​As the release of Samsung’s (005930.KS) Quantum Dot/OLED TV nears (likely at CES in January), Samsung Display (pvt) has released a promo slide show extolling the virtues of the technology, albeit mostly a simple comparison against LCD technology.  Since the promo was in Korean, we translated the 10 slides and put them in a PowerPoint below.

​Facial recognition is a topic that can be polarizing.  Some say it is an excellent way for devices to recognize users and others say it is a way for governments to track you wherever you go.  That said, the technology continues to improve both on the hardware and software side, and as that happens, the legitimacy of facila recognition improves.  It seems that Kakao Pay (035720.KS) has decided to test out that legitimacy by using facial recognition as a payment method in a number of unmanned stores early next year.  Shinhan Card (SHG), Korea’s largest card company has also begun using facial recognition for cardless payments at a number of retail locations in Seoul.
The Kakao system, which operates off-line so as not to be vulnerable to remote hackers, allows the customer to purchase items without any payment application or card using a 3D camera and anti-spoofing software that makes sure the face being scanned is not ‘camouflaged’.  The unmanned system requires that the user registers his/her face and card information at an separate unmanned register, after which you can use any register that accepts Kakao Pay by just looking at the camera..  The company is developing additional software that can estimate a customer’s age and gender.
Fortunately or unfortunately unmanned facial recognition systems for commercial applications are most common in China and Russia, with a fare payment system using facial recognition installed in 240 stations in Moscow, and a similar system in China, that allows the user to go through the system without taking their mask off.  That said, given the more restrictive governments in these locations, the odds are that those who might try to ‘game’ the system would be less likely than in an environment where such small violations might be overlooked. 
There are a number of bio-recognition tests that such systems have to pass in order to be certified by the government in Korea, with a number of banks and financial institutions having completed certification, but as is always the case with CE products, they tend to work differently in the real world, so we remain a bit skeptical about how quickly unmanned facial recognition payment systems will be deployed.  Retailers are certainly in favor as it would cut labor costs and associated benefits, but if it is found that the systems are more difficult for customers to use (think about the success rate for your groceries at the self-check registers “Item not found – Help is on the way”) or are not 100% spoof-proof, adoption might take some time, and could you imagine how many customers might be offended if facial recognition software identifies gender incorrectly?  The Twitter blowback would be massive….

​AnimalGo (pvt) is a private company in South Korea that specializes in pet products and services.  Based on the company’s ‘white paper’, the company’s AnimalGo App, which has seen more than 100,000 downloads, uses AI deep learning to analyze the pedigree of pets.  The service goes further with a Cologuard™-line pet stool sample analysis to predict pet health conditions, along with a number of companion apps that provide access to and discounts for the pet products that the ‘project’ sells, such as a pet hotel, pet kindergarten, pet swimming pool, pet beauty service (they come to you), pet insurance, and a pet café., all of which are available for franchising.
While we were fascinated by the AI Deep Learning pet pedigree analysis, which is said to be able to recognize the pedigree of a dog or cat through a photo, but is said to be able to pinpoint the ratios of contributions of various ‘blood types’ in your pet; sort of a ’23 and Me’ for pets.  By parsing the photo into RGB colors and separating features to compare to a database, your pet can be matched to ‘parents’ with similar features, giving a family history for your pet.  By submitting photos potential users can earn ‘points’ toward company products, and of course, block chain and AI algorithms and mentioned continuously.
But what is missing from this technology forward AI, block chain, and video company?  The Meta-verse, and to that end the ‘project’ has hired Jo Seong-ho, who is an adjunct professor at the AI Research Center and the Robot Interdisciplinary Program at the Korea Advanced Institute of Science & Technology (KAIST), a prestigious South Korean institution.  He will guide the company’s Metaverse project, which scans biometric data from your companion animal and creates an AI avatar for your pet in virtual space.  The data is collected through an IoT color placed on the real animal, which then allows the avatar to mimic the movements of the physical pet in the Meta-verse in real time.  This is said to help physical pet owners to better understand their pet’s movements and health, leading to possible health check-up services and insurance.  With 29.7% of South Korean households having a companion pet and ~64.1% of those said to be using ‘pet tech’, the company and the professor are expecting the pet Meta-verse to expand quickly once deployed.  Need we say more about Meta-verse hype?

