​Veeco (VECO) and Aixtron (AIXA.GR) are the two main players in the MOCVD (Metal Organic Chemical Vapor Deposition) market, with such tools being the crux of the LED manufacturing business.  Veeco is currently listed on NASDAQ, however Aixtron, a German company listed on the Frankfurt Exchange, terminated their sponsored US ADR program in 2017.  It seems that Citibank (C) has decided to create an ADR program for Aixtron by offering ADRs representing two Aixtron shares, however this is an unsponsored ADR, which means Aixtron will not be reporting results under US rules and is not involved with any aspect of the security.  The Aixtron shares underlying the ADRs are held exclusively by Citibank and are not new or existing shares being offered by the company.  The Citibank ADRs are currently being offered at a premium to their intrinsic value.

China is big on market share and in the LCD display space China continues to grow its share.  In some cases that growth is financed by the companies themselves, although that usually entails finding investors for a portion of the funding.  Other Chinese display producers have large holding company parents that can fund capacity growth through a variety of sources both public and private, while still others must rely on state or local funding.  In almost all cases however, some of the investor funding comes from state run organizations, and while they might be once removed from direct state oversight, they are still beholden to the government in a loose sense.  Most of the more direct funding is done through provincial or city governments who look at LCD production as a lucrative way to generate jobs, local publicity, tax dollars, and regional revenue, aside from political cache.   This system works quite well as long as two criteria are met.  First, the panel manufacturer has a talented enough staff to generate sufficient yield, and second, panel prices are above cash costs.
As the LCD production space is cyclical, one cannot always guarantee that either or both criteria can be met at all times, but when they are it sets the tone for the ‘gain share’ buzzer to go off, panel producers see themselves as the ‘leader’ in the space, and they start developing plans for further capacity expansion.   Local governments, looking at current profitability, seem to assume that this will continue both for all time and regardless of capacity and are happy to keep the local coffers open with just the suggestion that ‘the world has changed’ or ‘It’s not like last cycle’.  The problem is that nothing lasts forever in a cyclical world and the cost ($3b - $4b) to build a greenfield LCD fab is high.  Such construction, even in China, who is known for their ability to build fab structures quite quickly, takes at least 18 months from planning to initial production and during that period panel prices can and do fluctuate wildly.
China is still bent on gaining share and our updated LCD industry model cites a number of Chinese fab expansion projects that will continue through 2022 and beyond.  China will continue to gain share, particularly as South Korean producers back away from LCD panel production, but both demand and panel pricing must also stay at current levels or increase to keep capacity utilization high enough to maintain profitability, and that is a far more difficult question than ‘Should we finance a new fab in our city?’  Those decisions tend to be based on provincial competition and less on the effect it will have on the industry as a whole, but we doubt, even if the cyclicality of the industry were laid out in front of said decision makers, it would have much impact, as by the time the new fab is completed, local politicians would have moved on to more important state positions and the lack of profitability would become someone else’s problem.
It was easy for Chinese producers and financing sources to make decisions given the length of the up-cycle we have just seen, but even the short-term drops we have seen in TV panel prices should be a harbinger of what is likely to occur if unbridled capacity expansion continues.  Based on the projects underway, and those most likely to be added, we have charted the TV capacity share in Fig. 3.  China will certainly be the share leader going forward but the question really is, will that be the right place to be over the next few years?  It is easy to be optimistic when things are good, but it’s better to be realistic.  Bigger is not always better.

