All primary 5G indicators remained above trend lines in April and 5G smartphone offerings continued to accelerate, particularly since the beginning of the year, along with the number of overall 5G devices available.  We note also that we have added a new form factor this month, in-vehicle routers/hotspots, etc. which have begun to appear in new vehicles.  5G smartphones are still the dominant share of the total device count and have increased device share almost 3% since the end of the year, with the FWA (Fixed Wireless Access)/CPE (Customer Premises Equipment) category increasing in units from 26 at the end of last year to 37 presently, although its share of total form factors, 2nd only to smartphones, declined by 1.4%.  All in a strong month for the 5G ecosystem, despite any shortages that might have appeared, which thus far, have had little effect on the spread of the technology across CE and related devices.

​LG Display (LPL) is the de facto leader in the production of OLED TV displays and has been so since Samsun Display decided to opt out of large panel OLED display production in 2015.  LGD has large panel OLED production fabs in Paju, Korea and more recently in Guangzhou, China and has continued to add production capacity as it increases it OLED TV panel targets each year.  While parent LG Electronics (066570.KS) is their biggest customer, LGD has expanded its customer base to include over a dozen TV brands, with more added each quarter.  LG Display is expect to ship between 7m and 8m OLED TV display units this year, up from over 4m last year as the Guangzhou fab reaches full capacity and increases yields.
That said, we have noted that as LGD adds customers their existing capacity continues to tighten with sales more likely limited by capacity than demand.  To that end, when LG Display built the large panel OLED fab in Guangzhou, while it built two 30,000 sheet/month lines, it not only left physical room for expansion but also installed equipment for oxide back plane capacity in excess of what was them current production levels.  That equipment is installed, and along with a shorted tact time on key deposition tools will allow the fab to increase production capacity from 60,000 sheets/month to 90,000 sheets/month, increasing the company’s ability to meet expanding customer demand.
What makes this expansion unique is that the cost will be minimal as new deposition tools are not expected to be required and much of the TFT line is in place, albeit not operational.  We question when depreciation on the TFT line might begin based on Korean tax law, but the important fact is that LGD will incur little expense and considerable revenue from the capacity expansion, which is said to be starting in July.  The original plan for the fab was for the expansion to begin last July, but a delay in the plant’s original start-up and the effects of COVID-19 have pushed the expansion out by one year.  As LGD has been producing at the Guangzhou fab for a bit under two year, we expect the new line to come up quickly, especially as much of the equipment is either already in production or has been fully integrated to the fab, so we have little change to our LGD capacity timeline, which assumed a June capacity expansion, and while size mix will still be a factor in total unit volume this year, we expect the new capacity will help LGD to meet its unit volume goals this year, and will benefit key suppliers, particularly Universal Display (OLED) who both sells OLED materials to LGD but also receives a royalty based on the unit price of displays leaving the LGD fab.

The most popular Samsung (005930.KS) smartphone last year was the Galaxy A51, a mid-priced (~$400) phone that was released on December 16 in 2019, with a 5G version released in April of last year (~$500).  The higher-tier, Galaxy A71 ($550) was released in January of last year with its 5G version ($600) released in April 2020.  Samsung has announced this year’s 4G versions and one 5G version of those two popular Samsung phones, but there are some issues regarding when they will be released globally, with the A72 5G version rumored but officially unannounced.
The A52 5G is available in the US and some parts of Europe and while it is expected to be released in Korea next month, no hard date has been released to Korean retailers from Samsung as of yet.  While some believe that Samsung is staggering the release of all A Series smartphones this year to maximize sales, some Samsung suppliers are indicating that a shortage of application processors designed by Qualcomm (QCOM) are causing delays in the release schedule for both the A52 and A72 smartphones. 
Those application processors, while designed by Qualcomm, are actually produced by Samsung using an 8nm process.  Qualcomm’s CEO recently indicated he thought the semiconductor shortage would last through the end of this year and Samsung’s CEO stated that there was a serious imbalance in supply and demand of chips.  Given the shutdown of Samsung’s Austin fab, we expect that all of Samsung’s remaining fabs in Korea are operating as close to full capacity as possible tomake up for the capacity issue in Austin, but it looks like it is not enough to meet even Samsung’s own demand.

After an unusually strong January and March, smartphone shipments in China settled back to a more ‘normalized’ pace, similar to rates seen through much of last year’s 2nd half, but returned to negative m/m and y/y comparisons of -23.8% and -34.1% respectively.  While 5G smartphone shipments also declined 22.2% m/m they were still up 31.1% y/y and represented 77.8% of shipments for the month, the highest share rate seen since last June.   We note that while 1Q shipments in China were strong in both absolute and relative terms, last year’s 1Q was unnatural to say the least, so we look more at the average y/y decline in 2H last year, which was -23.0%, very similar to the y/y decline in April of this year. 
While the smartphone industry in China had high hopes for 2021 in terms of a return to growth, it seems that while certainly better than last year thus far, the industry is still facing growth issues this year.  While it won’t be seen in Chinese shipment numbers, Chinese smartphone brands will be feeling the effects of both component shortages and the massive COVID-19 outbreak in India, where Chinese brands hold considerable sway.

