Sanan Optoelectronics (600703.CH) has been a primary supplier of mini-LEDs to Samsung Electronics, and as a top 10 global supplier, has been able to remain cost effective as Samsung continues to expand its mini-LED offerings.  However, relying on a single supplier give Samsung less leverage than they are used to and has added South Korea’s Seoul Semiconductor (046890.KS) to its mini-LED supply roster, with the new entrant part of the supply chain for the Samsung NEO QLED TV line.  Seoul Semi is hoping to expand that relationship to micro-LEDs in the future and has been promoting its Wicop technology, (Wafer Integrated Chip on PCB) which, unlike flip-chip technology that must be bonded, Wicop allows an unpackaged LED to be surface mounted, reducing size, weight, and power consumption. 

​Samsung is also working toward changing its micro-LED production process from a single chip transfer system, which moves over 8.3m individual micro-LEDs of each of three colors to a product substrate (24.88m single transfers), to one where the red, green, and blue micro-LEDs are combined in one chip package, reducing the number of transfers to 8.3m instead of 24.9m.  While this sounds logical, it means making sure that each RGB die is placed correctly on each tri-color die, which in itself is a daunting task, for which there are a number of potential solutions that need to be scaled to mass production levels.
The fact that so many transfers are needed to produce a single micro-LED panel, and that the number of repairs needed, even at high yields, has kept the price of such devices in the six figure category.  As Samsung is trying to release true consumer micro-LED TVs this year, much will rest on the success of such development programs.  Toward that goal, Samsung has also partnered with AU Optronics (AUOTY), who is expected to be supplying glass TFT backplanes on which the micro-LEDs will be deposited.  Those backplanes will drive the micro-LEDs using TFT technology similar to what is used to drive LCD and OLED displays currently.

The abbreviation IP usually stands for intellectual property, a concept, device, material, or trademark that has value and therefore deserves protection against unauthorized use, however there is another IP that has both marketing value and importance to the consumer.  The IP we are speaking about is an international Protection Marking, aka ‘ingress protection’ or an IP Rating.  Most of the time, this rating will make little difference to consumers and lies low on the scale of features that consumer seek out, but if you have ever put your smartphone into a box of rice to dry it out or found that the ‘up’ volume button no longer works, the IP rating would have been of value.
The rating is based on tests performed on each device model to determine its ability to protect itself from the ‘ingress’ of solids, usually dust and similar particles, and water.  The IP rating is therefore two numbers, the first of which ranges from 1 to 6 (solids) and the second, ranging from 1 to 8, and occasionally has a 3rd number that is unofficial.
The ‘solids’ number is relatively easy to understand, the lowest numbers (1-4) indicate a device’s protection against objects ranging from a hand to a wire, so if you can’t disturb the operation of the device with your finger but you could pry it open with a screwdriver you get a 2 rating.  Dust becomes an issue with the higher numbers, where 5 means that the device will be protected for between 2 and 8 hours from dust that would affect the operation of the device.  That could mean that some dust does get into the device but not enough to ruin it for at least 2 hours.  The 6 (top) rating for solids means that no dust at all will get in for between 2 hours and 8 hours.  While we in the US are not used to environments where dust is that much of a problem, there are many overseas locations where particulate matter is a problem in everyday life.  It is those locations, and it could be from pollution or dust particles, that the Solids IP number has meaning.
Water is the 2nd part of the rating, and here is where marketing tends to focus as smartphones are routinely dropped in sinks, toilets, and run through washing machines, aside from use in unintentionally or intentionally wet locations.  The ratings for water start with the device being vertical and having drops of water on It, to full immersion in water for an extended period and under pressure.  Most of the in-between ratings are for various angles of both the device and the water, which allow for some water to enter the device.  Water ratings 6 and above do not allow any water to enter the device for the allocated time.
Most smartphones have such ratings, with a few examples listed below, however one of the negative selling points for foldable smartphones has been the hinge that allows the device to open and close.  Early models were subject to the entry of particles that caused the hinge to operate incorrectly and crease the display.  Such problems have been reduced or eliminated with updated hinge designs that keep dust particles out of the device, however no foldable has yet to be IP rated, meaning they are still subject to dust and water damage to an unknown degree.   That said, Samsung Electronics (005930.KS) has indicated that the release of the Samsung Galaxy Z Fold 3 and the Galaxy Z Flip 2 (expected June and July) will be the first foldable to have an official IP rating, giving customers some idea of how these devices will stand up to daily life.  No hint as to what the rating might be has been given.

