Once again panel price increases far exceeded our earlier expectations as shortages with both silicon and glass increased, especially at the silicon level with Samsung’s Austin plant shutdown.  While consumer demand still seems strong, we expect much of the price increases seen this month were more from buyers looking to grab any panels that met specifications, regardless of price, for fear that they will become increasingly scarce.  Given that Samsung has yet to reopen the Austin plant to full capacity and glass remains tight, we expect April will see the same or similar magnitude price increases as the only panel producers that are seeing increasing large panel revenue are those who are in the process of adding capacity, while those who remain at maximum utilization saw declining sales last month as component limitations reduced unit volumes.
On a general basis, Chinese large panel producers saw revenue increase in February, while all others saw revenue fall, other than Sharp () as large panel LCD unit volume was down 2.7%..Total large panel LCD revenue was down 1.4% in February, with panel price increases making up a portion of the unit volume decline.  We would have thought large panel sales would have been higher in February given the increasing panel prices, but that was not the case, although February in the display space is an odd month considering the Chinese New Year holiday, so we wait to see March panel sales before we can determine whether the February large panel sales decline was a one-off or a trend.

​Facial recognition software has become very sophisticated over the last few years and the use of facial recognition has gone far further than the searches seen on SVU or other police shows, and are used regu8larly by many law enforcement agencies to match camera photos to the vast databases that are built  by both independent companies and agencies like the FBI, who have access to some 400m non-criminal photos from the State Department and DMV records from the 16 states that allow access, along with the sharing of criminal facial recognition and the Federal, State, and local level.
But if you think the everyday rank and file US resident is worried about his or her face being mis-identified in a criminal investigation or could be associated with some radical conspiracy group through a blurry photo, that pales into comparison to those who are outraged that a camera in their Tesla (TSLA) is monitoring them as the car’s beta software is being tested.  This has nothing to do with the Tesla FSD (Full Self-Driving) beta software that some Tesla owners are using, and more to do with the fact that Mr. Musk has removed some owners from the trials because a camera that monitors the driver (who is supposed to have their hands on the wheel at all times in case the software makes a mistake) indicated that the drivers were not paying enough attention to road conditions.  This opened began a series of questions about whether the in-car camera was able to detect the ‘gaze’ of the driver, which Musk indicated was correct, raising privacy questions.
Surprisingly in China, privacy laws prohibit the use of in-car cameras and all Tesla vehicles on the Mainland have them turned off, excluding them from the software trials.  So it comes down to whether you trust beta software, for which the answer should be absolutely ‘no’, and if not, would it be better to be monitored and ‘reminded’ to pay attention when the software was in effect, so to avoid an accident if the software fails, or to retain your privacy and die in a fiery car crash because you were not paying attention?  We go with not dying…

3GPP is two things.  An organization comprised of over 700 members and a standard for global cellular service.  Since 1998 a number of standards bodies and individual companies (now over 700) have joined together to forge standards that allow global cellular system compatibility, without which all the broadband services enjoyed today.  3GPP is the basis for all digital broadband and enables everything from smartphones and streaming TV to autonomous transportation and IoT.  Without this protocol and the people and organization behind its development and evolution, the world would be a mess of disparate systems that were unable to communicate and global communication would be impossible.  The standard has led to the evolution of cellular systems that have evolved to 5G and will continue to evolve with 6G standards already being worked on for deployment in 2029 or 2030.  

As noted the 3GPP specifications for 5G are key to the development and deployment of the technology and while there have been many issues between the organizations top two contributing countries, the US and China, all of the specifications have been developed jointly, bringing in as much commonality as is reasonably possible on a global basis.  Without those standards, the quick list below would not exist and 5G would likely be incompatible not only from country to country, but potentially between carriers in the same country.  It wouldn’t be can my smartphone receive sub6, mmWave 5G or both, but can it even understand what is being broadcast, which could be different from one location to another.