3Q saw a continuation of the strong semiconductor equipment spending seen for this year and much of 2020, setting a new record of $26.8b in the quarter, up 7.7% q/q and 38.3% y/y.  This puts cumulative semiconductor equipment spending up 45.5% y/y through 3Q, already surpassing the 20209 full year total.  Semiconductor equipment spending in 3Q was driven by Taiwan (up 45.4% q/q and up 54.3% y/y) and North America (up 36.3% q/q and up 67.2% y/y), with both China and South Korea down on a q/q basis.  The ROW category was actually the biggest gainer (up 60.7% q/q and up 125% y/y) with equipment spending in Asia driving the category.
Industry estimates made as late as September of this year indicate that 4Q spending is expected to be down q/q but some of the regional forecasts were already exceeded by the end of 3Q.  Semiconductor equipment spending this year will come in between $93m and $96m based on industry estimates.  Earlier estimates for full year spending were to reach $71.9b, which was exceeded in 3Q.  Expectations for 2022 are for semiconductor equipment spending of $100m, with much of the growth coming from South Korea, Taiwan and Europe.  There were 76 ongoing fab construction projects underway this year (~$18b) with 60 continuing into 2022, while 16 new projects that will lead to volume production fabs are expected to be added in 2022 ($27b total).  

​What do these names have in common?  Decentraland (pvt), The Sandbox (pvt), Upland (pvt) and Somnium (pvt).  They all sell real estate but do not compete against Coldwell Banker (RLGY) or ReMax (RMAX) as the land they sell does not actually exists, or at least not in the physical world and companies like this are making money. And generating considerable interest among ‘investors’ that range from individuals to real estate ‘virtual’ trusts.  While virtual real estate might seem like the ultimate speculative bubble, especially given China’s recent abhorrence for anything bitcoiny, but that doesn’t seem to be stopping some from providing services necessary to develop such virtual worlds.
Alibaba (BABA) has just set up a subsidiary that is offering ‘platform-as-a-service’ to those seeking to develop their ‘parcels’ or build Metaverse sites and Alibaba’s gaming division launched its own brand in September, offering cloud-based games, and a developer platform within the PAAS framework.  This comes despite the government’s stringent new policies about speculation surrounding the Metaverse, which cites virtual real estate as an example of the type of speculation that could lead to significant losses, fraud, illegal fundraising, and money laundering.  The policies however are not quite specific enough whe  it comes to NFTs (Non-fungible Tokens) or similar virtual currencies, which has left the door open, at least a little bit, for some forms of virtual speculation, and an even wider door for those companies that will provide the tools for building virtual platforms.
The Meta-verse is so new that companies are popping up like mushrooms, ready to help you or companies do, well, whatever you want to do, in the Metaverse.  DAPPS (Decentralized Applications) from dozens of companies,  NFTs, DeFi (Decentralized Finance), and a vast number of tokens particular to platforms, make the Metaverse a complex and hard-to-understand concept for most, but a constant dose of ‘Meta-verse’ in the news, seems to keep things hopping and almost every tech company has to be looking at a way they can monetize the virtual world, especially when celebrities mention a virtual purchase or one of those virtual real estate trusts mention that they believe the value of their virtual land portfolio is 10x its purchase price.
The Metaverse Group’s (pvt) parent company Token.com (COIN.CN) recently indicated that a 116 parcel estate in the heart of Fashion Street (in Decentraland) sold for an equivalent of $2.5m, or 618,000 ‘mana’, the currency that Decentraland uses, with the idea that the new owner could lease out stores on the parcel, becoming a virtual landlord..  While this sounds like absolute foolishness to many, it does have roots in reality, or almost reality, as speculation in the domain name market when the internet was young, can be likened to what is similar in the Metaverse.  Speculators are hoping that their particular piece of virtual real estate will become infinitely more valuable, just as ‘Insurance.com’ ($35.6m) or ‘Hotels.com’ ($11m) became more valuable as the internet expanded.  That said, many domain names and tags have been bought for just that purpose, with almost as many essentially worthless today.  On sale currently are domain names with the tag ‘.boutique’ ($4.99 for 1 year), ‘.city’ (($7.99), or ‘.dog’ (($9.99), which give some indication as to the volatility that might occur with virtual real estate.  There is no free lunch.

