Semi.org has made their predictions for silicon wafer shipments (area) through 2025, with growth of 4.8% this year and growth each year through 2025 except 2023.  Simply put, macro conditions in 2023 are blamed for the decline in polished wafers, but growth resumes again in 2024 and 2025, which we have to assume means that inflation is brought under control in 2024 and beyond.  Hopefully, that will be true, but we expect estimates to come down a bit further  for 2024 & 2025 as we move through 1Q 2023.

​In June we noted that Foxconn (2354.TT) had been offering aggressive bonuses to workers at its Zhengzhou assembly plant known as “iPhone City”, where it employs over 200,000 to assemble iPhones and similar Apple (AAPL) products.  The incentives were offered to assuage worker fears that COVID restrictions might keep them at the plant for an extended period of time and limit their ability to travel to their homes when not working.  Last week we noted that the Zhengzhou government had done just as the worker’s feared and locked down the city, with many workers having anticipated the move and left the plant days earlier.  Foxconn provides for the workers that remain on campus with dormitories, food, and a few stores, but communal meals and other typical amenities are restricted.
Early indications pointed to Foxconn’s ability to transfer some of the Zhengzhou workflow to its assembly plant in Shenzhen, which faced its own closings in March and July, but was still operating at normal capacity.  In our note last week we calculated that while the issues in Zhengzhou could push out delivery times, they would not substantially impact 4th quarter iPhone shipments, though in worst case could push some sales into 1Q ‘23.  It seems that we were wrong as Apple made the following announcement this morning:
“We continue to see strong demand for iPhone 14 Pro and iPhone 14 Pro Max models.  However, we now expect lower iPhone 14 Pro and iPhone 14 Pro Max shipments than we previously anticipated and customers will experience longer wait times to receive their new products.” 
While we expect there might be a bit of the general economic macro malaise built into that statement, it seems more workers than expected anticipated the lockdown and took off before the restrictions began, leaving the factory at lower capacity than we had expected.  Apple did not give specific targets for 4Q iPhone shipments, but street expectations seem to be concentrated between 50m and 55m units for this year.  Apple’s previously had an optimistic view of iPhone shipments, as in June the company was said to have increased orders for iPhone components from suppliers by 5%, and during the pre-order period that started September 9, early pre-orders, especially for the iPhone 14 Pro and Pro Max, were said to be ahead of the corresponding iPhone 13 models released last year.  With this morning’s iPhone statement and what is likely another week+ of lockdown in Zhengzhou, the lockdown seems to be having a greater impact than we originally thought.  Our estimation of an 11.2m unit shortfall seemed high at the time and a bit extreme, but today’s comments from Apple make it less so.

VR, in our view, is hardware still looking for an application, and while gaming seems to be the application of choice for those who value the VR experience, the VR industry continues to tout its use in other, seemingly more legitimate applications, such as VR classrooms, VR in medicine, VR in business, VR in travel, and of course, VR in shopping.  That said, gaming is the largest category for VR application, and while there are great differences in estimates for the VR gaming market itself, we expect that VR gaming sales are second only to VR hardware, with all of the other applications being secondary, other than shopping, which, in what are somewhat stunted forms currently, likely become the largest VR applications down the road.
That said, we came across a VR application that struck us by its very strangeness, the use of VR in psychedelic psychotherapy, a field that must rely on clinical trials given the Federal illegality of many psychedelic drugs.  Psychedelic therapy, which can be used for PTSD, anxiety, OCD, and other cognitive and social disorders, is difficult on its own, as it is expensive and must be done in a strictly controlled setting with trained clinicians, but Enosis Therapeutics (pvt), a small firm located in Melbourne, Australia has developed four VR applications that the OVID Clinic (pvt) in Berlin is using as part of its psychedelic psychotherapy, an adjunct to the clinic’s Ketamine-based therapy.
The VR applications are (company words):

• Grounding – Used during the preparation, dosing, and ‘integration’ stage of the therapy session, providing a transition space from daily life to therapy.
• Surrender – Used during the dosing stage, providing an ‘awe-evoking’ scenario that helps the patient to engage more deeply with their trip and sets the focus on their intention.
• Rescue – Used during dosing, known as ‘digital diazepam and reliably and transiently pulls the participant out of the experience if it becomes counterproductive..
• Anchoring  - Used during all stages of therapy to create a multi-sensory memory library.  Expands tangible connections between insights and creates a multi-sensory mind map as the meaning making process unfolds and invites users to revisit and consolidate their recorded insights into a memory library.