Smart stuff is the new digital paradigm.  Smart cars, smart homes, smart appliances, and smartphones pervade our lives and collect endless amounts of data that is there to make your life (supposedly) better.  There are lots of smart things that really do help, especially in manufacturing, where smart tools sense when they need maintenance and smart visualizing systems can stop production when it senses a misaligned product is going to cause a problem down the line, but much of the ‘smartness’ we see on a daily basis provides only minimal help to our daily lives.  Yes, it is good that your refrigerator can tell you that your milk is two days from expiration or that your washing machine is leaking, but most smart products have little incremental value other than allowing you to sit on the couch longer than necessary.
That said, we came across a very unusual ‘smart’ item in our research on wearables.  This ‘smart’ category is an unusual one in that it is called smart textiles, a category that can include some medical devices such as active bandages that monitor cell growth in patients with wound issues.  But we fear that the ‘smartness’ in this particular application might be a bit unusual to say the least.  In the case of biometric sensors, they usually need to be in close and constant contact with the skin to register bodily metrics, and a number of companies have decided that the best place to put such smart textiles is a place that is not close to your heart, but close to ‘other’ parts, and thus the ‘smart underwear’ industry was born.
While companies in the consumer biometric monitoring space using smart textiles have been around for a bit, and include LIfeVest (Asahi Kasei (3407.JP)), a wearable defibrillator shirt and a knee brace that provides electrical stimulation and heat, a Canadian company, Myant (pvt) company also produces Skiin™ garments that feed data back to an app about your wellbeing.  Currently undergoing clinical trials their smart underwear that can continuously measure electrocardiography, stress levels, sleep quality, activity, temperature and other metrics, all through your underwear, although current products monitor and adjust only body temperature.  While this can be helpful for those that might be prone to random cardio events, the promotional literature for some of these products can wax a bit strange, citing “imagine a pair of pants that can play your party playlist when you get sad,” or wake you up if it senses driver fatigue (not sure how that works!), but in any case, consumer applications like these are a bit less helpful than a-fib detection or similar biometrics despite their potential appeal to those interested in being their own helicopter parents.  OK it’s a brand new world out there and we might be feeling a bit out of date when it comes to underwear, but perhaps there is a better way to promote the product…

There is an old adage in the car biz.  “Your new car decreases by 20% to 30% when you drive away from the showroom,” and that is a fairly common concept for most mechanical equipment used in factories, especially equipment that is highly specialized (small market) or customized.  But it seems that such is not the case when it comes to semiconductor equipment, where in some instances the cost of used semi production equipment is higher than the original price paid.  While this is a hard concept to grasp, there are two reasons why this has been happening.
First, last year saw a 19% increase in semiconductor equipment sales billings according to SEMI, and that data shows that China is now the largest market for semiconductor equipment, increasing its share from 23% in 2019 to 26% in 2020, and outspending Taiwan (#2) by 9.2% and given the Chinese government’s mandate toward semiconductor self-sufficiency, there seems to be no end to available financing for legitimate semiconductor projects.  The issue that makes this more difficult are the restrictions placed on Chinese companies by the US government, which limits their ability to purchase certain semiconductor capital equipment that is necessary for reaching that goal.
China has already been a purchaser of used semiconductor equipment over the last few years, but given that many vendors of such equipment are small and are not listed companies, the data for the 2nd hand market is hard to come by.  That said, in theory, the same restrictions that govern the sale of new semi equipment also carry to used equipment so even 2nd hand semiconductor tools face license restrictions, but that only works toward increasing demand, with China’s domestic tool suppliers not quite ready for prime time.
Despite the issues, Chinese OSATs are under pressure to deliver product and are pushing to expand capacity while the market is strong.  While they might not be able to buy EUV tools for 5nm nodes, there is plenty of business at 28nm and above, and there are few restrictions on such tools, but given the strong demand globally, lead times are long, if one can even get in the queue.  That causes OSATs to look for used equipment which is immediately available, and the competition for such tools becomes aggressive enough to bring prices for what was once almost valueless used equipment, to premium prices.  It is not easy to accept that used equipment costs more than new equipment in some cases, but if you are unable to fill customer orders during profitable times and lead times are measured in multiple quarters, the extra cost is part of doing business. 

