​Last month we noted that Chinese lens and camera module producer O-Film (002456.CH) had received notice from a specific overseas customer that it planned to terminate its relationship with the company in the 2nd quarter of this year.  The assumption we (and most others) made was that 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.  O-Film has already announced that 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. 
As O=Film was one of a number of companies that the US government had placed on the ‘entities’ list, they have faced difficulties with customers over whether the inclusion of their products would violate the US trade sanctions, which we suspect was the reason behind O-Film’s exclusion from the Apple iPhone supply chain.  At the time, the Wingtech deal had not been closed as Winftech was doing due diligence on the potential acquisition, and that Apple would continue to be supplied by the division once it was out of O-Film’s hands.
It seems that O-Film has received a bit more bad news as Samsung Electronics (005930.KS) did not include the company in its supplier list for camera modules in February, likely in anticipation of potential trade issues with the US that might require Samsung to specify a different camera set-up for smartphones destined for the US.   Rather than see the increased cost of such specialization, Samsung decided to exclude O-Film altogether, and the company has already discontinued producing what were relatively small amount of camera modules for legacy Samsung products.  

Another ‘unprecedented’ month for LCD panel price increases and once again those increases were higher than our expectations.  As we get further into uncharted panel pricing territory, we look at the longer-term data for a better understanding of the implications that such panel price increases will have on the CE space, but even our regular panel price charts show the rapidly increasing prices and the unusually high level of m/m increases.  In an additional set of charts, we look at how much of a price increase today’s panel prices are against yearly panel price highs and lows over the last 5 years.
While broad-based semiconductor shortages have plagued almost the entire CE supply chain, there have been some specific silicon issues that have continued to push panel production prices higher.  Timing controllers (TCON), which control which pixel is receiving data and when, are basic necessities for displays, and remain in short supply.  While there are a number of suppliers of TCONs and drivers for the display space, most are fabless, which means they are competing for foundry capacity with many other products and face the same supply issues as other semi products, which pushes foundries to both raise prices and give priority to their largest customers and products where they derive the most profit.
Glass substrate prices, which historically decline, have been relatively stable, and most recently, after a few mishaps and glass facilities, glass prices are rising, adding to the price of panel production.  That said, brands continue to hold onto aggressive CE targets, which gives panel producers, usually in the lesser bargaining position, almost free reign to raise prices faster than costs increase.  Chinese suppliers, who have become the dominant suppliers of LCD displays, are seeing this situation as a chance to justify their aggressive LCD capacity expansion over the last few years, and will continue to raise panel prices given their leading status.  This gives them much ammunition when negotiating with provincial and local governments for capital to continue that expansion into OLED and LED based displays, albeit in a potentially cyclical space.
The result of panel price increases of the magnitude seen here are higher prices at the brand level, and as COVID-19 still seems to be a lifestyle changing factor on a global basis, the raid vaccination process in the US is beginning to have an impact on demand.  In fact, demand continues strong for CE products on a global basis, but we expect a bit less demand from the US later this year if the rest of the population accepts the vaccine as readily as the first 20% of the population.  We expect there will be a demand surge for certain iCE items as the US population begins to step out of what amounts to quasi-quarantine, but higher CE prices and a lesser focus on filling time in front of a display will have some effect on demand.
The global CE space however will likely stabilize less quickly and demand for educational CE (notebooks, monitors) will likely continue until new case levels on a global basis, or at least in developed countries, begin to decline in earnest, and some kind of supply/demand balance returns to the space, but the rapidly rising prices do represent a deterrent to a return to normality for CE products, as high-priced inventory will built into CE prices as 2021 progresses, leaving CE brands with margins that might not support the aggressive volume estimates they have now.  While we would not expect panel producers to keep in mind such factors when deciding to raise prices, some restraint might help to sustain the length of their positive results.  Just a thought…

We note that the charts below show yearly high and low panel pricing and the difference between those and current panel prices in each of three categories.  We exclude mobile only because the products available change so often (size, resolution, etc) that the data in past years has less relevance to current prices than other panel categories.  Most telling is the difference between the current TV panel aggregate price and the low points in each year, with 2020 being the extreme due to COVID-19.  That said, TV panel prices currently are also up over 100% from the 2019 low and ranges from 39.0% to 73.1% above the hi/lo average for each year, which shows that while 2020 was an anomaly, TV panel prices have still increased by very significant amount using 5 years of data.