Whether you believe the company’s theories about how VR can be used in conjunction with psychedelic psychological therapy or not, you have to give the company credit for coming up with a novel use for VR.  While we are not sure that a VR environment is tolerable for anyone micro-dosing psychedelics, we can’t rule out the concept as we have yet to experience it, however we expect we would rely more on the clinical results from other studies rather than our own potential experience.  One typical aspect of those under the influence of psychedelic substances is the loss of reference points, sometimes relating to one’s body, and a similar loss of the sense of space, which are also analogous to the phenomena many experience in VR.  Whether the experience of using both psychedelics and VR intensifies or reduces those phenomena is an open question for most, but we expect the potential use of VR in psychedelic psychological therapy is not an application that will move the needle in the VR world.
Watch the Company promo here: https://youtu.be/l_B8OMsQ7xM
 

Only a few smartphone brands have the wherewithal to develop and use their own application processors.  Samsung (005930.KS) does and Apple (AAPL) has been doing it since 2010, while Google (GOOG) has its “Tensor” AP that it has recently used and Chinese brand Oppo (pvt) is in the throes of an AP development project, but most smartphone brands settle on one of the few AP producers, particularly Qualcomm (QCOM), Mediatek (2454.TT), UniSoc (state), or HiSilicon (pvt), or mix and match when necessary.
AP processor choice is a mix of price, performance, and availability that plays out in the year before models are released, with most consumers oblivious to what is running their smartphone.    Brands with models that vary widely in price, have to sacrifice performance for lower priced models, sometimes using an AP from an earlier generation in order to maintain margins within that price tier, but when it gets to high end flagship phones, the competition is fierce and processors can become a positive or negative to consumers.
Samsung went through such a scenario with the Galaxy S22, where it uses its home-grown AP (Exynos) processor in Europe and Qualcomm’s Snapdragon  chipset in the US and the rest of the world, something Samsung has been doing for years, although at times Samsung gets even more country specific as to which AP they use.  The Exynos processor, which Samsung designed, had some speed and power issues this year that caused a bit of a ruckus when compared to those same phones that used a Qualcomm Snapdragon AP, and recent indications from sources in Korea seem to indicate that Samsung is considering not using its own Exynos Ap in any of the Galaxy S23 flagship line to be released next year, regardless of its regional destination.
As the Samsung mobile division makes such decisions, and is ultimately responsible for the success of the smartphone line, regardless of whether it is the fault of the processor, we expect cost considerations were less important than another ping to Samsung’s smartphone reputation.  But this potential decision leads us to consider why Samsung has mixed processors on its flagship phones in previous years.  Understandably it is likely less expensive for Samsung to use its own processor in mid to low-priced phones, but why use it in flagship phones based on destination?  We have not been able to come up with a reasonable answer other than Samsung is trying to promote the Exynos processor whenever possible and the possibility that a self-designed processor will allow certain features to be enhanced, but we have been unable to understand (if that is the case) why Samsung would limit that to certain countries.
But Samsung’s AP design division is not taking the potential snub lying down and has split the processor design team in two, with one group working toward the development of a processor for computing and AI, and the other working toward the development of a processor specifically designed for communication applications.  Additionally the communications design team will receive input directly from the mobile division, allowing for a more ‘smartphone specific’ design.  We expect the reorganization will do little to change things for the Galaxy S23 line, but should have a positive effect on the S24 (2024), but even with Samsung’s massive resources there are a lot of moving parts to creating an AP that can outperform all other mobile application processors, and there are also questions as to how much Qualcomm being a customer of Samsung’s semiconductor business influences such decisions.  Ideally we expect Samsung would like to equip all of its smartphones with its own AP, but that has proven to be a difficult task for everyone other than Apple who has been at it for over 12 years and sells only top price tier phones.
 