• In understanding the number of systems that depend on these standards, here are a few facts:
• There are 807 operators in 240 countries operating LTE networks or FWA (Fixed wireless access)
• There are 327 LTE-Advanced operators in 148 countries that have launched such networks
• There are only 5 countries considered ‘not spots’ where LTE service is not available (Central African Republic, Djibouti, Eritrea, Mauritania, and North Korea)
• There are 428 operators in 132 countries that are investing in 5G mobile or FWA
• There are 153 operators in 64 countries where 5G service (mobile or FWA) is available

​Chinese lens and camera module producer O-Film (002456.CH) has indicated that it received notice from a specific overseas customer that it plans to terminate its relationship with the company in the 2nd quarter of this year, further indicating that revenue for the first two months of 2021 from ‘specific customers’ (assuming the same one) was up 5.52% y/y and will continue until March 31.  O-film was put on the US ‘entities’ list last year and was expected to apply for an exemption in November, but accusations of forced labor violations have been overhanging since the list inclusion.
The assumption most are making is the ‘overseas’ customer is Apple (AAPL), who has been a customer of O-film for the iPhone line, with the company indicating  that the ‘specific customer’ was 22.5% of net income in 2019.  Further complicating the picture is O-film’s announcement that it would be selling four subsidiaries to Wingtech (600745.CH), including GDIT (pvt), which is the division that includes the plant that produces camera modules for Apple, which O-film bought from Sony (SNE) for $234m in 2017, and was responsible for roughly 1/3 of the company’s profit in 1H ’20.  While a specific price was not given, the value of GDIT was to be 10x the subsidiary’s profit from overseas sales last year, which was $41.8m for the first nine months of 2020.  The deal however, did not include the unit that was placed on the US ‘entities’ list.
O-Film stated that “Due to the sudden occurrence of the incident, the two parties are still evaluating the impact of the termination of specific customer orders on the transaction, and communicating and consulting with the customer.  The order status after the completion of the transaction has not yet been determined,” which seems to indicate to us that if Wingtech cannot establish a supply relationship with Apple in regards to camera modules, the deal would have to be restructured.  Wingtech has given O-film $46m as part of the purchase intent agreement and has been conducting due diligence in the interim.

Gaming monitors have become a big deal, and while expectations around mid-year in 2020 were for a strong 35% y/y growth rate, it looks like those estimates were far exceeded by the end of the year.  Those estimates, which were for between 12m and 13m units, are history as the latest tally for gaming monitors indicates that over 18m units were shipped last year, bringing the growth rate to over 100%, with the top three brands, MSI (2377.TT), ASUS (2357.TT) and AOC/Philips (state-owned) all seeing greater than 100% y/y growth in the category.
Gaming monitor, which are loosely defined as monitors with a 100Hz or greater refresh rate, are a subset of the overall monitor segment, which typically thought of as desktop monitors associated with PCs or acting as terminals.  Gaming monitors however are designed to give the user a more ‘immersize’ experience and can be quite large and in some cases curved.  Relatively ‘typical’ gaming monitors are ~27” and run 4K displays at 120Hz to 144Hz refresh rates and cost between $300 and $500 depending on specs, but the high-end of the gaming monitor spectrum, where typical sizes run between 49” and 55” usually cost between $500 and $3,000 and include all sorts of bells and whistles that gamers appreciate.
The category, which is estimates to grow another 41% this year, is being driven by stay-at-home culture driven by COVID-19, and represents a more profitable segment to brands and panel producers than generic monitors, which are more typically based on PC sales and replacement cycles.  While not included in the gaming monitor category, more recently OLED and QD TVs have been cited as part of the gaming monitor market, as the specifications for high-end TV models are at or near gaming specs and provide the added function of acting as a regular TV (monitors typically do not have ‘tuners’ needed to receive TV signals).
That said, panel producers have been shifting production away from LCD TV panel production to ‘IT products’, meaning notebooks and monitors, as they are more profitable segments, which has tightened the TV panel market a bit more than normal.  Samsung Display, the leader in monitor panel production, has been reducing its large panel LCD production, and while that plan is temporarily on hold as panel prices spike, those plans will likely allow AU Optronics (AUOTY) and Innolux (3481.TT) to become the leaders in the gaming monitor market this year or next.  How sustainable such a high level of demand will be when COVID-19 is under control, but at least for the near-term, the category is of great interest to panel producers and brands.

