual growth of H2 for ST, which is 15%.

Industrial market, more clearly, Industrial Market, it is not a secret that what is related to power industrial. So all the power management or the thermal flow management for AI data center and power energy is very strong and clearly, our microcontroller general purpose analog and some Power and Discrete are taking benefits of this dynamic.

What we call smart industrial, so factory automation, as an example, China is very solid. Asia Pacific is growing. Europe, we believe we touched the bottom in Q1, and we see sign of improvement. Americas, so far stable. What customers told us that for consumer industrial, inventory are clean and they expect to have H2 growing. So clearly, for ST, Industrial in 2026 will really contribute to the growth year-over-year and to the growth H2 versus H1.

Personal Electronics. Personal Electronics, we don't expect to have growth overall year because the memory shortage impact, but we will have a growth H2 versus H1, maybe a little bit more moderated compared usually because the phenomena for our main customer would change a little bit, the introduction of new device profile because last week, they introduced a new one, and they will introduce foldable that has a deferred bill material for us.

But there remain 2 interesting driver for ST over the cycle. First of all, it is low earth orbit satellite communication. But here, we receive significant upside that will enable growth for ST and that's the manufacturing of ST can support. So this will be clearly a very solid contributor of the growth -- of the goal H2 versus H1.

More last, but not the least is AI data center. So AI data center, if we make it simple, you have 3 flow, you have the network flow. So from processor to data exchange. You have the power control management flow, where you know that ST is not yet really present, but this will come. And you have the thermal flow, so all the utilities, cooling and so on where we are.

Well, clearly, the network flow, we see really, really strong acceleration of the optical cable. That we're supposed to start next year, Q2, Q3, but that will start in Q2 and strongly accelerate in H2. So this will contribute to a very solid growth of ST, H2 versus H1. So that's the reason why, okay, when at the earnings, I told more H1 seasonality better and H2 15%. This I confirm and most likely with the current dynamic we are seeing and with a booster on top of the cycle, this is what we can deliver.

Okay. So a lot to unpack there. Maybe we'll come back to some of that. Maybe let's start with data centers. So you have the strategy to be a sort of foundry supplier on silicon photonics and BCD technologies, but obviously, off to a great start with AWS partnership. Can you maybe just help us understand the timing and the potential size of the revenues coming out of that partnership?

Well, here, okay, if we look overall, the AI data center, so the 3 flow and we had also the data storage. Basically, ST today, we have 380, 390 products part of the bill of material. So we have 250 from analog and power, and we have 135 from microcontroller and RF optical device. So compare what we said end of January, so that this year, we can deliver USD 500 million. So thanks to the first effect of the AWS contract, but amplified by the acceleration of the demand on optical cable, this year, I can say we will be nicely above $1 billion -- $500 million nicely. And next year, well above $1 billion. So we have really boosting effect from the start of the optical cable and from the benefits of our overall AWS contract that I repeat is a multimillion-dollar contract for the next 5 years that will start this year and moving forward will grow.

So this is the positive news, I have to say, since January, February, so I was last week in Taiwan, of course, discussing with a company and in China, working on optical cable. You know that ST provide as a foundry silicon photonics in 12-inch, which enable our capability to go fast. We provide electronic ICs. So it's the driver of the laser, in fact. And we provide the microcontroller. And we can start to see tension on this kind of microcontroller where the lead times are increasing.

So that's the reason why because of the situation on microcontroller and the industrial market coming back to a positive trend. We have a customer to put the order to give us the visibility because if not sooner or later, okay, microcontroller could come back to some tension. And by the way, the lead time for micro now are between 10, 14 weeks up to 28 weeks.

Okay.

So this is the situation.

So you can step in and relieve some of that tension on the lead time given your position. Maybe just an addendum to that, the split between maybe playing into transceivers market, pluggables market and the CPO that is emerging. Is that an interesting split for you? Does that obviously change over the years? And how does that change?

No, we received already some demand for programs moving to CPO, but more next year, more mid- of next year. But short term, I think it's important that the supply chain adapt all the device. Well, as an example, today, a limitation factor on optical cable is laser. Then silicon photonics, sooner or later, microcontroller, DSP possibly, but laser and silicon photonics, sooner or later will be a limitation factor.

Okay. Makes sense. And maybe we jump to automotive. It sounded like in there, we are going to the end of the inventory correction. It seems to be consistent with other speakers at the conference. So maybe to your platform, how would you talk to that? Are we at the bottom in Q2, and then we ship to end demand? And maybe could we talk about a recovery or is that too soon at this point?

No, on automotive, okay. For us, now is more to understand the regional approach because okay, America, as you know -- in Americas, [ the story ]. The carmakers, they come back on more thermal combustion engine technology. And for sure, the demand on electrical powertrain is slowing down. Our main customer for electrical car so far this year will not be a decrease for us, will be an opportunity of growth. This following recent discussion we had with them. On Europe, now the visibility in Europe looks better than 1 year ago, 2 years ago. So the strategy of the main carmaker is more clear now compared to 1 year ago, whatever BMW, Daimler, okay, and Volkswagen. So for us, it will be an opportunity of growth. And the Tier 1, by the way, they have no more inventory correction. And then sometimes, okay, there is escalation of shortage on some [indiscernible] technology.

APEC. So Korea, Japan, so far, so good because they are on hybrid car model mainly, so very stable. Well, what we have to monitor in China. So by in China, the forecast is 30 million, 31 million, 20 million vehicle -- new energy vehicles. So mix, hybrid and electrical. But we know that in China, the profitability is low. So here, we have to monitor exactly what will be the dynamic in China for exportation of car and the internal market. Again, I repeat for ST mechanically boosted by the fact that we have no more inventory correction to show a growth on automotive in 2026 between mid- to high single digit will not be a surprise.

Okay. And I think when you discussed earlier, you touched on some of the shortages affecting end demand, but you referenced handsets. You also referenced some design wins with your leading customer and how that might change the phasing through the year. Can you just repeat the phasing and the impact from the design wins, but help us understand really the impact of these shortages, particularly DRAM, as indirect on you guys?

I may bore you. I guess everybody has understood when you have the DDR4 price multiplied by 5 and DDR5 by 3 for the consumer or low end or middle end device, the bill of material is no more affordable because if they adjust the price, they completely consume the profit. So clearly, and in a shortage period, people will allocate the capacity to more high-end device and so on. But sometimes you have no choice. So yes, I know there is a number about smartphone that will decrease by multiple 10%, but we found it excessive, but as far as ST is concerned, as our main exposure is with our main customer, we do believe they are covered in terms of memory supply. So again, I repeat for us what is important is to follow their plan for the year. We know that they introduced at a different time compared to last year. So we have changed our revenue forecast accordingly. And again, the foldable phone has a different build of material.

So, so far as a matter of results, we believe ST Personal Electronics in '26 will be a transition year, means a very low single-digit growth, no more impact than the one we could expect on low-end smartphone.

Okay. And when we lap that, the subsequent year, growth should be stronger for the ...

Yes, Indeed.

Got you. Maybe if we jump to industrials, I think it's one area that we left out there. It looks like inventory normalization is progressing quite well. I think you do have exposure to consumer industrial, but also, I believe, factory automation as well where orders do seem to be improving. Maybe just help us understand what is the hard evidence for this to be a strong growth year? Or what would be the level of growth for this year for industrial generally?

You know the KPI are [ classified ] in inventory turn and distribution. So for us, in our model and distributor is between to have 2, 3 months of inventory where in China, we are below, well below. In Asia Pac, we are in the target or below the target. So this is the sign #1. The sign #2 is a POS increase. So POS is increasing. We still control the POP clearly to finish, okay, to adjust the pocket of inventory. We are here and there, mainly general purpose analog and some discrete.

But the positive sign is the POS. And again, I repeat what I said a few minutes ago, main distributor in Asia will see Q1 '26 better than Q4, including Chinese New Year and it's not because they have an effect before Chinese New Year of inventory buildup, it's because overall, this is something positive.

By application, I repeat Power and Discrete is going very fast, driven by AI, but also energy, supply and commercial and transportation. Smart Industrial in China are accelerating APAC as well. Europe is starting. Qualitatively, between distributor in Asia and some customers in America, they saw that H2 consumer industrial could increase again. In H1, inventory are clean. So far for the time being is soft, but they expect really something growing in H2.

That's the reason why we told to customers, please put orders, they commit that they will do it by April, May in order, okay, we avoid, okay, this panic that could happen because today, we have shortage in memory, we have shortage in passive. We start to have shortage in laser. So if we have this contamination of shortage, well after we have to pay attention to not go back to double booking and so on and so forth. So we have asked customers to say, please anticipate because we need to drive our manufacturing properly.

So slightly undersized channel, reinflation from an order book that has demand in the second half, but we're being mindful of double ordering at this point.

Yes, of course, we will monitor it carefully, taken the lesson of '22 and '21.

Makes sense. Maybe just staying with power then. Silicon carbide, I think you've called this out as being a decent year coming back to '24 levels. Specifically, it looks as though this is both EVs and in industrial. So maybe help us understand how are the drivers working as we started this year.

No, Industrial now with Tesla stop decreasing and the other program starting in Europe and in China, the priority for us is first, okay, to close our 6-inch ramp. Of course, in respect with the customer qualification timing and constraints, but it is a priority number one, okay, to go to a cost-effective 8-inch fab in China and in Catania or than to continue to introduce our upside in terms of technology. So now we start to give sample on the generation 5 and we prepare a generation 6 that will be a disruption, clearly, compared to the generation 5, even if we receive good feedback on generation 5.

And again, this is a monetary condition for us to go back to profitability breakeven in 2027 on Power and Discrete and then to contribute to the operating margin starting 2028.

Got you. Okay. Very clear. Maybe if we could touch on pricing dynamics. We -- it looks as though we could be getting to the end of that price to clear phase. Just maybe if you can get some evidence of that happening? And what are the sort of types of price erosion you're seeing maybe in general purpose designs, analog power and maybe discrete generally?

No. Now again, on pricing, we are going back to normal, what we call the low single-digit price decrease. Well, as usual, in Q1, we discussed with the car industry, the price, but we didn't see this year something specifically strong. As usual, in China, the price pressure is higher. I do not exclude that sooner or later, very selectively, we could be in position to increase price on some products specific -- driven by some specific application, but avoiding to increase price across the board like memory, but specifically on application, lead time increasing, we could increase price selectively.

Got you. Maybe just turning to investments. I mean, you called out a CapEx over $2 billion this year, $2 billion to $2.2 billion, if I remember right. I just want to understand how does that spend split maybe across the $300 million capacity build, silicon carbide, advanced packaging and now photonics as well. So help us understand how do you prioritize that spend.

Clearly, for the next -- last year, okay, I repeat, we have done well below 2, at 1.8. As a whole, we want to stick with the CapEx equal or below the depreciation we have at company level. And this is enabling the company to go back to $18 billion revenue. Clearly, since a few weeks, we are seeing an acceleration of this optical cable. So where we have some specific equipment that we have to have on top of our capacity in [indiscernible] mainly we need to invest and pull in because the demand is permanently increasing for '27 and '28, but in significant magnitude, I have to say.

So that's the reason why we say we can go $2 billion to $2.2 billion. And I repeat, we have the capability of microcontroller to use more foundry, we have 2 partners on microcontroller, it is TSMC and Samsung. On TSMC, they say that no, they will stop to invest on 14-nanometer or this kind of stuff. But here, we are covered.

But with Samsung, okay, I do not exclude that we will increase our microcontroller activity in Samsung, keeping place, okay, in [indiscernible] for silicon photonics, by CMOS 55-nanometer to support this fastest increasing demand on optical cable.

Okay. Maybe just -- I'll touch on cost restructuring and then I'll open it up to the floor. Can you just maybe give us an update on where we are with restructuring, what sort of annualized savings should we expect? You've already engaged in a headcount reduction program? How is that going? And what sort of onetime costs should we expect this year and next?

You know there is 2 parties that is what is related COGS. Well, this will not be too much -- visible, sorry, in '26 because we are dependent to the qualification time of customer, and it's really complex. It's thousands of product transferring, okay, between the various 8-inch to 12-inch or 6-inch to 8-inch. So the contribution of this cost will be more visible exiting '27, where basically, it's 400, 500 basis point contribution to the gross margin.

Expenses, we have already, let's say, in '25 and in '26 expense cost saving, so well above $100 million. But we have explained that on the expenses in '26, we have some specific headwinds. One was linked to the foreign exchange to dollar. But recently, we have seen is going down more, but I don't want to make a forecast on the dollar, I am not crazy. So we will see. But we have some inflation on the wage after 2 years where we contained a [ lot it ].

The nonrecurring additional costs we have is the cost related to the transfer of product and technology. But it is not recurring. It will cost an additional expenses for us. So means in '26, we will have a very few single-digit percent of OpEx increase, but we have the cost decrease linked to our program well above, it is 120, 130. And this will come again in '27.

So that offsets the inflation effectively.

Yes.

Got you. I did say it open up the floor, so I'll follow through on that. All right, not just now. Okay. So maybe I'll go back to -- we've seen some M&A from yourselves. You've recently acquired the NXP MEMS business. Just want to maybe understand how does that fit in as far as synergies, revenue, cost synergies and whether or not there is any dilution coming this year and next?

No. First of all, important to know that, okay, we were -- we cooperated with NXP as a foundry. So from a manufacturing point of view, the acquisition of NXP MEMS is zero risk, okay? Because we -- on the mechanical part, I have spoken. On the companion chips on the MEMS, we have an agreement of supply with them for the time we need, okay, to develop companion chips in our own factory. And as far as revenue is concerned, we have integrated and the backlog we have, okay, is consistent with this expectation. So it's a good deal for ST, it is accretive in terms of gross margin, will boost our revenue. From technology standpoint, there is really good technology block. And thanks to this acquisition, our MEMS will come back to the $1 billion club. We have capacity limitation, I have to say now. And you know that MEMS for the future what we call the intelligent sensing will open many opportunities linked to the physical artificial intelligence because sensor will be really instrumental.

So I am really optimistic mid, long term with this acquisition. Now we have everything to address this new path on top of automotive of this physical artificial intelligence. So what we call human robotics or robotics, where sensor will be instrumental including sensor with some AI on board. So this is where we are. So overall positive. A year of growth, let's see the order coming, confirming and ST is boosted by optical cable. Agreement of AWS will open many other opportunities, including the power management and multimillion-dollar contract. Well, let's execute them.

