STMicroelectronics N.V. (STMPA) Earnings Call Transcript

Ladies and gentlemen, welcome to the STMicroelectronics Capital Markets Day 2020 Analog, MEMS and Sensor Group Presentation conference call and live webcast. I am Alessandro, 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 Celine Berthier, group Vice President, Head of Investor Relations. Please go ahead.

Good afternoon, and good morning, everyone. Thank you for joining what is now the third session of our Capital Market day 2020 the presentation of our Analog, MEMS and Sensors Group, AMS. Jean-Marc Chery, ST's President and Chief Executive Officer, will start with introductory comments. Then Benedetto Vigna, our President in charge of AMS, will go through the presentation of his group. It will be followed by a Q&A session where Jean-Marc and Benedetto will be joined once again by Lorenzo Grandi, our President of Finance, Infrastructure and Services and Chief Financial Officer; and Marco Cassis, our President of Sales, marketing, Communication and Strategy Development. As we did for the last session, we will use as the Q&A using audio only. As usual, a few housekeeping items before we start. The slide webcast and presentation materials can be accessed on ST's Investor Relations website, and a replay will be available shortly after the conclusion of this event. This presentation will include forward-looking statements that involve risk factors that could cause ST's result to differ materially from management expectations and plans. We encourage you to review the safe harbor statement 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. With this, before I turn the floor over to Jean-Marc, let's start with a brief reminder of our key strategic objectives. [Presentation]

Good afternoon, or good morning to everyone. I am pleased to introduce our last product group strategy update with the Analog, MEMS and Sensor Group. AMS is one of the main contributors to our industrial and personal electronic strategic objectives and with a more focused approach to automotive and communication equipment, computer and peripherals. Its contribution is clear also looking at the 2020 revenue trends, 12% higher revenues in the first 9 months compared with the same period last year and this despite the pandemic that hit the industrial market very hard. Among our 3 product groups, AMS is a broadest exposure to hold the full end market we address. Since we covered the automotive market dynamics extensively a couple of weeks ago, let me provide you with a short overview of other 3 end markets. During Q3, the industrial market dynamics remain mixed, but gradually improved during the quarter. This trend was visible across all geographies. Demand was strong for power tools and home appliances, we continue to see positive dynamics in power-related applications, motion control and factory automation, all of which are focused areas where ST and specifically for AMS. Distribution is an important element of our go-to-market strategy in industrial. Here, the improvement of the situation in the channel is continuing to accelerate. In Asia, point-of-sale trends remained strong in Q3, up sequentially and year-over-year, with healthy levels of inventory in our distribution channel across all product families. In the Americas and Europe, recovery is still ongoing with point-of-sale up sequentially. Here, also inventory is back to [indiscernible] levels across all product families. In Personal Electronics, during the third quarter, there was a strong restart of consumer demand for smartphones and combined by steady growth in wearables, tablets, hearables and game consoles. This was driven in part by the stay-at-home effect and by consumer demand for health and fitness devices. In communication, equipment and computer peripherals, during the third quarter, we continue to see growing demand for homeworking-related products, while the demand for [indiscernible] drive was softer. In Benedetto's presentation, you will see a number of technology, product and application-related innovations, supporting the mega trends underpinning our strategy. From fully autonomous sensors, providing artificial intelligence at the edge -- of the edge for the industrial IOT. To optical sensing solution for direct and indirect time-of-flight for personal electronics. From smart mobility solutions with Motion MEMS and LIDAR technologies, to leading edge power and energy solutions for server, motors and wireless charging. You will also see how AMS is well positioned for balanced growth in both distribution and with key customers. Thanks to, first, its industry's broadest portfolio in MEMS and optical sensing solutions. Second, a differentiated and competitive portfolio of application-specific analog products, complemented by general purpose and alloids, addressing all ST's target markets. Third, a diversification strategy across customers and end markets, enabling profitable growth. And now let me hand over to Benedetto. Thank you.