Earlier this month Chinese smartphone brand Oppo (pvt) teased with a few words introducing Oppo’s first flagship foldable smartphone, the Find N, which was in development for 4 years and has morphed through 6 generations.  The phone is expected to be released tonight (Wednesday) in China as part of Oppo’s annual Inno Day.  While the picture (Fig 4) shows little, much of the speculation is that the new foldable will look similar to the Samsung (005930.KS) Z Fold, with Korean media indicating that Samsung Display (pvt) is the producer of the 7.1” folding display, while BOE (200725.CH) is the producer of the 5.45” external display. 
As SDC uses UTG (Ultra-thin glass) rather than polyimide films as a cover for the foldable display, the Find N is the first Chinese foldable to employ UTG, with previous foldables from Huawei (pvt) and Xiaomi (1810.HK) using polyimide.  The Oppo foldable is also the first Chinese company to use an LTPO (Low-temperature poly-silicon) backplane, again a characteristic of SDC’s most recent foldables.  While Chinese panel producers have supplied folding displays to Xiaomi for its previous foldable, it is expected that the upcoming Xiaomi Mi Mix Fold 2 will use an 8” SDC folding display (with UTG) and a Chinastar (pvt) 6.5” external display , while both foldable and external displays were supplied by Chinastar on the previous model.  Vivo (pvt), sister company to Oppo, is also expected to use the SDC foldable and a BOE external display for its foldable in 2022. 
Huawei was thought to also have been considering using SDC’s foldables for its upcoming upgrade to its Mate X2 foldable but seems to have decided to remain with foldable supplier BOE, along with Honor (pvt), a former Huawei brand, who is also expected to use BOE for its foldable next year.  Samsung Display is the share leader (3Q) in the flexible OLED market, with a 62.2% share by our calculations, but does face competition from BOE and LG Display (LPL), although combined they still have only a 28.9% share, but when it comes to foldables, SDC is definitely the leader with a share over 90% this year. 
Samsung’s push to spread foldables into the laptop segment has given them a new market, also one that they dominate, and while the unit volumes for foldable notebooks are still relatively small, the surface area is considerable large than for smartphones..  With parent Samsung Electronics’ foldable volume expected to double in 2022 from 7m to 14m units, and the overall market for foldable expected to double from 9m to between 18m and 19.5m units, we expect SDC to maintain a substantial lead over competitors in the foldable space despite the competition from BOE.  It might be a bit more difficult to maintain such a high share in 2023, as Chinese foldable producers sharpen their production skills, but SDC has been one step ahead since day one and will likely remain so in 2023 and beyond.
 

Estimates for OLED notebook production this year are for between 5.3m and 5.6m units, up from less than 1m last year, growing to ~7.5m in 2022 and 15m in 2023, representing not only explosive unit volume growth but significant demand growth on an m2 basis, one of the reasons behind OLED capacity expansion projects in China.  SDC will likely see it overall share of the flexible OLED market decline on a unit basis, but will likely see its actual share of OLED panels for notebooks on an m2 basis increase in 2022, as it has established relationships with almost all of the notebook brands that offer OLED displays, especially its parent, who is expected to have a 20% share of the OLED notebook market itself this year, second only to Asus (2357.TT).
​