 

In reference to passive components, we have mentioned Taiwan-based Yageo (2327.TT) a number of times concerning its leading share position in a number of passive component categories and its top 3 position in MLCC (Multi-layer Ceramic Capacitors), all of which the company has developed through acquisitions that target large vertical application segments and leverage those through direct sales and relationships with EMS providers and distributors.  Yageo is the share leader in SMD Resistors (34%), Tantalum Capacitors (43%), and 3rd in MLCCs (15%) and Inductors (13%) and continues to maintain a well balanced mix of components across the company as shown in Figure 1.

However, the company faces the same demand issues as other CE component manufacturers and is looking to bring up the share of its specialty products and lower its exposure to commodity components, with the goal of bringing up product segments that carry higher (premium) margins to 75% on a long-term basis.  That said, if we break down Yageo’s 2Q results by application category, we see that the ‘commodity segments (consumer, computer, & communications) represent 46% of sales, with 54% falling into ‘premium categories, which carry higher margins, longer order cycles, and have a higher barrier to entry.
What makes Yageo interesting is that last month the company made two acquisitions in a new product application segment that if included in 2Q results, would lower the ‘commodity’ share of sales from 46% to 36%, while increasing the premium segment from 54% of sales to 64% of sales (see Figure 2 & Figure 3), a major step toward the long-term goal of 75%.    Both acquisitions, which cost a total of $755.8m US, are in a new product category, that of sensors, including thin-film temperature sensors, electro-mechanical sensors, and electronic switches, all of which have applications in industrial and automotive applications (both premium categories) and putting the overall new sensor category at ~20% of sales based on 2021 sales from both acquired companies.

​All in, while we are not recommending Yageo to investors[1], we found it interesting that within a month the company’s two acquisitions changed the financial make-up of the company substantially and added another growth segment to the company’s component portfolio that is growing between 7% and 9% CAGR between 2021 and 2026.  This reduced the company’s exposure to cyclical /commodity components, with the company paying roughly 2x 2021 sales, and while the final closings will likely extend into 2023, we see it as an intelligent response to the volatility seen in many basic components over the last two years while not moving far from the company’s core manufacturing expertise, a far better response than the usual rhetoric of  ‘we will be focusing on more profitable product segments’, which usually translates into relatively minor changes to product structure and a quick return to ‘how it was’ as the cycle turns upward.


[1] We do not make specific company recommendations, nor do we have any stake in, affiliation with, or receive compensation from any company mentioned in our notes

​The Board of Directors of China’s BOE (200725.CH) has approved the capitalization of its latest OLED fab project, a $3.98b US project to be constructed in the Beijing Economic & Technological Development Zone.  The project is a Gen 6 OLED fab (1500x1850) that will sit on a 420,000 m2 site, and will be oriented toward the development and production of VR devices using micro-OLED technology.  The fab will also be capable of producing mini-LED backlight products and will have capabilities for both LTPS and LTPO backplane technologies.
The fab construction is expected to begin next year and be completed in 2025, with mass production scheduled for 2026.  BOE will be responsible for 11.5b RMB of the 29b RMB cost, with the remainder coming from a subsidiary of Beijing’s Economic Zone State-owned Asset Investment & Development program.  As is typical of such state financed projects, BOE will own 79.31% of the project, despite providing 39.7% of the financing.
This project will give BOE the ability to produce VR oriented micro-displays, a capability it does not presently have, and will further the development of its LTPO backplane technology when it is in operation.  LTPO is a key technology in that it combines the best aspects of LTPS and Oxide backplane technologies, but more to the point, is the backplane choice of Apple for OLED displays going forward.  Currently Samsung Display (pvt) is the only high volume LTPO provider, with LG Display (LPL) beginning to commercialize the technology, so if BOE is to compete with SDC, they need to spread their LTPO know-how to as many potential Apple projects as possible.   While Apple’s first foray into the XR world is expected to be produced by Sony (SNE) using micro-OLED, 2nd generation devices could be micro-OLED or micro-LED, both of which could be produced at the new BOE fab, according to the company, so it seems obvious that BOE is moving ahead with this project to gain ground with Apple.  That said, the competition for a 2nd generation VR display out a few years, will be intense and while Samsung Display is the de facto leader in small panel OLED, micro-OLED displays are far more difficult to produce than current smartphone or current VR displays, which are still primarily LCD displays.  While this project is an ambitious one, it will not be in production until 2026.  A lot can happen in the interim. 