Yes. Sounds great. I guess if I take a step back, we've seen good wins with sort of tentpole customers in different areas. So you've got Starlink in LEO. And you said, I think earlier, that's going very well. Obviously, you do very well with the iOS guys. Clearly, with Tesla as well. and it's -- you're a sort of strategic or technology strategic partner for a lot of these players. Where does this take you as a strategy over the next 5 years? I mean, where else can you go? And I guess I'm referencing the decision to be a BCD foundry here, doubling down on $300 million in Crolles. Should we always expect these tentpole customers and then maybe a fill out of the fat tail afterwards as a strategy or is there something else that we should be thinking?

No, I think about ST has demonstrated this capability to work in close cooperation-ship with this big customer. Maybe we can test different model with this big customer is to create ecosystem. So the big customer will enable us as a referral design for our product. We will put our product in their simulator. And by the way, they will deploy business, okay, making simulation and on robotics as an example, where we have no data, this is something that could happen in the future.

So this is certainly a business model we have to develop, to be sure that we develop the right product in advance and the right go-to-market to capture the full essence of the application and not waiting that the customer drive us, okay? So it's point number one.

The second point, I think ST is good performing to develop general purpose products like our microcontroller and so on and so forth. Where we have to make progress I say we are not bad, but we have to make opportunity of improvement is to better anticipate the application, the KPI, what matters for the application and to be sure that our product will enable KPI of the application performance. So it is what we call application specifics on our products.

But this is where -- okay, I want to push the company in the near months, and that's the reason why we will set up some system engineering approach not spread across the various group, but more concentrated in comparison with our competent center worldwide and our marketing to better anticipate and this is certainly a strategic axis. But to work closely with customers and having a reputation on manufacturing to be very solid and don't put the customer light on is something, of course, we'll continue to leverage.

So quite interesting. You see the initial big customer play is a pipe-clean moment for you to get to ASSPs for that relevant market. And then you touched on robotics there as a plan. And I think we have heard you say that there's a large customer in the U.S. and one in China as well. Is that still the state of play in robotics for you?

Again, the human robotics, you will have a different way, okay? You will have go-to-market with some small player distribution. You will have big player integrated that will drive us and then you will have in the middle, some people that will provide ecosystem, simulation tool, test tool in order for customers to develop their ecosystem. It doesn't mean the guy will make the robot everything, but it will be capable to build a robot and then subcontract to different Tier 1. But it is a model of electrical car. So pretty integrated, but with some Tier 1 on some specific subject, software defined, okay? And design in advance using digital twins. Here, the components provider must been introduced in advance in order to provide the simulation of all the components to be sure that you are a full match with the need.

Clock's running against us, but maybe we got time for one question from the audience? All right. In that case, Jean-Marc, thanks very much.

Thank you.

Thank you.

Ladies and gentlemen, welcome to the STMicroelectronics Full Year 2025 Earnings Release Conference Call and Live Webcast. I am Sandra, the Chorus Call operator. [Operator Instructions] The conference is being recorded. [Operator Instructions] The conference must not be recorded for publication or broadcast.

At this time, it is my pleasure to hand over to Jerome Ramel, EVP, Corporate Development and Integrated External Communications. Please go ahead, sir.

Thank you, Maura, and thank you, everyone, for joining our fourth quarter and full year 2025 financial results call. Hosting the call today is Jean-Marc Chery, ST President and Chief Executive Officer. Joining Jean-Marc on the call today are Lorenzo Grandi, President and CFO; and Marco Cassis, President, Analog, Power and Discrete, MEMS and Sensor Group and Head of ST Microelectronics Strategy, System Research and Application and Innovation Office.

This live webcast and presentation materials can be accessed on ST Investor Relations website. A replay will be available shortly after the conclusion of this call.

This call will include forward-looking statements that involve risk factors that could cause ST results to differ materially from management's expectations and plans. We encourage you to review the safe harbor statement contained in the press release that was issued with the results this morning and also in ST's most recent regulatory filings for a full description of these risk factors. Also to ensure all participants have an opportunity to ask questions during the Q&A session, please limit yourself to one question and a brief follow-up.

Now I'd like to turn the call over to Jean-Marc Chery, ST President and CEO.

Thank you, Jerome. Good morning, everyone, and thank you for joining ST for our Q4 and full year 2025 earnings conference call.

I will start with an overview of the fourth quarter and the full year 2025, including business dynamics, and I will hand over to Lorenzo for the detailed financial overview. I will then comment on the outlook and conclude before answering your questions.

So, starting with Q4. We delivered revenues at $3.33 billion, above the midpoint of our business outlook range, driven by higher revenues in Personal Electronics and to a lesser extent in Communication Equipment and Computer Peripheral and Industrial, while Automotive was below expectations. Gross margin of 35.2% was also above the midpoint of our business outlook range, mainly due to better product mix. Excluding impairment, restructuring charges and other related phaseout costs, diluted earnings per share was $0.11, including certain negative one-time tax expenses impact of $0.18 per share.

Q4 revenue marked the return to year-over-year growth. During the quarter, we further worked down inventories, both in our balance sheet and in distribution, and we generated a positive $257 million free cash flow. Looking at the full year 2025. Net revenues decreased 11.1% to $11.8 billion, mainly driven by a strong decrease in Automotive and to a lesser extent, in Industrial, while Personal Electronics and Communication Equipment and Computer Peripheral both grew. Gross margin was 33.9%, down from 39.3% in full year 2024. Excluding impairment, restructuring charges and other related phaseout costs, diluted earnings per share was $0.53. We invested $1.79 billion in net CapEx, while generating free cash flow of $265 million.

Let's now discuss our business dynamics during Q4. In Automotive, during the quarter, we grew revenues 3% sequentially. Year-over-year revenues declined, but with continued improvement in the trend. Automotive design momentum progressed with design wins across both electric and traditional vehicle domains for applications such as onboard chargers, DC-DC converters, powertrain and vehicle control electronics. These included design wins for power semiconductors, smart power devices, automotive microcontrollers, analog and sensors. These awards supported by engagements with various OEMs and Tier 1 ecosystems, strengthen our position as a key supplier to the automotive industry.

Regarding the acquisition of NXP's MEMS sensor business, the transaction we announced in July is still expected to close in H1 2026.

In Industrial, revenues were better than expected, showing increases of 5% sequentially and 5% year-over-year. Importantly, inventories in distribution further decreased and are now normalizing. In Industrial, our portfolio of microcontrollers, sensing technologies and analog and power devices is strongly positioned to support industrial transformation trends and the need of physical AI.

During the quarter, we saw design wins across industrial automation and robotics, building automation, power systems, health care and home appliances. In November, we held our STM32 Summit where we announced several key innovations, including the first microcontroller built on the 18-nanometer process, a next-generation wireless microcontrollers and an updated suite of edge AI software tools.

Personal Electronics, fourth quarter revenues were above our expectations, down 2% sequentially, reflecting the seasonality of our engaged customer programs. During the quarter, we strengthened our position in mobile platform and connected consumer devices, both with our engaged customer programs as well as our open market offering for devices such as our sensors, secure solutions and power management products.

Revenues for communication equipment and computer peripherals were up 23% sequentially, better than expected. In AI and data center infrastructure, we continue to reinforce our position supporting the increasing demand for higher power density and energy efficiency. During the quarter, we secured multiple design wins for silicon and silicon carbide-based power solutions, supporting next-generation AI compute architectures.

We also continue to work with customers to bring our silicon photonics technology to the market. The strong momentum in optical connectivity technologies for data centers also contributed to a significant rise in demand for our high-performance microcontroller used in pluggable optics.

The low-earth orbit satellite business based on our BiCMOS and panel level packaging technologies continued to progress during the quarter with shipments ramping to our second largest customer.

Moving to sustainability. We remain on track for our key 2027 commitments. Carbon neutrality in all direct and indirect emissions from Scope 1 and 2 and focusing on product transportation, business travel and employee commuting emissions for Scope 3 and 100% renewable energy sourcing. A major milestone this year was the launch of Singapore's largest industrial district cooling system at our Ang Mo Kio facilities in Q4. We also continue to maintain our strong presence in the major sustainability indices where we were honored to be recognized in the Time world's most sustainable companies list for the second consecutive year.

Now over to Lorenzo, who will present our key financial figures.

Thank you, Jean-Marc, and good morning, everyone. Let's have a detailed review of the fourth quarter. Starting with revenues on a year-over-year basis by reportable segment. Analog products, MEMS and sensor grew 7.5%, mainly due to Imaging. Power and Discrete products decreased by 31.6%. Embedded Processing revenues were up 1% to 2% with higher revenues in general purpose and automotive microcontrollers, offsetting declines in connected security and custom processing products. RF and optical communication grew 22.9%.

By end market, communication equipment and computer peripheral and personal electronics both grew by about 17%. Industrial grew by about 5%, while automotive decreased by about 15%. Year-over-year, sales increased 0.6% to OEM and decreased 0.7% to distribution. On a sequential basis, Power and Discrete was the only segment to decrease by 3.9%. All the other segments grew, led by RF and optical communication up 30.5%, while Embedded Processing and Analog products, MEMS and sensor were up, respectively, 3.9% and 1.1%.

By end market, sequential growth was led by communication equipment and computer peripherals, up 23%. Industrial was up 5% and automotive was up 3%, while Personal electronics declined 2%.

Turning now to profitability. Gross profit in the fourth quarter was $1.17 billion, decreasing 6.5% on a year-over-year basis. Gross margin was 35.2%, decreasing 250 basis points year-over-year, mainly due to lower manufacturing efficiencies and to a lesser extent, negative currency effect and lower level of capacity reservation fees. On a sequential basis, gross margin improved by 200 basis points. Q4 gross margin included about 50 basis points of negative impact resulting from a nonrecurring cost related to our manufacturing reshipping program. In the next few quarters, we expect a similar negative impact on gross margin from the just mentioned nonrecurring costs.

Total net operating expenses, excluding restructuring, amounted to $906 million in the fourth quarter, slightly increasing year-over-year due to unfavorable currency effect. They were slightly better than expected, reflecting our continued cost discipline and the initial benefit from our cost savings initiative. For the first quarter 2026, we expect net OpEx to stand at about $860 million, decreasing quarter-on-quarter. As a reminder, these amounts are net of other income and expenses and exclude the restructuring.

In the fourth quarter, we reported $125 million operating income, which included $141 million for impairment, restructuring charges and other related phaseout costs. These charges are related to the execution of the previously announced company-wide program to reshape our manufacturing footprint and resize our global cost base. Excluding the nonrecurring items, Q4 non-U.S. GAAP operating margin was 8%, with Analog product MEMS and Sensor at 16.2%, Power and Discrete negative 30.2% Embedded Processing at 19.2% and RF and Optical Communication at 23.4%.

Fourth quarter 2025 net loss was $30 million, including certain onetime noncash income tax expenses of $163 million compared to a net income of $341 million in the year ago quarter. Diluted earnings per share was negative $0.03 compared to $0.37 of last year. Excluding the previously mentioned nonrecurring item related to the impairment, restructuring charges and other related phaseout costs, non-U.S. GAAP net income stood at $100 million and non-U.S. GAAP diluted earnings per share stood at $0.11, including certain negative onetime tax expenses impacting of $0.8 per share.

Looking now at our full year 2025 financial performance. Net revenue decreased 11.1% to $11.8 billion. In terms of revenue by end market, Automotive represents about 39% of our total 2025 revenues. Personal Electronics about 25%; Industrial, about 21% and Communication and Computer Peripheral about 15%. By customer channel, sales to OEMs and distribution represent 72% and 28%, respectively, of total revenue in 2025. By region of customer region, 43% of our 2025 revenues were from the Americas, 31% from Asia Pacific and 26% from EMEA. Gross margin decreased to 33.9% for 2025 compared to 39.3% for 2024, mainly due to lower manufacturing efficiencies and to a lesser extent, the price and mix, lower level of capacity reservation fees, negative currency effect and higher unused capacity charges.

Operating income stood at $175 million compared to $1.68 billion in 2024. Excluding $376 million for impairment, restructuring charges and other related phaseout costs, non-U.S. GAAP operating margin was 4.7%. On a reported basis, net income was $166 million and EPS was $0.18. On a non-U.S. GAAP basis, they stood respectively at $486 million and $0.53. Net cash from operating activities totaled $2.15 billion compared to $2.97 billion in 2024. Net CapEx expenditure was $1.79 billion in 2025, in line with our revised expectation and lower than the $2.5 billion of 2024. Free cash flow was $265 million positive in 2025 compared to the $288 million positive of the previous year.

Inventory at the end of the year was $3.14 billion compared to the $3.17 billion at the end of the third quarter and $2.79 billion one year ago.

Days sales of inventory at quarter end were 130 days, slightly better than our expectation compared to the 135 days for the previous quarter and 122 days in the year ago quarter. Cash dividends paid to stockholders in 2025 totaled $321 million. In addition, during 2025, ST executed share buybacks totaling $367 million.

ST maintained its financial strength with a net financial position that remains solid at $2.79 billion as at end of December 2025, reflecting total liquidity of $4.92 billion and total financial debt of $2.13 billion.

Now back to Jean-Marc, who will comment on our outlook.

Thank you, Lorenzo. Now let's move to our business outlook for Q1 2026. We are expecting Q1 '26 revenues at $3.04 billion, a decrease of 8.7% sequentially, plus or minus 350 basis points. We expect our gross margin to be about 33.7%, plus or minus 200 basis points, including about 220 basis points of unused capacity charges. This business outlook does not include any impact for potential further changes to global tariffs compared to the current situation.

In terms of net CapEx for 2026, we plan to invest about $2.2 billion to support capacity addition for selected growth drivers like those for cloud optical interconnect and our manufacturing reshaping plan.

To conclude, 2025 turned out to be a challenging year for the end market we serve, characterized by continued inventory correction in automotive and industrial, in particular, the first part of the year. The second half was better with gradual improvement of the revenue trend and a return to a year-on-year growth in the fourth quarter. We are entering '26 with a better visibility than entering '25 with the inventory correction in distribution progressively improving.

Beyond the evidence of a cycle recovery, ST will benefit from the following company-specific growth drivers. In automotive, we see solid momentum in our engaged customer programs in ADAS, where we expect to grow this year and in the coming years. In silicon carbide power devices, following a significant contraction in 2025, we anticipate a return to revenue growth in 2026 with revenues projected to recover to 2024 levels by 2027. In sensors, we see strong demand, both in MEMS and imaging sensor and our planned acquisition of NXP MEMS business will strengthen our leading position across the automotive and industrial segment.