While Hannstar (6116.TT) has yet to release its November sales report, AU Optronics (2409.TT) and Innolux (3481.TT) reported mixed results for November.  AU Optronics reported sales of NT$ 30..88b ($26.14m US), up 1.6% m/m and up 19.0% y/y.  AUO no longer reports large or small panel shipments on a monthly basis.   Typically November is flat to up slightly m/m, so AUO performed well relative to its 5 year averages.  Innolux reported sales of NT$ 26.59b ($22.51M US), down 0.8% m/m and up 1.9% y/y. Large Panel shipments were 11.88m units, down 1.4% m/m and down 1.8% y/y, while small panel shipments were 27.93m units, up 13.6% m/m and up 2.5% y/y.  To characterize November for these two panel producers, we would say the month was a bit above the norm, but by a small amount. 
With declining sales across the panel space each month since August, some bounce could be expected but we presume that the general trend of declining large panel prices and weakening IT panel prices continues, although year-end orders might obscure that trend a bit.  December tends to be a volatile month for panel producers, with the 5 year average being down 2.5%, but 2020 saw a 3.3% industry sales gain in December after many years of December monthly declines, so our faith in the averages is a bit diminished. 
That said, assuming a flat December, AUO’s full year sales will be up 36.5%, with 48.3% of sales in 1H and 51.7% of sales in 2H, not far from the 5 year average split of 47.9% and 52.1%.  As TV panel prices have fallen faster than IT panel prices, AUO’s shift away from TV panel production and more toward IT panel production has helped to stave off the effects of the rapid TV panel price declines.  While we certainly give management the credit for such foresight, we expect some of that transition was also a result of the increasing pressure on AUO’s TV panel business from Chinese suppliers, who have continued to add capacity and dominate the space in terms of unit volume.
Innolux has been a bit more oriented toward large panel production, which has affected 2H sales a bit more than AUO.   2H share of the full year sales total was 49.4% vs. Innolux’s 5 year average of 53.1%, showing the effects of the more pronounced drop in TV panel prices in 2H.  The full year for Innolux was still up 29.71% (assuming a flat December) but monthly y/y comparisons are getting more difficult, just barely keeping pace with last year, while Innolux faces the same competitive issues from Chinese panel producers as does AUO.

Taiwan based Compal (2324.TT), a well-known OEM and assembler of CE products has announced a partnership with Nvidia (NVDA) to build a Metaverse “specifically designed for the enterprise segment.”  While Nvidia is just one of a number of Compal partners in its desire to join the Metaverse supply chain, the announcement is an add-on to the company’s announcement earlier this week about the release of its XR Cube Open Cloud Metaverse platform, which includes Compal’s 5G smart application solutions for IoT released last year.  The new Metaverse platform is described as a combination of Compal’s software and hardware and that of its partners that will “construct a virtual world that is fully integrated with the real world, and accelerates the realization of the ‘Meta-Universe’ in various industries and people’s life situations.”
Lots of statistics about XR (Extended Reality and its growth rate over the next few years and a mention of Compal’s own 5G ‘hybrid reality’ glasses (released last year) follow in the PR, but we were more interested in what the XR Cube Open Cloud Metaverse platform actual is.  Compal is trying to build what it calls a “Meta-cosmic ecosystem” (a new Meta every day keeps the doctor away) which is actually a service that is trying to consolidate all of the cloud or edge infrastructure needed by Metaverse developers.  In the case of new partner Nvidia, it would mean incorporating Nvidia’s Cloud XR platform which delivers streaming AR/VR across 5G and Wi-Fi from data center or edge networks to Android and Windows devices. 
Compal shows a graphic with a variety of names surrounding potential users and some of the structure needed to bring the world into the Metaverse, although some, such as Qualcomm (QCOM) were really part of the Compal 5G service, with the remainder being various VR/AR software developers and application providers.  Compal cites one of those partners, who will be able to create an interactive personal universe experience (aka UGC or user generated content) in real time using the XR Cube API, (we note the use of ‘will be’) and will not have to adapt the process to each VR platform, so what it comes down to is Compal is providing a development interface that would help to smooth the ability of Metaverse content developers to generate real-time Metaverse content.
Given that there are a multitude of Metaverse content creation tools like Unreal Engine (pvt), Sumerian (AMZN), Maya (ADSK), Blender (pvt), and Unity (U) and no basic platform for the Metaverse, like there are for search and social media, the field for helping developers do more than just create a game or virtual world is wide open.  That said, companies like Amazon and Google (GOOG) have such a significant amount of computing infrastructure that they will inevitably dominate the space.  In the interim, we expect the Compal’s of the world to provide as much connectivity help as possible to developers to help the Metaverse get real-time traction.  That said, most important would be improving VR hardware, as without high quality and physically usable VR displays, the Metaverse will be limited to a small subset of internet users.

