​As we noted previously, semiconductor foundry capacity has been constrained and chip shortages have become a topic for consideration in developing estimates in the CE space.  Aside from the inherent capacity issues, a number of recent fires at fabs have made the problem worse and now Mother Nature has also come into play as an earthquake of the coast in Japan over the weekend, caused the Renesas (6723.JP) Naki plant to close on Monday so it could be checked for damage and restart procedures begun.
The site of the 7.1 magnitude quake was ~40km north of the Fukushima Daiichi Nuclear Power Plant, the site of the March 11, 2011 earthquake and Tsunami that caused a nuclear meltdown and radioactive particle release.  The issue at the Renesas Naki plant was a short power outage, and while the equipment was restarted, it is expected that full production capacity will take at least a week to be restored.  This particular fab is the company’s only 12” wafer fab and 28nm process node, which we believe has been focused on automotive semiconductor production, which has been one of the areas that has faced the most constraints in the foundry space.

​Amid negotiations with a number of cities that are vying for Samsung’s Electronics’ (005930.KS) proposed $17b semiconductor fab, it seems that Samsung was forced to shut down its existing production plant in Austin, Texas due to local power issues caused by the snow storm that is blanketing the country.  Negotiations between Austin Power, a community run service that is ostensibly owned by the State of Texas, and the Coalition for Clean, Affordable, Reliable Energy, a group that negotiates on behalf of the city’s largest power consumer, initially ended with a request to customers to conserve power but escalated to power being shut off at those facilities that had back-up generators.  The load however was greater than the disabled grid was able to handle, even at the reduced rates and they were asked to fully shut down production.
Samsung says while they were eventually forced to shut down the plant at 1PM yesterday, “With prior notice, appropriate measures have safely been taken for the facilities and wafers in production.  We will resume production as soon as power is restored.  We are discussing timing with the proper authorities.”  That said, Samsung’s semi plant was not the only one in Austin to be shut down as plants for NXP Semiconductor (NXPI), Infineon (IFX.GR), and Skorpios Technologies (pvt) were also affected.  As far as we know, the NXP, Infineon, and Skorpios lines were all 8” lines, with small 12” capabilities, while the Samsung plant is a 12” fab. With capabilities down to the 11um node, although much of the fab’s output is at 14nm through 32nm.
While the fabs were shut down more gradually than during a full blackout, we expect the fab to now be in a ‘cold’ mode, where no equipment is kept running.  This means while the equipment will be able to turn on again when the power is restored, many tools will have to run a number of trial runs before they can be put back into full production, and the process can take as long as a month to complete, depending on how out of alignment the tools are.
In the case of Samsung’s fab, estimates are that Austin accounts for ~28% of Samsung’s total production capacity, so a full month shutdown would reduce Samsung’s full year output by ~2.34% or 0.3% of global semiconductor sales, which does not sound like a lot, however the specific chips being produced at the fab can make a big difference to global supply  A 30 minute shutdown at Samsung’s fab in 2018 destroyed 3% of the world’s supply of NAND chips and a power outage at Samsung’s plant in Hwaseong, Korea plant last month would up taking the company several days to restore power and resume production.  Given that the industry is already experiencing chip shortages, any new production reductions can only make things worse overall, and specific items produced in Austin, could lead to some specific parts being in even more short supply.