Thank you, Jean-Marc. Good afternoon, and good morning to everyone, and welcome to this presentation on AMS, the Analog, MEMS and Sensor Group. This year, we will not have the pressure to meet in person. But today, we can still be together, thanks to various technologies, many of which come from ST. I really hope that this digital experience will be equally effective. One year is gone by since the last time we met, and I'm happy to confirm you that we are in line with the plan presented during the last Capital Market Day. How did we do? I would summarize by saying that we maintained continuity of our strategic direction with continuous improvement in realizing and enhancing our unique value proposition, fully aligned with ST strategic goal. There are 3 points I'd like to address in this talk about AMS. First, a few financial figures, where I'm pleased to confirm that the 2020 business trend for AMS is good despite the pandemic; second, our strategy to continue to grow based on our distinctive product innovation DNA with the double helix of differentiated technologies and application know-how and focus; third, our product strategy to match the growing needs of each of our end market. Let's start with the financial. AMS is made up of a team spread throughout the world, and it represents about 1/3 of ST revenues. Despite the pandemics, I'm pleased to confirm that the 2020 business trend of AMS is positive. In fact, the first 9 months of 2020, AMS revenues were 12% higher than year-on-year. In this period, we built around $2.5 billion out of which $1.5 billion in MEMS and Sensors and $1 billion in Analog. In Analog, we grew less than in MEMS and Sensors. Why? Because on general purpose product, we experienced the negative impact of the pandemic on our mass market business. Our growth is driven by 2 underlying trends that are fundamental to the 3 mega trends that ST is focused on. The first is the sensorization of the world around us that enable analog signal to be converted to the digital data, bringing the real-world into the digital world. The second is the smarter and more efficient energy use. In the past 5 years, our strategic direction has been consistent. Product-wise, we have 2 pillars: Analog and MEMS and Sensors. In analog, we address 2 kind of products: first, application-specific product, mostly driven by key customers; second, general purpose product for the mass market. These are the glue of any electronic system. We have a special focus on signal conditioning like op-amp and voltage reference and on power management, like DC-DC and AC/DC converter. In MEMS and Sensors, we offer both MEMS sensors and actuators and optical sensing solutions. Market-wise, we address all the 4 ST end market. We have a strategic focus on industrial and automotive, with our sensors, general purpose analog, connectivity and power management solutions. In Personal Electronics, we are pushing high volume opportunities with optical sensing solution, MEMS and power management. In communication equipment and computer peripheral, we are very selective in our approach with our portfolio of sensors, micro activators as well as power management for server and storage devices. With all this product, we serve about 70,000 customers from large OEM to smaller customers that we address through the distribution. After this strategic overview, let's go now to the 2 AMS product pillar, and let's start with our MEMS and optical sensing solution. Optical sensing solution and MEMS are more and more pervasive in our industry and in our lives. They are key for consumer and for industrial IoT, for artificial intelligence for 5G networks, for biometrics and for the ADAS system in car today as well as autonomous car of the future. They are important today, and tomorrow they are going to unlock additional revenue streams. Indeed, thanks to this product, real-world signals like vibration, sound, pressure, proximity and ambient light we reach the digital world for data processing without the need for a keyboard, mouse or touch display. In the past 15 years, we sold more than 5 billion MEMS actuators and more than 20 billion sensors, both MEMS and optical sensing solution. We are, by far the world leader in time-of-flight sensor, a motion MEMS for Personal Electronics and automotive telematics. We started from Personal Electronics that still represent the largest market for us. And we then expanded into the other market. 