Samsung Electronics Co., Ltd. (A005930) Earnings Call Transcript & Summary

Thank you. Good afternoon. I'm Ben Suh, Head of Investor Relations at Samsung Electronics. First of all, I'd like to thank you all for attending this year's Investors Forum. Also, it's great to be back in Hong Kong and meeting with all of you attending in person as well as those viewing via video. Amid today's major uncertainties, actually today, we would like to focus more on the future. And we would like to discuss growth strategies from some of Samsung's key businesses. And after today's presentations, we also look forward to hearing back from you regarding your feedback and also additional areas of interest. We have 4 speakers with us today, representing Display, our DX business, Memory and Foundry. And each of our presenters will share key growth focus areas for their respective businesses. First, we will have VP, Gong Min Kim, from Samsung Display's Research Institute, and he will present driving the future with Samsung Display. Mr. Kim has an MS in Mechanical Engineering from Seoul National University. And after joining Samsung, he started his career in OLED process and equipment design. His deep and diverse experience in engineering and technology planning, includes establishing an operating Samsung Display Research Center in the U.S. and expanding corporate partnerships. Mr. Kim is currently the Head of the Technology Strategy team and in charge of technical planning as a Display Research Institute. Second, VP, TaeYoung Huh, the Head of eXperience Planning Group in the DX Division's CX-MDE center will give a presentation entitled Customer Experience Driven Growth Strategy. Many of you probably do not know what CX-MDE stands for and that's part of the surprise of his presentation later today. Mr. Huh graduated from UCLA and after joining Samsung Electronics, he led the product planning team in the VD business, contributing greatly to product planning and successful sales strategies in multiple areas, including Micro LED and Neo QLED TVs. And he has also played a major role in public relations. Mr. Huh is currently responsible for creating strategies and delivering differentiated experiences across multiple devices. Third, Harry Yoon, the Head of Marketing team in the Memory business will present the leap into the future, memory milestones for the AI era. Mr. Yoon has MS in Electronics Engineering from Seoul National University. After joining Samsung Electronics, he developed an extensive technical background in multiple areas. In particular, he played major roles as part of the product planning team for cutting-edge DRAM products such as DDR5, LPDDR5 and the world's first mass production of HBM2. As the head of the marketing team, Dr. Yoon oversees some market research, product marketing and other related areas. Finally, we will have EVP, Gibong Jeong, the Head of the business development team in the Foundry business, deliver a presentation on Samsung Foundry and AVP, We Give Life to Your Imagination. Dr. Jeong earned his PhD in Electrical Engineering from Stanford University. Before joining Samsung Electronics, he worked on Microchip Research and development as a senior engineer at Texas Instruments. At Samsung, he devoted himself to research and development of core semiconductors as a senior engineer in the System LSI business. Based on his diverse experience, he is currently leading the business development team and plays an instrumental role in identifying customer and market needs and establishing technology and marketing strategies. I'm confident that today's presenters will provide valuable insight into Samsung's growth strategies. And once again, let me remind you that we will have the presentations first. And if you have questions, we will have a separate Q&A session immediately following the presentations. So with that, please give a warm welcome to our first speaker, Gong Min Kim, VP of Samsung Display.

Thank you for joining us today. I'm Gong Min Kim, who leads technology strategy team at Samsung Display business center. It's truly an honor to me to speak on behalf of SDC, Samsung Display about the future display market that SDC will lead. Frankly speaking, I'm a little bit embarrassed when I heard this, I would be the first speaker. It's real just I expected to be the last speaker. But I'll do my best today just for you to understand the reason why after my presentation. Okay, here to start. Driving the future with the Samsung Display. Okay. This is the topic today. First, I'm going to talk about the general display market. And then just the smart Chapter 1, smartphone and TV market, I explained. That's the major display period, now the currently SDC is doing. And then next, just IT and auto markets. And that will be the biggest growth engine for Samsung Display in the near future. And lastly, the micro display I'm going to talk about that is really a new opportunity to us. So what is the display, as you might be aware, the people perceive and receive the information, mainly by the site. It's more than 80%. And I would like to define the display is the window that connects people and [indiscernible] and people and your work through the technology. And it means display would be everywhere at any time. So, that -- since the 2000, the display market grows dramatically for a decade. But from the 2010, the market seems like to be -- has been desaturated more than 10 years. So, what's going to be the next growth engine for display market? So -- but if you see the deep dive, the detail -- this graph. I'm sorry, it deepens the application by [indiscernible], especially OLED portion has been increased from the 8% to 34%. It's mainly just applied to the smartphone. And it implies that maybe we have the chance, the good opportunity in display market to grow. And SDC, it seems like similar to the whole market -- display market. We founded in 2012 and then more than 10 years, the revenue seems like to be saturated and it's not the big change. But if you see the OLED portion for this, in 2012, the LCD revenue is much more than the OLED one. But since last year, we exited the LCD business, so this year, just be just only doing the OLED business. But the revenue is not that this more, as you can see. So mainly we have -- we did our business in smartphone and TV also. But again, the -- similarly to the smartphone market, the IT and auto and other application area that OLED will be penetrated something dramatically soon. So this year, just the portion -- the share -- the portion of the new market, such as IT and auto is about 10% of our revenue or less, but it will grow more than 30% in the near future. So I hope that our revenue will just up to KRW 50 billion soon. So this is the snapshot of the SDC's display. I'd like to emphasize on that we have spent R&D, about 10% of our revenue and more than half of the profit. And also, just if you can see the number of the R&D engineers is also more than 50% of our total number of employees. And also, the -- our OLED market share is more than 70% of the economy. You may have heard that the word GOAT, it's usually used in sports things, the greatest of all time. I'd like to say that SDC would be the GOAT in display. If you see the information just last year, just our -- the market share in small and medium is about 70% again. And our operating profit, which is KRW 4.6 billion, but on the other hand, the other hours that play the companies, they lose [indiscernible]. And the number of the pattern, the SDC also is outstanding, if you can see. And next, I'd like to the market in smartphone and TV. These are the major period the SDCs are playing now. Let's see the video [indiscernible] [Presentation]

Okay. Let me talk about the smartphone market. It's the same as this, the OLED share is now about 75% in smartphone market. And it's worth noting that the adoption of OLED displays on premium and mid-range smartphone is distinct compared to LCD panels. And in 2007, we -- they have the succeeding in producing the OLED device world first. And then up to now, just we have developed and we have led the market with the flexible technologies and [indiscernible] technologies and portable and UPC. UPC means that under panel camera. And how will the SDC can keep our position and even strengthen? Here's the 4 key directions of it. The first, the board reduction, so to make the [indiscernible] screen. And second, the low power consumption and started the sense integrated display panel. And finally, the new [indiscernible] factor such as the foldable display and rollable display and slidable one. I'll expand that one by one. Every display makers wanted to be the 100% SBR display. So SBR means screen-to-body ratio. It's 100% SBR is the kind of dream for the OLED display makers. So we have tried to increase SBR by employing the narrow [indiscernible] technology as well as [indiscernible], the [ holding ] display and UPC. So in Galaxy S10, we reached about 88% of SBR. But you want it to be higher such as more than the 90% SBR. To do this, we have to -- we have been developing the circuit and module design. Those are the kind of key factors that reduce the border. So -- and as you can see here. And next, the low power consumption is always important. Many technologies are leaded to achieve low power concessions. But here are some of the flagship to low power technologies. The first one is TADF and second is OCF and third is adaptive frequency. The TADF is kind of the new [indiscernible] organic materials. We have used the kind of the fluorescence materials and phosphorescence materials. But especially for the blue color, we need more -- such a higher efficiency and the longer lifetime. So we have developed the TADF materials and recently just we have achieved dramatically the better data for that. And also the OCF, means the [indiscernible], the conventional display uses the polarizer, but it just cuts about the half of the life. So the transmittance and the light efficiency is very low if you use the polarizer. So we'd like to eliminate the polarizer and then we did it. So it's also the -- known as the Eco OLED Plus is kind of the trademark. So which enables for us to eliminate polarizer and to get the much higher transmittance. And the last, the adaptive frequency. It means that you want to refresh rate the -- to be able to [indiscernible] despite the contents. For example, the motion -- the fast motion sports where video clips needed a higher refresh rate, more than for example, 120 hertz. But if the display shows the still images or a text, you don't need to refresh very often. So we just drive the kind of 10 hertz or even 1 hertz. So it enables us the IC power to reduce or drive the kind of low refresh rate. So we have developed the kind of the adaptive frequency to do this. Next, sensor integration. Nowadays, more and more sensors are being implemented to smartphones. [indiscernible] is more than 30 sensors are embedded in smartphones, you may know. But I expect that these sensors to be integrated into the display for slim and lightweight smartphone with high functionality. So we are working on various all-in-one sensors like OPD, which stands for the organic photodiode. OPD just make us to get the all-in-one sensors. And it enables multi fingerprint touch and blood pressure and sugar -- blood sugar and so on. So for this technology, we have achieved dramatically the improved data we got recently. And finally, the new form factor. I think this is one of the most important factor for the smartphone market. We launched our first foldable phone in 2019. And then it has been evolving. And I don't need -- now currently just the foldable display is kind of the infoldable display. But you also just developing the half foldable and multi-foldable and even the slidable and rollable display. To achieve that, the most -- one of the important thing is you make the [ low stress ] structure as well as the materials. And then -- also the low stress structure ensures that each panel layer is made of stress resistant materials, making the panel withstand the multi stress. And stretchable panels, it's a little bit more challenging technology, which requires highly elastic materials, compared to other flexible panels as it has to be the stretch on all sides. I'm going to introduce our QD display and then also the other the video clip to demonstrate our QD space. [Presentation]