​Samsung (005930.KS) has had considerable success with what it calls the “Fan Editions” of a number of its smartphones, although the original “FE” model was a refurbished Galaxy Note 7, a phone that was recalled due to a overheating battery issue that caused some of the units to catch fire.  The popularity of the original Note 7, before the problems were discovered, led Samsung to replace the batteries in unsold Note 7s and sell them as the “Note 7 Fan Edition” about a year after the original release, which both attracted considerable attention and allowed Samsung to recoup a bit of the loss incurred during the recall.   The concept of a limited number of units that were similar to the original model, selling at a discount, gave a bit of cache to what might have looked like a heavily discount flagship smartphone, if the ‘FE’ moniker was not attached, so Samsung continued the idea with the release of the Galaxy S21 FE, a $700 version of the Galaxy S21 flagship, which sold for $1,000.  The S20 FE, released in October of last year (S20 was released in March 2020) also proved popular, with Samsung expecting to do the same with this year’s Galaxy S21.
Unfortunately, as we noted a number of times in June that the Galaxy S21 FE, which was originally expected to be released in August, was expected to be sold only in the US and Europe, due to a shortage of Qualcomm (QCOM) application processors, and later that production was halted for the same reason, with vague plans about pushing out the release until October.  It seems that Samsung has put the S21 FE back in production and while plans have yet to be finalized, the S21 FE is expected to be released some time in 4Q although likely in limited quantities at the onset. 
The good news is that Samsung’s return to production of the FE indicates that the shortage of application processors, at least for Samsung, is easing a bit, which could allow Samsung to have at least some chance of getting close to its internal smartphone shipment and share goals, especially given the popularity of the Fan Edition series.  That said, much will be determined by how many units Samsung can garner and when it is released, as it is already a month late and likely will be a bit capacity constrained going forward.  While overall foundry capacity remains tight, especially at 28nm and larger nodes, the Qualcomm AP is produced at Taiwan Semi (TSM) on 7nm (2nd Gen) so it is competing with high-end customers for fab time, rather than against every other foundry customer.  This obviously did not matter over the last month or so given the production halt, but Samsung seems to have been able to regain fab time at TSM recently, which could give them the ability to push FE production and squeeze out product up to and during the holidays.

​China’s smartphone shipments declined by 15.2% m/m and by 9.6% y/y in August.   Looking at the year thus far shipments have been 227m units, up 29.5% y/y but against the very weak early COVID-19 days last year.  The quarterly comparisons for this year are 1Q, up 100.2% y/y, 2Q down 26.9%, and thus far in 3Q up 7.7%.  But from a more qualitative point of view, the first four months of this year were volatile in the Chinese smartphone market, averaging 31.4m units/month and m/m swings between -45.6% and +65.6%, while the next four months have been relatively steady, averaging 25.4m units/month, with swings between -16.4% m/m and +11.8%.
There was a big jump in new models released in August, increasing from 26 in July to 57 in August, while this in itself is not unusual in that 3Q is usually the peak quarter during which new smartphones are released, but going back as far as 2017 September has always been the strongest month for new smartphone releases in China.  That would imply an even greater number of new models released on th Mainland in September (typically the increase of new models released in September on a m/m basis has been between 28.8% to 100%) unless the desire to anticipate the release of Apple’s (AAPL) iPhone 13 family pulled in many of those releases (we believe the former).
5G smartphone shipments in China were 17.69m units, down 22.5% m/m and up 9.4% y/y in August, a relatively weak month after a strong July, leading to a 72.8% share of total shipments.  While a bit on the low side in terms of absolute 5G shipments, 5G shipments have maintained a share above 70% since February of this year and we expect little change from that metric going forward, especially under the above assumption that September this year will see an increasing number of new model releases in China.  The share of 5G models out of all new models released in China this year has been averaging about 50%, which we also expect to stay consistent for the rest of the year.
 All in, shipments were as expected, to a bit lower, which reflects the increasing concerns over the economic effects of a resurgence of COVID-19 in China and a tight component market.  Based on our expectations for the remainder of the year, we expect total mobile shipments in China to be 333m units, up 8.1% over last year, but down from our original estimate (late 2020) of 364.3m units, which would have been up 18.3% y/y.    Based on our revised estimate, we expect 5G shipments in China to be between 248m and 255m units this year, which would be up between 53% and 56.7% y/y.