In industrial, in general purpose MCUs, building on market share gains 2025 and a road map of new product launch for 2026, we are on track to return to our historical market share of about 23% by 2027. In Personal Electronics, where we continue to see strong momentum in our engaged customer programs in sensors and analog, we should keep on benefiting from increased silicon content in 2026 and beyond. In communication equipment, computer peripheral, in data centers, including cloud, optical interconnect and power and analog for AI servers and data centers, with the current market dynamic, we believe we can deliver $1 billion revenue before 2030 with already USD 500 million in 2026. In low-earth orbit satellites, we are expanding our customer base, and we anticipate continued revenue growth as low earth orbit constellation projects expand globally and penetrate new applications such as direct-to-cell constellation.

Lastly, ST is uniquely positioned to address human wind robotics through our broad portfolio, spanning MCUs, MEMS, optical sensors, GNSS and power management. We are already generating revenues through engagements with major OEMs, and we estimate our current addressable bill of material at about $600 per system.

Thank you, and we are now ready to answer your questions.

[Operator Instructions] Our first question comes from Francois Bouvignies from UBS.

My first question maybe for Jean-Marc, I wanted to come back to what you said about the outlook. I mean, if we look at your revenue guidance down 8.7% quarter-on-quarter, this is below seasonal -- better than seasonal, sorry, of minus 11%. And if we take into account less days, it's actually significantly above seasonal.

So, I was wondering, I mean, this is looking quite interesting. And if we compare to other peers like TI yesterday or ADI and Microchip, you see a number of your peers talking about above seasonal. I mean what's your view on the trajectory from here? Do you think this above seasonal trend can carry on a little bit? Or we shouldn't get carried away like we did in the last two years where we have many fall starts? Do you see like a very genuine evidence of a cycle recovery from here?

Well, we will not guide for 2026 today, clearly, but we are confident in our ability to grow organically for next year. But it's clear that we enter in a better and healthier situation compared to '25. If you remember last quarter, okay, I already shared with you that we were seeing a backlog that were reading during the quarter better than the usual seasonality. And today, with the visibility we have on Q2 that generally speaking, okay, is plus, let's say, low mid-single digit, but we absolutely see no reason that we will not be at least capable to deliver it.

More important, I think, beyond the cycle is to share with you that we see for the company some specific growth driver. First of all, in automotive, clearly, we will have the sensor. And at a certain moment, when we will complete the acquisition of NXP, of course, it will bring additional revenues. This is obvious. But we see also positive momentum on ADAS ASICs and the silicon carbide after last year that was pretty challenging.

Well, in industrial, clearly, the dynamic is really strong, thanks to the inventory correction gone, but more important is our portfolio. So we have done a tremendous effort in introduction of new products in '25 and '26, and this will contribute beyond the cycle. For Personal Electronics, our engaged customer program, you know that we have the visibility, okay? So I confirm to you. So we confirm that it will support us beyond the cycle.

And last but not the least, data center. But clearly, in 2026, cloud optical interconnect, so means both photonics ICs and analog mix signal by CMOS ICs plus our high-performance general purpose microcontroller will contribute because you know that the connectivity engine of the server will move to optical one. So this will be certainly an acceleration. And as well, we will start to contribute to the power supply unit and to the server from the green to the processor.

Last but not the least, beyond the cycle in '26, we see also low earth orbit satellite communication with our engaged customer program, so with our ASICs really positive. This will be a bit offset by the capacity fee reservation. But all in all, I confirm really our confidence level to grow organically in 2026 and because we have, let's say, significant growth driver beyond the cycle of the market.

Very clear. And yes, maybe on the gross margin side, I mean, with it, I mean, obviously, it's a concern for the market. You delivered the guidance is in line on the gross margin, but 33.7%. But when I look at the consensus, it has 35.6% of gross margin for the year. So it would assume a recovery from here. So with the top line that you described nicely, should we see as well an improvement of gross margin from the level in Q1?

Maybe I take this one, Jean-Marc, about the gross margin. But today, of course, the gross margin will depend on the evolution of the revenue in the course of the year. As explained by Jean-Marc, we expect, let's say, to increase. But the gross margin today that we see in Q1, we believe is clearly the lowest point in the year, this expectation of 33.7%. So we will see some increase. This increase is also driven by the fact that we expect to have constantly reduction in our unloading charges during the year. So we expect some mild increase for the second quarter and then a more significant increase also driven by the seasonality of the revenues in the second half of the year. Yes, at this stage, we can say that the expectation for us is to have increase in our gross margin all over the year.

The next question comes from Andrew Gardiner from Citi.

I was interested, Jean-Marc, in digging a bit deeper into the automotive space. Clearly, your largest end market and the one where we're still seeing the most difficulty in terms of getting through the bottom of this cycle. There's a number of sort of end market data points out there that are, I suppose, still causing investors' questions in terms of the health of the market, tariff threats back and forth admittedly, but also not helping.

I'm just wondering how -- can you give us a bit more detail in terms of how you're seeing your customers behave? Do you think inventory is absolutely at a bottom in terms of the automotive channel and at the OEMs and the Tier 1s. What kind of confidence do you have as we look into the future quarters that we can return to stronger demand trends?

Well, first of all, clearly, when we see our Q4 revenue in automotive, it was slightly below our expectation and mainly, in fact, driven by the pulling from inventory a little bit lower than expected from some Tier 1 means that the automotive market for, let's say, legacy application, clearly is pretty soft. Inventory correction is certainly gone, but there is a kind of a softness of this kind of application.

What will be positive on automotive is clearly what is around, let's say, the electronic architecture, the new software-defined electronic architecture calling for more complex MPU, MCUs definitively. So this will be an important growth driver. But we know that the electrical powertrain will be still an important driver. But here, it is more the competition landscape that changed completely compared a few years ago because you see that out of, let's say, more than 30 million vehicles produced in China, more than half are battery based compared to America, where it is more marginal in terms of production. And in Europe, it is below 1/3. So here, it's more a question of the competition is in China. So you know that in China is more complex to compete. But the powertrain electronics, the demand is there.

So, all in all, I think the automotive market based on 90 million, 92 million, 93 million vehicles out of which 17 million to 18 million vehicle battery based and similar number in hybrid is still changing in terms of mix as well from the car classification is more middle end or premium car, even this car now embed some electronics. So the market is not yet stable. So that's the reason why we have to be, let's say, cautious to adapt ourselves. But we see a different situation compared entering in '25, where we faced very strong inventory correction in Q1 last year, if you remember, from our main customer, this will not be repeated.

It is more, let's say, a progressive stabilization of the market in terms of mix of car electrical hybrid thermal combustion engine and mix of car between high premium, premium and middle class and mix between China, APAC, Europe and Asia. So this is something we have to, of course, closely monitor and adapt ourselves with our supply chain. So this is how we see the automotive market.

Just a quick follow-up, given you mentioned China at length there. How is the partnership with Sanan progressing? Is that going as you anticipated? Is it helping your competitiveness in that market? Or is it still too early?

No. Clearly, so we will start to ramp up the facilities now, okay? We have modernized. We know exactly the efficiency of this fab. And clearly, it will be a key success factor in our capability to compete on the Chinese market.

The next question comes from Joshua Buchalter from TD Cowen.

I actually wanted to drill into the Personal Electronics segment a little bit more. I think there's some concerns of disruption or even pull-ins in the short term due to higher memory costs. It came in better in the quarter. Maybe you could walk through what the drivers you're seeing are there and if you're seeing any changes in order pattern. And I believe you called out higher silicon content in 2026. Was that referring to expectations for your largest customer this year?

Yes, you know that our revenue are mainly driven by our biggest customer and more on the high-end kind of product, which are, in some extent, less sensitive to the memory price. So, at this stage, with the visibility we have, first of all, we don't see significant impact detected by us. And I confirm that we expect to keep growing in personal electronics driven by our main customer in 2026, thanks to our increased device based on silicon and not module content increase in '26. So far, PE will be a growth driver for us in '26.

And then I think the last couple of quarters, you've been kind enough to give us your book-to-bill ratios in auto and industrial. It seems like things are getting better on the industrial side in particular. Can you update us, I guess, on those metrics and whether you're mostly done with the channel inventory clearing on the industrial side? Congrats on the solid results.

No. In Industrial, the book-to-bill was well above parity. Clearly. Also, beyond your question, I can tell you that the POS were growing, let's say, between low teens, mid-teens, which is a good news. So we continue to decrease our inventory.

But on automotive is the book-to-bill is a little bit more complex because we have some few key customers that are putting order in one shot for six months. So the book-to-bill must be, let's say, assessed on one-year moving average or six months moving average. So corrected from this, let's say, abnormal, let's say, process, the book-to-bill was parity on automotive.

The next question comes from Stephane Houri from ODDO BHF.

I just wanted to come back a bit on the scenario for the year, and I know you're not guiding. But historically, you've been saying that the second half is like 15% above the first, that's normal seasonality. And then on the top of that, you may have some specific programs. With the sting point you guide on Q1 and we look at -- when I look at the consensus for the full year, it seems to be banking on something lower than that because of the starting point in Q1. So can you just confirm that you see now that the inventory correction is done normal seasonality throughout the year and maybe give some comments about the adds of some customer engage program?

No. On the inventory correction, what we communicated, okay, I and Lorenzo and myself is to say by end of Q2, we believe we will be hold the excess of inventory. And this today, I can confirm -- it's already the case for many product family. We are still here and there some pockets of excess inventory versus what we see. But looking at the current dynamic, POS, POP by end of Q2, this will go. So now it's sure that in H2, we will be exposed directly to the end demand.

Now about again, what we consider engaged customer program be the cycle, let's say, we can split I have to say. One is the usual personal electronics, and why we say it's cycle is because silicon content increase, okay? So we have the visibility with the current visibility we have, okay? So this will help us to grow the cycle of personal electronics and assuming our main customer will perform in market share really well performed in 2025, okay? So this will drive our growth.

Moving to communication equipment and computer peripheral, well, communication equipment. Communication equipment, it is clear that for the lower or satellite communication is an important driver because thanks to our capability to supply and compete, our growth is driven by our largest customer in this field of activity. And as we see is pretty successful. And certainly this year, will be another demonstration of the success. Now since two quarters, we are supporting our second largest customer that is growing as well. So it is clearly beyond the cycle. So this will be a significant growth driver beyond the cycle for ST.

Last but not the least is AI data center. You know AI data center, okay, we were, let's say, a bit delay for what call the device addressing the power station we are in, let's say, process to close the gap and offer solution to our customers. But clearly, we will be at of the business dynamic, it is the optical engine or the cloud optical interconnect. So its photonics ICs, MOCs and high performance general microcontroller. And this will contribute to the growth of ST significantly in 2023.

Then moving to the more, let's say, traditional market focus we have, so automotive industrial. For ADAS, ASIC, last year was a challenging one because we saw some inventory correction on, let's say, some legacy ASIC. But this year, okay, clearly, with the visibility we have, this will be a booster of growth.

Finally, our SiC MOSFET, about the difficult year of '25 will grow again. And I can confirm to you that up to now in Q1, we have a good book-to-bill on silicon carbide that is very encouraging. And definitely, our sensor contribution with the acquisition of NXP MEMS plus the existing imaging sensor, existing MEMS we have. And I am very pleased that beyond the inventory correction done on general purpose microcontroller, the proliferation of our new products are really paying back very well. And I am really confident that in '27, we come back to our historical market share and '26 will be an important step to demonstrate it. So this is actually in a few words how we can describe '26.

I have a small follow-up on the gross margin comments. I think last quarter, you said that you think you would end up Q4 2026 above the level of Q4 2025 in gross margin. Do you still feel confident with what you see developing the mix, the underloading charges, et cetera, et cetera?

Yes. Yes, I confirm that at this stage, the expectation is that Q4 this year '26 should be better than Q4 '25.

The next question comes from Domenico Ghilotti from Equita.

A couple of questions. The first is on the unloaded charges. You are guiding for a significant drop in Q1. trying to understand despite the lower sales, I'm trying to understand if you see this number at the bottom and if you are already benefiting from, say, the efficiency plan that you carried out.

And second is some color on, if you can, on the second client in low earth orbit. So should we assume that it is a significant number or just starting entrance of new clients or an add-on, but not particularly relevant?

Maybe I'll take the one of the unused charges. Yes, unused charges are declining in the first quarter. There are -- the reason -- the main ingredient of the declining in this quarter is the fact that, as you know, we are progressing with our programs to reshaping our manufacturing infrastructure. This program is progressively reducing our capacity in 6-inch for silicon carbide, 150-millimeter for silicon carbide and 200-millimeter for silicon. And we start, let's say, to move ahead on this plan. So this is, if you want, is something that is mechanical. At the end, the capacity is reduced. We are now moving our product on the existing capacity on one side, 8-inch for the silicon carbide and the 300-millimeter for the silicon.

So that's why we see the level of unused capacity, notwithstanding that the revenue are lower in respect to the previous quarter to reduce. This trend will continue. Unused capacity will not disappear in the year, but will significantly reduce in the year and will be one driver for our improvement in the gross margin in the course of 2026.

About the second question, yes, it's significant. If not, we will not mention. But I can just confirm you to number in Q4, our CCP segment grew sequentially 23% and year-over-year 22%. Definitively, it is linked to the low or satellite business we have, and it is driven both by our first customer and then by the second one. So at 22%, 23% growth sequential and year-over-year, so you can conclude it is significant.

The next question comes from Sandeep Deshpande from JPMorgan.

My question is about your fab loading into the current quarter. Given what is happening with the gross margin in the current quarter, how is the fab loading going through in the quarter? And how is the mix shifting overall in terms of the gross margin? Because you have a revenue decline, but the gross margin is declining. So are you reducing your fab loading this quarter? Or are you increasing your fab loading?

And my follow-up question associated with that is how the mix, particularly associated with your better margin microcontroller products is shifting?

In the quarter, as I was saying before, the unloading charges is mainly related to the fact that we are moving out capacity, reducing capacity in certain specific fabs. where, of course, we are now moving production in different fabs, 300 millimeters, so reducing our capacity. So at the end, when you look at the level of loading, we are not overloading our production, let's say, in the quarter.

Clearly, if you look the inventory and you look where it will be the dynamic of the inventory in the quarter, as usual, you know that there is this seasonality in our inventory in which in the first half, our inventory is somehow increasing and then decreasing in the second part of the year. So, at the end, what it will be the impact is that now the expectation is end the quarter Q1 in the range of 140 days of inventory compared to the 130 days where we stand today.

But I repeat that this is more related, let's say, to the normal dynamic of our inventory over the year than, let's say, loading our manufacturing infrastructure in a way that is -- the impact on unloading charges is mainly related to the fact that we started with our programs to reduce capacity in some specific areas.