Earlier this month we noted the continuing battle between what is now II-VI (IIVI) and Lumentum (LITE) for Coherent (COHR).  Since then things have heated up further with a number of offers and counter-offers from both parties.  Since 3/8 the battleground looks like this:
3/8         MKS (MKSI) offers $135m in cash and $115m in stock for a $250m valuation
3/9         II-VI offers $170m in cash and 1.098 shares for a $260m valuation.
3/9         Lumentum offers $175m in cash and 1.109 shares
3/12       II-VI offers $195m in cash and 1.08 shares for a $272.68 valuation
3/16       Bain Capital (BCFS) to buy $1.5b in preferred of II-VI
3/17       Lumentum offers $220m in cash and .61 shares for a $275m valuation
3/18       II-VI offers $220m in cash and .91 shares for a $287.18m valuation
This leaves Lumentum until 3/22 to match or beat the II-VI latest offer.  Seems like the fat lady is still not singing…

While there have been innumerable article in the press about component shortages in the semiconductor space, there has been relatively little commentary on what the effect would be at the consumer level.  On a direct basis, basic component costs would be the source of CE device price increases, but some of that has been softened by long production times, overseas shipping, and price increase absorption by both brands and retailers.  That said, if you want to see the effects of shortages, and in this case not as much based on the actual price of silicon than the competition for foundry time, here it is.
It starts at bitcoins, which have been on the rise, rising from $5,830.25 one year ago to $59,893.95 today.  This has incentivized Chinese bitcoin miners, where power is cheap, to build out bitcoin mining capacity.  In order to do so, miners need processing capacity, more specifically computational capacity, and a common place to find computational power is graphics cards that normally perform the mathematical computations necessary to convert data into a format that can be displayed by monitors or other display devices.  True bitcoin mining systems tend to be based on customized ASICs and have a higher return than using graphic cards, but are even harder to find currently.
The leader in the graphics card space is Nvidia (NVDA) whose GeForce RTX line runs from the RTX 3070 to the RTX 3090, with the major difference being Clock speed and memory.  Since mining bitcoins is purely a computational function, faster speed means more computations, means more bitcoins mined, so miners always want the fastest card, but also that means higher power consumption, which is why much bitcoin mining is done in areas of China where power is the cheapest. 
That said, given the relatively finite capacity of silicon foundries currently, and the issue facing Samsung’s (005930.KS) Austin plant, foundry customers are finding that they must bid for foundry capacity, escalating prices further.  Graphic cards face both that problem, which limits their availability, and the very aggressive demand from bitcoin miners, which has caused prices to rise quickly as availability dried up.  As an example, the Nvidia RTX 3070 card, which is the low-end of the RTX line, was initially priced at $600 in China.  Over the last three months that price, if the card is available, has increased to $1,230.  In the US the mid-level card, the RTX 3080, initially priced at $900 is now $2,000, and the high-end RTX 3090, originally priced at $1,400 is now $3,300 and hit $3,575 recently.  Setting up a bitcoin mining server farm can use thousands of ASICs, and even then must carefully monitor prices to make sure it remains profitable, but small miners are the ones using graphics cards and the ones willing to pay up for same, as long as the price of bitcoins continues to rise.
So in this case there is a two-fold ‘inflation’ issue, first a shortage of components, in this case really a lack of available foundry capacity, and a specific increase in demand for the product.  While the second issue is particular to components associated with bitcoin mining, and therefore the rise in the price of bitcoin, being unable to satisfy that demand at the silicon level is the root cause, which is the same as in much of the CE space.  Increased demand from consumers for notebooks, monitors, and TVs during the COVID-19 pandemic and the inability to build out silicon or display capacity to meet that demand is winding its way through the CE space, and unless end-user demand begins to slow, we expect substantial CE ‘inflation’ this year.

​SMIC (688981.CH), China’s largest semiconductor fab, has signed a framework agreement with the Shenzhen government ot build a new 12” wafer fab.  Once final negotiations are completed for the $2.35b fab both parties will actively look for 3rd party investors for what would be a 22% stake in the new venture.  As it stands in the framework, SMIC will own 55% and the Shenzhen government will own no more than 23%.  The fab will focus on 28nm and larger nodes with a stated capacity of 40,000 wafers/m and is expected to begin production in 2022.  Given component shortages that are facing the CE space, it is no surprise that SMIC would push to expand capacity, and by focusing on larger nodes, will likely make it easier for equipment vendors to bypass US trade restrictions that would restrict SMIC’s purchase of 5nm or 7nm lithography equipment.  While production by the end of 2022 is an aggressive goal, we expect that the Central State government has already given its approval and the Shenzhen government will push through all of the necessary licenses and permits when 3rd party investors are found.