5 years ago, we started our diversification journey, along with 4 vectors: technology; product; application; and market segment and it is yielding results according to our plan. Now after this, let's focus on our MEMS business, an area where we have 2 solid leadership position. Number one, in motion MEMS for Personal Electronics and automotive telematics with a 50% market share and #1 in micromachine actuators for inkjet printing. Underlining this MEMS success, we have unique and flexible, differentiated technologies such as [indiscernible] that we have been developing over the last 15 years. They enable us to develop a broad product portfolio that is much larger than when the MEMS consumerization wave started. At that time, the human machine interface was the only [indiscernible] application. We are determined to keep this #1 position, supported by the development of leading-edge products such as accelerometer, 6-axis, inclinometer and pressure sensors we announced earlier this year. Our strategy is to continue to focus on the development of ultra-low power and higher accuracy sensors. For this, we are developing innovative sensing architecture, and we are embedding our latest generation machine learning core. In sensor computing, helps to reduce the overall power consumption substantially. We are talking about a power reduction of up to a factor of 100. Just look at our late 6-axis announced last week. The sensor, in full autonomy, I repeat in full autonomy, will be able to process raw data and to deliver to the processor only the relevant information. This is artificial intelligence at the edge of the edge. And it works in complete sync with the overall AI strategy of ST. After the MEMS world, let's talk about our optical sensing solution. Last year, when we met in London, I told you we were working on 2 axes: first, maintain our #1 position in Time-of-Flight; second, diversify our technology, product and customers. I am glad to share with you that also in this case, we are on track with our plan. On the technology side, as you know, we are developing all the technologies needed for structured light, direct/indirect Time-of-Flight and stereoscopic approach. We made significant progress in global Charter, fast photodiode and SPAD technologies. Today, we have a 14-nanometer SPAD pixel for direct time-of-flight in production, and we continue to invest in SPAD technology development. We also made significant progress in BSI technology to increase the quantum efficiency to shrink the pixel size. On the indirect Time-of-Flight, we transitioned to 4.6 micron pixel, and we plan to decrease even further. In global Charter, we are also moving from 2.6 micron to 2.2 micron pixel. These all represent important progress in a challenging field. On the product side, we continue to enhance the performance of our FlightSense family and ambient light sensor. In FlightSense family, we increased the detection distance, and the field of view went from 27 to more than 60 degrees, a big, big step forward. For ambient light sensing, we had the win with a major OEM in the field of Personal Electronics. And here, we displaced the strong competition. After technology, after product, the customer side. Here, we started to deliver our multi-zone 40-nanometer SPAD product to Samsung. We've been able to differentiate from major competitors with this unique product. It offers high performance and high level of integration. This recent success is a big, big step forward in the world-facing application. It complements very well our leadership in the Android market with SPAD for laser-assist out-of-focus application and in laptop with our presence sensor. Today, Personal Electronics remains still the main market for optical sensing solution. Diversification takes time, does not come so simple, but we are on the right track. Indeed, you can find our Time-of-Flight product in PC for present detection. And next year, you will find our global shutter technology in the car for in-cabin monitoring. But now let's talk about sensor for automotive more in detail. Automotive is a world where high performance and ultra-low power sensors are extremely important. In the car, whether electric or traditional combustion engine, you need an increasing number of sensors, such as accelerometer, gyroscope and lidar, but also in-cabin infrared sensor. In the past years, AMS has made good progress in automotive. First, we maintained our #1 position in motion MEMS for telematics, and we expanded our motion MEMS business in safety application such as vehicle stability. In line with our plan, we grew significantly in one major OEM. We also see good perspective for future growth because we are not competing with our customers. We do not sell module with sensors like some MEMS competitor do. Second, we believe that lidar will become more and more important for ADAS and autonomous vehicle system. Different technical solutions will be used in the market. That's clear. So on one side, we announced last April, a customer partnership with [indiscernible] tech to develop a miniaturized ultra-low power lidar using scanning MEMS mirrors. On the other side, we are developing solutions based on our latest generation of a FlightSense technology using SPAD in 40-nanometers. Lidar will