Clearly, the impact -- the positive impact, let's say, of this in terms of gaining efficiency and so on will come probably later, as you know, in our, let's say, manufacturing infrastructure. We do expect our program to be -- to start to yield a positive impact in our -- in our manufacturing efficiency more in 2027 than this year. But one of the impact that visible is the reduced level of unloading. Together also with the expectation of growth in terms of revenues. This we will see during the year, let's say, depending on the level of growth.

And my follow-up question is regarding about your microcontroller business, which is if you look at the Embedded Processing segment, it grew 1.2% year-on-year. I mean, many of your peers in this market are seeing better growth at this point. So why is ST growth in a key segment for ST lagging at this point or something else happening in that division?

So, embedded processing segment, clearly, the growth dynamic we have on the general purpose microcontroller is, let's say, at least consistent with our peers.

Why it is a little bit offset? It is offset by our automotive microcontroller because, okay, up to now, our automotive microcontroller are more the microcontroller that will be, let's say, for some model of car moving to the software-defined vehicle architecture removed clearly. And I already explained that we have done a strong effort in 2025 to rework the road map of our micro, but this will be paid back, okay, more, let's say, end of '27 and '28. For the time being, yes, we suffer on the automotive microcontroller that is, let's say, optically offsetting the real good health of the general purpose.

But the general purpose microcontroller, let's say, maybe I can share with you one number, okay, for Q1, the embedded processing solution segment will grow up low 30s. So above 30% year-over-year. So you can imagine that the growth of general purpose will be really, really strong more than, let's say, the secure microcontroller are growing a little bit less because driven by the market. And okay, of course, we have some offset linked to the automotive micro. But I can confirm to you that our general purpose microcontroller are performing or overperforming the market.

Mona, I think we have time for one more question.

Next question comes from Sébastien Sztabowicz from Kepler Cheuvreux.

Coming back to the transformation program, have you made any specific progress so far? And notably on the manufacturing front? And are you still on track to reach your savings ambition for the end of '27? And the second one is more on the OpEx trend. So Q1, we know where it will stand. But for the full year, where do you see OpEx trending? And how do you see the start-up costs impacting the OpEx 2023? Do you plan to accelerate a little bit further the cost-cutting actions for OpEx?

In terms of our reshaping programs, I would say that is progressing in line with the expectation. In the course of 2025, the main, let's say, impact was related to the savings in our OpEx that indeed, when you look at the overall are declining, notwithstanding, let's say, the negative impact of the euro dollars. So at this stage in the course of 2026, as I said, we will start, let's say, progressively to transfer some activity from -- in silicon carbide to 8-inch in silicon to the 300-millimeter.

As I was saying before, is now expected to yield the benefit in our manufacturing infrastructure efficiency of this program towards the second part of 2027 and 2028. So my short answer is, yes, we are on track in respect to what we have communicated previously. So, this is the situation.

In respect to the expenses of 2026, now the expectation remains substantially the same, means that at the end, at this level of exchange rate, including the impact of the hedging, we should be able to stay with a net OpEx means including other income and expenses on a low single-digit increase, something in that range, mainly driven by the fact that we will have a reduction in other income and expenses in respect to the one of 2025 due to the phaseout cost because, of course, let's say, from the one side, we reduced the capacity in our manufacturing 6-inch, 8-inch. But on the other side, we have a progressive phaseout from these steps that will be reported in this line. It's a temporary effect, but it will be there during 2026.

Thank you, Sébastien, and thank you, everyone. I think this is ending our call for this quarter. So, thanks very much all of you for being there, and we remain here at your disposal should you need any follow-up questions. Thank you.

Ladies and gentlemen, the conference is now over. Thank you for choosing Chorus Call, and thank you for participating in the conference. You may now disconnect your lines. Goodbye.

I think we're ready to start. Hi, everyone. I'm Simon Coles, Head of European Technology Hardware and Semiconductor Research here at Barclays. Delighted to have Lorenzo, CFO of STMicro with me here today. Thanks so much for joining us.

Thank you.

Maybe we'll start off with sort of setting the scene a little bit. So 2025 was an interesting year. The industry was hoping for some recovery in the analog space, and it never really materialized as hope. So maybe you could give us a sort of a review of how STM for the year. And then you could give us an update on sort of how you're feeling about the auto and industrial markets at this point?

Yes, sure. Good morning, everybody. Well, let's start to say that looking at the current quarter, I can confirm that what we said entering the quarter is what is going on. Let's say, we see the situation evolving consistently with our expectations. And maybe for some elements, even a little bit better than our expectation. But you know that this year for us in 2025, looking at 2025, clearly, the first part of the year -- the first half of the year was impacted in automotive and also in industrial by a significant inventory correction. Looking at automotive, clearly, we have a few customers, important customer for us that were significantly correcting this inventory in the first part of the year. And now I have to say that this is behind us. Indeed, when we look starting from Q2, we started to grow on a year-over-year -- to grow, let's say, on a sequential basis in automotive, definitely.

And still, we think that we see that also in this current quarter, we will continue to grow on a sequential basis on a mid-single digit. And the evolution is consistent and in line with our expectations. Industrial was similar in Industrial, the year has been characterized by an important inventory correction in 2025. And now if I look at the situation of the inventory and especially the inventory that we have in distribution, we see now a normalization. And in some regions, even a little bit better than what is the normal level of inventory that we should have. So I think also here, we see, let's say, a positive dynamic you see that our revenue in Industrial were growing on a sequential basis. In Q3 we were also on a year-over-year basis, improving. But we will continue to see some growth also in the current quarter, more low single digit. Why? Because we wanted to exit the year as clean as possible in terms of inventory.

But looking, for instance, some dynamic like the book-to-bill, we see that the company has a book-to-bill that is both parity and this is positive, let's say, especially when we look our booking in the industrial. So I would say that we are relatively optimistic on the trend at this point.

Okay. Great. You touched on some of the inventory levels there. Is it fair to say that you're cautiously optimistic that the destocking risk that -- or the destocking activity we've seen in '25 is pretty much done. And by the time you've got through Q4, you don't really expect to see anything else on that front. And then you mentioned book-to-bill is above 1. Are bookings sort of trending maybe a little bit better since sort of earlier in the year. And do you think you have sort of better visibility into 2026 than you did have going into 2025 at the same stage last year?

Now clearly, when we look at the visibility entering in 2026 compared to the visibility that we had entering 2025, yes, the visibility is definitely improving. As I was saying before, at this stage, we do not detect any inventory correction. Last year, we were impacted significantly. But today, clearly, what we said at our earnings release, Q3, we said about our expectation now there is no reason why to not to think that Q1 will be in the normal seasonality. Last year, we're well below the normal seasonality. Today, frankly speaking, with all the indication that we have, we think that the next quarter should be in that range.

So at the end, now I would say that overall, definitely, our visibility has improved. We enter with a level of backlog that is in the year that is better than what it was 1 year ago. Maybe we are not yet a situation in which we cover in terms of backlog the same level that was covered pre-COVID. But definitely, we see the situation improving, definitely improving.

Okay. Great. Maybe we touched on gross margins before talking about some of the revenue opportunities. I mean they're not where you'd like them to be. There are a number of moving parts that you've been very helpful sort of laying out in the last few quarters. Could you maybe help us how to frame how to think about those into 2026. I mean you're reshaping the manufacturing footprint so there's some costs shifting around there. Underutilization should come down as revenues are improving, FX, like lots of things going on. So could you help us sort of frame how to think about those?

No. Yes. Definitely, 2025 in terms of gross margin has been quite difficult definitely, let's say, at the end, at the midpoint of our Q4 guidance, our average gross margin in the year would be in the range of 33.8% or something like that. So clearly, this year, our gross margin has been definitely significantly impacted by what about, first of all, impacted by a significant level of unloading charges. This year, the unloading charges account for more than $400 million. That is a little bit more 300 basis points. So this was clearly a strong headwind. But when we look next year, clearly, at this stage, it's a little bit difficult to define exactly where we will position in terms of gross margin. But definitely, we can talk together with you about what it will be the dynamics of our gross margin.

For sure, one of the expectation is to have a significantly reduction in terms of unloading charges in respect to this year. Why we expect this significantly reduction? There are 2 elements to consider. On one side, we expect to an increase in terms of revenue. So this will better load our manufacturing infrastructure. And on the other side, we do expect to start to implement our reshaping plan for the manufacturing infrastructure. It means to reduce some of our capacity especially next year, we expect to reduce the one in silicon carbide, reducing capacity at 150 millimeter, and for some extent, also some reduction in the 200-millimeter of silicon. This mechanically will reduce the level of unloading because we will transfer products in silicon carbide 200-millimeter, in silicon to the 300-millimeter but the increase of capacity, it will be tuned based on the demand. We will try to tune based on the demand. So this will be a positive impact.

There is another positive impact always related to the manufacturing that is better loading, lower unused capacity, but also better manufacturing efficiency. This will come, let's say, of course, we do not yet exploit all the positive impact of our reshaping plan. But we will start to see something already in 2026.

Then I was talking before now, clearly, let's say, next year, we will not have any longer the impact of the inventory correction, especially in the industrial market. Industrial market for us is the most accretive in terms of gross margin. So it means that we expect a positive impact of the product mix. More revenues in our microcontroller, STM32, more revenues in analog. These are bringing a positive impact on our gross margin. Unfortunately, not everything is positive. There is also some headwinds. One headwind that we talk many times together was -- is the one-off capacity reservation fees. Capacity reservation fees this year are still meaningful means that impacting in a positive way our gross margin for around 100 basis points next year, this will decline significantly. Still some. But let's say, if we have a 20, 30 basis point impact positive is the maximum level that we can think about.

And then on the other side, we have not to forget that the dynamic of the exchange rate is not going in the right direction for us. Exchange rate this year in average for the company the effective one means considering the impact of the hedging has been 1.11. Next year, this level of spot rate most likely will be something between 1.15, 1.16. This clearly is a negative impact. All in all, anyway, we expect improvement in the gross margin and more significant in the second part of the year in respect to the first part of the year, most likely, also due to our seasonal dynamic in terms of revenues. But yes, at this stage, I'm quite confident to see some recovery in terms of gross margin next year.

That's great. That's very clear. Maybe just quickly on OpEx. It seems like after the Q3 commentary is set to go up next year. You have your cost-cutting plan. Anything we can do to sort of accelerate that to sort of try and manage this increase? Or how to think about that into '26?

In terms of OpEx clearly, we continue with our plan of resizing of our OpEx. Anyway, also here, we have a different dynamic. First of all, let me remind you that when we talk about OpEx, of course, excluding the restructuring cost at this stage, but is including SG&A, R&D and the line that we have in our P&L, other income and expenses. In this line, other income and expenses, we account mainly for the grants. And this year, this line is positive for us in the range of $200 million plus.

What are the dynamics of next year? Of course, next year, we will continue to resize our expenses. As you remember, we were saying that in respect to the cost structure of 2024. We wanted to achieve a savings of around between $300 million and $360 million. Let's say that my goal as a CFO is to be more on the high side than on the low side, but fine. This year has been more than $100 million in the range of $110 million something like that. Well, next year, we do expect it to be similar, maybe slightly a little bit slightly higher. So the savings will be there. On the other side, there are some negative impact. And the main one is related to a reduction of the positive impact of other income and expenses. Why we think that other income expenses will go down, not because there is less grants, more or less the level of R&D grants that we will have in 2026 is not so far from the number that we have in 2025.

But it's mainly due to the fact that we will account in this line the start-up cost, especially for our facilities in silicon carbide 200-millimeter. This is an impact that is onetime, but it will be accounted in that line. But then there is the FX also in the expenses, we will have the negative impact of the FX. You know that we are in our expenses quite exposed to the euro. We have a significant amount of expenses that are denominated in euro. So all in all, what is our expectation. All in all, it is an expectation of a slight increase when you take SG&A, R&D and other income expenses together.

Got it. Okay. Maybe moving on to some of the revenue opportunities that you have ahead. There's a lot of excitement around interconnect in DCs. And I think about a year ago, you announced a sort of silicon photonics foundry business. How is the progress on this opportunity? How meaningful do you think this can be for STM? Is it a significant revenue driver in '26 already?

Silicon photonics for sure, is a very significant opportunity for us. There is a strong traction here. You have to consider that we start now. And we are now qualifying our products and really, let's say, here, the opportunity, yes, I have to say that it's already in next year. We see some opportunity already next year, let's say, some meaningful revenues in silicon photonic. More we go more, we see traction on that. Clearly, let's say, 2027, it will be even higher and important than 2026. Anyway, we start to see already some revenues over there. We think that silicon photonics could be a significant driver in terms of our revenues but there is no reason why to not think that in 3 or 4 years from now, we may be, let's say, something in the range of $0.5 billion of this business. And this business by the way so far is also bringing more than decent profitability, I would say, for the company.

Okay. Great. Maybe you sort of linked to this then is the AI power opportunity. Some of your peers have obviously been announcing some big numbers recently, but you have all the same products as well, silicon carbide, gallium nitride. You talked about some of these opportunities at your CMD last year. I mean, are you expecting to generate revenue in this area soon.

As ST, I think we have a portfolio that is well fitting this kind of opportunity. As you know, we have also announced a cooperation with NVIDIA, let's say, with in this field in order to provide and especially to target architecture 800-volt for the artificial intelligence in terms of server for power. So here, I think this is an opportunity for us. We have the portfolio that is covering with the silicon carbide, gallium nitride, but not only we have a microcontroller, analog all these, let's say, products, high-voltage MOSFET and low power, low voltage power MOSFET. So we have a portfolio that can cover this kind of portfolio. We address -- we think that we are well positioned, let's say, to address the new architecture 800-volt. So it means that this probably a revenue that will come not immediately, not in 2026, but we will start to have revenues that are coming in the coming year '27, '28 and '29.

Also, here, we think that this is a market that is visible to everybody, growing significantly. No reason why we should not position ourselves with a similar level of revenues as for silicon photonics, this could be a significant driver for us, bringing revenues in the range of $0.5 billion and more by the end of the decade.

Okay. Great. So $0.5 billion potentially from AI in the next couple -- next few years. The satellite business, I feel goes under the radar a little bit and is performing very strongly for you. I mean how do we think about that business growing from here? Is there much competition in this space? How does STM differentiate? Can you give us a bit more color on the satellite opportunity?

You know that we are here in this area since a long time, I think something in the range of 10 years. Our, let's say, cooperation and work together with Starlink has been pioneering this area. We have, let's say, IPs, technology ability in design chips that are fitting very well this kind of application. But this is a significant growing area. We see, let's say, that is an area in which we have experienced in the last year's growth. We expect to experience growth also in the coming year, in 2026 and 2027. And on top to our, let's say, let's call it, historical customer. And now we are enlarging the customer base. There are new actors that are coming in some of them are quite important in which we are present. We start already shipping this quarter in 2026 will be a growth driver.