As we have mentioned in previous notes polarizers, a key but rather obscure part of almost all displays are not complex devices or made of complex materials, but they perform an essential function in LCD displays and to a lesser degree, in OLED displays.  In short, LCD displays don’t work without them, which makes the continuing shortage of polarizer capacity an issue that can influence the LCD business, and to a lesser degree, the OLED business.  Shortages in the polarizer space go back years (the note linked above was from April 2019) as the supplier list is quite narrow and the price of polarizers and polarizer materials is very dependent on supply.  In the abovementioned note, we expressed a hesitancy by polarizer suppliers to add significant capacity, as they were, at the time, unsure how long the expansion of Chinese LCD capacity would continue, especially as panel prices were declining and panel producers were operating at or below cash costs. 
Some polarizer manufacturers did add capacity in late 2019 and last year, but none expected the increase in display demand caused by the COVID-19 pandemic, especially after the very poor display industry results in 1Q last year.  Nitto Denko (6988.JP), among the top three polarizer producers, licensed their technology to others (at least two Chinese producers) rather than expand, and BenQ Materials (8215.TT) focused on increasing production efficiency, in lieu of new capacity.  Samsung SDI (006400.KS) and Sumitomo Chemical (4005.JP) added some capacity in 2018, while LG Chem’s (051910.KS) polarizer expansion came in 2019.  Cheng Mei Materials (4960.TT) (formerly Chi Mei Material Technology) while one of the smaller producers decided in 2018 to add enough capacity to increase its share almost 2X by mid 2020, but prices for polarizers continued to rise in 2019, as raw suppliers such as FujiFilm (4901.JP) and Konica (4902.JP) increased prices that were passed on to polarizer manufacturers and then to panel producers last year.
Jump forward to the present, and we see comments from the Chairman of Cheng Mei Materials (aka CMMT), despite their capacity expansion to four production lines (from two), citing continuing shortages of the films and other upstream materials needed for the production of polarizers.  In fact, he stated that the shortages were more severe than imagined, leading to the company being able to meet only 80% to 85% of customer orders, and that he expected the shortages to continue through 2021, particularly with certain polarizer types.  The materials shortage is not as severe for VA (vertical Alignment) type panels as it is for IPS (In-Plane Switching) panels, both of which are common in TV sets.  IPS panels have excellent viewing angle characteristics, meaning the picture quality remains high as the viewer moves off center, but such panels have lower contrast and mediocre black uniformity.  VA panels have poor viewing angle characteristics but are of higher contrast and higher black uniformity.  These two 20 sec. videos and the comparisons below show the differences using a Sony (SNE) IPS based TV and a Hisense (600060.CH) VA based TV.
The shortage of polarizers will have an increasing effect on LCD panel producers if demand remains strong, and given China’s increasing share of the LCD display market, the effect will be greatest on that region.  As long as demand remains strong, panel producers are able to justify panel price increases by citing the increased cost of materials, polarizers included, but we expect that polarizer capacity will remain tight even if demand slows.  This will make it exceedingly difficult for panel producers to justify such cost increases as demand slows in the future and can trigger a more rapid decline in panel producer profitability than might be the case if polarizers were not in short supply.
https://youtu.be/BS-daH4gk84
https://youtu.be/8uLoA8YxAuk