Dresden, Germany is the home to over 2,000 semiconductor related companies and fabs for Global Foundries (pvt), Infineon (IFX.GR), and X-Fab (XFAB.FP), along with about a half million people.   The region is considered the Silicon Valley of Europe and one of the most popular sightseeing points in Germany.  But it seems that last week the city and local region suffered the same fate as Texas, Taiwan, and Japan, and faced a blackout that halted production at at least one and possibly more semiconductor foundries.  Infineion, whose Dresden plant produces automotive semiconductor components, saw the 1 to 2 hour outage halt production, which inevitably will result in process loss and exacerbate the shortages in the automotive space that were just beginning to show signs of a more relaxed order environment, along with Global Foundries and Bosch (pvt), both of whom also saw outages that could lead to reduced production. 
Strangely, the cause of the power outage that affected over 35% of the area and over 300,000 homes and businesses, was a metalized balloon that drifted into a 110KV substation and tripped the system, shutting down power across the area.  The police are investigating the source of the balloon and whether it was radio controlled, however while it was recovered, it was quite burned.  If any control systems are found it would be considered an ‘attack’ rather than an unusual accident.   

Our obsession with Samsung’s (005930.KS) constantly moving pricing structure for their 4K/8K Mini-LED/QD TVs continues, but it looks like Samsung has reached a point where they either feel that they are selling enough sets from this new line, or the cost of producing same is beginning to affect the end-user price.  In short, prices for these sets, while still below Day 1 pricing, do not seem to be heading down, with only 3 of the 33 we track at their lowest prices, with 5 (15.2%) lower than our last check (average decline 4.7%) and 10 (30.3%) higher (average increase 6.2%).  Of the 15 Mini-LED/QD sets we track (includes 4K & 8K) 7 (46.7%) saw price increases, while 3 (20%) saw price decreases.  The table below breaks down the categories and compares the current discount to Day 1 pricing to pricing 2 weeks ago.

​In the Spring of 2012 a Pakistani citizen contacted an employee of AT&T (T) through Facebook using an alias, and convinced the employee to help him unlock AT&T smartphones.  He offered money and the incentive that he could earn more if he recruited other employees to join the scheme.  Muhammad Fahd, (real name) helped the recruits set up fake businesses and bank accounts, generate invoices, and got AT&T employees to use their company credentials to unlock phones for customers that were unable to qualify for an AT&T account.  In 2013 however, AT&T developed a more sophisticated unlocking system, making it more difficult to bypass the company’s protection software.  Fahd then hired a software developer to design malware that could be installed on AT&T devices with the company’s knowledge that would allow him to unlock phones more efficiently and in larger volumes, all done remotely.
Those AT&T employees ‘hired’ by Fahd gave him confidential information about the AT&T systems and company network access which he [passed on to his developer, allowing the developer to design malware that could enter and work on AT&T’s system computers.  Over a period of ~3 years, the indictment against Fahd, made in 2017, indicated that ~1.9m smartphones had been unlocked by members of the nefarious group, generating (AT&T’s number) over $201m in lost revenue for the company, as those whose phones where unlocked, stopped making payments on their phones and likely moved to another carrier with the unlocked phone.  Fahd, was arrested in Hong Kong in 2018 and extradited to the US and eventually pleaded guilty to wire fraud last year.  The judge in the case ordered him to repay ~$200m and to serve 12 years in prison.  It might take a while to make much of what Fahd owes to the government as the wages for in-prison work ranges from $0.16 to $0.65/hour. (147,929 years at $0.65/hr.).