And this is for us, let's say, enlarging the customer base. And then we have also, let's say, design win in other constellation. You know that both in Europe than in China, there are, let's say, traction on these kind of things. So I think this is an area in which we are well positioned. We are well advanced. We have, let's say, the right technology, both in front end and in back end in order to address and to enjoy growth and profitable growth here.

Great. Okay. Maybe we touched on silicon carbide that was seen as a huge growth opportunity. The last year or so has been quite difficult given what's happening in China. But it feels like the pricing pressure in places like China may be sort of bottoming out. How are you feeling about the silicon carbide outlook from here. You suggested growth next year already, I believe. So what are the drivers of that as well?

In silicon carbide, clearly, 2025 has been let's call it, a transition year for us. In 2025 seeing a significant decline in our revenues in silicon carbide this not secret. But clearly, we were impacted by fewer factors. Clearly, let's say, our main customers had inventory correction, also reduction in respect, let's say, the penetration in the market. This was somehow impacting our revenues. And we were not in the position in 2025 to compensate with new sockets that were there, but not at the level to compensate this impact. What moving forward, let's see. Moving forward, clearly, our expectation for 2026 is to have some growth in terms of silicon carbide. Why I'm saying so not only mainly due to our main customer, but because we have new sockets, both in Europe and in China that in the course of 2026 will ramping up.

Of course, we do not expect in 2026 to be back at the level of revenues that we were in 2024. There will be some increase, but still not at the level of 2024. Looking at what we have in our hands and looking what are the sockets, the platform where we stand today, we believe that in silicon carbide that we may be back close to where we were 2024 in 2027. Anyway, we believe that silicon carbide is still, let's say, technology product set that has an important development. Maybe has been a little bit later and slower than what was expected maybe 2 years ago. But still, there is a significant opportunities, not only in the mobility but also in the industrial market and to not be forget in the server for AI. So it means that at the end, I think here, let's say, our target in terms of revenue in silicon carbide is still there.

Got it. Very clear. Humanoid robots starting to get some more attention. STM has an extensive MCU portfolio. You also do lots in sensors as well. How are you feeling about this opportunity? Are there any numbers you could share about sort of content? Is this something that's exciting for STM in the next couple of years?

No, yes. This is an opportunity, definitely a significant opportunity in this market. Clearly here, our portfolio is fitting quite well this kind of application because, as you said, here is microcontroller analog and motion control, let's say, sensors, all these kind of things, sensors, both MEMS and image sensor positioning. There are a lot of products that are really fitting this kind of application. And we see also because already today, we are selling in this kind of market. But clearly, the point of this market that today is still quite small. Even if the content can be quite high because you may have as a minimum $200 to $300 of content in 1 humanoid and that could grow also in the range of $500 million, $600 million. So it means that at the end, this is a good opportunity. But the question mark is the evolution of this market.

The good point is that we are present. We are in more than one, let's say, customers for this kind of application. Then we will see how this application will move, how pervasive will be in terms of pervasiveness in the market. Yes, there are possible different opportunity. One is the industrial market. Why not defense because this could be another area for this. Maybe consumer a little bit down on the road. But definitely, this is a market that we look very carefully, and we think for us could be a good opportunity. And I repeat that we start already now to sell to some of these early, let's say, company in this field.

Okay. Great. You've got an acquisition in progress, NXP's MEMS business. Could you give us a bit more color on sort of how you think that's going to fit into the group? And then it's going to cost you not far off $1 billion. Are you happy with sort of the cash position post-acquisition?

To be honest, I think that this acquisition has been very, very positive for us. I'm quite excited about this acquisition. Clearly, let's say, it's fitting perfectly our portfolio. As you know, let's say, in MEMS, we are quite strong in -- when we look at the personal electronic consumer with the acquisition of NXP, we will rebalance our presence in the market with automotive and industrial. So means that at the end, for us, it is the perfect fit. The cooperation with the MEMS of NXP has been since a long time because we were working together. So it means that for us, this is a very good acquisition. By the way, -- but I think looking at the last years, let's say, M&A activity, if you consider the multiple on which we buy this business is more than reasonable.

At the end, we have enough cash on hand in our balance sheet to sustain this kind of acquisition that has been -- the consideration has been fully in cash, will be full in cash because so far, it's not yet closed, but I'm quite confident soon it will be. So I feel very, very excited about that because I think it positions ourselves in a very strong positioning in respect to the MEMS. So we will be, let's say, among the 2 top players in this field. And frankly speaking, this field is growing significantly. There are many applications, industrial, automotive. We were talking now about humanoid, all these kind of things. Think about -- there are many, many areas in which this MEMS are going. And to be honest, it's one of our most profitable product lines. So for me, it has been a great achievement.

Okay. Brilliant. And then signal times up. So Lorenzo, thank you so much for the time. It's very interesting, very helpful. I hope it is for everyone else. Thank you.

Thank you very much. Thank you.

Good morning, everyone. Welcome to Barcelona. Welcome to the 25th Morgan Stanley TMT Conference here in Barcelona. Welcome to, of course, Jean-Marc Chery, CEO of STMicro to the stage.

Jean-Marc, maybe just to level set the audience, if you could maybe help us or remind us rather, how did Q3 go? What was the main points? And how does that leave us going into Q4?

Okay. Q3, our revenue has been mainly driven by the usual seasonality of personal electronics, where we have benefited of both our increasing content with our main customer and the usual seasonality of the start of the new device. So personal electronics grew 40% sequentially in Q3. And -- except automotive, I will come back on it. All the other segments grew sequentially and good news year-on-year.

On automotive, we are still on a year-over-year decrease, but the main gap is related to one customer specific, an electrical car maker, well-known. And also you know that for ST, the progressive decrease of the fees related to capacity revolution. But clean on this effect that is not product related, that is more, let's say, the heritage of the shortage period, we can say that automotive is starting to grow year-over-year, very close. So this was the main point of Q3. And the book-to-bill was basically above 1, okay, which was okay, also good for Q4. Yes.

And maybe just keeping on Q4. Obviously, you guided sales, I think, 3.28 billion for Q4, slightly above the minus 10% that you would see for seasonality. So maybe help us understand, does that help us appreciate what the run rate will be into '26? And again, as reference to automotive, how should we think about that as we go into '26?

No, it's a very interesting question. '26, well, clearly, again, the usual seasonality for ST is to have a first quarter of minus 10% or slightly above minus 10%, mainly driven by the personal electronics and [indiscernible] because of the Chinese New Year impact for mass market and distribution.

Today, the dynamic of our backlog coverage for Q1 linked to the order entry we have, is clearly securing ourselves that we will be capable to deliver minus 10%, minus 11% in terms of revenue in Q1, which is positive news confirming that we are basically almost free of material inventory correction that was the case in Q1 last year. And by the way, if we deliver minus 10%, minus 11% Q1 versus Q4 year-over-year will be a nice 20% growth Q1 '26 versus Q1 '25. So this is, okay, what we can see.

But after qualitatively or directionally for next year, what will be the main contributor of the growth of ST above the cycle. First one will be a massive introduction of new product and microcontroller. So more clearly, next year, because now we are inventory free in terms of adjustment on distribution, industrial, we will have a legacy growth. But more important, we will introduce basically almost 45 new products on micro next year at 40-nanometer or even 18-nanometer. So micro, which will be fitting with the demand, okay, to have more computing power, more memory storage, hardware accelerator for AI at the edge and so on. So clearly, it will be 1 positive growth driver.

Second positive growth driver will be analog. Why? Because analog and spread between personal electronics, automotive, and industrial. Why? Because here is mainly for personal electronics new product, and for industrial and for automotive, it is, let's say, moderate growth, okay, we are seeing in the market and clean of inventory correction. More than as usual for ST, [indiscernible] by satellite with our new customer ADAS system. But personal electronics will be with our main customer, more moderated. But this is okay, what we see as a growth driver. So detractor will be the capacity reservation fee that will be basically close to 0. So at the end, we are not yet in position to confirm exactly the percentage of growth we have, but we can say Q1 has a seasonality. We enter the year with Q1 20% growth year-over-year, so we can expect a growth next year.

Very clear. Very clear. Maybe take a little bit of a detour here and just ask about AI servers. Obviously, a peer is reporting this morning a doubling of sales into next year. And I know you've got ambitions to get into power supply units, particularly using, I imagine, silicon carbide, silicon carbide modules, but you have the tie-up with Innoscience as well. So maybe help us understand how that opportunity plays out for ST. When would that come in? And could that be a contributor in '26?

Well, of course, I have seen like results and guidance and communication of our competitor. Well, what is the difference between this competitor and ST. What we have, what he has not; and what he has, what we have not. Okay. What we have and what he has not, we have today, now the silicon photonics that will start to contribute with the switch from copper cable communication connectivity to optical, let's say receiver and transceiver. This, for ST, will be a growth that should bring us revenue contribution in 3 years about USD 300 million -- at least $500 million at the end of the decade.

What we have is our general purpose microcontroller that are spread everywhere in the [indiscernible]. What we have as well, but not yet present because we focus on the market like the automotive, is a switch on silicon carbide. And what we are introducing is switch on GaN. What we have not yet is a low voltage capability on MOSFET or on a smart power switch.

We are working now to do it with the next generation of 800-volt architecture. So to make a long story simple, ST revenue in server in the short term, so means starting in '26 '28, will be mainly driven by silicon photonics and microcontroller. And more marginally, I have to say, by power and analog, that we started to be boosted beyond '27, '28 by the new architecture. This is a path of ST on AI server. Well why we are not present, okay? It was strategic choice, okay, mainly power automotive oriented and less industrial infrastructure. But this is a situation we have to deal with.

Got you. I appreciate that on short notice to answer that question. Maybe if we jump back to silicon carbide. I mean we did see revenues of $1.1 billion in '24. And I think you said this is a transition year '25, but '26, maybe '27, we get back to that sort of levels eventually. But maybe help us understand the profile and the demand perhaps by region, Europe and China in particular.

Well, clearly, you know ST has been really successful in silicon carbide, mainly driven by one customer. And we started, let's say, not late, but we started in the past 2 years, okay, the diversification with European first and targeting the China market after. Well, it's clear that in '25, when you accumulate our main customer decrease and correcting inventory massively plus the European player struggling more in their capability to grow, and with a Chinese success, but with a strong price pressure, but at the end okay, 25 is a significant share of decrease.

How we have reacted in front of this situation? First of all, we reacted the case since many quarters to diversify and accelerate ourselves both in Europe and in China. So next year, we will grow significantly, so well above USD 100 million compared this year. 60% will be driven by this time, the start of the European program, and 40% by the Chinese one. Moving forward, we expect to go back to the peak of our revenue in '24, in '27 and directionally go to the $2 billion between '28 '29.

Got you. Very clear. Maybe another area that investors are interested in is the power discrete area, Power and Discrete. Operating margin was weak, I think minus 15% last quarter. We're seeing that inventory normalization coming through the second half of '26, and you've got the manufacturing reshaping program as well. So how does that all drive profitability or the regain of profitability in that space?

In consistent way, we communicate that our reshaping plan, so cost base resizing and manufacturing reshaping. Yes, the cost base, the operating expense contribute now, okay? It's $120 million less per year starting '25, '26, '27. Unfortunately on manufacturing, the contribution will be '27. Why? Because in '25, it was the year to prepare the technology transfer, product qualification, and we are not completely controlling the agenda because it's customer, we control the agenda. Entering in '26, we have to ramp down the fab we want to close, and we have to ramp up the fab we want to open, specially on silicon carbide and [indiscernible]. We will be at a period of time where the 6-inch will start to degrade their performance and the 8-inch will not be at their optimum performance.

So for sure, in '26, I don't expect power to go back to breakeven. It will happen most likely H2 '27 and come back to stabilize, okay, where power contribute to the operating margin of SP in '28 and beyond. So for sure, that '26 will be still a difficult year for power discrete overall, but for good reason. The good reason is this reshaping of the manufacturing and will be okay, limited basically fourth quarter.

Got you. And so it's really that start-up costs into 200-millimeter or 8-inch that's a headwind.

Clearly, when you are making this reshaping, when it is related to digital or mixed signal analog fab [indiscernible]. You close Crolles 200, you go to Crolles 300, you have the immediate payback because the fab is at the scale at 12,000 wafers per week. So you have the immediate payback.

When you close a 6-inch power silicon carbide or 6-inch silicon and you go in a new fab, for sure, the new fab is not yet, okay, at the optimum scale. It will be the case on Sanan and Catania. So you pay the bill on the cost of goods sold, but you have to make a lot of expenses in terms of startup. So to qualify equipment, to qualify the process, the chemicals and so on. So for us next year, it will be $100 million of start-up costs that will disappear moving forward, but that will be present in 2026. And that's the reason why we communicated during our earnings that despite the cost decrease of ST linked to our plan, we will have this offset of start-up costs, but in '27 will be gone.

One other thing we noticed on the call, I think you called out the NXP MEMS sensor acquisition that you're doing, I think, $950 million, expecting that to close in first half. So how that -- once you get into sort of a full year of contribution, how does that roll through? How does that help the profitability growth in the business?

Historically, up to this acquisition, the MEMS business of ST was mainly driven by consumer and mobile phone, I have to say. And not by automotive, even if we started automotive, and less by the industrial. The objective of the acquisition of NXP was to better balance our MEMS product group to be copy paste of ST, so to have different paths. So in automotive, industrial, mainly with robotics, okay, and to continue on consumer, point number one.

Point number 2, okay, if possible to have also benefits in terms of technology. And we started discussion within NXP 4, 5 years ago. And we decided to start the cooperation with ST making the foundry for NXP, okay, a few years ago. And then, okay, during last year, we have a common agreement that, okay, for the MEMS of NXP, it's better to be joined with ST to have a scale business unit, addressing widely all the verticals. And I am pleased to say that our mission, and the mission of ST is to double our revenue on MEMS. And with this acquisition in the next 3 years plan or 4 years, we have a strong confidence, double our revenue on MEMS.

So it's the scaling of the operation sort of squeezing out under utilization in that business area effectively.

It is not the primary objective. The primary objective of the acquisition is to balance our exposure from consumer to better balance between automotive, then robotics and consumer to have technology block acquisition. And of course, for manufacturing is basically transparent because we were already the foundry for the mechanical part of the MEMS for NXP. Then, the companionship is done in the foundry. We will not rapidly add these companionships. It's a waste of money. Okay, we don't make an acquisition and this I would like to clarify, ST will not make acquisition to load the fab. ST will make acquisition to boost the revenue and to boost the product portfolio or the technology portfolio, but not to load the fab because it's engineering bond with consumption, it's cost of transfer, it's customer qualification.