enrich the information flow in the car to the advanced processing unit, such as those developed by our colleague in Automotive and Discrete Group. Third, we won a business for a driver monitoring system application, supporting legislation in various countries, especially in Europe from 2022 onward. Our global shutter sensors offer dynamic range and the low noise required by in-cabin optical sensing. Here, we are working with multiple OEMS, Tier 1s and partners such as specialized software companies. We are engaged in multiple programs with key OEM and sense forecast to start next year. Whatever the technology the customers select, we are not competing with them. And this is a key competitive advantage. All of these sensors, motion MEMS, global shutter, image sensor and lidar are also very important for industrial application. Can you imagine the benefit sensor will bring to the industrial world? Indeed, in these days, we see a lot of emerging application for industrial IoT, asset tracking, predictive maintenance, real-time monitoring and metering are word that are in our daily jargon. A few years ago, these were just words known only in academic circles, many papers, 0 business, today, this is different. We are talking about application to optimize the supply chain, to track the asset to fully digitalize the factory and to meter critical resources, such as electricity, water and gas. For all these applications, AI at the edge of the edge and connectivity with or without wires, play a fundamental role. Here, we have multiple solutions to satisfy the various needs, and we made the progress with technologies like Bluetooth Low Energy, SubGiga radio, PLM, KNX and IO-Link. We also start to see sensors used for structural health monitoring. Several years ago, some universities start to use our accelerometer for vibration monitoring of the Milan football stadium. At that time, power consumption was way too high. Now thanks to ultra-low power consumption, we see sensor mounted on elevators, wind mills and on critical infrastructure, such as bridges, dams and tunnels, but this is not all. Industrial application can also benefit from the adoption of new technologies originally conceived for entertainment and gaming. I'm talking about augmented reality technology, designed to merge the digital with the analog world. One key enabler is small, high resolution and low power MEMS mirrors. These tiny silicon devices can pave the way to new services in ways to work socialize and interact. We are already in production with electrostatic and electromagnetic mirrors, and we are ready to launch in 2021, the next-generation based on piezoelectric mirrors. Together with 4 partners, we also launched the LaSAR Alliance to shorten time to market for augmented reality eyewear. You can find already our MEMS mirror in 2 products on the market, HoloLens from Microsoft and RealSense from Intel, 2 products with 2 different use cases. Our mirrors in Microsoft HoloLens can help a surgeon to be more precise in patient-critical surgery or help a fab manager to plan better in advance for an optimal layout of all equipment he has to install in his fab. Our mirror could also help us to feel closer to our colleague when we find ourselves remotely working because of the pandemic-related threats. Not the same feeling as the face-to-face situation. But for sure, much better than the current one. Our mirror will enable new applications in the industrial world by adding the third dimension to the digital world of today. Okay. So far, we talked in detail about MEMS and sensors. We talked about the product, the technology, the market. Now let's move to the second pillar of this group, the Analog. First, I will cover general purpose analog and then application-specific product. General purpose analog product are very important for us. Since with this product and together with our distribution partner, we are addressing the mass market with a special focus on the industrial segment. Our product are used for power management, interfaces, signal conditioning. Because of pandemic, in the first 9 months of this year, we experienced a revenue drop year-over-year, but we are definitely on a positive trend in the last quarter. And this is even more true for our application-specific product. Here, all of the key business programs from May 2019 are well on track. And in addition, I would like to highlight 3 new product families we recently announced. First, the MasterGaN product family already in production, allowing to shrink the size and the weight of power adapter for PC, a smartphone by a factor of 4. This is possible thanks to the intrinsic benefit of the gallium nitride and our unique BCD technologies. We are at the dawn of a new phase in power conversion, and we want to play a leading role here. Second, power management chips for our STM32 microprocessors. These