​As we have previously noted, a fire broke out on the first floor clean room of the building that houses the Renesas (6723.JP) 300mm semiconductor line (N3) in Hitachinak, Japan on March 19.   That fire took about 6 hours to be confirmed extinguished by the local fire department and shut down the N3 line, while adjacent buildings (N2 & test) remained in production.  The following day an inspection was made and it was determined that plating equipment overheated and caught fire, damaging about 5% of the clean room area, along with 11 units, or about 2% of the N3 line’s equipment.
After further inspection it was determined that an additional 12 units had been damaged (~50% of total), and the financial impact of the N3 halt to Renensas would be ¥17b ($157.2m US)/month and was originally expected to take about a month for repair and equipment replacement, which would be ~mid-April, with the company expecting to receive replacements for 11 of the damaged units at the same time.  Roughly 75% of the N3 WIP was salvageable and can be used when the fab reopens..  Based on the repair work and equipment delivery schedules, the company expects to begin to ship product from WIP from the 1st floor processes on or about May 19 and WIP from 2nd floor processes on or about June 19, with expectations for a 100% recovery of the N3 fab within 100 days of the fire, which would be around June 28.  The company revised their impact estimates to ¥17.5b if the recovery was to take 45 days and ¥24b ($222m US) if 60 days, excluding equipment procurement costs, inventory disposal, fixed asset impairment, and repair costs.
The company indicated that the clean room resumed operation on April 9, leading to a resumption of production around May 9, a bit ahead of schedule and on April 17 the company reported its production capacity at N3 was ~10% of production before the fire. And added an additional 7 to 10 days to the final recovery period, likely due to delays in equipment delivery.  Unfortunately yesterday at 2:30 PM smoke was discovered in the N3 plant, and while the potential fire hazard was extinguished by plant staff, firefighters were called and 1st and 2nd floor production was shut down for about 3 1/2 hours.  Company officials have stated that the short-term shut-down will not affect any existing production or shipment schedules.  Sounds like Deep Purple 1972…

​5G is not easy for many countries.  Geography and power resources can play a big part in how communications develop in many countries, while GNP and the cost of financing communications across sometimes large and diverse regions can add to the difficulty of broadband access.  But in Malaysia, a country of ~32.7m (a bit less populous than California) that is spread over two large islands in the South China Sea, covering 127,350 m2 (a bit bigger than New Mexico), the biggest issue holding back the implementation of 5G has been the government.  Its not that the government is not interested in allocating 5G spectrum, but more that it keeps changing its mind.
In 2019, the Malaysian government tasked the Malaysian Communications & Multimedia Commission (MCMC) to begin a public inquiry into the allocation of low-band spectrum in the 700+MHz range, and mid-band spectrum between 2300MHz. and 2600MHz.  According to the study, the best option for allocating these 5G spectrum bands was through a tender process.  Despite the study and agreement on the results, the MCMC announced it had awarded the low-band spectrum to 5 companies in an Official Ministerial Directive, surprising all involved in the tender process.  In June of last year the MCMC reversed its decision on the allocation, however in August the government announced that its initial plan to introduce 5G in 2020 was being put on hold until 2022 or 2023, and a new plan to upgrade the country’s 4G infrastructure was introduced in its place.
 