Thank you for watching. Samsung Display launched QD display in TV application last year. So currently, the current TV market is mainly divided into the LCD and OLED panels. And there are 2 types of OLED displays for TVs. The one is the White OLED and the other is QD OLED. As you can see, the LCD relies on the color [ pewter ] and PLU, the [indiscernible] unit, and White OLED is just the White OLED materials and color filter. But QD OLED employs color conversion through blue emitting layer and QD, quantum dot. As you just saw in the video, QD OLED is superior in brightness, contrast ratio and color [indiscernible], especially color and [indiscernible] is much better than the other displays. And you made that -- since that we are succeeding in the manufacturing the QD display last year, the QD display TV just won a lot of awards from the display institute. And also this year, as you may know, that Nobel prize in Chemistry, just the scientists who discovered the QD, they won the Nobel prize in Chemistry. So -- and we will continue to develop the QD OLED technology to make it more durable and cost-efficient and an eco-friendly thereby further increasing its competitiveness. Next, let's talk about the IT and auto market. That would be the major growth engine for Samsung Display. As you can see, the -- like the -- in smartphone market, currently OLED -- the share OLED portion in IT and auto is relatively small. But as you can see, the growth rate will be very high. It's about 34% in IT OLED market and 35% in auto OLED. It's very promising, I think. So firstly, in 2019, we just launched the laptop with OLED display. And then this tier just extended our scope and feel in IT area. So that we are preparing those kind of 4 key technology to extend our area in IT field. The first one is the larger model glass and then higher luminance and longer lifetime and variable frequency and [indiscernible] light display. I'm going to explain one by one. Samsung Display just announced that we're going to invest the 8.6 glass for the IT business. The current existing lines consist of the Gen 4.5 and 5.5 and Gen 6, but -- if we just operate the 8.6 mother glass, the glass efficiency is much higher than the previous one. And also process costs will be much lower and it just affects if the display size would be bigger and bigger, these [indiscernible] -- the efficiency will be higher, too. Next, the high luminance and longer lifetime. Especially the IT market is compared to the smartphone market, need more higher luminance and the longer lifetime. So to do that, just we have developed in the tandem structure. The tandem structure is the strategy that forms the diode OLED layers in series. You can think of it as 2 batteries being placed in series. So it makes the display much brighter and much longer lifetime. And we're also working on the backplane technology that supports the variable frequency. As you may know, the amorphous silicon backplane uses for LCD in feature phone. And for smartphone, the LTPS backplane, low-temperature polycrystalline silicon backplane, should be needed because it needs more higher mobility. And nowadays, the HOP, it means hybrid oxide polysilicon, HOP is adapted in the high-end smartphone to give the both advantage of the LTPS and oxide. And then finally just in IT business because it's mother glass is bigger and bigger and oxide technology is essential, we believe. So we have been proved that the oxide technology could be adapted in IT market. So -- and then it's about to adapt in IT, the product. And [ selenium ] light, the conventional LCD display is consists of the [ color pewter ] and TFT glass and PL units. But if you just convert to the OLED, the digital OLED, especially, it's -- we don't -- we can -- we don't need the PL unit. So just it's much lighter and much slimmer than the LCD. But even if we just to do -- to get the lighter and the slimmer display panel, we just eliminated [indiscernible] glass and then instead of that, we used the thin film encapsulation. So we call it the UT technology. And this technology will be applied -- adapted in the IT business system. Also, finally, in IT, the market IT displays also we are considering the full flex where the OLED could be applied for the high-end market. And next, our auto technology and auto markets. Our first -- our auto products, make debut with the adoption of the [indiscernible] to a European company in 2018. And then also we extended our scope, PID, RSE technology, and finally the curved and [ long ] display. PID means that passenger information display and RSE stands for the rear seat entertainment. I'm going to explain them more. Despite low OLED adoption in auto display market, we are dedicated to carve out a new OLED market by carrying to customer needs. So that's -- those are the 3 technologies you can see here. The seamless display and switch over to privacy mode and shy technology. I'm going to explain one by one. The seamless display, it's a -- you may know that the display area in the car is getting larger that means that the front dashboard CID and PID are seamlessly incorporated into the single large panel. To this end, we've been developing a technology that attaches multiple panels to one large cover glass. And [indiscernible] technology, we also be needed to fit into the various compactors. In privacy mode. The PID is becoming available for more and more cars and due to a wide view angle, the PID screen can -- inadvertently is seen by the driver. It's interfere with driving and breaching the passengers privacy. To address the issues, we have employed the privacy mode that uses [ LCF ] to make a narrow viewing angle. And also, we've been developing a technology that can switch between wide and narrow viewing angles. So ShyTech is not the [indiscernible] one. It's kind of use slidable display and rollable display. And this tech is a feature that can be made less [indiscernible] when not in use and is often used for rear-seat entertainment. This new technology will let consumers tailor their [indiscernible] design to their needs. And last, the micro display. Please note that the definition of metaverse can vary. The XR is short for the extended reality. The XR just includes the VR, AR and MR. The first of VR, the virtual reality is lies on the 360 degrees videos to create an immersive virtual world. So the display devices seems like that the [indiscernible] is kind of the party to give the immersive images. But AR, augmented reality is kind of the -- it provides a real world with the digital information overlay. So it's the real world remains centered and then the experience and then enhanced by the visual details. So the display -- the users should see the real world, the display can be light -- should be light and transparent. And finally, the MR, mixed reality, it's kind of [indiscernible] proposed. It's recorded as kind of interim, but now there's kind of people think it could be the mainstream. The mixed reality is the device that [ real and virtual sides ] intervene each other. And it's kind of the -- it could be the personal TV and then it provides lots of the values to the users. And what's the requirement? What's required features of the micro display? The key point is that one is the higher duration and the other is higher luminance. Imagine that seeing the image on TV sitting on the couch and seeing the same image across on your smartphone. The displays comes near and near to eye that should be the higher resolutions. And also just if you think about the micro displays, if the -- uses in the outdoor circumstances, it should be very high luminance needs. And also the optical system in micro display, the efficiency is very low. I'm going to explain more of this next. So anyway, that's why just micro display needs the key features, which is that high -- much higher resolution and much higher luminance. If you can imagine the smartphone, the highest resolution would be the 500 or even the 800 PPI. PPI means the pixel per inch. But again, for the micro display, the resolution should be more than 5,000 or 6,000 PPI. To get that, we cannot risk that kind of the high resolution with the glass technology. We need the semiconductor technology. So that's why we have to develop the silicon based [indiscernible] technology. And also just the [indiscernible] material, the OLED materials, it's a little bit -- there's some limit just to give some higher luminance. To reach the -- for example, to reach the higher luminance, inorganic material should be adapted such as the LED. So now the -- to develop the micro display, the new paradigm should be adapted. The one is silicone backplane and the other is inorganic materials. So in [indiscernible], those technologies, we can think that those kinds of type of the micro display, the micro OLED and micro LED. So it can be called that if it uses the silicone backplane, we call it just the OLED on silicon and LED on silicon. But still the low-end micro display or for the watch applications, we can -- also it can be -- can use the glass backplane technology, because it's not that require such a higher PPI. Let me explain it more. So there you see -- just I share some of the technology that tied to micro display, interpretation for a new paradigm shift. Such integration the compact optical system, high luminances and LED transfer. Especially for LED -- micro LED technologies, LED transfer and wafer-to-wafer bonding is needed. First, the circuit integrations. As I mentioned earlier, the silicon backplane technology has to be developed for higher resolution. Under the conventional glass backplane, pixel transistor and gate drivers are integrated. And source driver and power and then others [indiscernible] are tested in the form of the components, according to the IC. But for the micro display we have to develop the new circuit technology to realize complete circuit integration once the silicon backplane is ready. And compact optical system. To make the virtual images for the micro display, we need some optical systems. For the VR currently with many players uses the pancake glance. But again, the Pancake is very low light efficiency and very bulky and heavy. So it's very important to -- for us to develop the kind of light and higher efficiency optical systems. And the light efficiencies for the pancake is around 10%, that's kind of the theoretical just the upper limit. And also for the AR, if you imagine that the Google Glass -- AR glasses, the [indiscernible] should be needed. But the [indiscernible] light efficiency is 1 digit. It's about just 1% to 3%. So that's why we needed a higher luminance as the display side. And also we have developed the kind of more efficient optical systems. That's the kind of the examples. And then again, the -- for the outdoor uses and the lower -- the light efficiency in optical systems, micro displays needs much higher luminance. For the OLEDoS, currently just White OLED technology is trying to be adapted but finally, the RGB OLED technology should be adopted for the OLEDoS, because it's much -- it gives -- it can give much higher light efficiency. But to do that, we need kind of the new technology just using the silicon technology to make the [indiscernible]. Here, there is time pattern mask. It should be like -- again, we have to achieve like more than 5,000 PPI is fair changing. And LED efficiency, currently, just if you think that LED can give very high luminance, but as you can see in the graph here, if the LED size comes down to under the 10 micrometers, the efficiency decreased dramatically. So to be adapted in AI and [indiscernible] the size should be under the 10 micrometers. So that's one the biggest challenge when we got the LED in micro display. Sorry. And definitely, this is just working on the technology. And finally, in the LED technology, LED should be transferred from the wafer to the glass backplane or the wafer-to-wafer bonding is needed. If you think about the 8K display, the number of transfer should be kind of 100 million times, okay? So it's very costly and it takes a lot of time. So a lot of technology should be tried just to -- by using the lasers and fluidics and other stamps just to less the tech time and cost. And on the other hand, it requires very high accuracy. Again, as you can see, you can imagine and needs higher resolutions. Okay, it's almost there. We announced that we merged, Imagine the U.S. companies for OLED -- for OLEDoS, the market. We announced about in May and then recently just closed the deal. So [indiscernible] the #1 micro display technology they have. And I expect that it's very big synergies between the 2 companies work together. Okay. And in conclusion, SDC will cover the all sides of display. And we are mainly just focused on the smartphone and lastly just something to the TV application. But we are extending our business to IT, auto and monitor. also just again the VR, AR market, too. And then Samsung Display the [indiscernible] display makers, we try to cover all the size of the display centers. And then we will be leading the market in all the applications. Okay. This is the last slide. So in summary, you can see the market condition -- display market conditions here. The smartphone, the OLED penetration rate is currently about 89% and it will increase up to 94% in [ 2008 ]. And similarly, in tablet, not the IT market, the OLED penetration rate would be increased dramatically from 6% to 48% and TV also. And not only that, we are trying to extend our scope to the smartwatch and auto and XR the application. These markets are growth very rapidly. So -- and we are pursuing some opportunity in that area, too. And so finally, the SDC just we covered all the application. And would like to lead everything. Thank you.