Samsung Electronics, as the world’s largest producer of smartphones, has assembly plants across the entire globe, but over the last few years has farmed out both the design and assembly of some of its phones to outside sources under its Joint Development & Production program (JDM).    While Samsung shares in the design of these phones, the four outside assemblers work with Samsung on the design and are responsible for purchasing all components directly with suppliers, as opposed to OEMs, who are subcontractors and are told who they can order from.  Samsung has been increasing the amount of JDM it uses each year, moving from ~6% of total phone production in 2019, to over 10% in 2020 and is expected to reach almost 20% in 2023 in order to reduce the cost of Samsung’s low and mid-priced phones.
At the same time Samsung still produces the majority of its phones through its own facilities around the globe, utilizing lower-cost labor for the manually intensive assembly process.  Given the high cost of labor in South Korea, Samsung does little assembly work in its home country, with the majority of phone assembly going to factories it owns in Vietnam, but this year Vietnam has been plagued with COVID lockdowns and rapidly rising labor costs, up 6% this year after a one year freeze in 2021, so Samsung is reducing Vietnam’s share of total production from 50% to 60% this year to 45.8% next year, with much of the shift going to India and Indonesia, two regions where Samsung has aggressive long-term production plans.
Of course, in terms of actual production, much will depend on whether Samsung is able to reach its mobile phone production target for next year, which is 320m units, as it will likely fall short of its original 310m target for this year by ~10%, but the rise in JDM share will remain on the increase to contain costs.  There is however some downside to the JDM program, the loss of leverage for Samsung itself with suppliers.  As the JDM is responsible for choosing suppliers and purchases parts directly, Samsung will lose a bit more of the buying power it has with suppliers, and that could result in higher costs, so there is a difficult evaluation that has to be made if the JDM program is to continue to grow, and the difficult task of determining whether the cost of components is going up because of macro trends or due to a loss of leverage, will have to be made by the company, a complex task at best.  It seems the world learned much about the global supply chain over the last year, but it is infinitely more complex than most people think, with billions of dollars at stake with each new factory or supplier relationship
 