Samsung is big on foldables and has taken that to include rollables, meaning those devices where the screen can be coiled inside of the device.  Taking it to another level, Samsung has filed a patent for another rollable device; this time a watch.  With typical watch displays ranging from ~1.4” to 1.9”, Samsung seems to feel that the user might need more room to see what time it is and has developed (IP) the mechanism for a rollable watch display that can increase in size just by pulling the top part of the watch frame.  The increase in size was not specified but from the drawings it would seem to have double the area.
But Samsung did not stop with just a rollable display, but also includes S-Pen capabilities and the ability to turn the display from vertical orientation to horizontal without removing it from your wrist and being a ‘smart’ device, the watch itself can adjust it size according to the user’s content.  So if you want to see the time, the display would stay normal size, but when reviewing an e-mail or text, the display would elongate to show more of the message or allow for larger type.  When extended, the user will be able to use the under-display camera, allowing the watch to be used for taking pictures (“Bond, James Bond”) and of course, there will be no need to pair the device with a smartphone as it will operate as a standalone cellular device.
While Samsung is pushing the limits of current technology with such a device, they are not the first to patent such an idea as China’s Oppo (pvt) received a patent for a foldable watch all the way back in 2019, although it was a very different display set-up.  Given the complexity of such a device (the patent is 113 pages), we expect it might be some time before Samsung makes such a watch available to consumers however the patent drawings were quite specific on how the mechanical components might work, likely more to keep others from duplicating the idea than Samsung having proved out the mechanics on a scalable basis.  Interesting idea though.

Ceramic Capacitor) fabrication facility that will be completed in March of 2023.  As one of the few volume producers of MLCCs in an industry that remains conservative about capacity growth, the expansion project, which is expected to cost ~$150m marks the first major capacity expansion project for the company in many years.  With competitors Yageo (2327.TT) and Murata (6981.JP) adding capacity (production dates) in 2020 and 2019 respectively, the industry has been relatively careful about adding capacity despite the increasing demand for MLCCs.  We believe this is a function of problems associated with a demand cycle back in 2016 – 2017, when the industry added capacity quickly and additional demand was slow to materialize, leading to a significant drop in MLCC prices.
However MLCCs, which as small and hard to produce capacitors that appear in many CE, automotive, industrial, and defense products, continue to see strong demand despite supply chain issues that have slowed orders for some CE products.  While the automotive industry has had its issues during the COVID-19 pandemic, demand for MLCCs continues to grow and has offset the slower demand from CE.  As hybrid and fully electric cars become more entrenched, MLCC content in that space will continue to grow and we expect the industry to remain tight on a long-term basis, especially as entry into the production side of MLCCs is quite difficult for those that do not have the considerable expertise to make these small components. 
Inventory levels for MLCCs have been declining this year, and while this would not be considered a ‘shortage’ situation, lead times have been extended as automotive and 5G demand has picked up, despite the silicon supply chain issues that have limited broader CE demand late this year.  We are less concerned with what are relatively short-term movements in MLCC lead times, inventory levels, and pricing, as we are in longer-term demand, and industry leader Murata has provided internal research that gives a picture of current MLCC usage in a number of products, giving some indication of the volumes needed for this component.
5G is a driver for increasing MLCC content in smartphones and the growth in Hybrid and EV is the driver for the automotive segments.  While we believe some of the growth estimates for MLCC demand in 2022 are quite aggressive (25% is the highest we’ve seen), EV unit growth alone in just the first half of the year, has been averaging over 170% y/y for light vehicles on a global basis, and Fig. 2 shows the historic sales growth of Hybrid and EV vehicles in the US, the 3rd largest market for plug-ins behind China and Germany last year.