Hello, everyone. It's really nice to see you here and it's really honor for me to explain about Samsung CX-MDE strategies. So the division that I belong is under DX and then it's the CX-MDE Center and it's very recently formed center. Actually, we have started these businesses actually from last year, so it's only less than two years. And it's really great that why the Samsung is so much focused on CX-MDE. And actually I'm pretty sure that most of you guys are not really aware of what this CX-MDE means. So let me first explain about what's CX-MDE. So I'm pretty sure that you have heard about CX. CX means customer experiences, actually on our diligence like mobile and VD, we used to have the product planning. It was more like a planning origination, focusing on developing the product, right? But then we recently changed all of our organization's name from product planning to CX. Why is this so? That's because the most important thing is how users are using our product and what kind of really good customer experiences that we can bring. So CX becomes really important. Then what is MDE? Actually, we have many different divisions like mobile and video and also the home appliances. But nowadays, people are using multiple devices. Then by combining all those devices, they are actually implementing these connected devices into their actual life. So we really believe that all this is a real-time and opportunity that Samsung make a differentiations by leveraging our multiple devices. So since this is a CX-MDE session, I first wanted to talk about why then CX-MDE is important to the customers. Why then Samsung become a pioneer of this CX-MDE business opportunities? And then what's the real needs and then wishes of the customer's expectation regarding the CX-MDE? Based on these customer's needs, we came up with the CX-MDE vision and also we came up with some very nice interesting ideas that we today will talk about the future of the Samsung, right? That how these CX-MDE experiences can benefit the future of Samsung businesses. That will be my conclusion. So then how do customers like changing their lives and how do they really perceive these CX-MDE as more like a very important value? Within the recent studies that we had internally, we found out that the user experiences across multiple devices become the number one regional brand innovation factors. So when they're talking about the brand innovation for companies like Samsung, they believe that these multiple-device experiences are the far most important. And obviously, we are selling many different brands. For one customer like even if he or she is the mobile customer, we wish she or him to also purchase other product like TVs and also home appliances. Then why then this MDE is really important? Because controlling multiple smart home appliance with one app, which is like revolutionary process, will become a key issues of customers to choosing the same brand. That's why CX-MDE is so important to Samsung. So that previously we are more like a product focused and then now everything is our customer focused. Their experience is first. And then within our recent study we found out that about twice more people actually have Samsung account are having multiple devices of Samsung brand. And about the average our users have average of more than four devices per person. And especially in U.S. area, we surveyed how many connected devices you have at home. And we found that from the previous 10 to 11 devices, now more than 20 devices they have at their home with the connected devices. It's very fascinating. Then why then Samsung has the advantages of having these CX-MDE scenarios? That's because Samsung is the top brand which has all those kind of very variety of the range of the connected product lineup including mobile, VD and also all those digital appliances. And that we have a very strong digital platform called SmartThings. And probably you guys are aware that we continuously have the number one positioning in VD for more than 10 years. And then even for the mobile and home appliance divisions, in terms of the varieties and in terms of the already sold connected devices, we are far most the #1 in global. It's not just the connected devices, number isn't really important. But what kind of value can we bring to the customers? Actually this is the experiences. Then how do we understand the customer's experiences? Because with the connected apps and also connected services that we provide to the customers, which is preloaded to the devices, we kind of guess what kind of lifestyle and then what kind of needs they want to have at their home life. It's really important. The example that I show is just a vacuum cleaner. What kind of user benefits can we get out of these usage datas? It's not just the number of times that you use. Actually we can understand the footages of our customer's cleaning habits. And which time and in which condition. All those kind of various conditions that we understand, we can provide the very different and very fresh ideas and values to the customers. Okay. Everybody will kind of wonder if you understand those kind of device activities. And by understanding that, if you can guess the customer's lifestyle, then what about the privacy and security? Obviously with Samsung, privacy and security is the far most important factors. And then since we have very designated technologies called a Knox Vault, which means that not only the software , but in terms of the hardware for all of our connected devices, we have a very high-end security solution. So that many top government agencies around the whole world actually, they have satisfied a Knox as a most valuable security solution. At the same time, if you have more than one devices, the more devices that you have, actually the more security levels and privacy levels that we can provide. Even at home, without connecting to the cloud, with the Edge cloud among the multiple devices, we can very clearly enhance the technologies of security and privacy. [Presentation]

So this is the interview of probably you guys are aware of Sonny. So we have interviewed him and then asked him. After using the Samsung SmartThings what's the benefit for him in terms of the value of the SmartThings? It's actually what he's doing when he's not being at home. It's not just when you're at home, actually when you're outside of the home, you can manage so many things. He clearly defined the definition of SmartThings, it's so easy, right? It should be very easy for even if you're not a technology geek, it's actually you need to use it without any effort. That's possible with automatic connections. But if you purchase Samsung devices and then if you are a member of Samsung, then actually all the devices that you purchase will be automatically connected to your system. And also in between different devices and apps, when we try to use the mobilities and then the locations, the experience that you have must be continued. So It needs to be easy to transfer. Then we need to provide the continuous experiences. Then all those kind of usages of your devices, if you're moving from the A to B, actually you need to have the consistent experiences. That's really important. And then also, everyone's lifestyle is different. That's why we need to personalize the experiences and then tailor the experiences which is the rightly fit into your lifestyle. We understand the customer's needs. Based on this customer's needs, we have recently announced the vision of CX-MDE. So as I previously mentioned before, CX-MDE's vision is the more you own and use the Samsung devices, there's a whole wide range of the devices. It start from your very personal mobile devices to even variety of the home appliances, which is located to all different position of your house. Then the devices get smarter and then it can understand you better so that they become making your life much richer and simpler and more enjoyable. With this kind of benefit, then what can we do? Actually it could increase our customer's desire for their lifestyle. Here's the key. We used to have a different customers for the mobile, different customers for the TV and different customers for the home appliances. But once users are using our same branded devices in much smarter way, actually we really wants to help them to help bond with the Samsung more closely emotionally. And not only our customers, but even to their extended families, we want them to be the part of the Samsung family. It will provide the personalized experiences tailored to your lifestyle, to you and also to your family. So under this vision, we kind of very carefully defined the experience area. So when people are thinking about SmartThings, they only think it as more like a home management. Saving your energy is, of course it's really important. Security and also the usability is really important. But with all these conditions, what do you do at home? Actually, you enjoy your lifestyle. You listen to the music, you're watching movies, and then you're also playing games. At the same time at home, actually probably after the pandemic area, I'm pretty sure they're used to working very hard at home. And also, your kids also need to have the learning experiences at home also virtually. On this kind of very changed environment, I think Samsung, CX-MDE can bring much benefit to the customers. And also keeping your health is really important at home. So Samsung has a different way of defining the health and also the care for the loved ones. It's not only for yourself, it's also for your families. It's really important that we need to focus on care area and also the health areas. Today, let me just briefly focus on this care area and health area. What's the definition of Samsung's care? It's an experience where my devices and things work together so that it could keep everyone I care for safe, healthy, and connected. It include everyone. It include your parents, the senior, and also your kids. At the same time, if you have families with some this kind of discomfort, we also included all those family members with these inclusive experiences. And that list, your lovely dogs and cats, pets are also the member of Samsung family. What kind of categories can we bring with these care scenarios? It's a preventive care so that before anything happens, we can alert the situation so that you can be preventive. At the same time, it's a relational care so that you can check in your loved ones even if you're not at the same position -- same places. You can remotely understand your parents or your kids' conditions. And if anything can happen, you could become the first one to be connected and make sure that everything is safe so that it could lead to the managing care. This is an example of senior care. With this presentation, you will understand why Samsung really wants to keep the goal of the caregiver and caretaker. Here are these scenarios. Only with the acceptance of your parents and actually she could limit the coverages in order to actually admire her privacy. And then with all the Samsung devices connected by understanding the usage of the refrigerators and also all those kind of sensors, actually your family can easily understand what's going on. It's not just when something is happened. Actually in everyday life, when you're starting the life -- the daily life, and then when you're ending the daily life, actually all those kind of your situation can be shared with your parents. In emergency, you don't have to actually call. All those kind of immediate situation will be directly connected to your caregiver so that he can actually connect to you without any problem. Does it make sense? Okay. And another thing is the health. For us, health means you eat well and you sleep well, and then you exercise well. And every experiences need to be connected also at the same time. Based on our wide range of the product and health, Samsung food, SmartThings apps. Actually we can offer a balanced healthcare solution of just, I mentioned sleeping, eating and exercising and it could provide a different experience. So probably you guys wonder what kind of benefit then Samsung refrigerators or their air-conditioners or even TVs can deliver when you are sleeping. So I prepared a short video clip. [Presentation]

Especially in areas like Hong Kong, in summertime it's very humid, so obviously air-conditioners must have devices. Sometimes it's noisy and sometimes a very small but very strong LED light can distract your sleep. Each Samsung device has control abilities of even controlling even small light, but it's really hard to control those functionalities. So with our SmartThings engine, the features are set up with very simple mode such as sleep. Even a tiny light of your connected devices can be controlled and help you to get sleep well and also awake well. Then how we come up with all those kinds of valued experience product. We called it sometimes offering, we called it sometimes customer scenarios. Actually, we gathered all those kinds of needs and also the ideas of CX-MDE scenarios from many different parts, from our colleagues, from our partners and also from our third-party vendors, and then it was more than a thousand of the applications that we gathered together. We grouped it together and then we want to test it. So later on I will explain about what CXI lab means and X1 Suwon means, but those are the facilities that Samsung is operating in Suwon office and at the same time, we have a local CX-MDE experience joined in every regional headquarters. There we have validated all those scenarios and we made it prioritized and then also we grouped the similar experience together, so for total we have more than 120 experience products and then all those more than 120 products grouped under the five pillars that I have just mentioned, care, home management, health, enjoyable life and work and learn. So these kinds of ideas, some of the scenarios apply to the global level and some of the ideas apply to the regional levels and I will explain the very interesting ideas of the regional level ideas and also at the same time, it can be fitted more well into a seasonal holiday season like Christmas season or the spring home cleaning season and stuff. So we made these scenarios and try to leverage the customers by introducing not only the product but also the customer's experience based on the multi-devices. So what is then CXI lab? This is also very recently formed. Actually, we