China saw unusually strong mobile phone shipments in January at 40.1m units, up 50.8% m/m and up 92.7% y/y.  The 5 year y/y average growth for January is -14.5%, so January 2021 represents by far the biggest y/y increment since the data became available and the first positive January since numbers were supplied.  Of course, the assumption by the press is that this indicates that China is now in full recovery mode and has returned to the device growth that China has seen in the past, but we are a bit more cautious given the volatility seen last year.  We give some credence to the idea that Chinese smartphone consumers were buying in front of the New Year holiday, especially given that an increasing amount of device sales have been on-line rather than through brick and mortar sales, but less credence to the idea that such sales were stimulated by a large number of new smartphone models being released in China in January, as the data indicates that both overall new models and local brand new models declined 11.1% and 10.5% respectively on a m/m basis.
Another reason cited for the jump in Chinese smartphone sales was that of the new models released (see above) many were lower priced, allowing a greater base to purchase new phones.  In order to check this assumption, we looked at 53 new models that have been made available since January 1 of this year. And compared that data against last year’s total shipments by price.  We broke the data down to three price tiers (all data shows in the table) and compared each tier to the tier percentage from last year.
While we understand that the data shown below only represents only 12.9% of the year, it does give us some understanding of how new model smartphone pricing compares to last year.  Since we do not have granular data concerning new models for the January/February period last year, we converted the unit data into percentages, which allows for a better comparison.  Based on those comparisons, we can make the assumption that pricing has come down thus far in 2021, with the lowest price tier (0 - €300) showing a big jump from 68.3% to 82.3% and the highest price tier (€600 - €1000+) seeing a decline from 11.7% to 6.9%.  The one tier that did not follow the trend was the mid-tier (€300 - €600) where the share declined from 20.0% last year to 10.3% this year, however if you combine the mid and lower tiers, the aggregate (prices from 0 to €600) the share of those lower-priced models increased this year, lending some credibility to the idea that lower priced models helped to stimulate January Chinese smartphone sales.

That said, we believe that more of the jump in Chinese smartphone shipments in January is a result of a combination of the broad-based lower priced offerings, the continued increase in the share of 5G smartphone new models, which has been averaging above 50% since last October, and the increasing share of 5G smartphones as a percentage of all smartphones shipped in China, which has been just under 70% since last November.  When we take these considerations and the less traveling that Chinese citizens will be doing for this New Year holiday, we can see that consumers might want a new smartphone, especially a 5G smartphone, to communicate with relatives that are not traveling this year.  The bigger question however, is how sustainable is this bump, and with the understanding that things could change quickly, we look at Fig. 2 2020 data, and expect something similar this year, especially when looking at the long-term shipment picture shown in Fig. 4. 

​In the normal scheme of things microprocessors perform functions on data.  Since those functions can have more than one processing step, the processor sends the interim data to a memory bank until it is needed for the next calculation, at which point it is sent back to the processor.  While this is standard practice for typical systems, there is a great deal of traffic between the processor and the memory block, and depending on the type of calculations, the sequential nature of how the data is processed, the traffic, and the time between when the data is stored and when it is returned for more computations, all contribute to the amount of data that can be processed in a given timeframe.
Samsung, the leader in the memory market, has come up with a solution that should increase the throughput of existing systems, without adding to the processing load.  This is accomplished by adding a ‘mini’-processor (PCU or Programmable Computing Unit) to each bank of the memory (typically 32 per memory chip), in the memory itself, which can be programmed to do some of the data processing that would normally have to be done by the main processor.  As these ‘mini-processors are programmable, they can be set to perform functions based on the system’s application, making the device ‘intelligent memory’ or they can be turned off, allowing the memory to operate in ‘normal’ mode.  The PCUs utilize standard memory commands for the PCUs, so Samsung says early adopters would not have to make any hardware or software changes when using the new memory configuration.
Of course, everything comes at a price, and in this case we do not mean the actual cost of the chip, but in real estate.  Since the mini-processors are part of the memory chip itself, they take up some of the real estate that would normally be used for memory.  In fact the PCUs take up about half of the memory space of a standard 8Gb die so Samsung offsets some of that loss by combining both regular memory and PCU memory, returning about half of what was lost.  All in, Samsung says that the new chip, which is currently being tested by customers (expected for commercial release in 2H), provides 4x higher processing bandwidth than an off-chip solution, reduces power consumption by 70%, and does not need and changes to conventional memory controllers and their command protocols.   Not addressed thus far is the cost and the ability to keep such stacked chips cool given the internal processing, but if the device lives up to its promotional material, we expect the heat challenge will be taken care of, while the cost will likely depend on how quickly Samsung wants the new devices to become widespread.
 