Once again, the Malaysian government changed it mind and in February of this year announced its plan for the Malaysian Digital Economy Blueprint that would roll out 5G spectrum in stages starting this year, but will be allocating the spectrum to a new ‘special purpose vehicle’, the Digital Nasional Bhd (aka SPV) that will build, own, and manage the country’s 5G infrastructure, leasing spectrum to telcos that will provide 5G service to customers.  The details of how the mechanics of this will work are still unknown, but the SPV has just announced that it is inviting telecommunications equipment vendors to begin bidding on the $3.6b US project, with the bidding process to be completed by mid-year.  The SPV is hoping to complete 5G coverage in two small (population ~100,000) cities and parts of Kuala Lumpur (pop. 1.8m) by the end of this year and expand coverage to 17 major cities in 2022 and 2023. 
The companies that have been invited to bid are Ericsson (ERIC), Huawei (pvt), ZTE (000063.CH), Cisco (CSCO), NEC (6701.JP), Nokia (NOK), Samsung (005930.KS), and Fiberhome (600498.CH), a number of whom have been singled out by the previous US administration as companies that should be banned from 5G networking.  While every country is different, with different geography and financial resources, given that governments are typically responsible for broadband spectrum, 5G roll-outs are quite easily affected by politics, indecision, and red tape and Malaysia is fairly typical in that its plans have changed a number of times over the past two years.  Even technically sophisticated countries like South Korea have seen radical changes in 5G policy, sometimes within very short periods, so we don’t focus on Malaysia as a place where things are worse than other countries, but one that happens to be a bit more visible.  Such confusion about 5G happens in many other countries.

Back on July 7, 2020 we noted that LG Electronics (066570.KS) confirmed that it would replace the power boards in 18 models of its OLED TVs produced between 10`6 and 2019 in South Korea, roughly 60,000 units.  An unnamed component on the power boards that was used to minimize ‘noise’ from the power supply was said to face possible failure leading to the potential for overheating on a few of the models indicated but the company decided to replace all of the power boards on the indicated models regardless of any previous problems.  At the time the company indicate that sets outside of South Korea were not affected, however the official notification, issued 9/28/2020 did not specify sets in South Korea or any specific region.
It seems that the problem was not limited to South Korea and that LG is offering to replace the power boards in other regions and has begun to hold discussions with regulators about the issue and its obligations to its customers.  Electrical safety regulators and LG (not surprisingly) have categorized the issue as ‘low risk’ with no incidences been reported to warrant any safety concerns and that they use non-flammable materials that meet international safety standards for TV products, and any overheating is extremely unlikely to lead to fire or result in any injury.
LG has set up service partners in various countries that will replace the power board, which LG says should take about 30 minutes, and some regions have mobile service that can come to a customer’s home for the replacement, rather than bringing the set to a service center.  The company has not indicated whether the problem will affect other OLED TV brands that use the same OLED display, given LG Display (LPL) is the de facto supplier of OLED TV displays, however we believe the power board is more likely to be particular to the brand.  If you have an OLED TV that is specified in the table below, we urge you to call your local LG dealer and find out the procedure for having the board replaced in your area.

​We are getting tired of noting semiconductor price increases, as foundry resources remain under considerable pressure, but while the increases are usually a positive for foundries, they have a cascading effect on the CE space and are therefore important to note.  Taiwan based Trendforce has been among those predicting a rise in DRAM prices, having set its expectations for 2Q DRAM price increases of between 13% and 18% only a month ago, after a relative small increase of between 3% and 8% in 1Q.  In just a month, Trendforce, who we might say tends to be a bit toward the high end on many of its estimates, has just increased its estimates for 2Q q/q DRAM price increases to 18% to 23%.  Strong notebook demand, expected to be up between 7% and 9% is also pushing up expectations for PC DRAM to see a 23% to 28% q/q increase this quarter and server DRAM pricing is now expected to rise by 20% to 25%, against previous estimates of 134% to 18% and ~20% respectively.

Apple’s (AAPL) virtual event yesterday introduced a number of new and updated products, many of which are based on the company’s own silicon SOC, the M1, the successor to the A12 Bionic.  The new iMac base model is larger (24” vs. 21.5”) and has a much higher display resolution (4480 x 2580 vs. 1920 x 1080) than the previous 21” model, and has a few new bells and whistles such as touch ID on the accessory keyboard and a better camera.  But most of the improvements come from the M1 chip, which reduced the size of the system board and heat dissipaters (fans), and increased the speed at which much of the image processing is done.  One nice feature is a single cord that attaches to the unit (magnetically, which could have its challenges) and connects to a small box on the floor, which connects to both the AC power and Ethernet, which means less cords to the desktop.