have built the place last year and then there are so many interesting places. So for all those kinds of experience zones, like I have mentioned, health and gaming and smart work and cooking, actually we gathered not only the Samsung product but even our competitors' product and also at the same time our partners' product and then created this place as more like actual living places so that anyone can come and actually find the problems or find the new ideas. So there many new ideas are coming up and then all those kinds of ideas will be validated in these places. At the same time, we have developed a new facility called Experience Home. One is in Palo Alto and one is in Suwon, especially the one in Suwon actually we invited many important customers and also our own employees to enjoy all those Samsung connected device experiences. So there are about more than 120 devices that are connected and there actually you can sleep and then you can spend your time and they can get some ideas out of those kinds of experiences. And the regional offices also have things called Smart Homes. They're using these facilities in many different ways. So those are the examples or the snapshots of the locations that we provide that's more like a SmartThings home. So especially in this experience zones, they use these facilities for evaluating the customer's expectation and also getting their feedback. At the same time, our whole employees will have the opportunities of having these hands-on experiences and test not only our devices but also our partners' devices. At the same time, they leverage this place as more like a marketing event and workshop places. So then let me just briefly explain about how then the regional pieces regarded the importance of the scenarios. We collected all those kinds of ideas in each regional levels. We found out that energy saving is by far the most important and common in all regions. So we found out that customers really care about the SmartThings and that they really believe that energy saving is the most important reason. But not only the energy saving mode different by region, actually they came up with many interesting needs and ideas. So for an instance, for Korea actually they're concerned so much about their pet care so that even our Korean office decided to have this pet care as more like the most important strategies by leveraging the CX- MDE zones and then in North America and Europe, they're interested about this new terminologies called Care Services which we will greatly announce next year and provide quite unique services. And these are the examples of the local ideas so that in Europe actually reducing the electricity could be and then also having this to reflect to energy concern was by far the most important needs and especially in China, I'm pretty sure that the automotive industries are really important in China. So on the board services by connecting your personal devices with your car and home was far most important factors. At the same time, countries like Middle East actually they try to develop their -- enhance their experiences by adopting a new mode called Prayer Mode or the Sandstorm mode. I think everyone is wondering what kind of benefit then can it deliver when you pray. So for an example, I will just briefly explain how they developed this product concept and how they actually implemented to their actual businesses. So obviously religion is really important for the Muslim customers and then there are so many different Muslim kinds and then their way of praying is different by each different religions part. So for them keeping the rules is really important, so that SmartThings can really benefit when they are having this kind of praying time. At the same time, they want to keep their home safe by providing this kind of Sandstorm Mode. At the same time, after the Ramadan season, actually one of the biggest problem is to take care of the leftover food. So they want to wisely manage their leftover food. So with all those kinds of the actual needs of the customers, we came up with some ideas of the application. This is the example of the Prayer Mode. So concept locally developed and they partnered with the local partners providing the accurate training time. So it's not only touching the regional needs, it actually touches very emotional scenarios such as care as I just mentioned or the sustainabilities and accessibility scenarios. So all these scenarios are already existing right now. Then you can find it in Samsung's IoT apps called SmartThings in many different areas. Let me just give you some examples of other fascinating ideas. So the first thing is family care. So as I previously mentioned, easy to use is really important. Then, how difficult is it to connect multi-devices. Actually for the first who initiate this, he would need to connect many devices by himself. What about your family? Actually, it doesn't need to be complicated. So let me just give you a quick clue. So by inviting your family or even your guests or even your kids, you can set up the level of device control abilities. And then with this kind of setup and sending the invitation with one click of the QR sense, your families or even your guests can have the easier use of the services and this kind of application actually can be used in many different B2B areas like hospitality services like hotels or hospital. It only activates when you're staying there and after you leave the places, every connection will be disconnected. And then also another interesting thing is the usability between TV and mobile. So that if you are under the same account, if you're near to the TV areas, actually even on lock screen conditions, you will have this prompt and then there you can use your device very wisely. So then you can like share your contents and then you can mirror your screens and then you can control your TV with many number keys and many different keys. Also, energy is so important. So what we tried to achieve is we tried to make a partnership with the energy companies in global and tried to provide the customers the chance of actually controlling your whole home energy bills. So it's going to be easy to use. And then obviously this is my wish that everyone just uses Samsung devices, but it's not the reality. So with using this energy matter, actually this can control also other partners' devices and then you can have the whole view of the electricity and wisely manage the bills. So with this kind of applications, we use it by entrusting it under our IoTs and also even some markets like U.S. and Korea, they use this for the CRM activities means that once our user is purchasing one device, we now have the expectation of what kind of other devices he or she is willing to use or purchase more by the researching and analyzing the big data. And then also, we currently adopted onehour.com and also our own stores and even for TV, it will directly indicate what kind of benefit you can have by having this TV with IoT hub built in. So these are some examples of the materials that you use for communicating the SmartThings. And then we used this and also we testimony this importance of the CX-MDE on many different occasions like exhibitions or the conference and also, later I will explain that is the actual offline shop experiences. So that -- with this kind of experiences how can we benefit it to the future services? So I just mentioned that by understanding many connected devices actually we can get customers' insights and we can share some of these customer insights with our partners so that we can make a very strong and a different level of the partnership with them. And by partnering with our great partners, actually we can propose new business areas such as not only B2B areas but also the huge energy sectors or the security sectors or even the house areas and even the entertainment areas. First, let me explain about the retail opportunities. Retail used to be the place that everyone is checking The product, comparing the price and trying to check whether the product is working well or not and then if you really want to see the actual design, then you go to the shop. But we want to transform all those retail shops into the place where you can actually have this personal experience of your lifestyle. So we have created a quite interesting site so that you could recommend your interested area such as energies or sleeping and stuff and after selecting this, you can set up the level of the understanding of the IoT, then we provide the experience product what you can enjoy by having our products and our partners' product. It will list directly into our partner website or even the things that you can purchase at the retail shop. So once we developed it and then launched it in our dotcom and also in exhibition, many of our partners, best partners, they really want to exploit this kind of application at their retail shop. The current one is only directly connected to our D2C shop, so obviously they want to sell their product so it needs to be tailored in each retail mode so that it can be extended to even not only our shops but also our partners' shops so that we are developing this system. So by next year, I think not only Samsung shop, but even shops in your neighborhood can have this kind of experience zone. So that this is the first shop that we employed this experience zone, it's the first experience shop launched in Korea, actually last week for our investors. Actually I visited there and then recorded a small experiences, so why don't we take a small look? So I took it with my own mobile. So there with the QR you understand where the experience zone is. There you can choose your interested area. So for her, I think she's interested about energy and pet. Then it provides the examples of what you can do with these pet scenarios and then actually it lists what kind of items you need to purchase to have these experiences. At the shop, actually you can purchase not only the Samsung product but also the Samsung partners' product and there you can test, you can control the pet feeders and then without your video leave, don't worry to your kids because your kids will play around with the Samsung TV. At the same time, when you try to leave actually you can set the mood quite easily and also for your pet care, this kind of pet finding could be become a perfect solution. So with all those kinds of our initiative and approach, we have strengthened our partnership. So there are now more than 300 brands, which have partnered with us on the SmartThings app. At the same time, it includes more than 3000 devices and these 3000 devices used to be sold online or in very limited shop areas, but by enhancing with the Samsung, actually this kind of product and services and values of their services can be an appeal to not only Samsung shop but also the partner shops. So I have mentioned about the new business opportunities. Obviously, in terms of the scale, energy saving and then this security automated-related home management is by far the most interesting sectors. And not only that, we will lead all those kinds of applications and transform it into the B2B ideas so that we will enhance our B2B total offering services, and as I mentioned before, the senior care. Once we develop these scenarios, it can be connected to a scenario such as more like tele-care marketing and also many different segments of the healthcare domains. And not only that, even for the music and game, changing your places to a different level of experience is far mostly important. So even today we are partnering with so many IoT things partners so that when you're playing game and when you are listening to the music or when you do the partying, actually not only the Samsung product but with our partners' product and our solution, that can enhance their experiences much better. So let me just briefly explain one of a few cases of energy saving partnership and also the security automation partnerships. So this energy saving mode. Actually, with this kind of solution, with our Samsung energy AI solutions, we tried to employ it in more wide ranges with a more wide service coverage, so that currently we cover about 68 countries, but by next year we want to expand it to more than a 100 countries. And especially countries like U.S. and Korea, we call DR Service. DR Service means demand response services. What that means is that actually all of your devices is connected to the energy matter and this energy matter information will be shared into your electric companies, so that you can control your electricity bills, especially in hot energy saving time. Actually, it's really important that you manage your energy skills in much detail so that you can save money. And with new solutions like solar power energy systems, with the variety of solution providers because there are so many different solution providers in world for the solar energies, we try to enhance our technologies or partnering with this kind of solutions. Another thing is home management. So ABB is one of our very important partners. So we jointly participated in a government project such as -- I'm pretty sure that you guys are aware of NEOM Project in Saudi Arabia. So we participated in this project together and became the far most important first vendor for this project. At the same time, with the same ABB in different regional offices like Sweden, there was a huge rural area development project, we also participated and made a huge success. Not only that, with our local partners like Acora or Fibaro, they are more like very important partners. We try to extend our business' coverage to a consultation services and also installation services and retail services. So it's more like a win-win because for them the coverage of the retail becomes a problem and for us, this kind of consultation services and installation services was more like a new area. So by partnering this, both our partners and Samsung can expand our business coverages. So as I've mentioned, according to our vision, the more you own and use Samsung device, actually the smarter the devices become. They make your life much better and richer. It applies to our new business areas. So the more enhanced experience we provide, actually we can empower our customers more, and also it will help us to move into the extended ecosystem. So this is my presentation, and if you have any questions about Samsung CX-MDE strategies, then I'll be more than happy to answer.