​The trend in the TV business is toward bigger TVs.  TV buyers have options for TV sets ranging from ~ 18.5” to the massive 110” TV, whose screen is produced by China’s BOE (200725.CH), or if you really want to go big, you can buy Samsung’s (005930.KS) 292” *The Wall”, as long as the six figure price tag is not an obstacle, but sometimes, it turns out that small is better.  LG Display (LPL) is the only volume producers of OLED displays for TVs and as such they have control over what is available to consumers in terms of OLED TV size, but given the popularity of large format TVs the generic 55” model has been the mainstay of the OLED product line since its release in 2012.
As noted, bigger OLED sets are certainly available, but unless you were looking for some sort of specialty product, 55” and up was your venue, so why would LG come out with a smaller OLED TV set and why would anyone want one?  Last year LG Electronics (066570.KS) announced a 48” OLED TV and has been rumored to be releasing a 42” OLED TV panel in the near future, and as it turns out, the 48” OLED TV has proved far more popular than most thought.  So what prompted LG to go in this opposite direction?  There were two reasons.
First was gaming.  Gaming is visual and as the space picked up legitimacy over the last few years, gamers were willing to pay up for a better visual experience.  In the past gamers were limited to monitors, while TVs with lower resolution and slower response times were less of an option even though they came in larger sizes.  But as TV specs improved gamers saw the larger sizes as a way to enhance their ability to see both detail and a more realistic view, and began adopting TVs as a gaming platform.  Since such could also be used for watching regular TV or streaming, the cost was justifiable spread across a wider base.
OLED TVs are known for their high contrast, making dark areas where opponents might be hiding a bit more defined, and high response time reduces the blur or trails that might follow a fast moving object, so the gaming elite began to choose OLED TVs as the perfect monitor replacement.  However the fact is that OLED TVs are more expensive than regular LCD TVs and 55” LCD TVs are just a bit too large for a gamer to sit close to, without sustaining neck injury from constantly twisting from left to right.  The 48” OLED TV released by LG was the ‘Goldilocks’ solution, not too large and not too small, and in theory a bit less expensive than a larger set, making it an ideal choice for those gamers who wanted the best possible experience, without a specialized (and very expensive) custom solution.
But there was another reason for 48” OLED TVs, and one that had less to do with demand and more to do with supply.  Production efficiency is a key to display panel profitability and substrate efficiency, or the ability to utilize the greatest amount of the display substrate, is a large part of that.  As LG Display produces its large panel OLED displays on Gen 8.5 platforms employing a 59.2 ft2 sheet, utilizing as much of that substrate is a key to fab efficiency.  Unfortunately not every size panel can be efficiently cut from a Gen 8.5 substrate as shown in the table below, and as consumers migrated toward 65” OLED TVs and LGD shifted production toward larger sizes, efficiency declined.
This was not just a problem for LG Display but for all panel producers who found themselves in the same situation with LCD substrate efficiency, but panel producers are a wily lot and came up with a solution called ‘multi-mode’.  Instead of producing only one size panel on a sheet of substrate, fab production engineers figured out a way to mix panel sizes on the same sheet of substrate, and while this meant a slower TACT time, it was much more efficient for panels 65” and larger.  By using Multi-mode to produce 77” TV panels, panel producers are able to bring the substrate efficiency up from 59% to almost 83% by producing both two 77” panels and two 48” panels on a Gen 8.5 substrate.  This gave them the 77” panels that were in increasing demand, but also gave them 48” panels which were a novelty at the time.

As it turned out, the 48” panels created smaller OLED TVs that were initially popular because they were lower priced than larger OLED TVs and were more directly competing with LCD TVs at least from an absolute price point.  As 48” TVs became more widely available, the gamers mentioned above began to pick up on the possibilities for their particular application, and demand increased.  In fact, at times the demand for 48” OLED TVs and the high volume of 55” TVs produced has led to the 48” models being higher priced than 55” models, but all in, demand for LG Display’s 48” OLED TV panels has been higher than expected, to the point where LGD is allocating additional 48” capacity at its plant in Paju, Korea, and this time exclusive for 48” panels rather than mixed capacity.
While this trend is antithetical to the norm, the desire for larger TVs, it seems LGD has created a price point that is attractive to consumers who want an OLED TV but are somewhat constrained as to budget.  As the efficiency of producing a 48” panel on a Gen 8.5 fab is the highest among all sizes 32” and above (actually 49” is slightly higher), it seems that LGD has latched onto a hot product and we expect that 48” OLED TVs, now a difficult size to garner, will become more commonplace over the remainder of this year.  This will not only open the base of potential OLED TV buyers but will help LGD to increase the overall number of panels its produces, which while only a ‘numbers’ game, will help to increase OLED TV share, making it a more visible part of the TV set business.
Here’s what is available on Best Buy (BBY) as far as OLED TVs.  We note that other sellers will have different brands and prices.  We have excluded all models that are either unavailable within 300 miles (1) or are open-box only (2).