devices offer high efficiency DC-DC converters to power the MPU with a flexible and scalable architecture, optimizing power consumption of the entire system. Third, the family of power management IC for SSD and server. We call them cloud power and SSD power. Reliability is key for this application. With our proven BCD technology, we are among the few players capable to deliver a power-efficient solution, meeting the reliability needs. Managing energy more wisely more efficiently is, for sure, a key priority for us. That's why we continue to invest in application-specific product for Industry 4.0. Our chips enable higher safety and increased level of automation in synergy with our broad offering sensors. Attention is often focused on artificial intelligence algorithm required by Industry 4.0, but we should not forget the hardware that enable 2 important features of Industry 4.0. First, more distributed control for increased flexibility; second, safer, cheaper and more reliable processes. Moving from centralized to distributed control, sensor data from the real-world must be available in real-time, and everything must be connected according to different standards. We already addressed the importance of the sensor as well as wired connectivity like IO-Link and KNX. Now let's talk about our metering and motion control solution for Industry 4.0. In the world of metering, we are market leader in Europe with our Power Line management solution. And our solutions are now being deployed also in Latin America. Our chips are in electricity meters and also in solar panel, enabling remote monitoring and safer operation. In the field of motion control, with our STSPIN and STDRIVE product families, we are able to drive any kind of motor, either DC or stepper or brushless, over a wide range of power from a few hundred watts to tens of kilowatt. Depending on the customer needs, we can pack, in the same package, one or more [indiscernible] such as the drivers, the controller and the high power output stage, either in silicon or in gallium nitride. An important proof point for us, in the last years, our rate of innovation increased substantially, 25%, 1/4 of industrial sales in the first 9 months of 2020 came from product less than 2 years old. But in the factories of the future, people and robot will work closer and closer together. So-called cobots will be more and more common. Thus, industrial processes must be made more reliable and safer for human beings. With the digital transformation, high voltage and the low voltage section of the system will come closer and closer. So safety and protection are becoming more important. And in this direction, one technology play an important role. It's the galvanic isolation. Galvanic isolation enable new system architectures. It allows the integration of analog, power and logic in the same package without compromising human safety. Our galvanic isolated gate drivers and intelligent power switches can go up to 6 kilowatt. They are already in production for factory automation. And our chips are driving silicon carbide transistor in well-known electric vehicles on the road already today. But that's not all. Before closing this section on analog application-specific product, let's talk about an important growth driver, wireless charging. All the key business programs for wires charging are in production as we planned. Here, we replaced 2 strong competitors, thanks to our BCD technology and design IP. We believe that today, we are the only company able to provide such a broad range of solutions from a few watts for wearable and hearable devices to more than 40 watts for high performance is 5G smartphone. And that's not all. We continue to develop efficient wireless charging solution up to 100-watt for both transmit and receive. All in all, we are committed to maintain a leadership position in this application. Let me now conclude. I'm very pleased to confirm that despite the pandemic, the 2020 business trend for AMS is good. In the first 9 months of this year, AMS revenues were 12% higher than year-on-year. Going forward, with continuity of our strategic direction, we believe we are well positioned for balanced growth in both distribution and key customers. And this is thanks to the 3 factors: first, our industry broadest portfolio from MEMS to optical sensing solution, which we are rapidly expanding through diversification; second, our differentiated, highly competitive application-specific analog product complemented by general purpose analog addressing all market segment from Personal Electronics to automotive, from industrial to communication equipment and computer peripheral; third, our distinctive product innovation DNA with the double helix of differentiated technologies and application know-how and focus, thanks to a strong customer intimacy. Thank you so much.