Okay, 3 o'clock. Okay, so hello everyone. My name is Harry Yoon, Memory Marketing in Samsung Electronics. Actually, when we talk about memory solutions for AI application, so most people just recall HBM is the only memory solution for AI, but as you know, that's not true. So data is many and also various technologies and solutions for AI application. Today, actually I'm very happy to introduce Samsung is leading the new technologies and solutions for AI applications. I'd like to elaborate on the innovations that we are making along with this to create a new feature. First, let's watch a video clip for AI expansion in our lives. [Presentation]

Okay, what do you think? As you saw in the video, AI is already all around us and it is expanding by the minute. Behind these AI revolutions, there will have been enormous improvements in fundamental technologies. Computation capabilities have grown dramatically and AI models are becoming increasingly sophisticated and advanced. As a result, services like ChatGPT have come out and generative AI has rapidly become the new paradigm in our lives. Stories about generative AI are flourishing in the media and numerous AI services are emerging to be the next big thing in AI. So without a doubt, we can say it is the AI Big Bang. Through the AI Big Bang, AI will fundamentally transform every aspect of our lives. It will make us more productive at work and it will make our lives at home more convenient and enjoyable. It will also completely change the areas of healthcare and manufacturing. With generative AI, we can be anything we want, analyst, a graphic designer or a composer, anything, you name it. How cool is that? These possibilities are just the tip of the iceberg and I believe the changes that are yet to come will go beyond our current imagination. The current AI innovations and services have started in cloud, but they will soon penetrate into other platforms like Edge devices and automotive. These devices are starting to be equipped with computing and memory capabilities sufficient for solving AI models on their own and this trend will drive requirements for memory and storage as well. As my main focus today, I will explain the trend and requirements for each of these three platforms. Okay, first let us look at the cloud. Data centers are the central part of AI services. ChatGPT, Bard and Bing Chat are all services that are provided through data centers. One of the major concerns for these service providers, as you may already know, providing these services is extremely expensive. So TCO is the number one concern. In addition, there are increasingly more concerns related to multi-tenancy, QoS, security, power and thermal. Let us look a little more deeply at the memory system and see how the system drives memory requirements. Simply speaking, what memory has to do is to supply the parameters of an AI model and what storage has to do is to supply training data. That's it. Is it all it does? Very simple need, but when it comes to supplying billions of parameters and tons of training data in a short time, it becomes an extremely complicated task. It is not only complicated but also very important because memory performance becomes the bottleneck of the whole system. A new era of AI-driven memory world has started. Without high bandwidth and density, you cannot fully utilize the expensive GPU you paid for, no bang for your buck. So ironically buying the fastest and largest memory in the market will drive the whole TCO down. As a result, the AI-driven requirement for memory are growing at an unprecedented rate. DDR memory bandwidth has increased exponentially. And the amount of HBM in the server market is growing fast. The need for higher capacity and lower power are greater than ever. Based on this requirement, we have innovated our product. First, we have developed a DDR5 32-gigabyte product, the first of its kind in the industry. We are currently sending out samples to our customers. It has the largest mono density and can support 128-gigabyte high-capacity modules without using TSB. This means these 128-gigabyte modules will be available at a more attractive price. They are just what you have been waiting for. With TSB, on the other hand the DDR5 can support modules of up to 1 terabyte, which is a density you've never seen ever. In addition, it offers 7.2 Gbps, which marks a 50% improvement compared to the previous generation. To extend performance and capacity folder, we've developed MRDIMM product. It is an innovative solution that can support twice the capacity and performance of the existing all DM. The first generation of MRDIMM will offer 8.8 giga bps, while the second generation will offer 12.8 giga bps. This makes it a suitable solution for high-performance computing and AI applications. Next, we have developed the HBM3E, which we have adopted as a Shinebolt. Compared to existing HBM3 product, the bandwidth and capacity have been improved by more than 50%. It offers bandwidth of up to 1,250 gigabyte per second and density of up to 36 gigabyte per cube with Samsung's 12 High technology. As the name implies, it will shed a light on our shared journey into the future of AI. The last on my list of server DRAMs is the GDDR7. This product is another one that we have developed as an industry first, and it offers the fastest per pin speed of up to 32 Gbps with a 20% improvement in power efficiency. Now I'm going to move over to Cyber SSDs. The PM1743, which we are already mass-producing, is the industry's first PCIe Gen5 16 Channel cyber SSD. It offers lovely twice the performance of the previous generation with a 40% improvement in power efficiency. Plus our advanced security telemetry and multi-tenancy features make the product on optimal storage solution for high-performance computing and AI servers. This year we newly added a key server SSD product into our lineup. It is on a channel PCIe, Gen5 SSD. With our groundbreaking controller design. It brings remarkable improvement of 1.8 times in Performance and 1.6 times in power efficiency. We are also investing a lot to explore all kinds of pathfinding avenues, from incremental and sustaining to radical and disruptive. We are on course to be the industry's first provider or product that use next generation interfaces like DDR6, HBM4, and PCIe 6.0. All these technologies will help to provide the bandwidth requirements of the future. In addition, we are exploring the prospect of utilizing new interfaces like CXL and D2D, and even Memory and Logic to enable the shortest reach interfaces. We are also actively exploring radical and disruptive technologies such as Custom HBM, Processing in Memory, and Smart SSD. These technologies have the potential to dramatically improve performance and power, thereby lowering the memory wall. Okay, next is on-device AI. Sorry. Okay. On-device AI is also a very interesting topic these days. Please allow me to cover the technology trend, memory requirement and solutions regarding Edge devices. The need for on-device AI is driven by the pursuit of personalization, better security, high performance and lower cost. This will eventually improve the user experience and bring benefits to service providers and device manufacturers. AI systems in Edge devices will store multiple AI models. Each and every model will be tailored to a specific function it performs. Chatbots, image generation, document summarization and photo editing are examples of such functions. Let's suppose I want to see a portrait of a cat painted by Van Gogh. So upon my request, the model first must be loaded to memory in less than a second, otherwise it'll take my patience to the brink. Then it must be supplied to the computation unit from memory with high bandwidth to make the picture pop out right away. Actually, as I said, definitely is that must be supplied to the computation unit from memory with high bandwidth to make the picture pop out right away. In that respect, do you know what the problem is here? So given the performance of existing memory and storage, this requirement creates a specific limitation for the size of the model and therefore the accuracy of the result. Okay, the requirement just mentioned strongly demands high performance of nearly 100 gigabyte per second for memory and 10 gigabyte per second for storage. It also makes the device a very hungry for capacity and power, so it warrants high capacity and low power memory and storage. So here is what the Samsung memory is preparing to offer for Edge devices. Okay, the first is the LPDDR5X. We are developing a new LPDDR5X product that supports a per pin speed of 9.6 giga bps and bandwidth of 76.8 gigabyte per second. We are planning to invest in production in next year. In addition, we utilized advanced high-K metal gate technology. That is, which is that it improves 30% in power efficiency compared to the previous generation. We also provide LPDDR5X in the form of LPCAMM. The low power nature of LPDDR and the detachable form factor will come in very handy for PCIe. We are currently sampling the first version of LPCAMM, for the first time in the industry. This product supports performance of up to 120 gigabyte per second and it is expected to present a new paradigm in the PC market. We also offer at LLW, which stands for Low Latency Wide I/O DRAM. It is especially designed to provide ultra-high performance of 128 gigabyte per second at an extremely low power 1.2 picojoule per bit which is 70% lower than even the LPDDR5. It is a perfect solution for running an AI model that requires an almost instant response. On XR device, it is such an example and we expect to find more applications in the course of ushering in on-device AI. UFS 5.0 standardization is expected to take place in the first half of 2027. Ahead of this, we are preparing to provide our 4-Lane, UFS 4.0 product. So usually USF 4.0 is using 2-Lane. These 4-Lane devices has a sequential read speed of up to 8.4 gigabyte per second. This will enable the first loading of AI models required in flagship smartphones. We are also preparing PC SSD solution for the AI era. Samsung's first PCIe Gen5 PC SSD is lined up to be released next year and it'll double the sequential rate performance and improve power efficiency by 33% compared to the previous generation. Just like in the cellular segment, we are also exploring many different pathfinding avenues for Edge devices. Our goal includes not only being the industry's first provider of product with LPDDR6 and UFS 5.0 interfaces but also pioneering System in Package and Processing in Memory solutions for future AI requirement. Okay. Now we have reached the time to discuss the future of AI in automotive. Today's autonomous driving technology has arrived at level three, as you well know. At this level, you can take your feet off the pedals and hand off the fuel at your own risk. The standard hardware requirement for this level is around 24 TOPS. To achieve level five and offer the full self-driving capabilities that everyone has dreamed of, the number has to increase to more than 1,000 TOPS. This is a huge jump from what we currently have. Okay, so the desire for level five self-driving and advancement in entertainment systems will eventually drive memory requirements AI has a few thousand gigabyte per second for bandwidth and a few hundred gigabyte for capacity. This is three to four times more than what we currently have. To meet the future demand of the automotive industry, Samsung will provide a diverse portfolio of automotive grade product. First, the industry's first LPDDR5X in our 561-Ball package is under development. It'll only be half the size of the existing 441-Ball package. We will release this product by the first half of 2024. In addition, we are planning to launch an auto-grade SSD, with an SR-IOV feature and on auto-grade GDDR7 in 2024 and 2025 respectively, which fulfill the demand job of our customers need. So far, is that I covered all our technologies and solutions for cloud server Edge devices and automotive applications. As you see this slide, we all know that there are tremendous challenges ahead of us. And the problems we are facing are incredibly complex. These challenges often the strategy across multiple dimensions and cannot be resolved with an isolated approach. Such complexity necessitates a great deal of collaboration in which diverse talent comes together to tackle the issue from multiple angles. We are working to partner with many entities in both industry and academia to find solutions together. And last, when I talk to my friends and colleagues, I often say we are at an inflection point where we have to face the unknown. However, with strong collaboration and partnerships, together, we can transform uncertainty into opportunity. Let's get together and lead into the future. Thank you.

