In March of 2020 we noted that Epistar, now part of Ennostar (3714.TT), who was the holder of considerable IP surrounding filament LED light bulbs, had begun a lawsuit against Lowe’s Home Center (LOW) for infringing five of its patents, all concerning filament LED light bulbs.  Epistar has also sued Adamax (pvt), who settled and is now paying royalties, and V-TAC (6229.TT), who agreed to stop selling the infringing product, but the Lowe’s suit has been in the courts for so long that we expect some of the legal staff has aged out.  That said, on 12/6 a jury in the US District Court – Central District of California, ruled in favor of Epistar in the case and awarded $707,000 for each of the three (of five) patents that it found Lowe’s had violated, for a total of $2.12m.  While this is likely less than the legal bills going back to the original filing in April of 2017, Epistar is now seeking to increase the compensation as the jury found that Lowe’s had intentionally infringed Epistar’s patents.  The bulbs were produced by China based Yankon Lighting (Zhejiang Yankon Group (600261.CH) who was not named in the verdict.

Driver ICs are an integral part of all displays and are the interface between signals generated by a computer, smartphone or other device and the display as they translate that data into a visual image.  Display driver sales typically track closely to display unit volumes, but over the last year, display driver IC prices have increased rapidly as utilization rates at silicon foundries has been running at close to 100%.  As these foundries allocate tight capacity to device types a number of segments in the CE space have seen shortages, including driver ICs.
We noticed that the Taiwan trade press had indicated that November, while down for Taiwan’s driver IC companies, wasn’t ‘as bad as expected’, so we dug into the data to see if this was just a headline or a real change.  The commentary in the press was that TV set sales had been improving as TV panel prices declined, buffering display driver companies from declining sales, but we were suspect in that it takes time for both lower priced panel inventory to move through the supply chain, and for real TV set prices to see significant declines.  IT device drivers, which include those used in monitors, notebooks, and tablets, are the segment that is likely going to be the most volatile, with panel producers having shifted production away from TV panels and toward IT panels, where prices have declined only recently.  That said, we compared both TV and IT panel industry sales to a number of driver IC companies in Taiwan, who report sales on a monthly basis.
The chart shows the sales of Novatek (3034.TT), Raydium (3592.TT), Focaltech (3545.TT) and Fitipower (4961.TT) over the last 23 months against panel industry sales broken out into IT and TV segments.  We have scaled the monthly results of the driver IC vendors and the panel sales so they can be compared, so there are no y axis amounts, but the disconnect between driver IC sales and panel industry sales is quite easy to see.  As driver IC scarcity became an increasing issue late last year, the price of drivers ICs increased, boosting sales, while panel unit volumes remained relatively flat, constrained by component shortages and raising prices.  All in, it is hard to look at driver IC sales and not expect at least some ‘normalization’ to closer tracking to panel volumes.  As with panel producers, driver IC companies have been shifting capacity toward drivers for the IT segment, which puts more pressure on IT product sales going forward, an area that has seen demand weaken over the last few months.  While not perfect pure plays in the IC driver space, Taiwan’s driver companies are a good indicator of the sensitivity the CE supply chain has to relatively small changes in demand and provide monthly rather than quarter sales.

​One of the more visible responders to the COVID-19 pandemic has been Apple (AAPL), who has made public much of its communication with employees about its plans for remote worker mandates and store closings.   In early March of 2020 Apple closed its stores in China due to the rapid spread of COVID-19, which were to remain closed until March 27.  While they reopened those stores in March, they closed 450 stores in 21 other countries at the same time as the virus spread globally, with the expectation that those stores would reopen before the end of the month.  The stores in various countries have faced local closings on occasion as infection rates forced restrictions however the company had reopened all of its 516 stores globally by June of this year, being the first time in just under a year and a half that all of its stores were open.  It seems that Apple has decided to once again close a few stores in the US and Canada to prevent the spread of the Omicron variant, after mandating the use of masks for all stores in the Apple system last week.
Apple has faced similar challenges concerning its office workers as far back as February 2020, when it reduced guidance due to a slower than expected return of Chinese workers to factories in the Apple supply chain and reduced demand overall in China., and in a March 13, 2020 statement indicated that “In all of our offices, we are moving to flexible work arrangements worldwide outside of Greater China. That means team members should work remotely if their job allows, and those whose work requires them to be on site should follow guidance to maximize interpersonal space,” with management indicated that all hourly workers will continue to be paid on a business-as-usual basis, and expanding sick leave, mandatory quarantining, or childcare policies going forward. 
In August the company notified employees that any return to offices for Apple staff employees, which had been tentatively scheduled for September and then pushed to October, would be postponed until January 2022 due to the Delta variant, with employees given a month notice before being required to return, although only for an initial 3 days/week.  However yesterday, Tim Cook e-mailed staff, indefinitely delaying that return date and giving all employees, including retail workers, $1,000 that can be used for “work-at-home needs”.   Aside from their donations and direct help during the COVID-19 pandemic, Apple has been quite willing to take whatever action necessary to protect its workers from the COVID-19 pandemic, not something we can say about some in the CE universe.  Altruism is a very rare commodity in the CE space and should be recognized when it occurs.