Thank you, Benedetto. We are now ready to take your questions.

The first question comes from Achal Sultania from Crédit Suisse.

Just trying to understand on your 3D portfolio, you mentioned you're working on a various different technologies, be it structured light, Time-of-Flight stereoscopic vision. Can you help us understand what are the challenges or the limitations of Time-of-Flight today if it were to be used for facial recognition in front of the camera? Are we already there with the technology? Or we still need to resolve a few of the issues to make it as accurate as structured light?

So Benedetto will take your question. Thank you.

Thank you so much for your question. I think that today, let me say, you can find several solutions from a technological point of view on the market. But the one that is providing the best resolution in terms of let's say, pixel size and those in terms of accuracy is by far the structured light. The structural light because of its intrinsic approach in the short distance is providing a better and higher resolution. The Time-of-Flight is good, but especially good from a medium-range to farther range.

So just trying to understand, can -- are there any work going around, Time-of-Flight to make it as good as structured light for resolution as well? Or is it something that probably will take much longer than maybe the next year or 2?

Yes. I think that -- let's put it this way. The structured light is a big advantage of being used in higher volume. So everyone in the industry, all the player in the industry gained more experience how to cope and how to achieve higher resolution. The indirect time or direct Time-of-Flight does not -- cannot count on the same amount of volume of the phone with structured light. So if you go to see that the -- [indiscernible] going behind theoretically, they can reach more or less the same limit. But on one side, there is a much more manufacturing experience that is helping everyone, all the players, to push much further the structured light in terms of resolution and accuracy. Now in the future, let's say, other solution can appear. And this is the reason why, as I said a few minutes ago, we are investing on technologies that enable both stereoscopic, structured and Time-of-Flight direct or indirect.

Okay. And maybe one follow-up on your partnerships with customers in China. We understand that Huawei was obviously one of the key customers you had for both wireless charging and also for optical sensors. Now with the ban on Huawei, how are you positioned with some of the other local brands in China like Oppo, Vivo and Xiaomi, is the relationship improving there? Or you think that there is still room for that partnership to evolve further with those 3 vendors in China?

I think we need to remember a little bit what is AMS, Analog, MEMS and Sensors. We are working with some -- with Chinese players since many years, okay? And with some of the ones you mentioned, we really started when they were very, very small. So let me say in this way. We are addressing with our broad range portfolio with analog product, with optical sensing and with MEMS, we are addressing all the Chinese players.

The next question comes from Andrew Gardiner from Barclays.

Benedetto, if I can ask one on the topic of ambient light sensing. You mentioned -- you guys have been talking about this for sort of a year or 2 now in terms of it being a target market for you. You mentioned one win with a major OEM in Personal Electronics. I'm wondering if you can give us a bit more detail, a bit more of a hint as to sort of where within Personal Electronics that might sit? You also referenced that you displaced strong competition. What was it about your product, your design or your technology in particular that you think caused you to win that? And then finally, how are you seeing other opportunities? What -- can you give us a sense as to the addressable market in the next couple of years for you within ambient light?

You remember very well, I said already 1 year ago in London that we were working on ambient light sensors and we were going to make an [indiscernible] in this -- with this product. And indeed, as I said before, we are on track with our plan. So next year, you can buy some product with our ambient light sensors. What is unique? You asked, no? What is unique? I think that the performance of the product allowed us to win against some, let's say, competitors. And this performance is related to the ambient light sensor per se, but is also related to the fact that by making different kind of sensors, as I said, MEMS and optical sensing solution, we can optimize the ambient light sensors looking at, let me say, all the dimension of the product. So we started to invest, we will start to have the business next year as planned, and we plan to continue to grow in the years to come. And the beauty is that by having different kind of sensors, we can also think combination of different sensors together with ambient light.

You have a follow-up, Andrew?

Just on that last point. I mean, when you're saying the different types of sensors you have within your portfolio allow you to optimize the solutions. So is that -- you're suggesting both in terms of stand-alone ambient light or MEMS? And then you went on to say, well, you can combine them into a module depending on what the customer needs. Just on the first point, what is it in terms of your -- the sensor know-how more broadly that is allowing you to optimize, can you just give us a bit more detail there?

I think I can record the numbers I mentioned in the presentation. We have put, let's say, on the market more than 20 billion sensors. And sensor has a little bit different from standard semiconductors. And when you manage and you -- in high volume, the production of sensor, you discover a lot of features that are important both on the performance side and also on the manufacturing side. And with this experience that we gained with all these sensors, we've ordered the sensor, let me say, in quantity, but also in variety. We have been able to optimize the product in performances and dosing costs. And this is coming, I have to say, really, with experience. If you were making me this question 7, 8 years ago, I would not have been underlining a lot, or properly, the experience we gained with all this manufacturing volume. So this is very, very important. More you do like everything, but the sensor is even more important. How you do more you learn to optimize the 2, variable performance and cost.

The next question comes from Matt Ramsey from Cowen.