​There has been a great deal of speculation as to the effect US trade sanctions have had with China, with some pointing to difficulties faced by Huawei (pvt) and ZTE (000063.CH) and a general slowdown in Chinese semiconductor projects that were looking to bring Chinese fabs closer to 5nm nodes used by Taiwan Semi (TSM) and Samsung.  China has been trying to develop its semiconductor expertise for a number of years, and while it has not yet brought its foundries down below 28nm for commercial production, the trend toward development of its own semiconductor independence began before the trade sanctions became as serious as they are currently.
That said, one thing that the trade sanctions did seem to do was to light a bigger fire under provincial and local governments to initiate semiconductor projects, and while most are not oriented toward competing against TSM and Samsung at 7nm and 5nm nodes, there are a number of projects that are oriented toward more practical chip production with a bias toward power semiconductors, necessary items for the electric vehicle industry.  Not all of the projects below are specific to power semiconductors, but all initiated construction last year, and while timelines for such projects can change radically, as were have illustrated in the past, just in sheer numbers and spending, the success of even a few of such projects will give China a bit more standing in the semiconductor space.
The projects below are not fully constructed or in mass production, and some might never achieve that status but if nothing else, we give the provincial and local Chinese governments a nod toward making a concerted effort to establish at least a modicum of their own needs.  If you don’t have the means to procure equipment that takes you down to the 7nm/5nm node then fill in the blanks in areas where there are already shortages and will likely continue to be so as the markets for electric vehicles continues to grow, particularly in China. 
Ganzhou Mingguan Microelectronics Power Chip Project
Initiated by the Jiangxi Provincial government
Phase 1 – 8” Power chip wafer production
$930m US (Phase 1)
Phase 2 – 12” Wafer line
$2.2b US (Phase 2)
 
Changsha BYD Semiconductor
               R&D for Automotive electronics
                              $155m US
               8” wafer line
                              250k/yr. target production
 
Western Semiconductor IC Project
               Initiated by Youxian District government & Singapore Topu Electronics (601689.CH)
               Power Chip, RF, and 5G related production
               8” wafer production line
                              $1.25b US
 
Ganzhou CRRC Shengyilum Industrial Park Project
               8” wafer production line
                              Phase 1 - $1.24b US
               Power Chip and packaging
                              500k/yr. target
               Total Project Cost - $4b US
 
Xian Tongxinyuan Power Device Project
               Initiated by Xian Hi-Tech Zone
               8” power device line + power management
                              $457m US
 
Anhui Bengbu MEMS Manufacturing Project
               Initiated by Bengbu government
               8” MEMS (Micro-electro-mechanical systems) line
                              Smart Sensors
                              $155m US cost
               Target   50k MEMS wafers/yr.
                              400m MEMS chips/yr.
              

​Apple’s (AAPL) iPad is the global leader in tablet sales, shipping 58.8m units last year and reaching a 37% share, but in one region during the 4th quarter of last year, Apple came in 2nd to Samsung, an unusual event.  In EMEA in 4Q last year, Samsung Electronics outsold Apple by 14.7%, making it the 3rd year and the 5th quarter in a row that Samsung was able to beat out Apple tablet shipments in EMEA.  Normally we would not focus on such data, as it is a relatively small part of the global picture, but we note that while 4Q outperformance by Samsung in EMEA tends to be a function of price, with the Galaxy Tab S7 being between 20% and 24% cheaper than the iPad Pro, Samsung shipped more tablets in EMEA for every quarter in 2020, which had not been the case in previous years.
It seems that the COVID-19 pandemic has made the government voucher programs installed to assist students with remote learning have not only increased tablet and notebook sales, as we have noted previously, but have focused consumers on lower-priced alternatives in order to be covered by voucher programs.  While this is not particular to EMEA only, the data is quite obvious in EMEA, with y/y increases from other well-known but lower priced brands also seeing large increases in shipments in 4Q, with the only loss coming from Huawei, where US trade sanctions have made life difficult for the company outside of China.  On an overall y/y basis 4Q tablet shipments were up 11.7% as shown below.