​In our note of 6/25/21, we indicated India’s desire to become a producer of LCD panels based on the development of a local fab.  We noted that in 2015 the Indian government was said to have contacted a number of LCD panel producers under the auspices the 2014 “Made in India” plan, trying to convince them to build LCD panel production fabs in the country.  Panel producers however were not convinced that the country could support a local panel fab and such a project was never developed.  India has had better luck with CE assembly, having attracted Foxconn (2354.TT) and a number of component suppliers to build out facilities by providing cash incentives and tax breaks, but politics, a low-productivity workforce, and a lack of stable resources and infrastructure, have kept display producers from building local panel fabs, despite the growing local demand.
It seems that India is trying again, this time also trying to attract semiconductor foundries to build out production fabs on the sub-continent.  This week the Indian government approved a project that will invest ~$30.2b US to create an Indian semiconductor and display that will support two greenfield semiconductor fabs and two display fabs, with about 33% of the funds going toward a co-funding arrangement for those fabs.   The remaining capital allocation will be divided between developing large scale electronics and hardware manufacturing clusters (24%) and battery, solar panel, and white goods production centers. (43%).
Companies that are involved in compound semiconductor production , silicon photonics, and sensor manufacturing will be eligible for up to 30% of their capital expenditures, while IC design companies will receive up to 50% of their expenditures, based on sales volume, with the government creating a ‘semiconductor mission’ to oversee the ecosystem, which will include global experts in order to smooth the implementation of such projects.
While this all sounds like a solid plan, India has yet to show panel producers (and less likely semiconductor producers) that it can actually provide the massive infrastructure that would be needed to support either industries.  Aside from the need for a skilled workforce, reliable water, power, and transportation sources are an about necessity for such multi-billion dollar commitments, but the real problem is still the chicken-and-egg situation around supporting suppliers that are key parts of the semiconductor or display fabs themselves.  For display producers, glass substrate suppliers tend to be close by or co-located on fab grounds and fluorine production is also an on-site requirement for an efficient display fab and semiconductor fabs rely on similar gas or chemical resources to maintain continuous production.  The establishment of these key industries is a primary part of setting the path toward convincing either company type to establish a production base in the country and India has yet to prove they can do so and the competition from other regions who have well-established infrastructure makes it an exceedingly difficult task.