The iPad Pro has also been updated to include the M1, with similar improvements in speed and power efficiency and 5G connectivity has been added, with support for both Sub6 and mmWave in the US (Sub6 only in most other locations).  The iPad Pro includes both a LIDAR scanner and the True Depth system, and together they are used in a new feature (‘Center Stage’) that allows the ultra-wide camera to keep the subject in the center of the camera’s view by automatically panning.  If additional subjects are added, the camera will recognize them and zoom out to include all in the frame. 
That said, the most significant change to the iPad Pro is the display, which now includes mini-LED backlighting.  The 12.9” display uses over 10,000 mini-LEDs to drive 2,500 zones that can be individually controlled as to brightness.  This allows the display to reach 1,600 nits peak brightness (1,000 typically) and is a far cry from the 72 LEDs that were in the previous iPad Pro display.  While this certainly was not unexpected, Apple’s entry into the mini-LED world legitimizes further a technology that helps to maintain the huge investment the industry has made in LCD technology, and adds a bit of cache to the LED space, at least for those producers that have anticipated the mini-LED transition.  While producing mini-LED backlighting has its differences from standard LED BLUs, the basic technology is the same, which gives it a shorter adoption timeline, offset by the potential for a shorter ‘first mover’ advantage, with a boost in quality that should compete more effectively with OLED.
Apple also introduced its Air Tag, a tracking device that is very similar to the Smart Tag+ that Samsung (005930.KS) released recently.  Both devices operate using Bluetooth to find tagged items that are within a few hundred feet, but take lost item discovery a step further using UWB (Ultra-wide band) technology to find items out of Bluetooth range.  There are a few differences between these two competing tags, but by and large they are the same, with the most likely choice for users being the one attached to their current ecosystem.  The table below points out the similarities and differences.
Given the relatively limited number of devices that are equipped with UWB, much of the ‘finding’ will be done within Bluetooth range, which can accurately find objects within a theoretical 400 foot range, although most Bluetooth trackers are more practically limited to ~200 feet.  Adding UWB opens tracking up to far more precision, down to centimeters, although Apple’s system shows you how far and which direction from the lost object you are while Samsung’s system overlays the object’s location onto the phone’s camera image, which gives a visual location of the lost item rather than a distance measurement.  The interesting part of these ‘finding’ networks is that any device so equipped will sense a lost tag and notify the owner that the tag has been located without the intervention of the ‘finder’. This allows for tagged items to remain hidden from the finder unless the owner wants the tagged item recognized by the finder.  This, along with encryption and shifting Bluetooth codes, keep the tag user anonymous and (hopefully) protected from being tracked. 
Networks such as these are only as effective as the number of ‘network equipped’ devices increases, giving a greater chance that someone in a particular network will eventually come close to the missing item.  There are ~260m active smartphone users in the US, with ~113m active iPhone users, so Apple likely has the larger installed base and therefore a better chance of finding a lost item, however Samsung has a larger global installed base, so each has some advantages, but the biggest question will wind up being how inexpensively such tags can become, as their cost will determine their usage.  Paying $30 or $40 for a tag means the item you are tagging has significant worth, but a $1 or $2 tag would be justifiable on a much wider variety of items.