Good afternoon, everyone. Thank you for joining us today. My name is Gibong Jeong. I'm leading business development at Samsung Foundry. We live in the era of generative AI and accelerated computing, that's what Jensen Huang from NVIDIA said this year. I'd like to add one more thing, near memory computing. That's closer to more reality. In other words, the computing paradigm is shifting. Samsung Foundry and AVP, AVP stands for Advanced Packaging. We formed a new organization called AVP. We make customers' computing ideas to reality. As a matter of fact, Samsung has a very unique competitive advantage. I call it GDP, G stands for GAA, D, DRAM, P, advanced packaging. To make innovative AI device or network switch device, you need 3 key enabling technology. That's what GDP stands for. Advanced node processes, DRAM high bandwidth memory, and packaging. Samsung has all 3 and I think that's very unique. Only one company in the world today. That's not what I said. Several customers approached to us and showed their ideas to us. Can you make them? I asked them, why did you come to us? And that's exactly what they said, because you guys have all 3. You need all 3 to make innovative devices they want to make. So today, the theme of my talk is that the synergy between our GAA, DRAM and advanced packaging technologies will ensure innovative computing devices for our customers, including AI. This is the agenda. Let me start with business update. Since our -- wait a second. Our revenue surpassed $18 billion last year and continue to rise. And $18 billion is not a big money, but still second in foundry ranking. It might change if the American -- the U.S. company joined this foundry business, but we are #2 today. And our customers increased, more than doubled or tripled since we break out in year 2017, and then it will increase more than 5x, 5 years from now. In 2005, Samsung entered the foundry businesses and play a pivotal role in pioneering in FinFET technology. Our significant milestones include initiating mass producing using 14-nanometer process in year 2015, which led to the widespread adoption of FinFET technology in the world back then. Building on the success in FinFET, we established an independent business unit in year 2017, into independent business unit. That's where I belong. We achieved another industry first by implementing EUV lithography, now, we're 7-nanometer. That was the first time. A lot of you know guys, I mean, they're one of the choke points of supply chain of semiconductor. That's EUV from ASML. We were the first to use that in semiconductor manufacturing. Once again, we became the first GAA gate all around, manufactured last year, last June. And this year, as I said, we introduced AVP, Advanced Packaging organization. It is advanced packaging business team, and today, I'm going to talk about foundry and AVP's growth potential. Okay. Let me begin with the manufacturing sites. Samsung Foundry offer foundry services across U.S. and Korea. In South Korea, we operate several fab, 8-inch, 12-inch, including like legacy and advanced nodes. In U.S., we have a 12-inch fab in Austin. It's called S2. It's operating more than 10 years. And after 8 years of stable operation, we have enjoyed experience with -- experienced curve effect with FinFET. Our 14-nanometer FinFET is mass-produced in Austin, Texas. We are currently constructing a new manufacturing facility in Taylor, Texas, with production expected to commence in late 2024, which is next year. Okay. Let me move on to the market opportunity. This will be very interesting. I mean, this is not company's vision. It's more of a -- my story. How do I see these things? Let's see. Look at these 2 curves. The lower curve illustrate the foundry market success. The annual growth, on average 11% from year 2000 to 2028, and that growth rate exceeded that of the overall semiconductor market. The semiconductor market about the same time period grows only 4%, and reaching about around $600 billion last year. In contrast, foundry market, it's only a part of the semiconductor market, will grow 11%. Last year, it was around $134 billion. It will grow $187 billion year 2028. This means as time goes on, the importance of the foundry within the semiconductor ecosystem is getting bigger and bigger. That's why you guys are here. I mean, in Korea, ordinary people know about the foundry. That was not that last year. Okay. Let me look at the more details. In year 2028, the projected revenue in the foundry is $187 billion. And if you break down into the segment, the Automotive sector and HPC sector stand out as the fastest-growing segments. In year 2022, HPC sector represent $40 billion out of $134 billion '22, and projected to grow an annual rate of 12%. Meanwhile, the Automotive sector, last year, only a $5 billion market that -- but it will be grown. It doesn't see any absolute value, but with a growth rate of 17%. So those 2 segments are the target segments of Samsung Foundry and AVP, and that's where the computing paradigm shift is happening today. The growth of both HPC and Automotive sector is driven by the rising demand for advanced computing. Advanced computing is integral to the AI accelerators, in-vehicle infotainment system, autonomous driving semiconductor. Okay. This is -- I like these pictures. The computing events is drawn from '80s or '70s until year 2040. And if you look at the words at the top, centralized, distributed, centralized, decentralized, it seems that the idea of the history repeating itself applies to the computing as well. Before the 1990s, I guess most of you guys are too young to remember those guys, the IBM mainframe computer were the norm, and it represents centralized computing. However, in the 1990s, distributor network, decentralized network were prevalent driven by the need for scalable, resilient and decentralized solutions. It leveraged the growing popularity of personal computers, the Internet back then. And again, 2010, centralized computing came back with the rise of cloud computing and big data analytics, centralized computing suits AI technology like deep learning, GPT, which requires substantial computation power and memory. However, experts predict the resurgence of decentralized computing because we need secure, private and low-latency solution, even for AI. That's anybody says. On-device AI, Edge devices. That's why we are talking about that. Jensen said 2 trends are emerging at once: generative AI and accelerated computing. The computing paradigm is shifting toward what I say decentralized Edge device plus energy-efficient computing and near memory computing, and I'm going to explain what does it mean later. Samsung Foundry, memory, AVP, we bring customers computing ideas into reality. The synergies between GAA and wide I/O, high bandwidth DRAM and advanced packaging will make their dreams, their ideas, reality. Let me begin with data center. What do they have? What kind of concerns they have? Number 1 is energy. The HPC industry within the HPC sector, the data centers -- I'm sorry, the significant energy consumption of data centers poses one of the most urgent challenges we must tackle as a semiconductor industry. The significant energy consumption data centers raised environmental concerns, with projections suggesting if you look at the numbers, fourfold increase from year 2012 to 2028, from scale 0.4 to 1.6. The IEA, if you look at the World Energy Outlook 2022 report by the IEA, they suggested, proposed a net zero emission by 2050 roadmap. And if you look at that road map, it accounts for the increased electrical power generation through more power generation factory. And it predicts, without any further policies intervention, government intervention, data center energy use may surpass electrical power generation by 2028. A lot of people are concerned because today, data centers are in the advanced countries where more and more people develop underdeveloped companies, and in third world, the data -- more and more data center will be constructed there, and the energy consumption will increase if you don't do anything politically. Let me take an example. This is the ChatGPT. How do we make 1 data center to provide a sub-second response time? What do you need? It requires significant computing power to run large language model. It runs GPU chip, today is H100 NVIDIA, with big die size, almost over the 100 square millimeter per die per chip. And each consumed, they call it TDP, thermal design power, peak 700 watt. And to make it work, you have to put in 8 GPU into 1 module. They call it a server PCI Express card module, 8 module. To support 1 sub-second response time, you need around 20,000 servers. If you multiply 1, 8, 20,000, in terms of cost and one H100 is around $25,000, right? That card is $25,000. If you multiply all those numbers, it becomes $4 billion just for the hardware cost. $4 billion, that supports ChatGPT sub-second response. In terms of power, if you multiply 700-watt with this, you need more than 100-megawatt data center. We're talking about 100-megawatt of power consumption plus $4 billion hardware cost. We should do something, that's why there are a lot of opportunities are rising. Some people want cheaper chips. Some people want energy -- more energy efficient, more energy-efficient hardware, and that's where we are headed. Interestingly, similar things happened in automotive. The automotive industry is also seeking advanced computing solutions that offer even higher performance and energy efficiency. Why energy efficiency? Today, vehicles, if you look at -- this is the generation of EE electrical/electronic architecture. Today is on the -- around the third generation is mainstream. Vehicles rely on 70 to 100 distributed ECU, engine control units, all over the place in your car. Manage the function such as drivetrain, connectivity, infotainment, body and comfort. In the fifth generation we are headed -- by the way, in the middle, we call it domain centralized. All of those ECUs are classified according to functions, the jargon is domain. So for example, drivetrain-related circuit will be assembled and connectivity is assembled onto one single domain. So in the middle stage, it will be centralized to the domain. We call it domain centralized. But eventually, the fifth-generation central compute units will consolidate this function into one, a link to the control unit sensor actuator through [indiscernible] gateway. The drive -- why to do that? By doing that, you can get economy of scale. You can do a software-defined vehicle because, let's think about it. Let's -- you want to -- let's assume that you want to update the software for each ECU. It's impossible today. But if you consolidate it into one high-speed CPU, you can do an over-the-air updates