Yesterday we made a reference to Yangtze Memory Technologies’ (600345.CH) effort to identify and replace foreign components to reduce its dependence on off-Mainland sources.  While the company is publicly denying that it has set up a special task force to replace US suppliers and is committed to becoming part of the global semiconductor value chain, the response to US sanctions against Huawei (pvt) and other Chinese companies alleged to be associated with the Chinese military has been both obvious and well-documented.   The use of EDA software developed in the US is widespread in China and is a source of concern for Chinese design companies, as are parts sourced in the US or produced with US tools, and while a public exhortation of plans to reduce dependence of US based semiconductor tools, components, or software only serves to incite additional ire in the US, you don’t need a weatherman to know which way the wind blows (“Subterranean Homesick Blues” – Bob Dylan – 1965 https://open.spotify.com/album/1lPoRKSgZHQAYXxzBsOQ7v?highlight=spotify:track:6k9DUKMJpWvu6eFG3O64Lg).

​The last year has been both a difficult one and a good one for Samsung Display’s (pvt) large panel LCD business.  A radical decision to end all large panel LCD production by the end of last year, made during a period of falling LCD panel prices, was postponed as panel prices began to rise after 1Q last year.  Samsung Display did however sell its Suzhou, China fab to TCL(000100.CH)/Chinastar (pvt) for $1.08b US, and took a 12.3% stake in Chinastar to ensure access to production volumes, but kept its remaining large panel fabs in South Korea in operation as panel prices continued to rise.
It seems however, that Samsung Display is sticking to its original plans with Samsung Electronics affiliate Samsung C&T (012260.KS) representing SDC in the sale of two additional Samsung Display large panel fabs.  The two fabs that are said to be on the block are SDC’s Gen 7.5 fab known as L7-2 and L8-2-1, part of SDC’s Gen 8.5 line in Asan.  The lines have been in mass production mode since 2008 and 2010 respectively and while currently running much smaller amounts, had peak production in their prime of 90,000 and 100,000 sheets/month.   If both fabs are sold it would end SDC’s large panel production, leaving only fabs dedicated to small panel and QD/OLED production.
As both lines are relatively old, we expect it might be rather difficult for SDC to find a buyer that was willing to ship the fab to another location and upgrade the equipment, but the price will likely be the determining factor.  As both lines are used for panel production between 20” and 86”, but perform more efficiently at sizes more suited to IT displays and TV displays in the 55” category, a buyer would likely be purchasing same to enhance their IT panel production capabilities, which is currently a market that faces a capacity shortage, which makes this an ideal time for such a sale.  That said, it will also leave SDC more at risk to the success of its QD/OLED panel project and it small panel flexible OLED business, where it dominates the industry.
While the sale will include all production equipment, all production information will be removed and no IP will be involved in the sale, which should be of little concern for any LCD producing entity, but might lessen the value to provincial or local Chinese governments, who see the purchase of an already producing fab as an easy and quick way to enter the display market, which looks far more attractive to outside entities than it did a year ago.  The Chinese government however has instituted policies over the last year to afford itself more control over the purchase of what would be 2nd hand LCD production fabs.  This additional control gives the central government a larger call in the supply/demand balance that is currently working in the Chinese display industry’s favor and is justified by the government bailouts of a number of Chinese LCD production fabs over the last few years, particularly when panel prices are in decline.

Intel (INTC) is getting the band back together, really restarting its foundry and contract chip production business.  Money is not an issue or is talent for Intel, but building out a supply chain is a more daunting task as it involves negotiations that lead to financial commitments and living up to those requirements.  One area that Intel needs to develop is in the packaging space, as raw semiconductors need to be connected to PCB boards, a process that has become exceedingly more complex as processors and other semiconductor devices grow larger.
It seems that Intel has already approached one of its packaging suppliers, Samsung Electro-Mechanics (009150.KS) and is in discussions about investing what is expected to be hundreds of millions of dollars in FC-BGA, a connection technology used in high pin count devices such as server and PC CPUs.  However it does not seem that the two have reached an agreement of financial considerations, despite Intel offering to invest its own money first.  There are only a few FC-BGA suppliers of note, which limits Intel’s potential choices, with some assuming the company has already had discussions with its two larger FC-BGA suppliers Japan’s Ibiden (4062.JP) and Shinko Electric (6967.JP) but no agreement has been announced. 
While Samsung EM is a smaller supplier to Intel than Ibeden and Shinko, there are other smaller players that could potentially expand capacity over time, such as Daeduck Electronics (353200.KS) and LG Innotek (011070.KS) in Korea, Nanya (2408.TT) and Unimicron (3037.TT) in Taiwan, and AT&S (ATS.AV) in Austria, but competing against the Japanese and potentially Samsung EM, whose largest customer is parent Samsung Electronics (005930.KS) could prove difficult.  Negotiations with Intel have continued since March.