While announcements from key foundry operators, Taiwan Semi (TSM) and Samsung, along with Intel’s (INTC) ‘return’ to the foundry business, have made headlines, and smaller foundries such as SMIC (688981.CH), UMC (UMC) and Global Foundries (State-owned – Abu Dhabi) have stepped up plans for more mature node expansion, there now seem to be another tier of semiconductor suppliers getting ready to add capacity.  The Chairman of Korea’s SK Hynix (000660.KS) indicated that he has an interest in ‘investing a lot in there (foundry)”, with what is likely an investment in adding to its 8” capacity.
What points to the fervor of such a statement is the fact that SK Hynix is currently in the process of moving its 8” foundry capacity from its plant in Cheongju, South Korea to Wuxi, China, a daunting task, and last year took a 49.8% stake in Key Foundry, a company formed to purchase Magnachip’s (MX) Fab 4 foundry last September.  With the real shortages facing foundries that are not focused on 3, 5, and 7nm nodes, but on more mature processes, one would expect SK Hynix to have an interest in building out those nodes to fill customer demand, but side comments by SK Hynix’s Chairman seem to indicate that his interest lies in competing with Samsung and TSM at the 12” wafer level, which implies more aggressive nodes.
Given that the semiconductor industry as a whole is convinced that shortages will continue through 2021 and into 2022, we expect there will be even more talk of the need for additional capacity throughout the year, with those like SK Hynix, who have been focused on more narrow semiconductor production (in this case memory), looking to widen their customer base with foundry capacity outside of their traditional wheelhouse.  Whether this proves to be a correct assessment in the long-term is far from clear and industry pundits citing a ‘new digital age’ and ‘this time is different’ is music to the ears of equipment vendors.  That said, the semiconductor industry is cyclical, and while the trend line is certainly heading up, indicating that whatever undersupply or oversupply is the outcome of these moves toward new capacity across the industry, there will be times when the current enthusiasm will prove less then prescient.  In the chart below, which represents semiconductor industry revenue over the last 20 years, 50% of those years have seen sales below trend line.  

​Almost everybody was wrong about OLED last year, and while we have yet to make our own OLED predictions for this year, we have seen a few that are quite optimistic.  That said, on a unit basis, 2020 saw OLED units decline by 3%, likely a result of the COVID-19 pandemic and the associated changes in both physicality and limited monetary resources.  Huawei (pvt) was also part of the shortfall as US trade sanctions slowed the company’s smartphone business, and even smartphone leader Samsung saw an impact on smartphone sales, particularly in 1H.  Chinese smartphone brands also saw demand decline affecting OLED display sales, and missed their combined regional goal of 100m units by 20m.
This year is expected to be better, with a general recovery in smartphone sales and an increasing share of OLED displays in that mix.  Most important however is OLED relative newcomer, Apple, who is expected to be the largest buyer of OLED displays this year.  As Apple has transitioned from LCD to all OLED for the iPhone, the company’s combined OLED display purchases are expected to rise 47.6% from 114.5m units last year to 169m units this year, including both legacy and new iPhone models.  As older LCD models fall out of production Apple should continue to move toward 100% OLED smartphone penetration.  The OLED penetration rate for Apple’s smartphone business is expected to be ~78% this year.
Following Apple this year would be Samsung itself, who is expected to see a ~3% increase in its OLED panel purchases for smartphones this year, with Xiaomi (1810.HK) the 3rd largest OLED smartphone display consumer according to OMDIA.  All in, expectations are for a total of 584m smartphone OLED units, up 28% y/y, with 57.5% being flexible OLED smartphone displays.  As Apple uses only flexible displays, they would represent ~50% of the total flexible unit volume this year.  That said, Samsung Display (pvt) will remain the primary supplier to Apple this year, and while share will decline from 78.3% to 65.1%, unit volume will increase by ~23%, with LG Display (LPL) doubling its OLED unit volume to Apple this year for a 29.6% share, up from 21.8%.  China’s BOE (200725.CH), who has been attempting to break into Apple’s OLED display supply chain, is expected to ship ~9m units for a 5.3% share of Apple’s OLED smartphone display business.
While OLED smartphones are usually the biggest share of OLED shipment area, LG Display’s increased OLED TV panel production has changed that share during the last few quarters, but OLED smartphone revenue still outpaces OLED TV revenue by almost 5 to 1, making OLED smartphones the more profitable business for panel producers on a per m2 basis.  As the volume leader in the OLED smartphone display space, and the producer with the highest yields, they should see a strong year for the small panel OLED business as things stand currently.