​LG Electronics (066570.KS) is the leader in rollable OLED TVs.  LG is also the only manufacturer of rollable OLED TVs, and this very special product has only been available in South Korea, the company’s home.  That said, LG is now offering the  set in 15 countries, or sort of, as the furthest you can get toward ordering the ‘OLED R’ is selecting your country and calling or e-mailing an LG representative in New Orleans.  No mention yet of the price, which in South Korea is around $87,000 and in the UK for $138,000, but expectations are for an even $100,000, which should come with a gold ‘sucker’ emblem on the back.
South Korean media has suggested that only 10 of the units have actually been sold, so opening sales up to the global market seems a good idea.  That would imply 64 units in the US or 1,500 globally based on South Korean sales, so we doubt there would be any difficulty in getting one if you so desired as we suspect there are still a few lying around from the initial run.  Not that there are lots of estimates for how many rollable OLED TVs will be sold, but OMDIA predicts that 3,000 units will be sold in 2022, rising to 74,000 in 2024, and 672,000 in 2027.  That’s a cool $300m next year at an average price of $100k each, but even when compared to regular OLED TVs, that sold ~2.2m units last year, it is far below insignificant.  If LG Display sells its expected 7.5m displays this year and all get sold at an average price of $1,500 (just an arbitrary number), that would be $11.25b and OLED TVs are only expected to capture ~3.3% of the TV market in terms of units.  Time better spent promoting other TVs…any other TVs.

​In September of last year, Softbank (9984.JP) agreed to sell its controlling interest in ARM Holdings (pvt) for $40b to Nvidia (NVDA).  Such a large deal and the fact that ARM architecture underlies many processors powering everything from supercomputers to smartphones, the deal has been and will be under extreme scrutiny, particularly by the UK, given that ARM is based in Cambridge, UK.  Today, the UK Department for Digital Culture, Media, and Sport issued a public interest intervention, which is essentially an investigation into the merger to see if the merger is in the best interests of UK citizens and meets with all applicable UK laws.  A decision is required by July 30 of this year.  The FTC is already investigating the merger as is the UK’s ‘Competition & Markets Authority’, who is evaluating the effect of the merger on competition.

TOF, or Time of Flight is a process that measures the time it takes for a signal to reach and be reflected by an object.  TOF has a number of forms and names, particularly LIDAR, which gets much press related to autonomous vehicles.  As we have noted in the past, there are two ‘flavors’ of TOF, direct and indirect, having different characteristics and cost parameters.  While a few smartphone brands experimented with TOF sensors in 2018, Samsung Electronics (005930,KS) was an early proponent of TOF, using it in its Galaxy S10 5G flagship smartphone in February of 2019, following up with a number of other models such as the Galaxy Note 10+ (8/2019), the Galaxy S20+ (2/2020), the S20 Ultra (2/2020), but did not include a TOF sensor in the Galaxy S20, the least costly of the S20 line.
This omission led some to believe that Samsung had abandon TOF technology, either for performance or cost issues, however Apple’s (AAPL) use of the technology in the iPhone 12 those questions began to dissipate.  However Samsung did not use TOF in any of the three flagship Galaxy S21 line (January 2021) the controversy began again, with the speculation that Samsung will not use the technology for the next generation of Galaxy smartphones, the S22 line due out next February.
In understanding the circumstances around Samsung’s previous decision it is important to understand that there is a difference between ‘types’ of TOF sensors, each with different characteristics, which we believe has more to do with Samsung’s decisions about TOF than the technology itself.  In the table below we have mapped out a simple comparison of the characteristics of the two types of TOF sensors currently being used.  From a technical standpoint, range is important, but accuracy and cost of production are more so.  There are two primary suppliers of TOF modules, an affiliate of Samsung and Sony (SNE), and as noted, Samsung produces the indirect version (measures distance using reflected ‘phase’ information), while Sony produces the direct version (measures distance by calculating the time it takes for the actual reflected signal to return).
As Apple settled on Sony’s solution, it became obvious to Samsung that if it were to continue to directly compete with Apple (Galaxy S Series vs. iPhone series), lower cost was less of an advantage than accuracy, especially accuracy over a greater distance.  In order to change the characteristics of the Galaxy line, Samsung had the choice of buying TOF modules from its biggest competitor or abandoning the technology.  We expect that Sony would be less inclined to sell to Samsung, preferring customers that it can retain, and Samsung would likely be low on the preferred customer list, leaving them open to potential delivery shortfalls, so the issue has been moot for some time.
That said, Samsung has released its own TOF sensor, the Vizion 33D (March 2021) which we believe retains the indirect technology, but is said (by Samsung) to capture more phase information, making it comparable to the Sony direct TOF product.  If this is so, the logic would be that Samsung would promote its own TOF device in its premier smartphone lines, so recent speculation that TOF will again be excluded from the S22 is questionable, especially this far from production.  That said, one limiting factor is production capability, which could have been a factor in the decision not to use TOF in the Galaxy S21, as the S21 has seen very strong sales relative to the Galaxy S20, but we expect Samsung LSI will have ramped up production by the 4th quarter of this year to a point where there should be little potential for capacity issues.
That leaves marketing, and while we expect Samsung LSI is certainly hoping to sell the new TOF to parent Samsung Electronics, it will also try to garner outside customers, who might need some reassurance that the largest smartphone brand worldwide would be using such a device in its most visible line.  As we have seen before,  Samsung Electronics does not always look kindly on its affiliates products, as indicated by its early reticence to adopt Samsung Display’s (pvt) QD/OLED displays, but without at least some ‘cred’ from Samsung Electronics in upcoming models, it will be hard for Samsung LSI to justify continued TOF product development.