OLED displays have a number of qualities that make them inherently superior to LCD displays, among them high contrast and response time, however these distinct improvements come with a price, burn-in.  OLED materials have a finite lifetime and that lifetime is a function of how much a pixel or sub-pixel is ‘used’, meaning ‘on’.  When OLED displays are showing normal content, the images and therefore the brightness of each sub-pixel changes constantly.  Each sub-pixel therefore ages at an ‘average’ rate.  However when an image, such as a logo remains located in the same position for an extended period, such as on news content or gaming, those pixels stay on at a high level of brightness and age more quickly than others.  As they age, the OLED material loses its ability to respond fully and those areas retain part of the image regardless of what is one the screen.
In order t compensate, OLED TV designers have come up with a number of schemes to reduce or eliminate burn-in, with the most common being a periodic shift of the image by one pixel.  Given the number of pixels on an OLED TV, this is not noticeable to the user but will move the aging of each pixel closer to the average and reduce potential burn-in.  Unfortunately if there is static imagery on the display OLED displays will burn-in as they age, perhaps to a lesser degree with shift compensation, but the materials are finite, so if you use your OLED smartphone for something that causes burn-in, like gaming, you could face some ghostly images as the phones ages.
Rather than live with OLED burn-in issues or having to buy another phone, there is a fix available from a small Canadian company (80mi from Niagara Falls) that has created a device that can fix burn-in on smartphones.  The company, IGNIS Innovation (pvt), has released a portable device that can reduce burn-in on OLED smartphones to an imperceptible level in about 90 seconds.  The system is based on the company’s algorithms that measure the characteristics of every sub-pixel in the phone’s display and sets up a ‘compensation table’ to adjust the brightness of each.  The table acts as an add/subtract number for each pixel that brings areas that have been burned-in back to normal and adjusts the overall display to a uniform brightness for each of the three primary colors.
Using the IGNIS device, a burned-in device can be repaired quickly by local or brand repair shops, although we do not yet have the cost of the unit.  The company says the platform and the technology is scalable which means it should be able to perform the same functions for OLED notebooks and monitors, and eventually for OLED TVs.  While the adjustment made by the IGNIS device will compensate for the burn-in, more typical repairs would be the replacement of the screen, depending on how severe the issue has become.  Under Samsung’s (005930.KS) direct replacement program screen replacement can be an expensive proposition, ranging from $549 for the Galaxy Z Fold 2 (Inner Screen) to $79 for the Galaxy A01, representing 37.9% and 53% of the selling price respectively as quoted by Samsung, so pricing for burn-in repair would likely be a subset of the full screen price.
Ctl+Click for the video:
https://youtu.be/oKSaoipIAqE
​

Over three years ago the Australian government asked the Australian Consumer & Competition Commission to examine the impact of Facebook (FB) and Google (GOOG) on competition and advertising in the country, setting off an 18 month investigation that not surprisingly found an imbalance in bargaining power between the news media and the social media platforms.  At the end of the inquiry the ACC issued a draft of a new ‘Code of Conduct’ and the parties were asked to comment.  Also not surprisingly, news media companies viewed the legislation positively while Facebook and Google were afraid such rules would set a precedent that would be copied on a global basis.
Both media companies used their platforms to rally an outcry against the potential rules, with Google embedding a Yellow Caution Sticker in its Australian home page and went further by telling content creators on YouTube (GOOG) to flood the ACCC with objections, while Facebook took a more direct approach, threatening to block Australians from sharing news items with friends and family if the bill was passed.  Things continued to heat up until the Australian Prime Minister stepped in and warned Facebook and Google against using coercion to influence the proposed plan.   The bill was tabled at the end of December and things began to quiet down, until last week when a third giant thundered into the fray, with Microsoft (MSFT) taking the side of the government and offering to fill the potential search gap that would be created if Google abandon its Australian customers in protest.
Microsoft President Brad Smith spoke to both the Prime Minister and Communications Minister and issued a statement assuring the government and population of Australia that news publishing is vital to the country and while Microsoft is not subject to the legislation, it would be willing to ‘live by the rules’ if the government were to designate Microsoft’, while helping small businesses and advertisers to transition to Bing instead of Google search.  While Microsoft’s altruistic sounding intentions are a bit hollow when looking at search share in Australia, where Google has a 94.5% share and Microsoft has a 3.6% share, there is considerable evidence that the two social media giants have an advantage in terms of bargaining leverage, with $81 out of every $100 spent on online advertising going to the pair, Generating $4.3b for Google and $700m for Facebook in the country last year.
The bill would not directly influence the dollar amounts that either could generate in advertising revenue, but would create a framework for negotiations between local news and media companies and the two giants.  If an agreement cannot be made the code assigns an arbiter to determine a fair payment level for the use of that news by Facebook and Google, and requires them to give advanced notice of any algorithm changes that might affect the new media business.  If either social media company refuses to negotiate with local news sources, they face a penalty of $10m, or 10% of annual revenue, or 3x the benefit gained, whichever is greater.  According to the Australian Prime Minister the bill is expected to pass into law ‘fairly soon’, with many countries watching to see how Facebook and Google react.