So it's a huge work by [indiscernible] program, our reshaping program. We have 2,000 device, and we have basically 150,000 product change notice to manage because of this restructuring program. So it's a huge work. So don't add the acquisition that will ask us to transfer in our fab. This is not the objective, okay, our objective is driven by revenue, profitability and technology.

Perfect. All right. That's very clear, actually. Maybe jump to microcontrollers. I mean we've got a class-leading family, STM32s. But if we look at general purpose microcontrollers more generally, it is starting to show recovery. It's starting to see that you're maybe on a path back to that historical market share of low 20s, 23%. So what do you think specifically maybe by launches or by regional take-up? What is going to drive that maybe through '26 and '27?

Clearly, the good news for ST is our inventory is good. So it means, okay, over market share and revenue trajectory will be no more polluted by this inventory digestion that has been built up between '21, '22 and first half '23 because okay, customer ask a warranty of supply. So we have put in place non-consumable order and so on and so forth. Well, this is clean, point number one.

Point number 2, the microcontroller now, they have to follow or to anticipate the request for more computing power, sometimes to be capable to have a hardware accelerator to enable machine learning or including, okay, some AI at the edge, connectivity and security features. So that's the reason why ST in '26, will introduce 45 new products, now at 40-nanometer technology and 18-nanometer technology. So I am really pleased to say that in '26, we will produce and sell the first 18-nanometer microcontroller which will be a nice machine. And moving forward, we have about 60, 70 new products in the next 3 years that will be overall introduced.

And this will be driven by the industrial market for sure, definitively. But okay, now, okay, some start in consumer because inventory is good. But one of expectations we have as a booster of growth will be the robotics. Clearly, industrial robots or humanoid robots. But we don't know exactly yet what will be the success. You have many [indiscernible] saying 1 billion of robots. It will be maybe like [indiscernible] the story. But whatever, okay, you have microcontroller, MEMS switch and driver everywhere in the humanoid robotics. And by the way, we are working with the best player, American player or including Chinese player.

So microcontroller, okay, after this terrible clear story of other inventory buildup and then correction now are coming back on the healthy pace, so driven by innovation, cost, of course, shrink of the technology and new application driven like the robotics for industrial or the humanoid for both consumer and industrial. So we plan to go back to between 21%, 23% market share next year, '26 and to go back at our historical market share in '27.

Got you. Very clear. Maybe you touched on humanoids, and we've got obviously robots out in the lobby there. Help us understand the addressable market here because if we think about AI vision, the brain and the compute that's going in there, but the huge amount of sensor array and in particular, the degrees of freedom in the hand, all of that seems to be relevant to ST. So I mean, if we go through each of those, how does that play out? Is it a microcontroller plus sensor story? Or is there something else, maybe GaN?

No, it is -- you have, for sure, the sensing -- sensing part, so the MEMS. But we have as well the optical sensing solution. So you have the 3D sensing with time of like kind of technology. Most likely, you will have some face or object recognition clearly microcontroller everywhere, and you have motor control everywhere. And where you have a motor control, you have a driver, you have a switch. So for ST, we can expect that the consumer like robots, we will have a content between $300 and $400. And for the most sophisticated humanoid industrial level, it will be more in the range of $600 to $700.

Interested. And would this include like power semis because I assume you're going to need a power inverter, onboard charging, that sort of thing. Does that -- in that 600 to 800, I think you said, does that include power semis as well?

You will have a battery. So if you have a battery -- you have a battery management system. So you will have a charger. So you will have all this feature. You need to make some move to take some weight to make some act. So you need to have power, power activator; sensor, sensor activator and microcontroller and imaging sensor. I cannot disclose the name, but we are working with the 2 best American player on robots and with the most well-known Chinese one and including some European.

Fair enough. Very good. Quite widespread already. I'll maybe pause it there just to see if there's any questions in the audience. Anybody would like to ask anything? Not at this point. Maybe we'll come back to that in a second. You've got a manufacturing reshaping program, obviously, you're talking about triple-digit millions of saving. I think as we exit '27, we're looking at that. And I think you saw 360 million -- sorry, $300 million and $360 million savings in OpEx. Can you maybe walk us through that? How does that linearly progress? And now that we're end of '25, what does it mean for '26 and then out to '27 and how it splits between COGS and OpEx? If you could just remind us how that works?

We -- on OpEx, it's pretty smooth, okay? It's basically $120 million per year, and basically, this what we deliver in '25. And it is mainly related to labor census reduction that we are managing mainly on a voluntary basis. So creating some complexity, but for the time being, okay, is working according to our expectations.

Of course, okay, there is a streamlined of some product development, okay? So we have already decided to stop some project programs because you cannot reduce sensors and especially in R&D, if you are not streamlining your R&D. And it will have also to be compensated progressively, but accelerating moving forward, but the business processes that will be automatized okay, with the AI development project we have. So here, I am pretty confident moving forward that we will deliver $120 million, '25, '26, the same and '27.

Well, on the manufacturing is another story. On the manufacturing, first of all, '25 was the year to prepare the job. So it means, I repeat, 2,000 device, 150,000 product change notice, okay, with many customers and prepare the technology transfer and the product transfer. So it was more the preparation work. Then to discuss, okay, with a representative of the personnel about the plan to have the agreement in the various region where we are operating. So it was more preparation work.

Then '26 will be the transitory period where we will have to ramp down and ramp up. And according to the customer acceptance in timing to have the ramp down because, okay, you are not protected against customer wake up and say no, I forget that I need to have this excess of inventory or blah, blah, blah. So this will be a transitory period. And that's the reason why the full benefits of the reshaping plant where you will have the 6-inch fab in Catania, Singapore, 6-inch activity into close. Agrate 200 closed, Crolles 200 closed. The full benefits will happen in '27 when the fab will be closed and the receiving fab at the scale. Crolles 300 will be at the scale. Agrate 300 will go at the scale. The most difficult one are the ramp-up of silicon carbide, Catania and Chongqing, that will be at scale end of '27. So that's the reason why, okay, on power discrete, okay, we will still struggle in 2026 in terms of economical performance.

Got you. And with all the shifting movements on the manufacturing base and your China for China strategy as well, surely has implication for CapEx. And I think you're moving closer to that sort of $2 billion level as an annual run rate. Do we see that as a sensible level for the next couple of years? Or could we see upside based on growth trajectories, for instance?

Clearly, our CapEx are mainly driven today by this reshaping program. So means 300-millimeter capacity increase in Agrate for the short term, not in full, because we are on use capacity. So we will load the capacity first. Moving forward, we will have to increase the capacity to support the growth of the silicon photonics, as an example. So the CapEx will be dedicated on it and will be dedicated on the silicon carbide 200-millimeter. This will be valid for the next 3 years. So for sure, in terms of CapEx, we expect to contain the CapEx to sales ratio, okay, well below 15%, okay, for the next 3 years. And driven by the reshaping program, our capacity will be driven by mix, okay, but just what we need, no more.

Okay. Maybe again, if we take a step back and sort of think about things at a strategic level, you have things like humanoids coming. We have the AI server story coming into play as well. And we're being very clear about how we're moving the manufacturing footprint around as well. I guess if you think about things through an M&A lens, however, you've been active in the market. You've acquired MEMS capability from NXP. Is there anything else out there? I mean, obviously, you can't give too much plans away. But what would you say here is would you -- there's a desire for end market or a JV capability that you could see that really sort of flip the switch in some of these big new opportunities coming?

I think during the past 2 years, you have seen the panel of what we can do out of the pure stand-alone organic growth and operating model because we have done this acquisition of MEMS NXP. We have done this joint venture with Sanan, with a material incentive from China. And we have done this strategic agreement with Innoscience of GaN, okay. What it brings to ST? So MEMS, in '24, this business was USD 300 million. We can expect moving forward that could represent $400 million, okay, I have to say. So additional revenue, better balance in terms of verticals and additional technology. Then the joint venture with Sanan, what it brings to us. What it brings to us is, first of all, the access of grants of capital that mitigate a lot our effort and our risk, access to local asset at a lower price. And I have expectations that, okay, it will bring to ST some recognition about to be considered as a local player, okay?

Then the deal with Innoscience, okay, at a certain moment, it enabled us to streamline our R&D effort on GaN to take benefits of the high level of know-how of Innoscience, to [indiscernible] this technology in our fab to deliver the Western world and to have the China for China strategy for free. And then we could develop -- so you see this is a panel of what we can do. But I would expect that we continue in this direction in order to do what, to continue to address, okay, all the vertical we address with a pretty wide portfolio because to sustain power analog, analog for power, analog for digital microcontroller, some [indiscernible] digital, silicon photonics and sensing. Believe me, it's a huge effort in terms of technology R&D and in terms of product development. Without the right scale and the right market share and the right partnership, basically, it's really difficult to sustain.

So now this is where we want to go. But then we have to let's say, we have still our ambition, okay, to reach $18 billion, then $20 billion, but maybe with a way which will be complementary to the pure organic growth way.

Very full answer. Maybe just staying on the point. So I think you mentioned silicon photonics in there, clearly, a growing footprint in the AI server or the data center space. How would you rank your positioning there? And how does that -- because it doesn't feel like it's full size yet. We haven't really seen all of the applications that you can address. So maybe help us understand where does this go to beyond sort of just a change to the plug-ins on the interconnect.

No, we claim that on power and analog, okay, we want to reach 10% market share, that -- of this addressable market, that -- the overall content on the AI server are addressable by ST is $2,000 of components, out of which you have $500 on silicon photonics and the rest is power analog and microcontroller.

So on this power analog and microcontroller, we want to achieve 10% of market share, okay. On silicon photonics, who are our competitor? Is TowerJazz and GlobalFoundries, okay, of course, I have a huge respect for this company, but TowerJazz is an 8-inch in silicon photonics. We are on 12-inch, okay. I don't know how TowerJazz will move to 12-inch without an agreement. So for sure, okay, on silicon photonics, I expect our market share to be well above 10%, and that will benefit to the growth of ST.

Very clear. And I think maybe just on GaN as well because I do think it's a very interesting tie-up within a science, given your pricing power, given you sort of regional prowess as well. But also, there's a systematic change in the data center. I think your interim bus conversion moving to GaN or your second stage. And that feels as though it's a market that's ready for you guys to address. Is that part of the long-term plan sort of '27 and beyond when we get [khyber] racks?

It is clearly part of our plan. It is -- I repeat, okay, compared to the big competitor in Infineon and MPS, okay, our focus using a wide-band gap technology like SiC or GaN were more exclusively driven by automotive up to 2 years ago, okay? 2 years ago, I decided to break it, okay? And that's the reason why I changed organization. I say I want to have product organization, not driven by automotive, but more broad range, okay?

So this 2 block analog power sensor and micro and digital. And to support the 4 segments in order to diversify ourselves, not because I was anticipating a disaster on automotive, but I was feeling that it is the protection of the company to be well balanced. Unfortunately, we know what happened in automotive. That will come back. But for the time being, it's a little bit struggling.

So yes, okay, we will participate this market. The point is today, we are not participating because you have already 2 or 3 players in good position. So we have to prepare the next step because if not we will come with income bond position that we will have to pay the margin for sure. So it's better to play the next step will be the 800-volt architecture and when our product will be ready, so either our SiC our GaN, but the low voltage. So the low-voltage MOSFET and the smart switch -- smart power switch that we have not yet developed because automotive focused.

Unknown Speaker.

Got you. Okay. That's pretty clear. Maybe one last one for me. Just on the gross margin guidance for Q4. I think you guided 35%, including the 290 bps headwind on unused capacity charges. I think there's an additional 30 bps. I'm reading this from a sheet of paper, so 30 bps with process migration. I could almost pass this to you. Really, just how should we think of the evolution on unused capacity charges we go through next year?

Next year will not completely disappear. For sure, we will be at least -- we expect compared to Q4 this year, at least divided by 2 in terms of negative contribution to the gross margin moving forward, mainly Crolles 300 will be normal with used cost, I guess, I got as well. Silicon carbide, we will close the 6-inch. So we'll set up just what we need in terms of capacity on 8-inch. But we will have still maybe some legacy capacity available on 8-inch that will generate unused capacity cost still Q4 '26, but I hope at least divided by 2 compare, okay, to Q4. So this is how we see moving forward, and of course will be related to the revenue we will have in H2 next year and in H1 in '27.

Perfect. Jean-Marc, I see the clock ticking down. Thank you very much for coming. Enjoy Barcelona.

Thank you. Thank you very much. Thank you.

Ladies and gentlemen, welcome to the STMicroelectronics Third Quarter 2025 Earnings Release Conference Call and Live Webcast. I am Myra, the Chorus Call operator. [Operator Instructions] The conference is being recorded. [Operator Instructions]. The conference must not be recorded for publication or broadcast.

At this time, it's my pleasure to hand over to Jerome Ramel, EVP, Corporate Development and Integrated External Communications. Please go ahead.

Thank you, Myra. Thank you, everyone, for joining our third quarter 2025 financial result call. Hosting the call today is Jean-Marc Chery, ST President and Chief Executive Officer. Joining Jean-Marc on the call today are Lorenzo Grandi, Creditor and CFO; and Marco Cassis, President, Analog, Power & Discrete, MEMS and Sensor Group and Head of ST Microelectronics Strategy, System Research and Application and Innovation Office.

This live webcast and presentation materials can be accessed on ST Investor Relations website. A replay will be available shortly after the conclusion of this call. This call will include forward-looking statements that involve risk factor that could cause ST result to differ materially from management expectation and plan. We encourage you to review the safe harbor statement contained in the press release that was issued with the results this morning and also in ST's most recent regulatory filings for a full description of these risk factors.

Also to ensure all participants have an opportunity to ask question during the Q&A session, please limit yourself to 1 question and a brief follow-up. Now I'd like to turn the call over to Jean-Marc Chery, ST's President and CEO.

Thank you, Jerome. Good morning, everyone. And thank you for joining ST for our Q3 2025 earnings conference call. I will start with an overview of the third quarter including business dynamics. I will then hand over to Lorenzo for the detailed financial overview, and we'll then comment on the outlook and conclude before answering your questions.

So starting with Q3. We delivered revenues at $3.19 billion, $17 million above the midpoint of our business outlook range with higher revenues in Personal Electronics, while Automotive and Industrial performed as anticipated, and CECP was broadly in line with expectations. All end markets, but Automotive are now back to year-on-year growth. Gross margin of 33.2% was slightly below the midpoint of our business outlook range, reflecting product mix within Automotive and within Industrial. Excluding impairments, gross recurring charges and other related phase on costs, diluted earnings per share was $0.29. During the quarter, we managed to work down inventories, both in our balance sheet and in distribution and we generated a positive $130 million free cash flow.