​According to local South Korean press, there was an outbreak of COVID-19 that started with one office worker who tested positive at LG Display’s (LPL) Gumi production complex in South Korea.  That led to further testing which revealed that 7 co-workers and the wife and daughter of the original positive case had tested positive.  The office in which they worked has been closed and all employees moved to remote communication only, and testing all of the 800 employees in the Gumi facility, which houses a number of LCD and OLED lines has been completed.  The company is currently waiting for results and the lines remain open.  Hopefully the infection is limited to the office in which it was originally found.

​We all get them, calls about extending our car warranty, or winning a free trip to somewhere, or the more onerous ones about a relative who is being detained or that someone has been using your social security number.  Scammers are alive and well and don’t care about the US ‘Do Not Call List’, particularly as most are off shore and are likely either paying off local authorities to look the other way, or know that there is little any legal organization is going to do because the sums are relatively small.  Well, a new record was set in Hong Kong, where the number of mobile phone fraud cases increased from 648 in 2019 to 1,193 last year, an infinitesimal number compared to the 2.1m ‘fraud incidences’ reported to the FTC last year in the US, with about $3.3b in losses, although not all of these were from phone scams.  (On-line shopping scams were the largest overall).
An elderly woman living in a residence on the very exclusive ‘Peak’ in Hong Kong, (Christie’s has 19 homes for sale there, with the cheapest a mere $8.2m and the most expensive ~$451m) was contacted by a self-proclaimed government official who stated that she had been involved in a serious case of money laundering on the Mainland.  The scammer went as far as to send someone to the woman’s address to give her a ‘special’ mobile phone for her to use concerning the investigation.  The woman was instructed to send HK$7.9m, roughly $1m US to a particular account so it could be examined to see if it had been involved in the money laundering, which she did.  That was followed by an additional request for a total of HK$2509m ($32.2m US) and the scammer even had the woman’s personal driver bring him and her to the bank together so she could make the transfer.
The woman’s helper was smart enough to call her daughter to report the suspicious transactions who asked relatives to go to the bank to monitor the transaction, however they did not follow up on the issue and the transfers continued until the entire HK$250m had been transferred to the scammers account over a few weeks.  Eventually the daughter followed up and was able to freeze what was left in the scammer’s account, about HK$9m ($1.16m US) and police arrested a 19 year old university student.  Of course this is an unusual extreme (and a new record), but it happens all the time.  There are a few phone tools that can help and a recent FCC ruling makes it easier for carriers to filter out the more obvious scams, but the best way to avoid being scammed is by not answering calls from numbers you do not know.  If they don’t leave a message, it wasn’t important, and if it is an odd sounding one, don’t call back, just call the institution represented to verify the call was real, rarely if ever government agencies call on the phone….