In order to understand where the display industry is focusing its attention on a long-term basis, investors need to have an understanding of the new technologies that are being developed.  Currently the industry is divided between existing LCD infrastructure and OLED infrastructure, with the former dominating the large panel space and the later just recently becoming the dominant technology in the small panel space.  From a technology and process standpoint these two technologies are quite different, with LCD technology based on a combination of backlight, liquid crystal, and color filter panels that convert the LED backlight to tri-colored dots (pixels) creating an image. 
Over the years there have been many developments that have kept this relatively older display technology from disappearing, such as quantum dots and LED backlights, along with a very substantial infrastructure, with competition from an alternative display technology, OLED, challenging LCD’s leadership, at least at the small panel level.  OLED displays differ from LCDs in that they do not have a backlight, as the pixel material itself emits light, and in the case of small panel OLED, there is also no color filter.  That said, the process for producing OLED is newer and more challenging, and while OLED displays are far more versatile in form than LCDs, OLEDs are still more expensive to produce.
As we noted, LCD stakeholders have a vested interest in maintaining the technology’s dominance, and to that end have continued to enhance the LCD backlight, a key part of the LCD display.  While there are almost 25m pixels in a 4K display the LED backlights in medium and large panel LCD displays have anywhere from a few to a few hundred LEDs to provide light to these pixels.  This means that if an area of the image displayed is relatively dark, the LEDs behind the dark pixels should be off, while the LEDs in areas where the image is bright should be on.  Given the mis-match between the number of pixels and the number of LEDs there are many instances where the light from those LEDs that are on leaks into areas where they are off, causing blacks to become grays, causing ‘halos’ around dark areas, or washing out colors.
In order to remove these issues, display and backlight manufacturers have continued to increase the number of LEDs in large panel displays, offering high-end displays with hundreds of LEDs to help to manage those problems.  Last year however, manufacturers took it one level further, introducing mini-LED backlights, which are LED backlights using smaller LEDs enabling 10,000 or more LEDs and giving display designers more control over trying to match pixel counts with LED counts.
Along with this concept of smaller and more LEDs come some issues, particularly bottlenecks in producing these mini-LED backlights due to the smaller sizes and higher LED count.  Aside from the inherent cost of a larger number of LEDs and the driver circuitry needed to make them work, tools that move these mini-LEDs from wafers to a mini-LED substrate need to move these tiny LEDs at a rate that makes the process cost effective.   Given that mini-LEDs tend to be ~.1mm or smaller, with the smallest ‘standard’ size LED being twice that (smaller than a typical grain of sand), standard pick and place tools are less than effective from a time and cost perspective.
We have noted that Kulicke & Soffa (KLIC) has been marketing a higher speed mini-LED transfer solution (Pixalux) and is expected to release a faster version later this year, which gives mini-LED producers a shot at bringing mini-LED backlight costs down relatively quickly.  This will allow mini-LED displays to move down the price curve and into high volume production as adoption increases this year and next, but display producers understand that even with the display enhancements that mini-LEDs produce, the technology is still based on LCDs, and the inherent limitations in LCD technology will eventually lessen the impact of any upgrades.
While panel producers are as eager to make money in the near-term, they also understand the limitations presented by LCD technology, especially given how OLED has impacted the small panel portion of the display space.  To that end, a number of panel producers, LED manufacturers, and CE companies are developing micro-LED displays, which is a self-emissive technology that competes directly with other self-emissive technologies such as OLED.  While mini-LEDs and micro-LEDs sound similar, they are quite different, not only in size, as micro-LEDs are an order of magnitude smaller than mini-LEDs, but are not dependent on LCD technology.
This has good points and bad points from an industry standpoint in that mini-LEDs capitalize on existing LCD infrastructure and are therefore not overly capital intensive, while there is no industry infrastructure for micro-LED display production.  As the process for producing micro-LED displays is more like semiconductor production, the initial capital costs to build out the technology will be relatively high, but more focused on assembly rather than process, but the bottleneck problems mentioned earlier for mini-LED backlight production are multiplied more than a thousand times, given that every one of the ~25m sub-pixels in a 4K display would be a single micro-LED.