​In July we noted that Samsung Display had been having difficulty with it’s ink-jet printing tools on its QD/OLED line and had reconsidered replacing those tools, which had been supplied by Samsung Electronics (005930.KS) affiliate, SEMES (pvt), after a bake-off with US based IJP supplier Kateeva (pvt), despite the better results from the Kateeva tools, which SDC already uses to encapsulate OLED displays.  We were surprised when the SEMES decision was made, as Kateeva had considerable experience with SDC and was the performance leader, but we expect there were other criteria, likely price, that could have influenced the decision at that time.
Since then it has become known that SDC is either replacing or adding IJP capacity to its QD/OLD line using Kateeva IJP to deposit quantum dot materials, with South Korea’s HB Solutions (297890.KS) as the provider of software that sits on the Kateeva tool, and delivery of the combined tool set is expected this month under a $10.2m contract with HBS.  SDC was said to have already purchased the Kateeva tool late last year and provided it to HBS for the software modification.  As HB Solutions made a $13.5m investment in Kateeva earlier this year and also purchased rights to additional Kateeva patents in the US, there was a point at which, if the deal between Kateeva and SDC did not go through, HB Solutions would become the owner of some of Kateeva’s collateralized patents, but it seems that any recent delays had little to do with Kateeva.
While it will take some time to bring the Kateeva tool into full production, we expect that the addition of this new IJP tool is part of the increased capacity SDC has been touting in relation to its QD/OLED fab.  Earlier yield issues are now assumed to be at least in part, related to the SEMES IJP tools, and the use of the Kateeva/HBS tool should allow SDC to bring QD/OLED yields even higher than the recent 85% rate that has been estimated.  The battle over who should supply IJP to the QD/OLED project has been going on since early 2020 (likely even before that date), and now seems finally resolved with Kateeva the winner after Kateeva had spent considerable capital developing the tool for SDC.  While SDC’’s original choice of the SEMES IJP tool almost cost Kateeva its existence, it seems SDC has had enough sense to admit they were wrong and make the change, something a bit less common than one might think in the display space.  All in, it is a positive for HBS and Kateeva, but more so for SDC, who needs to expand QD/OLED capacity with as little capital as possible. 

Theoretically tablets have been around for almost 30 years, although we look at the tablet ‘era’ as starting with the Apple (AAPL) iPad, which sold 2x predictions in the first 3 quarters after its release in April 2010.  Since then tablets have had their ups and downs, originally focused on  the workplace, and accused of potentially destroying the notebook market, tablets remain a functional part of the CE landscape.  On a long-term basis (see Figure 2), the category has been in decline in unit volume terms, since 2015 but saw a resurgence in 2020, the result of COVID-19, as have a number of CE categories.

As the effects of the pandemic have eased over the last few quarters, tablet shipments began declining q/q in 2Q of last year and on a y/y basis in 3Q ’21.  That has continued with only the 2nd quarter of this year seeing a q/q increase, only to be followed by a q/q decline in this year’s 3rd quarter.  Comparing the first three quarters of this year to 2021 (same period) shows shipments down 17.5%, while 2021’s first three quarters were up 25.1% over 2020, and the first three quarters of 2020, when the pandemic started, were up 58.3% over the pre-pandemic first three quarters of 2019.  This brings us to the conclusion that while tablet shipments have been on the decline, they have not yet returned to pre-pandemic levels.
Does this mean tablets have found a new ‘normal’ level, or are they just returning to pre-pandemic levels more slowly than other CE categories?  Unfortunately we believe the later based on Figure 3 & Figure 4.  Figure 3 shows quarterly tablets shipped by brand and the quarterly number of panel produced.  We note that we zeroed the panel production numbers to shipment numbers in 1Q ’17 to show the difference between tablet shipments and tablet panel production, so there are no panel units shown in Figure 3.  However in Figure 4 we show cumulative tablet shipments and cumulative tablet panel shipments, along with the spread between the two. 
Figure 4 indicates that tablet panel shipments were relatively closely matched to tablet device shipments until the 4th quarter of 2018, when increasing tablet shipments more than offset tablet panels produced, leaving tablet brand buyers unable to meet quotas.  Given that this was an untenable situation for tablet brands, panel producers began to step up production, building to both their own and brand inventory.  This increasing level of tablet panel production continued through much of the pandemic, with the spread leveling off between ~22% and 23% as tablet sales began to slow last year.
This would suppose that in order for tablets to return to a more steady growth pattern, brands need to reduce inventory levels, reducing the spread between shipments and panel production, while panel producers could lower tablet panel production further, with both parties working toward a more JIT supply chain that will reduce the spread and carrying costs.  But this is a lot to ask an industry that has relatively little visibility and is quite sensitive to macro events, which leads us to believe that the slow return to pre-pandemic levels for tablet shipments will continue into 2023, with the typical positive bumps from yearly new model iPad launches that occur in the 4th quarter.  Hopefully the long-term trend line curve seen in Figure 2 will pan out to be a harbinger of a more rational tablet market, but we expect more of the same for the next few quarters.