While display is still the most costly single component in most visual CE products, semiconductors, in a variety of forms, represent an increasing larger basis for BOM and potential pain points as device complexity increases.  Over the last year an unusual spike in certain CE products due to the lifestyle changes wrought by COVID-19, have strained semiconductor capacity, which has little room to adjust to short-term changes such as we have seen.  This has led to high utilization rates at semiconductor foundries, foundry capacity allocation, and most recently wafer price increases or the elimination of more traditional foundry rebates to customers.
Given that Taiwan Semiconductor (TSM) is the largest global semiconductor foundry, we have spent a bit more time looking at 1Q results, and more importantly guidance for some hints as to what to expect for the remainder of this year and next.  Putting aside the typical company-speak, we summarize TSM’s comments and Q&A to give some perspective on what investors should expect from TSM going forward, not only as the largest foundry, but as a company with a large variety of customers.  We refer back to our comments made yesterday for more data.
2Q Guidance
               Revenue flat q/q
                              HPC up, offset by smartphones down – HPC up no surprise given both bitcoin and PC/Notebook demand, but smartphone weakness seems to not be reconciled with recent strong smartphone demand in China, which would likely lead to smartphone brands having more confidence in full year goals.  2Q smartphone foundry weakness could be just a lull before real 2H smartphone production demand sets in, or it could be the ‘Huawei effect’.  If 3Q guidance for the smartphone segment is not better then it would give us doubt as to the conviction smartphone brands have toward the 2nd half.
               Inventories
                              Up in 1Q – Typical seasonality story doesn’t play as well currently because of high utilization rates unless TSM is pre-building against 3Q delivery orders already on the books and is getting compensated for insuring full order delivery guaranties to large customers.
                              Should decline going forward – TSM also expects overall customer inventory levels to be higher than ‘historic seasonality levels’, which is certainly a positive for TSM demand, but also raises the risk  at the customer level if there is any slowing in end-user demand.  While the company was careful to restate how closely it works with its customers on order and inventory management, they did note that they cannot rule out the potential negative effect that overbooking might have, although they were not suggesting that currently.
               Automotive – Expect automotive shortage to be ‘greatly reduced’ by next quarter – A bit hard to imaging how that would be the case unless automotive customers have already agreed to price increases that are higher than other components, or slower smartphone demand has freed up some capacity in 2Q that was not available in 1Q.
               Margins – Pressure from increasing proportion of 5nm business, which currently has higher cost and lower efficiency than more mature nodes.  As they expect 5nm to revenue to be 20% of revenue this year, that would imply considerable strength in the 5nm node for the remiander4 of the tear, as 1Q saw 14% od revenue from 5nm.  Increasing the cost efficiency of the 5nm process will be key in seeing margin improvement since high utilization on other nodes makes it more difficult for TSM to improve process
Full Year Guidance
               Semiconductor Market Growth (ex memory) – up 12%
               Foundry Growth – up 16%
               TSM Growth – up 20%
               Expect overall demand to remain strong and for shortages to continue through this year and may extend into 2022.
               Expect HPC and automotive to be higher than corporate average for 2021 – The company has seen demand strength from both segments over the last 3 months.  As we noted yesterday, HPC strength is relatively easy to understand, but we wonder if automotive strength includes any overbooking as desperate customers vie for capacity.
               Capex up to $30b from $26 previously
                              80% of 2021 capex on 3nm, 5nm, 7nm nodes
                              10% on Advanced packaging
                              10% on specialty technology
               Will spend $100b over 2021 – 2023 period – Unusual in that they gave long-term capex guidance.  Likely due to spending announcements by competitors Samsung (005930.KS) and Intel (INTC).
               Customer interest in 3nm – 7nm nodes is increasing – Regardless of whether that translates into revenue as soon as TSM hopes, they have little choice but to spend on 3nm – 7nm nodes in order to try to stay ahead of Samsung, but moving up the learning curve will determine whether they can continue to generate record sales and profitability id utilization stays high.
All in, it was a good quarter for TSM, who should be generating peak sales and profits, even as 5nm grows.  While the company was obviously optimistic about the next 12 months and the three year period encompassing 2021 to 2023, there were some points that would likely keep expectations from getting out of control.  Aside from the margin pressure from 5nm, the slower pace of overall cost/process improvement because of the high utilization, which is a good problem to have, and a 30% increase in depreciation this year, along with the mention of over-ordering, point to some of the limiting factors that should keep estimates under control.  On the other side of the equation demand remains strong and the positive effect of wafer price increases should allow TSM to meet its 20% growth target this year, barring any further exogenous variables or natural disasters.