As we have previously noted, Samsung Display (pvt) has been shipping samples of its Quantum Dot/OLED display panels to parent Samsung Electronics and other potential customers for evaluation, with the hope that they will find enough demand to justify the production of such panels at their Q1 fab in Asan, South Korea.  While the project has faced an initial lack of enthusiasm from Samsung Electronics, more recently the project seems to have been accepted wholly, with SDC committing to starting production in November and sets shown at CES 2022 in early January.
While we are optimistic about the prospects for QD/OLED, we are a bit conservative in what we believe can be produced in 2022, based on production limitations and yield more than demand.  We expect production to be 1.32m units over the next 14 months ending 12/22, with 1.28m being produced in 2022 itself.  We base these estimates on capacity and yield, using 55” and 65” displays only, as these are a logical combination using MMG (Multi-mode glass) on a Gen 8.5 line.  When actual set shipments will begin and where the sets will be released is still quite speculative, and we note that our estimates are a bit lower than most we have seen.

​On April 20 ‘Foxconn (2317.TT) Wisconsin Official Twitter (TWTR) Account’ announced that the Wisconsin Economic Development Corp (aka WEDC) had endorsed an amendment to the initial 2017 agreement, “…that will give Foxconn the flexibility to pursue business opportunities in response to changing global market conditions,” while also noting that “…original projections used during negotiations in 2017 have at this time changed due to unanticipated market fluctuations.”
The new agreement provides for:

• Up to $80m in performance based tax incentives to Foxconn based on the creation of 1,454 qualified workers earning an average wage of $53,875 and target capital expenditures of $672m by 2026.  The original agreement was for $2.85b in state tax credits and the hiring of ~13,000 workers over 10 years.  None of the original goals were met to date.
• Less product restrictions as to type of products or goods manufactured.  The original claims were for a Gen 10.5 LCD production fab, which changed a number of times, decreasing to plans for vending machine production at the facility.
• Preserves benefits of the EIT Manufacturing Zone, including market based energy rates and use tax exemptions.  The original agreement bypassed many environmental and water use restrictions.  No indication as to whether they have changed but not likely.

Foxconn sponsored a job fair on September 8 at the ‘Advanced Manufacturing Facility’ where they expected to hire 100 employees to fill a variety of manufacturing, information technology, and supply chain positions, despite the fact that it remains unclear what Foxconn currently expects to produce at the facility, although the company announced a partnership with Fisker Inc (FSR) to produce parts for electric vehicles including semiconductor products, along with ‘plans for electric vehicle manufacturing’.
That said, today Foxconn has disclosed plans to acquire a 6.2m ft2 assembly plant in Ohio, as part of a joint development agreement with Lordstown Motors (RIDE).  The agreement, which is non-binding until completion, includes:

• Purchase the Lordstown facility, excluding the motor assembly line, battery assets, and certain IP, for $230m based on Foxconn agreeing to manufacture Lordstown Motors Endurance full-size truck at the facility, along with employment agreements for Lordstown operational and manufacturing employees.
• Purchase $50m in Lordstown Motors stock @ $6.8983/share and the issuance of warrants covering 1.7m shares exercisable for three years at $10.50/share.

While the Wisconsin “Wonder of the World” project was never intended for automotive production, there were local hopes that the talks with Fisker might evolve into something relating to the upcoming Fisker Ocean electric SUVs due out in 4Q or 1Q 2022.  That possibility still exists to a degree, but the deal with Lordstown seems to have taken Foxconn’s focus away from Wisconsin, where little, if anything, has been produced, despite Foxconn’s insistence that they have spent ~$900m in Wisconsin, including the  almost 1m ft2 ‘Advanced Manufacturing Facility”, a ~300,000 ft2 ‘Smart Manufacturing Center’, a
120,000 ft2 ‘Multi-purpose Building, and a 100 ft tall “High Performance Computer Data Center Globe’. 
The company still touts their ‘3+3 corporate vision – electric vehicles, digital health, and robotics using AI, semiconductors, and 5G’ for the Wisconsin site, saying that that idea has “caught the attention of many businesses and investors from around the world who share their vision for the site, making the site and the region a new hub for next gen technological product design, jobs, investment, and smart manufacturing.”  The problem is that we go by results rather than ‘vision’, especially one that includes all of the key buzzwords that investors seem to like to hear.  The aerial images of the Foxconn site taken on 9/17/21 and 7/20/2021 don’t seem to indicate that anyone is ‘sharing the vision’, nor does the fact that little has been produced at the site other than construction jobs.  With the Lordstown deal in the works, it looks like Foxconn might not be focused on its vision for Wisconsin at the moment.