​In November of last year former President Trump signed Executive Order 13959 invoking the International Emergency Economic Powers Act to prohibit all US persons from making equity investments  in all “Communist Chinese Military Companies”, under the definition that the entity “is owned or controlled by the People’s Liberation Army and is engaged in providing commercial services, manufacturing, producing or exporting.”  On January 13 of this year the executive order was amended to require all US persons to fully divest and holdings of shares in CCMCs, and the next day the DOD added an additional list of CCMCs that brought the total to 44 companies.
On March 12, the US District Court for the District of Columbia granted a preliminary injunction against the implementation of the EO, in favor of Xiaomi Corp (1810.HK), one of the named CCMCs.  The court cited the Administrative Procedure Act, finding that the DOD did not develop sufficient evidence to support its designation of Xiaomi as a CCMC, particularly after Congress over-rode the President’s veto of the National Defense Authorization Act, adding more stringent requirements for inclusion in the DOD’s list.  While the exemption only includes Xiaomi, it weakens other CCMC designations and leaves them open to legal challenge with the potential to cause the Executive Order to prove ineffective. 
Such designations by the DOD and other agencies that were granted during the previous administration also have implications for businesses and contractors involved in the US government’s supply chain, as ‘procuring or obtaining” “any equipment, system, or service” that utilizes “covered telecommunications equipment or services” for certain critical technology or a “substantial or essential component of any system” is forbidden under the Federal Acquisition Regulations Act, although the Secretary of Defense has the ability to apply the prohibitions to an entity that he/she believes to be owned or controlled by, or otherwise connected to, the government of a foreign country.  Overall federal agencies have broad discretion in evaluating national security risk however if CCMCs are involved in transactions regulated by the US government, legal challenges to CCMC status could limit the use of the CCMC designation when agencies are doing risk assessment.  By overstepping the use of the designation, the previous administration will likely have reduced or eliminated its effectiveness going forward.

​The China Academy of Information and Communication Technology is the supplier of China’s smartphone shipment data, and provides a monthly look at China’s domestic mobile shipments and more granular data.  Along with the February data, CAIST adjusted some numbers, particularly those relating to new models shipped in 2021 (January).  The total number of new models reported last month was 40, however, with the addition of 28 new models being released in February, the total number was adjusted to 81, meaning January’s new models increased from 40 to 53, or 32.5%.  Similarly the number of new 5G models released in January was revised from 23 to 32, or 39.1%, and local new models (meaning from Chinese brands) was increased from 34 to 49, or 44.1%.  No explanation was given for the changes, which did not influence shipments, but we felt it necessary to disclose given last month’s charts and this month’s charts would be inconsistent.
In terms of February Chinese mobile shipments themselves, while February 2021 shipments were up 241.5% y/y, reflecting the extremely poor shipments during the initial COVID-19 lockdown in China, and puts the 1st two months of 2021 127.6% above last year’s first two months.  Given the unusual circumstances seen in 2020, we went back to see how the first two months of 2021 compared with earlier years.  As seen in the table below, the first two months of 2021 have been strong on both an absolute and relative basis, with unit volume stronger than any year since 2018 and Fig. 1 points out that January ’21 was close to April of 2020, which was a snap-back month after the initial COVID-19 lockdown in China. 
But the question remains as to whether January/February 2021 represents a return to growth in the Chinese mobile device market.  Some have cited the desire of Chinese consumers to move to 5G, given China’s aggressive base station installations, but 5G Smartphone shipments, shown in Fig. 2 look quite similar to the overall shipment chart in Fig. 1   In fact we track both 5G smartphone shipments as a percentage of total smartphone shipments in China and the same comparison of new 5G models against total new models in China, and the share of 5G smartphones shipped relative to total shipments has remained roughly the same for the last few months, just a bit under 70%.  While the number of new 5G models has been increasing, it has varied between 45% and 65% over the last few months, and is less of an indicator of adoption than 5G share of total shipments., so we have to rule out that a surge in 5G customers is the root for the current (2 months) recovery in the Chinese mobile market.
When we look at the long-term chart of Chinese Smartphone Shipments (Fig.4) the recovery looks a bit less sanguine and more in-line with the long-term trend, and as Glenn Close famously said, “I’m not gonna be ignored”, so we wait for March data to gain a bit more insight into whether the population of China has decided that they need to upgrade their smartphones this year, whether the smartphone penetration rate (83% last year according to GSMA) is increasing significantly, or whether this is another period of pent-up demand that cycles back to the norms seen previously.  We still believe that 5G is certainly a factor that has helped to sustain the smartphone replacement rate in China, but whether 5G alone can account for the early strength this year still seems problematical. .