​If you were not convinced that the Metaverse is ‘real’ in its ability to attract both attention and capital, look at Hong Kong based Arpara (pvt), a relatively new company that released the industry’s first micro-OLED High resolution VR headset and provides ‘arparaland’, a virtual social platform.  What makes this unusual, aside from the micro-OLED display, is the fact that on 12/14/21 the company began a kickstarter campaign to raise $50,000 in 60 days to begin mass production (and get pre-orders) for a March 2022 shipping date.  As of today the campaign (1 day) has raised $421,351 from 714 ‘backers’, who as a reward, get a discount on the “Basic” VR headset, from its $700 list to a $400 reward price or a discount on the ‘All-in-One” VR headset from its $1,100 list to a $600 reward price.  The $50,000 goal was met within minutes of the campaign’s start.
Aside from the interest, the Arpara VR headsets are 2560 x 2560 resolution (each eye), higher than the Oculus Quest 2 (FB) the most popular of the VR headsets (2020) resolution of 1832 x 1920, with one model being tethered to a PC and the other wireless. And far more expensive, even at the discounted price (Oculus Quest 2 starts at $300), but the promise of higher resolution and the high contrast of OLED seem to have done the trick.  Arpara is just one of many small companies that have or are producing VR headsets, all of whom compete with much larger CE companies with vast resources, so the support lifetimes of some of the devices has been rather short.
We are not commenting on the specs or how the Arpara headsets compare on a technical basis, which would be a more technical note, but on the excitement generated by anything VR, AR, or Metaversey.  While there will be many new products released by CE companies at CES 2022 in January, we expect ‘Meta-verse’ to be one of the most common search words relating to CE products in 2022.  While that doesn’t mean there will be an real understanding of the space, but rather infinitely more eyeballs looking for information as to what it is, how it works, and whether it is real or just another CE hype.  Hopefully we can sort out the details in our notes, giving some understanding of how the Meta-verse might work and who is really involved.
Here’s a list of those companies that have released a 2021 or 2020 VR/AR model, giving a better understanding of the type of companies involved in the space that have ‘recent’ products..

​As mentioned yesterday, Chinese brand Oppo (pvt) released it first foldable smartphone last night, the Find N.  As noted the phone looks similar to the Samsung Z Fold, although a bit smaller when opened at 7.1” vs. the Galaxy Fold 3’s 7.6” main display and a 5.5” cover display vs. the Galaxy Fold 3’s 6.2” display.  The Oppo phone has 3 main cameras (50MP, 13 MP, and 16 MP) and two selfie cameras (both 32 MP), while the Galaxy Fold 3 has the same number but with lower pixel counts (12 MP, 12 MP, and 12 MP for main and 4MP and 10 MP for selfie).  Other than those differences the two devices are quite similar, and while an official price has not been revealed for the Find N, it is expected to be ~$1,200.  The Galaxy Fold 3 5G with no trade-in (unlocked) is $1,650.  At least for now, the phone will only be sold in China with deliveries before the end of the year.

“We will control the horizontal.  We will control the Vertical”.  Such were the words of the ‘control voice’ before every episode of the 1962 Sci-fi series “The Outer Limits”, but it seems they are again apropos to the semiconductor space where a joint development effort between Samsung Electronics and IBM (IBM) has developed a way in which the fragility of Moore’s Law can be challenged.  The device type is called a VTFET (Vertical Transport Field-Effect Transistor) as opposed to the common FinFET (Fin Field-Effect transistor), where transistor gates are places around the transistor source and drain, both to allow the circuit to be opened or closed, but also as ‘dummy’ gates to isolate the cell from others.  As FinFETs are laid out horizontally on a silicon wafer, those dummy gates and interconnect wiring all take up room, limiting the density of FinFET devices as per Moore’s Law. 
Samsung and IBM have found a way to stack source, drain, and gate structures in order to further increase the number of transistor cells on a wafer.  According to both companies, the technology will allow for two times the performance of current FinFET technology and up to an 85% power savings without reducing the space (pitch) between each transistor, which also reduces the distance power has to travel and the distance between vertical cells, which improves performance, along with the ability to be less constrained as to the size of each sorce, drain, and gate structure.
All in, this is a step toward another iteration in the ability to produce circuitry that is required for faster and more complex computing, without the issues facing sub-nanometer node production.  Neither company has set a timetable for products using the technology, but we would expect early versions to be shown next year, although real implementation will likely take a few years to further define production processes.  It is certainly a step forward in IC design. 

​Magnachip (MX), a top 5 producer of display driver ICs, has officially ended its merger deal with Wise Road Capital (pvt).  The $1.4b deal for Magnachip has met with opposition from various US and foreign government agencies, who see the merger as a way for China to gain access to US semiconductor technology.  In particular, the US Committee on Foreign Investment (CFIUS), run jointly by the Department of the Treasury with members from a variety of US government departments, put the deal on hold in June to examine details further, with the Treasury Department indicating that the deal was at risk in August.  Without the CFIUS approval, the deal could not be completed.  As part6 of the merger deal Wise Road is required to pay Magnachip a total of $70.2m, $51m to be paid immediately and the rest by the end of 1Q 2022. 

​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….