Let's now discuss our business dynamics during Q3. In Automotive, during the quarter, we grew revenues about 10% sequentially, in line with expectations, driven by all regions, except Americas. Our book-to-bill came above parity. We expect to grow mid-single digits in the fourth quarter compared to the third quarter, which would be the third consecutive quarter of sequential growth.

During the quarter, we continued to execute our strategy for car electrification. We had wins with both silicon and silicon carbide devices for electrical vehicle applications, such as traction inverter and onboard charger designs. On new application where we see silicon carbide being used is investors for full active suspension. Here, we have a design win with a module solution for our key Chinese electrical vehicle maker.

Another key element is a switch to electronic fuses to support zonal and domain architectures both in 12 volts and in 48 volts. Here, we added to our pipeline of designs for our eFuse controller with leading electrical vehicle makers and qualified our products for volume ramp-up.

Other wins in the quarter included microcontrollers for DC-DC management in electrical vehicle powertrain, body control modules, and HVAC systems across multiple vehicle models.

In car digitalization, we are executing our microcontroller product roadmap, with a strong lineup of new solutions across both our Arm-based Stellar and STM32A product families. Design-in activity continues globally, with engagement from both large-scale automotive OEMs and Tier1 suppliers.

In legacy application, we had several significant wins based on our smart power technologies in application where we lead such as airbags, steering, and braking solutions.

With our automotive grade sensors, we continue to see strong design-in momentum and growing opportunities. Wins in the quarter included MEMS sensors for road noise cancellation and door control, and both MEMS and imaging sensors for in-cabin monitoring.

Shortly after our results announcement in July, we announced that we entered in a definitive transaction agreement for the acquisition of NXP s MEMS sensor business for a purchase price of up to $950 million in cash, complementing and expanding our current leading MEMS sensors technology and product portfolio. The transaction remains subject to customary closing conditions, including regulatory approvals and is on track to close in H1 2026.

In Industrial, revenues were in line with expectations, showing increase of 8% sequentially and 13% year-over-year, back to year-on-year growth for the first time since the third quarter of 2023. Importantly, inventories in distribution further decreased. In Q4 we expect to grow revenues low-single digit sequentially, as we continue to decrease inventories in distribution.

During the quarter, we saw strong design-in activity for our power and analog portfolio across a range of applications. These included factory automation, power systems, medical equipment, motor control, white goods, solar inverters, and metering.

We also continued to expand the use of our industrial sensors in robotics, including robots and cobots and humanoid robots, an area where we see demand for significant numbers of sensors. We also had wins in medical devices like insulin pumps and fall detectors.

In Embedded Processing, we continued to win designs with our STM32 microcontrollers for a wide range of industrial applications, with products from all parts of the portfolio, from high-end to wireless to specialized functions. These included power supply and optical modules for AI servers, industrial automation and robotics, energy storage, home appliances, metering and white goods. We have a full pipeline of new products and software coming to market in the next quarters and you will hear more about this during our STM32 Summit in November.

For General Purpose MCU, we grew revenues both sequentially and year-over-year and we are on the right trajectory to return to our historical market share of about 20% -- 23%, sorry.

For Personal Electronics, third quarter revenues were above our expectations, up 40% sequentially, reflecting the seasonality of our engaged customer programs but also increased silicon content - which also translated into year-over-year growth.

Further strengthening our unique position as a sensor supplier with both MEMS and optical sensing solutions, we signed a new license agreement with Metalenz. This new agreement broadens our capability to produce advanced metasurface optics, leveraging ST s 300mm semiconductor and optics manufacturing capabilities. This opens up new opportunities from smartphone application like biometrics, LIDAR and camera assist, to robotics, gesture recognition, and object detection.

Revenues for Communication Equipment and Computer Peripherals were broadly in line with expectations and up 4% sequentially.

For AI data centers, we had multiple wins with silicon and silicon carbide devices for high-power solutions. Also, last quarter we announced that we are working closely with NVIDIA on a new architecture for 800V DC AI data center leveraging our power portfolio, by combining silicon carbide, gallium nitride and silicon-based technologies with advanced custom design at both chip and package level. I am pleased to underline that we recently completed full-power testing on a prototype GaN-based solution, successfully demonstrating over 98% energy conversion efficiency.

Silicon photonics is another key technology for future data center and AI factories. ST now heads the STARLight consortium, a collaborative R&D programs across the full value chain, with key suppliers and customers to develop high-speed optical solution for datacenter, AI, telecommunication and automotive, from the substrate to the final products.

During Q3, we have seen an increased demand for Photonics ICs prototypes to be launched in the next quarter and beyond in our 300mm wafer fab. This confirms that Photonics ICs will be a revenue growth driver for ST in the near-term.

In Low-earth orbit satellites, we have further strengthened our leadership position in the rapidly growing low-orbit satellite broadband market by beginning shipment to a second global customer. Leveraging our winning combination of BiCMOS technology for front-end modules and Panel level packaging for user terminals, our business in this segment is well-positioned for steady growth, driven by several satellite constellations.

This confirms that photonic ICs will be a revenue growth driver for ST in the detail. In low earth orbit satellites. We have further strengthened our leadership position in the rapidly growing low broadband market by beginning shipment to a second global customer, leveraging our combination of biosimilars technology for front-end modules and paddle level packaging for user terminals. Our business in this segment is well positioned for steady growth delivered by several satellite constellations.

Now over to Lorenzo, who will present our key financial figures.

Thank you, Jean-Marc, and good morning, everyone. Let s start with a detailed review of the third quarter, starting with revenues on a year-over-year basis.

Thank you, Jean-Marc, and good morning, everyone. Let s start with a detailed review of the third quarter, starting with revenues on a year-over-year basis.

By reportable segment, Analog products, MEMS and Sensors was up 7.0%, mainly due to Imaging. Power and Discrete products decreased 34.3%. Embedded Processing revenues grew 8.7%, mainly due to General Purpose MCU. RF & Optical Communications declined 3.4%.

By end-market, Industrial increased by about 13%, Personal Electronic by about 11%, Communication Equipment and Computer Peripheral by about 7%. Automotive was still decreasing, by about 17%, but showing some improvement in respect to the 24% decline recorded in the second quarter.

Year-over-year, sales to OEMs decreased 5.1% while revenues from Distribution increased 7.6%, back to year-over-year growth for the first time since the third quarter of 2023.

On a sequential basis, Power and Discrete was the only segment to decrease, by 4.3%. All the other segment grew, led by Analog Products, MEMS and Sensor, up 26.6%, with Embedded Processing up 15.3% and RF & Optical Communications up 2.4%.

All our end markets grew, led by Personal Electronics, up by about 40%, followed by Automotive, up by about 10%, with Industrial and Communication Equipment and Computer Peripheral up respectively by about 8% and 4%.

Turning now to profitability. Gross profit in the third quarter was $1.06 billion, decreasing 13.7% on a year-over-year basis.

Gross margin was 33.2%, decreasing 460 basis point year-over-year, mainly due to lower manufacturing efficiencies, negative currency effect, lower level of capacity reservation fees and, to a lesser extent, the combination of sale price and product mix.

Total net operating expenses, excluding restructuring, amounted to $842 million in the third quarter, broadly stable on a year-on-year. They were better than expected, reflecting notably our continued cost discipline, with the first benefits of the resizing of our global cost base.

For the fourth quarter of 2025 we expect Net OpEx to stand at about $915 million, increasing quarter-on-quarter, due notably to calendar days effect. This would lead Net OpEx for full-year 2025 to decline by 2.5% compared to 2024, despite unfavorable currency effect.

As a reminder these amounts are net of other income and expenses and exclude restructuring.

In the third quarter we reported a $180 million operating income which included $37 million for impairment, restructuring charges and other related phase-out costs. This reflect impairment of asset and restructuring charges predominantly associated with the previously announced company-wide program to reshape our manufacturing footprint and resize our global cost base.

Excluding this not-recurring item, which is partly non-cash, Q3 non-US GAAP operating margin was 6.8%, with Analog Products, MEMS and Sensors at 15.4%, Power and Discrete at -15.6%, Embedded Processing at 16.5% and RF & Optical Communications at 16.6%.

This quarter 2025, the net income was $237 million, compared to $351 million in the year-ago quarter. Diluted Earnings per share were $0.26 compared to $0.37. Excluding the previously mentioned non-recurring items, Non-US GAAP Net Income and diluted Earnings per share were respectively $267 million and $0.29.

Net cash from operating activity decreased 24.1% on a year-over-year basis in the third quarter to $549 million. Third quarter Net CapEx was $401 million, compared to $565 million in Q3 2024.

Free cash flow was a positive $130 million in the third quarter, compared to $136 million in the year-ago quarter.

Inventory at the end of the third quarter was $3.17 billion, a reduction of about $100 million compared to the end of the second quarter. Days sales of inventory at quarter-end were 135 days, slightly better than our expectation, compared to 166 days for the previous quarter and 130 days in the year-ago quarter.

Inventory at the end of the third quarter was $3.17 billion, a reduction of about $100 million compared to the end of the second quarter. Days sales of inventory at the quarter-end were 135 days, slightly better than our expectation, and compared to 166 days for the previous quarter and 130 days in the year-ago quarter.

Cash dividends paid to stockholders in the third quarter totaled $81million. In addition, ST executed share buy-backs of $91 million.

ST maintained its financial strength with a net financial position that remained solid at $2.61 billion as of End-September 2025, reflecting total liquidity of $4.78 billion and total financial debt of $2.17 billion. It is worth to mention that in the course of the third quarter we repaid, fully in cash, $750 million for the first tranche of our 2020 convertible bond.

Now back to Jean-Marc, who will comment on our outlook.

Thank you, Lorenzo. Let s move to our business outlook for Q4 2025. So we are expecting revenues at $3.28 billion, an increase of 2.9% sequentially, plus or minus 350 basis points. We expect our gross margin to be about 35.0%, plus or minus 200 basis points, including about 290 basis points of unused capacity charges. This business outlook does not include any impact for potential further changes to global trade tariffs compared to the current situation.

The mid-point of this outlook translates in full year 2025 revenues of about $11.75 billion. This represents a 22.4% growth in the second half compared to the first half, confirming signs of market recovery. Gross margin for the full year is expected to be about 33.8%.

Finally, to optimize our investments in the current market conditions, we have reduced our Net CapEx plan, now slightly below $2 billion for full year 2025 compared to a range of $2 to $2.3 billion previously.

To conclude, in the fourth quarter we expect to report further sequential revenue improvement, with revenues now broadly stabilized on a year-over-year basis, as well as an increased gross margin, while continuing to decrease inventories in distribution. We are on the right path to improve our gross margin in the medium-term through the reduction of unused capacity charges, the reshaping of our manufacturing footprint and definitively out product mix improvement.

In a context marked by sign of market recovery, our strategic priorities remain clear, accelerating innovation; executing our company-wide program to reshape our manufacturing footprint and resize our global cost base, which remains on schedule to deliver the targeted savings, and strengthening free cash flow generation.

Thank you, and we are now ready to answer your questions.

[Operator Instructions] The first question comes from the line of Francois Bouvignies from UBS.

My first question is on the top line. I mean, you guided plus 3% quarter-on-quarter, 2.9% to be precise. It seems to be below your seasonal at plus 7% quarter-on-quarter, if I'm not wrong. I mean you can remind us maybe the seasonality. Can you explain us as to why you are a bit below seasonal in Q4 for the top line and the drivers?

And then secondly, on the gross margin, I mean, it's nice to see this improvement of 180 basis points quarter-on-quarter, how sustainable it is this gross margin? I mean, if you have any seasonality, product mix, should we extrapolate this dynamic of 35% into the first half of '26? Just trying to understand the work you have done on gross margin, how sustainable it is at least in the first half of '26 would be great?

So we'll take the revenue seasonality and Lorenzo, the gross margin. No, on the revenue seasonality of Q4, basically, there is two effect. The first effect is on automotive. Because on automotive, even if we will grow on a quarter-over-quarter 6%, but year-on-year, it is still minus 12%. And why? Because, okay, 80% of this performance gap is explained by 2 reason. It is a decrease of our capacity reservation fees compared to last year.

And you know it is overall volume of one important customer of ST in the field of electrical vehicle. So this is what is explaining why in Q4, we are below the seasonality. The second explanation to be below the seasonality in Q4 is because in Industrial, we continue to decrease inventory in distribution. So our POP revenue recognition is significantly below the POS. However, on the other, let's say, verticals like Personal Electronic, Communication Equipment, Computer Peripheral and other legacy on Automotive or Industrial in the field of power, energy; basically, okay, we are at the seasonality we expect.

About gross margin. In Q4, the gross margin, the main positive driver, let's say, when we look at the sequential increase of our gross margin moving from the result of Q3 and the expectation of Q4 is clearly improved manufacturing efficiency. That is -- if you remember, let's say, in the first half and also in Q3, we were impacted by a significant negative impact on the efficiency -- manufacturing efficiency that was due to the very low level of production that we have, especially in the first half of the year. There is also some improvement in terms of new charges.

When we look, let's say, to how we will move moving in the first half of next year, but we have to remind that clearly, there are two negative effects that will be impacted moving forward. The one effect is related to the fact that there will be some reduction entering 2026 of capacity reservation fees. And definitely, you know that in the first half of the year, there is some seasonality in term of our revenues, let's say, in respect to the second part of the year. And then don't forget that there is also the renegotiation of the pricing that will impact even if we see today [indiscernible] significant drop.

We think that it will be something in the range of low single digit, mid-single-digit decline. On the positive side, we will have, let's say, still continued positive impact on manufacturing and reduce -- continue reduce level of unsaturation. At this stage, it's a little bit too difficult to size, let's say, the level of gross margin because it will depend also on the level of the revenues. But this is directionally the trend that we will have moving -- entering in the next year.

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

Maybe to follow up on that last one. Could you maybe spend a couple of minutes talking about how you're thinking about managing utilization rates right now? It seems like you're taking things back up. Are you at the point where you feel comfortable building a little bit of inventory downstream and/or on your balance sheet given the comments. You mentioned you're going into some negative seasonality into 1Q, but it sounds like utilization rates are going to be up in the fourth quarter and the first quarter. Could you maybe just spend a couple of minutes talking about what you're seeing there?

Now for the inventory, clearly, let's say, as you have seen, we try to keep control on the level of inventory in the current quarter, we think to stay substantially stable in number of days. This is our expectation in respect to Q3. But the positive point is that entering in the next year, clearly, let's say, as I said, there is our seasonality, the normal seasonality that means that, in general, the inventory in the first half of the year is a little bit higher also in number of days in respect to the second part of the year.