of your software. So this central compute architecture will open the way for software-defined vehicle. That's why the fifth generation is the way. All these car companies are heading led by the Tesla, and by doing that, you can save the way too. Today, those ECU have a wire, a bunch of wire, in terms of length on average 5,000 feet, on average, 30 to 50-kilogram just for wires. And by going in that direction, the wire will be much simpler, use a high-speed like Ethernet or dedicated high-speed cable. By doing that, you can save weight. Demand for energy-efficient computing is primarily fueled by the development of autonomous driving. The journey, the left-hand corner is the 5 levels of autonomous. I think Level 1, I think you already heard of it. Level 1 is basic driver assistance. Level 2, feet off the gas or brake. Level 3 is hands-off the wheel. And then Level 4, eyes off the road. Finally, Level 5 allows all passengers disengaged from driving. Level 5 requires massive computing performance. This data show -- it's different from 1 to another, but Level 5, we required more than 4,000 teraflops. One teraflop is today's H100. That means you need more than 4 H100 inside your car. That means your car becomes server, it is a moving server. That's why we call, okay, we are solving the same problem. It's a computing problem. But at the same time, data center, they can use a 100-megawatt electricity, but car is running on battery, right? You want to have a longest driving distance, so you have to save the power of battery. That's why you need energy efficiency much more in different reasons. And right-hand corners, this slant line showed 0.1 TOPS per watt, 1 TOPS per watt, 10 TOPS per watt. So this direction has higher performance per watt. If you listen to these days announcement from Jensen rather, [ Elisa ], they're talking about energy efficiency. Even Intel say energy computing. Why do they call it? And they talked about performance per watt, suddenly we are talking about performance per watt. The reason is this: We are not optimizing only performance. We're optimizing both performance and performance per watt, and that is another aspect of the paradigm shift. And let me summarize what's the similarity between HPC and automotive. Just in data centers, next-generation vehicles will rely on -- in the middle on the right hand, advanced process node because they need energy-efficient processors, and GAA is a typical solution for that. And they need higher memory bandwidth, you need a high bandwidth DRAM. And then if you look at the down, advanced packaging, you have to consolidate a lot of different things for central compute. CPU, GPU, NPU, GPU command accelerators, you need advanced packaging technologies. And later, I will explain another reason why they need advantage packaging, this automotive sector. Because in terms of the quantity, you're talking about 50,000 -- 50 million new vehicles every year. It's nothing, right, compared to 1.1 billion of handsets, right? So these people are worried about economies of scale. Automotive, ADAS, semiconductor chip lacks economies of scale. So why do they do have to that? They have to collaborate with other semiconductor companies so they -- some people of the automotive architecture, they prefer chiplet approach. They want to focus on ADAS chiplet. And all other chiplet, they want to buy from off-the-shelf from the chiplet marketplace. So that's why they want to create, they want to join -- I mean a lot of automotive guys tried to recently joined the UCI Express, which is chiplet. The reason behind that is economies of scale, they lack that economies of scale. They want to work with the chiplet vendors from the off-the-shelf. And again, in summary, HPC and automotive share the key enabling technologies in common, which is advanced process node G, high-bandwidth memory DRAM, and advanced packaging. That's what I'm saying, GDP is the key enabling technology elements for both HPC and Automotive. Okay. Let's delve into one by one. Memory wall problem. A lot of people know that. The green line is the performance of the processor, especially on floating point processor. Pentium 2, MMX expansion is around 0.2 gigaflops in year 1996. And then NVIDIA GTX 550 was the first achieved 1 teraflops, 1 teraflops GPU. And then 800, 800, it's around 300 teraflops. And this year, H100, they say they achieved 1,000 teraflop. And if you plot it, it's about 60,000 improvement in 20 years, about 3x in every 2 years. However, DRAM memory bandwidth, it's only 1.4x in every year. Even though we work hard, but still it's not as fast in terms of it. The gap is widening that's -- there a lot of people call the memory wall, or some people say we are wasting GPU. Expensive $25,000 GPU sit idle because of memory, they're waiting for the data coming from memory. So how do you solve the problem? I call it, we call it near memory computing. Near memory computing, or in other words, high bandwidth memory is explained very well in Google's papers, I think, a few years ago. This graph is called Roofline model. X-axis is operational intensity, it's a key keyword. Y-axis is the performance, the gigaflop or teraflops. The X-axis, operational intensity, means data reusability. So larger number in this axis, in X-axis means for the same data, how many operations can do that? Then if higher data reused, you don't have to fetch data new again, right, because you can do -- operate many times on the same data. On the other hand, the smaller operational intensity is very data intensive. You have to read every time you have a new operation, you have to read it. So that means you need a lot of memory exchange transaction, right? And it depends on the application. Some application, for example, calculation projection, it is application 3 type. But certain things like large language model, ChatGPT, they need to fetch upon each time, each time from data. It's like application one. In those applications, data intensive application, your performance is limited, limited not by the performance of the processor but by the memory bandwidth. Look, application 1, performance is limited by memory bandwidth. In that case, if you develop higher bandwidth memory, then this limitation will go up. Your roof will be raised. And then without buying a more expensive processor, you can improve your performance, right? Similar why, that's why these companies like Apple, AMD, NVIDIA, they've developed their own technology for higher bandwidth, near memory. In other words, they put DRAM near to the CPU. Look, Apple M2, they call it LLW, low latency wide aisle. They put DRAM in the same packaging closer to their M2 processors. AMD, they stack SRAM cache on top of their processors, so very close for the same reason. NVIDIA, they use HBM memory, 6 of them next to their H100 processors. Another trend, which is very interesting to me, is big techs, big technology companies, Apple, like Apple, Amazon, Google, Tesla, all of them want to develop their own chip recently. Why is that? They want to choose their own architecture, they want to choose their own process node technology, they want to choose their own custom memory solutions, tailored and optimized for their distinct needs. They have all different needs. Hyperscalers, or even automotive OEM, not only Tesla but other guys, all of them are designing their chip and want to their design chip and they came to us, can you build that for us? Can you build these, guys? Well, it's hard, right? Sometimes it's -- I mean, foundry is a matter of volume of scale. We are dealing with 10,000 wafers per month, not a few thousand per wafers. So even though some ideas are very innovative, we are reluctant to support them. That's the problem we have today. A lot of innovative ideas come to us, but it's hard. It's hard to invest because of the economy of scale. Well, you never know. You never know. There might be a second NVIDIA coming out, third NVIDIA coming out. If these things work and change and win in this computing paradigm, they may order a few thousand wafers, a few thousand wafers per month like NVIDIA today. That's why we are promoting emerging start-up. We kicked off the emerging programs, start-up program this year and then support their ideas and then change our cost structure to promote this new idea. The thing is, computing paradigm is changing. We'd like to capture the rise of this paradigm shift. Okay. This diagram, I didn't draw it. It's done by the CEO of the some of the -- one of the top semiconductor companies. Actually, they think it as a threat. It is one of the fabless semiconductors. Previously, it's all divided like set manufacturers, fabless semiconductors, IT company, foundry. We were doing business. And now, as I said before, the cloud companies and enterprise, they want to design their own chip, and then they want to bypass this semiconductor fabless companies and then directly work -- try to work with foundry. So foundries are getting bigger and this cloud getting bigger, and then traditional semiconductor fabless might struggle. In fact, as a matter of fact, couple or 2 or 3 traditional semiconductor, big conglomerate semiconductor company, fabless companies are struggling these days. The reason is cloud, and other guys want to design their chips. And foundry getting bigger and bigger. Okay. I talked about -- I listed up the trends, what interests me so far. So what's the Samsung strategy? What do you want to do with -- in this environment? Let me talk about that. Once again, I summarized it as a 3 letter, GDP. By the way, the foundry belongs to DS. DS stands for Device Solution, which used to be a semiconductor department of Samsung. Our mission is -- our semiconductor including foundry and memory will give life to your imagination, and my -- the title of my talk came from here. Okay. Let's talk about GAA. Why do we need GAA? As I said before, in earlier days, we talked about shrinking. Moore's Law. How can you make a smaller die size in the performance. One is the max frequency, CPU frequency. How much can you gain? That was the old key performance metric. As I said before, recently, because of data center, automotive, now energy efficiency is coming to the picture again. So if you look at the Y-axis, it's called EEP. EEP stands for energy-efficient performance. Actually, that was defined like 20 or 30 years ago with Stanford professor, Horowitz. It's -- the unit is -- if you look at the bottom, 1 over Joule per second. It's energy times delay. Energy times delay should be the metric. That's the conclusion of his paper, Professor Horowitz. Because if you -- if you optimize only the energy consumption, meaning battery consumption, because of the circuit design theory, you can achieve the optimal battery performance