We have mentioned that a number of Chinese exchanges have an open platform where investors can directly ask listed companies questions about a wide range of topics.  Most questions are somewhat specific and refer to statements the company has made, statements competitors have made, or in some cases ‘suggestions’ as to how the company might do better or make changes.  Not all company’s respond and in many cases responses are boilerplate and give no answer to the question, but while slogging through many irrelevant or downright ridiculous Q&A sessions, occasionally we come across something significant or important to a company or an industry.  In most cases it is the former, but the type of questions asked, especially over a relatively short period of time can also be revealing and we focus on such a question here.
If we had to typify the question it would be, “How will your company participate in the meta-universe?”  Originally this threw us in that it was difficult to understand how the concept of multiple universes existing in time but in different planes would have anything to do with a panel manufacturer or hardware producer, but we were thinking like physicists or science fiction writers, when the answer was far more simple; sort of.  The meta-universe is a concept that is an outgrowth of what we call AR (Augmented Reality) where you wear a pair of AR glasses from the moment you wake up until the moment you go to sleep.  Said glasses provide information on just about everything and provide a visual overlay on top of whatever you happen to be looking at.
Purveyors of this concept liken it to being connected to the internet 24/7, allowing you to point to an object and pull up scads of information about it without having to connect or use a browser, so looking up at the sky could open up a local forecast or pointing at an item in a store window could bring up item and price information and even allow you to buy it without entering the store.  Sounds cool, although a bit like walking down a street covered with neon signs, but it doesn’t stop there.
The meta-universe concept goes further, especially if you ask those involved in the more immersive VR (Virtual Reality) space, with the creation of a multitude of meta-verses that are not specific to a game or application, but are shared by the global population.  Some of these universes might be specific or dedicated to a particular theme, while others might be more general but perhaps less stressful than today’s reality, but they are all constantly evolving and exist in real-time.  While these idealized universes might seem like playgrounds for those with free time or those looking to de-stress, there is one underlying commonality to all, and that is everything in any meta-verse is digital, including and especially money, so while it might be wonderful to stroll down a street made of rainbows and purchase a unicorn that you can ride every time you visit that world, someone is checking to make sure you have enough digital currency to purchase the item, and someone else is logging where you were, what you bought, how long it took you to make the purchase, and hundreds of other metrics about you that they will either use to target advertising in that meta-verse or sell to some other organization that will use it for something else.
There are many supporters of the meta-verse concept, and here we are not speaking about scientists but companies.  Those that produce imagery for animation and gaming would certainly have a big stake in such a new reality, but as we noted, underlying all of this shiny and futuristic technology is the same old motivation, selling stuff, and finding more ways to grab the attention of consumers.  Vendors will of course eventually latch on to any idea that they think will help them sell more units, but who would really want consumers to have access to a multitude of shred environments?  Social media companies.  What could be better than collecting information about every person, the universes they visit, the one’s they like the best, and what they do when in those universes.  Such data could be worth infinitely more than what you posted on Instagram (FB) last Tuesday, and since those meta-universes (univeri?) will likely be maintained by a consortium of massive server farms owned by companies like Amazon (AMZN), Facebook (FB), Tencent (700.HK), and Sina (SINA), everybody in the social media world will be happy campers.
More interesting however is how such a global meta-universe environment will affect traditional social media.  As pressure continues to build about the emotional consequences of social media and its goal of keeping users attached, an alternative such as shared meta-verses will draw users away from smartphones and more toward AR/VR.  If this eventually slows the growth of traditional social media, it will give those companies some respite from the scrutiny that is developing, while they shift their focus to the developing a meta-universe world, where they can say ‘we are all working together to create a more positive social environment’.  Of course we are likely overly paranoid about such a scenario, but that lack of paranoia has led to a society that wants information only if it agrees with what they think and what could be more enticing than spending much of your day in a world that was developed and maintained for folks with a similar mind-set?  Most shared environment multi-player games are addictive, just think of how much more addictive a shared environment could be if it were fully immersive?

​Private Mobile Networks are just like public mobile networks but they are not accessible to individuals unless you are part of the organization that owns them.  Inherently they would be smaller and in many cases company specific, but they have characteristics similar to large public mobile networks, without many of the regulatory issues and requirements that public networks are burdened with.  There are lots of potential investors in private mobile networks, and by that we mean potential customers that could build such networks for themselves or others. 
Industry organization GSA (Global Mobile Suppliers Association) has identified 626 organizations in 55 countries or territories where private network deployments (4G or 5G) have been made or licenses granted for same have been issued, including 71 network operators.  As part of the same database there have been 187 active private mobile network projects identified and at least 50 vendors who are providing equipment for such networks and projects, so we see the segment as one that not only has its own cost/return characteristics but is not growth limited by competition.