Then you have to consider that entering next year, let's say, we start to have some decrease in term of overall capacity, linked to the fact that we started to have some benefit coming from our reshaping of the manufacturing infrastructure. This will somehow mitigate the level of unused moving in 2026. This is, let's say, one of the drivers that we see in terms of progressively improve in terms of the utilization rate, together, of course, with some growth in term of revenues.

I was hoping to ask about the Industrial segment. So it looks like book-to-bill went back to parity. Anything major going on there? Any geographies that are better or worse? And maybe how would you categorize the health of the general purpose microcontroller business underneath there? Basically, should we assume or sort of shipping back to normal now?

No. In industrial, we see a different dynamic when we grow on some subsegments. We see a growth and dynamic more pronounced for power energy, basically all subsegments, okay, of this one are growing. And it is growing more definitively than the smart industrial, it means the factory automation. We can say that robotics is so far good, but overall, the factory automation is really, really soft. More than all the industrial, which are volume-driven, means consumer-driven, the hub cycle is pretty soft.

So the takeaway we can have on the Industrial is what is related power energy infrastructure and robotics is now upcycle pretty solid. What is related volume and consumer is a very soft upcycle. It looks like inventory are digested, but the visibility is pretty short, it's pretty low. So that's the reason why the customers are still putting order on short term.

But here, our decision is to continue to manage the distribution very closely and continue to adjust our POP below their POS forecast to continue to decrease inventory. Inventory and general purpose microcontroller came back what we classified normal, means a level of months of inventory that enable short-term business. Well, we have still some pockets of other inventory on some specific products like Power & Discrete or sometimes general purpose microcontroller, but we are going in the right direction. So this is a dynamic, okay, we are seeing on the industrial market.

The next question comes from Tristan Gerra from Baird.

I wanted to see how linear is the reduction in capacity reservation fees that you expect in '26 from the $150 million, $200 million reduction that you're looking at for this year. Is there a big drop in Q1? Or is it going to be pretty linear throughout all of next year?

In terms of capacity reservation fees, it works in this way, let's say, substantially, the capacity reservation fees that are ruled by contract with the carmakers are quite constant over the year in term of million dollars. But yes, you can have a little bit higher, a little bit lower during the various quarter of the year, but they are not linearly going down. Let's say, they are substantially quite flattish, I would say, quarter after quarter. Clearly, when the contract expires, that is, at the end, for instance, of 2025, many of these contracts are expiring.

But then, yes, you have the decline. And then the decline remains the level that you get in the first quarter will remain substantially similar all over the other quarters. So this is the way that it works. So what we will see in Q1 will be this reduction? And then that -- after that, we will stay stable, more or less stable during the course of 2026 at the level of capacity reservation.

Just a quick follow-up. Of course, it's going to depend on end demand, but any sense of -- or when you think POP can get back in line with point of sales in Industrial next year?

Globally, POP will be aligned with the POS each time our product line reach the target of inventory, we didn't want to exceed. This is okay, a lesson we learned from the past. And now, we are really disciplined on this point. So you cannot see the POP overall. We have to look the POP in detail by product line. And I repeat, our microcontroller is pretty well aligned. So our POP is really driven by the end demand POS and by region, I have to say. While China, APAC, America are pretty okay, but Europe is still soft. And for the other product line, okay, we are still in a mode where the POP is below the POS; however, we expect to go back normal in H1 2026, most likely Q2.

Next question comes from Stephane Houri from ODDO BHF.

Yes. I have a first question about the CapEx budget because you're adjusting downward the CapEx for the end of this year. I guess this is in the course of managing your capacity by the end of the year also an expectation of 2026. But what are you reducing at the moment? And how do you look at 2026 in term of CapEx at the moment where you're transforming your tool from 200-millimeter to 300-millimeter?

No, we reduced the CapEx. In fact, there is 2 dynamics. There is a dynamic driven by [indiscernible] where you know we want to close the 200-millimeter [indiscernible], so I got an hold. And of course, okay, we need to put the CapEx to increase the capacity at the right level, and I got [300-millimeter] and [indiscernible]. But here, we have not especially limited the dynamic because the demand is pretty solid. But then the other main important action is the CapEx for 200-millimeter conversion on silicon carbide because we will close the 150-millimeter. But here, we have limited the CapEx delivered by the demand, which is below what was -- we expected 1 year ago.

So the main impact of the capacity limitation is on, let's say, silicon carbide. But then after it's more spread across test assembly, where we clearly adjust the capacity of what we needed the more and generally speaking, is more adaptation to mix rather than volume increase.

Just to ask you, if -- with the Nexperia situation, you do receive phone calls or kind of rush orders from your customer? Or you see nothing for the moment?

No, I mean, we are sure that the carmaker has [indiscernible] Tier 1 of the automotive industry have clearly taken the lesson of the previous shorter period, and they have enabled many source to prevent such issues. And of course, okay, as the other semiconductor player, STMicro is part of this process. More than that, I have no comment.

Next question comes from Didier Scemama from Bank of America.

I have first question maybe on your inventory and related to that, on what you're thinking about in terms of factory loadings for the first half, I think, one of your U.S. peer already announced last week or earlier this week that they would reduce factory loadings to reduce inventory, especially in the context of a shallow recovery? So I think it looks like your inventory are tracking about 30, 40, and 50 days above where they used to be. So are you thinking about taking down further factory utilization in the first half, I guess.

In term of inventory, I would say that, yes, you're right, it's a little bit higher in respect to what was our historical ending of the year, that is a little bit higher. But at the end, I think that when we look next year, I think the dynamic of our -- we will continue to keep under control the inventory. The dynamic of the inventory will, let's say, be, as usual, a little bit increasing during the first half of the year to go back and to decrease in the second part of the year.

In term of that, let's say, unloading factory utilization I think that moving in 2026, there will be an improvement. Notwithstanding, we will continue to keep under control our inventory. This improvement that I was saying before is due to the fact that we do expect some, let's say, increase in term of our revenues, so looking at the evolution of the market. And the other element is that we start to, let's say, reduce capacity in some of our fabs. The one that we aim, let's say, to progressively close in the course of the -- by the end of 2027.

So we will start, of course, to move out some equipment, and this will reduce the capacity, and this will reduce the level of unused then.

And then I think last quarter, you said that the gross margins were impacted by, if I remember correctly, roughly 70 basis points of the 140, at 70 basis points of FX headwinds and 70 basis points of related to basically the manufacturing transition from 6 to 8 and 8 to 12. Is there any of that in Q4?

No, no. Let's say, moving from Q2 to Q3, let's say, the FX was overall an impact of 140 basis points. Q2, Q3, let's say, related to the combination of these 2 effects, but very different. Let's say, something in the range of 120 basis points was the FX and around 20 basis points was the impact of these extra costs, let's say, related to our programs. Now, let's say, in this quarter, clearly, the FX is a minor impact because it s quite stable. It's a little bit negative because we moved from 114 to 115, isranging in the range of 20 basis points negative impact.

It's not so material, while these extra costs related to the activity to reduce the capacity and to start to move products from one side to the other is impacting our gross margin expected for Q4 between 30 to 40 basis points. This is -- impacted by something ranging between 30 to 40 basis point of extra cost.

Understood. And just a clarification, because it wasn't clear, your OpEx guide for Q4 is 915, right? It's not 950?

No, no. It's 915. And this is driven by the fact that we have a negative calendar days impact for 2 reason. The calendar is longer. And that the vacation in Europe is, let's say, lesser than what we benefited in the course of the previous quarter. On the other side, we will continue with our, let's say program to reduce account in expenses, and this will bring us some benefit.

The next question comes from Sandeep Deshpande from JPMorgan.

My question is regarding the trends into the first quarter. I mean, you normally have a weaker first quarter. And thus would you expect the utilization rate to go down? And given all the other factors you've talked about in the earlier factors, which were there, there is a downtick associated with the capacity reservation fees. Should we expect that your gross margin in the first half of the year to be weaker than where it is at the moment? And I have a quick follow-up after that.

Yes. In term of gross margin, it's true that in the first half, the seasonality is not favorable. And yes, there are the lower capacity reservation fees. On the other side, in respect to where we stand today, our expectation is that the level of a new charge will decrease. The decrease is not due to the fact that we aim to increase our inventory. There is some seasonality in our inventory, but the decrease, as I was trying to explain before, it's mainly driven by the fact that we start to reduce the capacity. So it means that we will start to some transfer of equipment.

And this or, let's say, not utilization of equipment due to the fact that we progressively in some fab, we started to reduce the capacity aimed, at the end, let's say, to move -- to close this spec. So we will start, and this will progressively impact our capacity and, for some extent, our unused capacity.

I mean, a follow-up to that essentially -- quickly on that would be, is the number of days in Q1 lower than Q4 or is it anything different. And my main question is about 2026 overall, I mean, on the revenue, do you have any new engaged programs with your customers, which will improve revenue significantly either in first half or into the second half, particularly?

No, I confirm, Sandeep, that in Q1, Q1 will be shorter in terms of number of days, than Q4, Q4 is longer in term of days than normal 91. And the calendar next year, Q1 will be shorter than the normal 91. It's a little bit the same trend that we have seen this year, let's say, in term of calendar. So yes, I confirm that there is a shorter calendar in Q1.

Well, first of all, okay, about next year 2026, Q1. With the current visibility, we have of the loading of the backlog we have seen in Q3 and we are seeing today. But we don't see a specific reason why we will not be at the usual seasonality of Q1 revenue versus Q4. This is generally speaking, really slightly above minus 10%.

Well, then moving forward, of course, we will -- but it's depends on the market dynamic. But I would like to say that for 2026 -- well, first of all, okay, in the second half, we will clearly see the normalization of inventory everywhere. We really expect that in H2 2026, we will have no other inventory, point number one.

Point number two, next year compared to 2025, the silicon carbide will be a year of growth because 2025 is a year of transition where basically, okay, we have cumulative headwinds related to one specific customer, some program not going at the expected speed in Europe and you know we are not specially still present in China. But -- next year will be a [indiscernible]. Then after we have our exposure to fast-growing segment. Clearly, that already give us a sign of growth like ADAS with our main customer that already provided some let's classify upside and MEMS as well.

And definitively, one point is our increasing content in terms of value and silicon in our main customer. So all in all, we do believe that Q1, we have no sign that the seasonality will be impacted by other factor that we do not control. And in H2, we will be as well as usual seasonality of growth H2 versus H1. Do we grow more like here because this year, we grow at 23% and the usual seasonality is 15%, H2 versus H1. Well, here, we need to have a little bit more booking in Q1 and in Q2 to confirm.

So my takeaway is, yes, we will have, let's say, idiosyncratic growth driver on top of the, let's say, up cycle of the market that we are seeing today even if this up-cycle market of automotive and industrial should be classified at this stage, soft, okay? And with subsegment pretty dynamic like the one related to infrastructure.

The next question comes from the line of Janardan Menon from Jefferies.

I just wanted to go back -- go to the Power & Discrete business where your margins are still very weak at minus 15% in the third quarter. So what can be the drivers to improve that? You talked about silicon carbide improving in Q3 -- I'm sorry, in 2026. But would that revenue come mainly from your Sanan JV to Chinese customers? And will that help your overall profitability given low utilizations in Europe? And do you need to take any further action to try and improve the profitability there in Power & Discrete, given the kind of competitive environment in that industry?

And then my follow-up is just a small clarification on a previous answer. Your 30 to 40 basis points of manufacturing inefficiency from the conversion and shutting down, et cetera, does that continue till you reach the end of that journey, which is when you fully close down your 200-millimeter transition to 300 millimeter? Or does that drop off before that?

So Lorenzo will comment about the improvement driver of Power & Discrete profitability. While Marco will comment on the dynamic of Power & Discrete revenue because as I have already anticipated, in my last answer, clearly, silicon carbide for us in '25 is a transition period. And then Lorenzo, [indiscernible]

Yes, I can take it. Clearly, well, I will let Marco to explain what are the drivers. But at the end, let's say, clearly next year, we do expect a recovery in terms of the top line that is -- this year, we were impacted by a significant inefficiency in our manufacturing environment for the Power & Discrete in general and for the silicon carbide, in particular, due to the fact that we were working a very low level of saturation for these steps.

Clearly, there are the following drivers that we expect to recover in term of profitability. Having a higher level of revenues clearly will help to better load our infrastructure. Then don't forget that silicon carbide, it will be the first to move, let's say, in the course of next year from the 6-inch to the 200-millimeter to the 8-inch, and this will bring clearly, let's say, some positive in the medium term in term of profitability.

Moving up in term of revenues, will improve significantly our expense to sales ratio that today clearly has been impacted by the fact that revenue are quite depressed. So at the end, these are the main drivers that we see together with the fact that we are improving, and we are moving to the next generation of silicon carbide that give also some benefit in terms of performance for what concern, let's say, the profitability.

Before to give the -- to pass to Marco, I just clarify the point of this extra 30 basis points on gross margin. Yes, this is mainly related to the duplication of mask related to the, let's say, qualification of processes. But this will continue, the amount will be more or less this range over, for sure, the next part of 2026 and probably also in the second part because we will continue with this program. This will be probably peaking in the first half 2026 then it will go down.

But yes, this is something that we need to expect to have -- as we have this activity, let's say, to migrate our products from one fab that is going to be closer to another fab.

Okay. So we take on the dynamics. So we'll have basically 2 dynamics in 2026 that will help to start to grow. First of all, well, as Jean-Marc has said, during the first half of 2026, we will keep reducing and will be clean in terms of inventory in Power & Discrete; here, speaking mainly about the nonsilicon carbide portion. And this will allow the market dynamics next year to restart having a year-over-year growth.

Specifically, on silicon carbide as Jean-Marc has already anticipated, 2025 is a transition year, meaning is that we are experiencing lower volumes and inventory collection from our main customers. I would like to underline, this is happening while we still are maintaining stable our commercial contractual level of market share. This is happening since the beginning of 2025. And during 2025, we are -- these dynamics is not yet offset by Europe and China.

So there is yet no strong contribution from electrification from the electrification programs in Europe and China. During the next year, we will start seeing growth in these 2 regions that will help the 2026 overall growth of the silicon carbide versus 2025. I hope that this answer your question.

Thank you, everyone. This is ending our call for this quarter. So thank you for being with us today, and we remain here at your disposal should you need any follow-up questions. Sorry for the one that you don't have time to ask a question there. Thank you very much.

Ladies and gentlemen, the conference is now over. Thank you for choosing Chorus Call, and thank you for participating in the conference. You may now disconnect your lines. Goodbye.