A German security researcher (code name ‘Stack Smashing’) indicated that he had successfully hacked Apple’s (AAPL) AirTag microcontroller and was able to reprogram the device for the first time.  Once a device is hacked it can be reprogramed to do whatever the hacker decides, such as tracking a device or opening a website application on a smartphone. Despite the security measures that Apple has incorporated in the AirTag, the operating system for such devices is far less sophisticated than iOS or Android, and even Apple indicated that the security of the AirTag is, “the industry’s earliest powerful preventive containment method” and said that AirTag tracking and prevention measures can be further strengthened and improved, which seems a bit less than reassuring.  It is just a matter of time before the same fate awaits Samsung’s recently released ‘SmartTag’, which will undermine the security tag space a bit further.   The video below shows the regular AirTag and one hacked to open a different website.
https://twitter.com/i/status/1391165711448518658
​

April panel sales in Taiwan declined for all three primary producers, however that should be looked at against both the strong March that all three saw, and the overall y/y strength that the panel industry has seen.  That said, based on the data from AU Optronics (AUOTY),  Innolux (3481.TT) and Hannstar (6116.TT), large panel production declined, with AUO, seeing shipment area -6.3% m/m (no data for y/y), Innolux saw large panel shipments decline 10.6%.  Hannstar, while it saw a large drop in large panel shipments (-66.7%) is a small player in the large panel space and saw a large jump in March, returning to a more normalized large panel shipment rate in April.  As AUO no longer gives direct large or small panel sales of shipments monthly, we rely on Innolux, who saw a 6.4% decline in small panel shipment, and Hannstar, the leader in small panel of the three, seeing an increase in small panel production of 16.8% m/m.

​As we expect panel prices rose across the board again in April, lower sales indicate an overall slowdown in production after a strong March.  We expect that panel producers pushed out as much product as possible in March before they faced depleted component or materials inventory, which began to limit production further in April.  We note that panel sales remain substantially above historic levels as panel pricing has risen dramatically since the beginning of last year, and that even if sales level continue to fall, y/y comparisons will remain strong at least until August/September when pricing stabilized somewhat.
At least for now, component shortages will cap out unit volume, with overall sales dependent on panel price increases, and AUO pulling down some production for yearly maintenance.  As a result of the shortages, we expect panel prices will continue to rise, offsetting lower panel volume, at least for the next month or two, but the reduced COVID-19 infection and hospitalization rates in North America and a number of other key CE markets could help to relax global stay-at-home demand during the summer months.  While there are a number of wild card countries still facing major COVID-19 outbreaks, the overall global infection and death rate has been almost flat since the beginning of the year.

5nm semiconductor process technology is at a premium and with only two foundries operating at that process level, others, or at least the few that could potentially afford the R&D and equipment cost of such a node are working toward similar production.  That said, both Taiwan Semiconductor (TSM) and Samsung remain the only mass production fabs at 5nm.   While semiconductor roadmaps allow for 4nm or potentially 3nm projects, IBM (IBM) has pushed the limits of Moore’s Law and announced it has developed the world’s first 2nm chip.  Using current terms however to describe transistor dimensions is a bit wonky, as earlier generations considered the two dimensional size of transistors, while today’s terminology describes three dimensional components, which means a 2D comparison of actual transistor size would not show the expected size reduction as structure stacking (3D) adds to component counts and gives an ‘equivalent’ to what would be a 2nm chip.
That said, the transistor density is till the measurement point for process nodes, and while they vary from manufacturer t manufacturer and fab to fab, the table below gives an idea of the density of transistors from various manufacturers at each process node, shown as ‘millions of transistors/mm2, and how they can differ.  IBM is using a 3 stack configuration to achieve higher densities (assumed) while Samsung will introduce a 3 stack cell at the 3nm node, while TSM is expected to wait until the 2nm node.   While IBM uses Samsung’s foundry services for much of its actual chip production, it is likely a jump ball as to who will be first to use the technology, which based on the estimates, would see a performance increase of 45% and an energy consumption reduction of 75% relative to the 7nm process.  

While this sounds very significant for IBM to be the first to develop chips at 2nm, we say develop because IBM will not be a producer of such a device given they sold their manufacturing facilities to Global Foundries (pvt) 7 years ago, with the R&D work being done in their Albany, New York research facility.  That said, they have partnership agreements with all three potential producers and we expect are in discussions as to how the technology implementation will progress, despite the necessity for mass production EUV and other process tools that need to be developed for such a device, so an actual 2nm processor is still a 2024 (or later) project.