​In most cases such private mobile networks are local, meaning single sites or campus locations, with 33% of those identified and an additional 21% assumed to be LANS, while city-wide or National private networks make up ~13% (33% unidentified), which is not surprising as these are in many cases, specific use networks, meaning attached to a particular company or type of organization, but when viewing the breakdown of industry segments (Fig. 2) that are deploying private mobile networks, there are some such as rail or O&G that would be candidates for WAN networks to maintain control over assets that are widely dispersed.  LTE (4G) is still the most common private mobile network technology but while we expect the share of WAN networks to remain relatively constant, we expect the share of 5G networks to continue to grow as new deployments would likely be 5G based or a combination of both.

Manufacturing is by far the sector with the most private network deployments, and factory or campus networks that allow personnel, autonomous vehicles, or stationary devices to utilize location data, IoT, and individual communications networks that are factory bound, are relatively commonplace, but there are many other industry sectors that have need, either for security reasons or for logistics, that need to be ensured that they will be able to communicate with critical infrastructure, be it local or otherwise.  Some can be critical, such as mining, where local nodes are in flux as locations change, or logistics, where a waste number of assets must be tracked, while others less so, such as government networks, which can be used for specialized announcements or notifications for a select group of members or constituents. 
Breaking down the manufacturing segment further, the automotive sector has been a strong proponent of private mobile networks, given what can be a vast facility such as the Volkswagen (VOW.GR) plant in Wolfsburg, Germany, that covers 6.5 square miles (3x the size of Monoco) and has ~60,000 employees (48% of the entire population of Wolfsburg), or the Hyundai (005380.KS) Motor plant in Ulsan, South Korea that covers ~5 square miles, employs ~34,000, and has its own fire station, hospital, sewage treatment plant, and roads.  However we see few limitations and progressively more opportunities for same as data networks evolve and the need to build out concurrent mobile networks increases proportionally.  Collecting lots of data is a good thing but the objective is to use that data to make meaningful improvements in operations.  Those improvements are made by management, engineers, and line works who must be informed as to how those changes are being manifested, and getting all interested parties in one place for a meeting can be quite difficult.  Getting information to the proper parties makes private mobile networks an obvious option.

​Rainbow (600707.CH), aka Caihong Display Devices, aka Irico[1], has announced that its board has approved the construction of a large Gen 8.5+ LCD glass production facility in Xianyang.  The project, which is to cost $1.41b US will consist of 8 hot lines and 4 cold lines, ~4 times the size of the company’s existing two Gen 8.5 glass lines in Hefei.  While the company is financing the entire project itself (including outside funding), the announcement gave considerable warning that full project funding has not been reached and might not be available when the project is to move into full scale construction, leading to possible delays.  The given timeframe for the construction is 36 months, so the earliest completion date would be 4Q 2024.  As a new site, we believe the company will build the shell and necessary infrastructure for the facility and then add kilns and processing lines individually, which would slow the in-country share gains that are the goal of the investment.
Rainbow competes with the three major LCD glass producers, Corning (GLW), Asahi (5201.JP), and NEG (5214.JP) and a few small local producers, but is still dealing with improving the quality of its large format glass substrate products and remains a small supplier of Gen 10 LCD glass, especially when compared to Corning, who has a co-located glass plant next to BOE’s (200725.CH) Gen 10.5 glass plant in Hefei.  We believe that most of China’s demand for large (Gen 7+) glass is supplied by non-Chinese suppliers, while local producers have a higher share of Gen 6 and smaller glass sizes on the Mainland. 
This has been a sticking point for both the Chinese government and the Chinese display industry, who has been under pressure to develop local glass production, but as has been the case with a number of other display technologies, just throwing money at a technology does not mean success is guaranteed.  Quality and guaranteed supply are of greatest importance to panel producers when it comes to substrates, as unspotted defects can cause major problems further down the production line and can cause major yield issues, an ongoing problem for Chinese panel producers.  This seems to be the type of project that should have been instituted two or three years ago, as the Chinese large panel LCD industry was developing, but we note that China has made a few attempts to ‘dominate’ the local glass market in the past, none of which proved successful, so it might have taken a bit longer to convince the CEC that such a large project might work, and even then we won’t know until late in 2024.


[1] IRICO (Rainbow,etc.) is owned by IRICO Group, which is owned by China Electronics Corp, the state-run entity that manages Chinas electronics industry.