Benedetto, thank you for the presentation. I have, I guess, one question for you on the business and then sort of a bigger picture question for Jean-Marc. On the business, I noticed in the slides that you called out, I think, 200 different wins for the Time-of-Flight sensor business. And then you mentioned a number of different product categories for the wireless charging business from, I guess, hearing aids and earbuds all the way up to really high-end smartphones. For those 2 product categories, together, how would you characterize the customer concentration of revenue today versus what's potentially in your pipeline going forward? And then I'll just follow up with Jean-Marc.

Okay. So yes, I repeat, for optical sensing. So for the Time-of-Flight and for the wireless charging, how many customers we address? Well, we address basically -- you can find our product, wireless charging as well as optical sensing solution in big customer as well in small customers. Here, in the chart that you have seen, we were mentioning specifically some product in the FlightSense here, we were talking -- I was talking about the Samsung. But I can tell you that there are other customers that are using our Time-of-Flight as well as the wireless charging, either in the low power range or in high power range that are going from small scale to big scale. Really, I have to say that we are improving a lot our presence in the mass market with this wireless solution and also with Time-of-Flight. I cannot be more specific because of confidentiality, but there are really a lot of new application of our Time-of-Flight. One of this I just got a few minutes ago, it's pretty much interesting. You will see soon on the market.

I appreciate that. Jean-Marc, on sort of a bigger picture question. The -- your company has a really established footprint in microcontrollers and has a number of analog sensors to handle sort of both sides of the the A to D transfer of data collection at the edge. Our teams spend a lot of time around the edge computing market and both on the server side and on the client side. I wonder how you think about this strategically, the evolution of the edge as a business driver for the company across the divisions going forward?

Well, the strategy at the edge, I think, will follow various direction, okay? One obvious direction is computing power increase. It is clear that we have to find the way to, let's say, continue to improve the computing power, but not at the detriment of the power consumption. So it is clear that the famous spread of PPC. So power -- computing power, power consumption and area of the chips, okay, which was basically the key driver of smartphone, will be exactly the same for IoT and what is happening at the edge. That's the reason why here, at a certain moment, we do believe that for sure, we will have a traditional architecture, okay, either very powerful microcontroller or you will have some architecture with, let's say, kind of standard microprocessor with a memory [indiscernible], okay, which is the architecture, okay, of the smartphone. But for sure, much less complex. Long term, clearly, we could see happening some new architecture and the edge, which really solve this equation between more computing, but less power consumption. But we have clear ideas, but as it is at level of pure innovation and R&D, okay, we do not communicate on it. We will communicate timely, but we have in our pipe of innovation, this kind of architecture. Well then definitively why more computing power? Because the data that will be processed and the specification request of this kind of device to be self-managed and to avoid to send in the network and to the cloud, useless data will be more and more important. And that's the reason why for either values of interoperability or, let's say, to be self managed, artificial intelligence at the edge will be monetary. And here again, either it will be compatible with microcontroller architecture encompassing some hardware accelerator, okay, to be capable to process a complex neural network and complex data flow. Or you will have, okay, some architecture with some kind of standard microprocessor and later on, okay, new innovation. Well, the second big axis that, okay, we will follow-on IoT or edge device. Of course, it is connected, and here, we will need to have the full set of connectivity indoor, outdoor. And definitively, the connectivity related to the 5G network. And here, I have to say immediately that the company's strategy, of course, will be to enable our embedded processing solution with adequate connectivity for the network. So either case ELCM or [indiscernible] IoT. But ST wants as well to capture the value of this kind of device also be ready to offer a front end module, RF front-end module in order, okay, to be present in the full architecture and the full bill of material of this kind of device. So as a takeaway, our strategy for ST will be, okay, to enhance the computing power of our microcontroller. Or moving by step, okay, to standard microprocessor. And later on, in new kind of architecture some artificial intelligence inside this device either putting hardware accelerator in our microcontroller. But later on, again, moving to standard microprocessor or moving to new architecture. And in parallel to offer the full set, let's say, of a critical device, connectivity, but module as well. It is obvious that to offer a full set of, let's say, reference design device sensor are very important, but I will not come back on the presentation of my friend, Benedetto and analog as well.

The next question comes from Sandeep Deshpande from JPMprgan.