by reducing your performance, meaning wait longer time and then lower the power, lower the power supply. But that's not what I wanted. I mean I want to save battery power, at the same time, want to use the peak memory, so that's why he suggested we have to optimize product of energy and delay. You have to minimize energy, at the same time, process the delay so we didn't have delay. So this performance is 1 over energy delay product, and this is the plot of the GPU. And planar transistor -- by the way, GPU, this performance of this is not only determined by the circuit and process but also the architecture, software is all combined together. And this is the performance like 2.2x per 2 year. This is a combination of architecture, circuit, process technology together. But as a foundry player, if you look at only the process technology only, transistor architecture only, planar transistor already. If you look at the dot, it starts to curve downwards. That means there's a wall. So with the planar, you cannot go beyond the EE performance. That's why we introduced FinFET around year 2013. Now, we introduced GAA last year. In other words, to achieve energy-efficiency that data center and automotive guys want, GAA is the way. So the G of the GDP. GAA, it will be the key element of this computing paradigm shift. That's why every company, every foundry company, even though the name are different, we call it GAA. Our competitor call it nanosheet. Some other people say [indiscernible]. But the GAA will be the key element of the new paradigm shift, and we are the first. We mass produced last year. What is GAA? So GAA is gate all around. I wish I had a pointer. If you -- right-hand most, the channel is surrounded by 4 sides, right, so we can control the channel by 4 sides. On the other hand, FinFET, you can control mainly your 2 sides. You look like fin of the shark, and you can control the channel by the voltage of the 2 sides. And before, it was a planar where you can control the channel with 1 side of the above gate, gate above the channel. By doing that, you can shut off the current in off state. So you get running water, you can cut off supply. By doing that, you can say you can reduce the leakage current, and then you can reduce the supply voltage, and supply voltage is directly related to the power consumption. So that's the GAA. But look at it, it's very hard to make, right? So that's why the process technology out of the 3 is most complicated. And then wafer cost suddenly goes up and up. I mean you're hurting about $20,000, $25,000 per wafer because of that process complexity. So let me show you the video. How does it, GAA work. [Presentation]

Okay. We provide FinFET, and GAA -- our GAA start from 3-nanometer. Okay. That was the G part of the GDP. Now, I'll talk about D, DRAM part. I told you, we think it's a near memory computing paradigm shift. Near-memory solves -- I mean catches 2 rabbits at a time. One rabbit, energy efficiency; second rabbit, latency or high bandwidth. So the 3 stages. One is typical DIMM -- DRAM card. If you open up your PC, there's a DRAM card. Typically 2, right? You buy DIMM cards and said, your processor Intel, AMD processor is far ahead, right? You have to -- your signal has to go all the way around. By doing that, you spend time and you consume a lot of power -- consume -- you can heat. That's why you need a fan, right? To solve this energy consumption problem and latency problem, you have to put closer to processor and memory sits closer and closer. You saw that 3 companies, right, Apple, AMD, what was that? NVIDIA. They do this kind of near memory approach with a different technology. One example is silicon interposer. They call it 2.5D or CoWoS packaging, so your DRAM stack HPM sit next to the lodging H100 and connected through the silicon interposer. And by doing that, your energy transfer energy is decreased from 12% to 3.5%, latency decrease from 7.5% to 6%. Even further, if you stack in 3D, 3D, you can save it further. That's where we're headed. So today, 2.5 the HPM, it sits on a silicon interposer, and then they are talking about 2.5D. And -- okay. The next step will be the 3D on topic of memory and logic. By doing that, you can minimize latency and minimize power. Okay. Also for automotive, we are doing a nonvolatile memory, and then our solution is embedded MRAM. And we recently announced 5-nanometer advanced node FinFET MRAM, skip it. Okay. Now third part, P part of the GDP, advanced packaging. I'm sure you guys have heard of a lot of things about packaging, and there are multiple motivation. One motivation is the cost. If you look at the curve, that is the yielded wafer cost for the same area. Suddenly, it's like hockey stick. Suddenly, cost went up from 7-nanometer. The reason because they start to use EUV. EUV is like, how much, $50 million per machine. That raised the wafer cost, right? And instead of using that advanced node, they want to split it. They use N-1, older legacy wafers in doing this way. That's 1 motivation, cost reduction. But more, for example, automotive guy because they lack economies of scale, they want to reuse design from others or they want to reuse older design for a new development, if for the same company. So design reuse will be collaboration with the -- all your work or other people's work. Another motivation. Third motivation, like NVIDIA, AMD, they want to put more transistors into their system. But today, 100 billion transistor was a limit, even though you use GAA. So if you put more than 150 billion, like 500 billion transistor, you have to put more. For example, MI300 AMD, they used 150 billion with several chiplet, the GPU, CPU. So those are the third category of motivation. And our ADP organization, provide diverse technology offerings, advanced packaging. The technology they have, ADP has, both HPC mobile segments and a bunch of a different configuration of logic and memory, near memory configuration. Side by side in the posture, cache DRAM on logic, bufferless HBM on logic, SRAM on logic, like cache or LLW, LPDDR options for the mobile. And another problem we have to solve is even though you have a technology, you need to support design flow, design technology. And the methodology to test, when you design a single mono die, you can do an EDA. We have already developed tools. But when you have a multiple dies, you have to simulate not only signal integrity but also thermal problems and other reliability factors. So we work together with the EDA and OSAT, the 7 categories of our ecosystem. Our ecosystem has grown. So we introduced and announced multi-die integration lines this year. So about 20 companies joined it, even our competitor joined it. And then eventually, this is my view. I mean becoming a -- democratization chip design. Democratization of semiconductor manufacturing. That was the concept of foundry, right? Even though you don't have a fab, you can make a fab. That's democratization of manufacturing. But once they're further, not only manufacturing, even design. Now, once you have any system ideas, you can use this chiplet foundry. And they design services, you can reuse others' design. By doing that, you can design the chip itself. So it's 1 step further next to the democratization of chip manufacturing. Now, the era of democratization of chip design is coming. System innovator, who has great ideas, they can easily build a very complex design with this foundry ADP, memory and design services. Okay. And to make this happen, I talked about EDA and the simulation and verification too. At the same time, multiple time vendors are supplying it. It has to be interoperable, so we need a die-to-die interface standardization. That's the third piece. So chiplet standard, interface standard is a must. So UCIe is a mainstream interface. It was formed 2 years ago, Intel is a partner. A lot of big guys already joined it, and Samsung is chairing this UCIe standards. So let me show you some clip. [Presentation]

So I talked about GDP, 3 components one by one. This is not just a plan. We're already shipping, producing AI chip today. Some customers use our 14-nanometer in the -- selling AI accelerator, then developing 4-nanometer accelerator. Even for the automotive, #1 EV company use our 14-nanometer ADAS. So you -- a lot of our chips are running today. They are developing 5-nanometer next version, fully self-driving chipset. So we have pretty high market share today of the automotive driving, and 8-nanometer IVI is all over the world. It's one of the 3, top 3 automotive CPU companies. Okay. This is, I think, the last slide. So I was -- okay, what should I do for the next slide? I asked ChatGPT about the future of Samsung semiconductor. The context, I gave them -- I gave it as energy-efficient computing, computing paradigm, near computing, I give some cues, ideas, and then in response the answer is as follows: Samsung Foundry is heavily investing in advanced process technology like GAA and positioned to assist companies in designing energy-efficient computing chips. Additionally, Samsung is optimizing high bandwidth DRAM solutions and delivering customized DRAM solution. Even I didn't know, but it said delivering customized DRAM solution to meet individual customers' computing needs. Advanced packaging plays a pivotal role in enhancing energy efficiency and boosting overall performance. The synergies between GAA chips, custom DRAM and advanced packaging will ensure innovative solution for our customers in the evolving computing paradigm shift. That's -- I don't know, maybe I gave too much hint, but that's exactly what I wanted to hear from ChatGPT. Samsung Foundry and AVP Memory, together with our partners, we will give life to your imagination. Bring customers' computing ideas to reality. Thank you.

So thank you very much for spending your valuable afternoon with us today, and I'd also like to thank JPMorgan for providing the venue for today's event. So hopefully, after the sessions today, we're able to share with you our growth strategies for our -- based on our technology leadership in display to expand to new areas including IT, automotive, and going in down in size all the way to micro-LEDs. In our memory size, our leadership will expand our roles in AI to not only HBM, but even much beyond that in the future as well. Also, our last session about GDP growth in foundry based on the GAA, DRAM and packaging. And also, as the world's largest provider of connected devices, how we aim to enhance customer experiences based on multiple connected devices. And for those who attended that session, you even learned what CX-MDE stands for. So with that, I'd like to wrap up today's 2023 Samsung Investor Forum. Once again, thank you very much, and I hope you have a great evening. Cheers.