​Sanan (600703.CH) has announced that it will sell shares in a non-public offering to raise capital for a Mini/Micro LED project in Hubei Province.  While the amount of the offering has not been specified, we believe it will be complemented by a subsidy from the local government.  The project is said to composed of an R&D center, a line(s) for Gallium Nitride (blue or green), Gallium Arsenide (red) chips, and a line for packing the LEDs into 4K displays.  The project, will be situated in Ezhou Gedian Development Zone, where a number of Chinese semiconductor and panel producers have existing facilities.  More to come. 

Samsung Electro-Mechanics (009150.CH) has decided to make a big investment in semiconductor substrates and is currently waiting for board approval to start the ~$845.3m expansion project.  Given that Samsung EM spent $609m last year for facility maintenance and expansion last year, across the entire company, which also includes production facilities for passive components and camera modules, the amount being spent is significant.  In fact, the company asked the board to approve a $1.44b investment but the board lowered that to between $845m and $930m.
As almost all semiconductor substrate products have been in short supply for almost a year, made worse by a series of fires at Unimicron (3037.TT) in October of last year and again in February of this year, and the potential closing of SEMCO’s RF PCB business later this year, delays in deliveries of PCB based components have been high enough to cause product roll out pushbacks and rising prices.  There are a limited number of substrate suppliers, and while the global market is large (Estimates range from $53b to $64b last year), many of the suppliers have a particular product niche.
In that regard, Samsung EM is expected to spend much of the capital on Flip-Chip BGA substrates where there are few suppliers, and is expected to be dedicating some or all of these lines to a “US Chip Company”, likely Intel (INTC), who will also be making an investment in the lines.  FC-BGA is most commonly used for CPUs, GPUs for servers and mobile devices (smartphone application processors).  Samsung EM currently generates ~$423m from FC-BGA, which is expected to grow to over $1b when the expansion is complete.  No timetable for the expansion has been set, although we expect more detail once the board gives final approval.

Smartphone brands have to put a lot on the table when making estimates of how many units they expect to sell each year, as that data becomes the jumping off point for the product supply chain.  While modifications can be made after the new devices are released, in order to build enough stock, hard unit numbers must be supplied at least 3 months in advances, and given the shortages the industry has been facing over the last year, more likely those numbers must now be locked in 6 months ahead to avoid delivery delays.  However, it gets even more complicated as companies like Samsung have a multitude of smartphone series, and a number of models in each with different component line-ups and variations down to the basic component level.
The real work for component buyers comes before assemblers have any idea of what they will be producing as brands must also make sure either they or the supplier can access the number of components needed for each variant, and that those supplies meet brand unit volume goals.  Taking the example of Samsung’s (005930.KS) Galaxy S line, which is the company’s most visible flagship smartphone product, the current (S21) series has three basic models, the S21, the S21+, and the S21 Ultra.  All three have different displays, case sizes, cameras, batteries, and many, many differences in internal components, so not only must suppliers get an order specifying the total number of units overall, but it must be broken down into every potential variant.  If those orders, based on both the brand’s estimated unit volumes and estimated component and labor costs are wrong, in either direction, the profitability of that line will suffer.  At an estimated $415 BOM for the Galaxy S21, incorrect estimates leading to delivery issues or excess inventory add up quickly.
On a broad basis the table below shows what Samsung has, and is expected to order for the three Galaxy S2x variants in terms of percentage of total (Note – Where ranges are given we take the center point which means the totals do not always add up to 100%).  The upcoming S22 will be the first time Samsung has ever allocated more than half of total production to a single model, which we take to indicate some concern for both the differences between the models and the price of the more expensive plus and ultra versions, and while Samsung had placed early orders for 30m units this year (S21), it is expected to be ordering only 20m initial units for the 2022 series.  We do note that Samsung’s initial expectations for the Galaxy S21 series was 26m units, the company upped that to 30m around the January release date.
Last year Samsung was surprised at the initial popularity of the Galaxy S20 Ultra and found that they had not ordered enough of that model to fill orders leading to shipment delays and cancellations.  It is easy to see that they expected a similar circumstance with the S21 series, more than doubling the previous year’s order, while some of that expected enthusiasm is reflected in the order breakdown for the S22 series, but the total number of S2x Ultra units declines from 9m this year to 5m next year, even with what we expect will be a flat or slightly lower BOM, which indicates Samsung’s lower risk profile for smartphones going into 2022.