Aside from the capital cost mentioned above, the two biggest issues that face micro-LED display development are that same ‘pick & place’ bottleneck faced by mini-LEDs, now magnified from ~10,000 to 25m, which means that standard pick and place tools, even the faster and more sophisticated ones developed for mini-LEDs, would not suffice for micro-LEDs.  While high-speed P&P tools could produce 5 or 10 mini-LED backlights/hour, it would take about 25 hours to produce just one micro-LED display.  In order to bring such numbers into the world of eventual profitability, a number of companies are trying to develop ways to move vast amounts of very small LEDs at much higher rates, with a number of very different and novel approaches to the problem.
Along with the issue of placement speed, there is a secondary problem that also must be addressed, that of reliability.  Even at 5 9’s, there would be 248 bad sub-pixels on each display, and that is assuming there was no damage to the LEDs during transfer, and that they were all placed correctly.  There can be no ‘bad’ pixels in a display, so there must be some way to either eliminate bad LEDs before they are transferred or replace non-working LEDs after the transfer and no matter how fast a transfer system can operate, removing and replacing bad LEDs will markedly slow the production time.
Each of the development paths that companies have taken for the development of micro-LED displays has plusses and minuses, with some requiring pre-processing steps or post-processing steps, some faster or slower than others, and some are still in the realm of the unknown with tools and processes still in development.  The companies supporting each of the various processes are believers that their technology will be the winner, and while most are small, just as KLIC found themselves in the mini-LED beneficial ‘first-mover’ situation, it is still anyone’s game.
The table below was taken from IP surrounding micro-LED tool development and shows most of the current processes being developed for micro-LED production.  While this is the simplified version of the table there are still a few points that need to be clarified.  The first is ‘active’ or ‘passive’, which indicates whether a system allows a mass transfer to take place unassisted, or whether the process is adjusted as it is happening.  Taking ‘Self-Assembly’ as an example, passive self-assembly would be the equivalent of ball bearings rolling across a surface that had indentations wherever a ball needed to be placed.  As the balls roll across the surface they fall into the indentations without guidance, but any that reach the other side without filling a hole must be returned to their initial position and the process repeated until all spaces are filled.  An active representation of that same system would be one where the indented surface was tiltable, which would move on axis until all the balls that did not initially find a hole were placed.
There are also characteristics such as the mechanism for actually removing the micro-LED die from the wafer onto the new substrate.  This can be done mechanically, either individually (slow) or as a group (faster), or can be done with a laser, which would tend to be faster but could cause heat damage, and as noted above, the number of steps for each process and the amount of pre-processing and post-processing are big factors that would contribute to the cost effectiveness of each potential process.  In reality absolute speed is not the only factor that could make the difference between a costly and cost-effective micro-LED process tool, especially when one considers that some of these processes, usually the passive ones, require that any defective die be removed before the transfer takes place, while active processes can keep from transferring those micro-LEDs that are already known to be defective.
Once the transfers have been made the entire array must be tested and any defective or misplaced LEDs must be replaced, which in some systems would mean returning the array to the tool and running it again to fill in the blanks.  Some tools have the pre and post test stages built in, which allows the system to repair and replace without moving the wafer or substrate, but this also makes the tool larger and more complex, so the jury is still out on which process will wind up as the ultimate winner.  That said, we expect it will take at least two years to commercialize any of these systems, which means in the interim, micro-LED displays will be an expensive novelty rather than a mass produced product.  That said, once these challenges are met, we expect micro-LEDs will find their place in the world of displays and hopefully will improve quality while maintaining or reducing cost.