After my article last week on the possibility of Foxconn building an LCD fab in the United States (DCSM 01.23.2017), I’ve had a number of e-mail exchanges from people wanting to understand some of the economics, so I’ve run through a few calculations to ponder. The starting point for this idea is that the USA is the biggest market for big-screen TVs, 60” and larger. Although the China market has a bigger average TV screen size and China has larger volumes than the US up to 55”, in the 60”+ category the US constitutes more than 50% of all global volume. So it’s worth asking the logical question: if the market for the biggest TV sizes is in the US, why make all these LCD screens and TVs in Asia and ship them all the way to the US? Why not make them in the US and save on all that shipping?
I have a side story on this question, going back about 10 years to the high growth days of LCD TV, when I was the Director of Marketing Intelligence at Corning Display Technologies. We would meet with TV makers to understand market dynamics and exchange forecasts, and during this time we met regularly with Sharp-USA, based in Mahwah, NJ. Sharp at that time had recently completed their Gen 8 fab at Kameyama, and were considering locations for the next big investment, that ultimately became the Gen 10 fab in Sakai City. The Sharp-US team told us that Sharp considered exactly the question above – why not build in the US to save on logistics? The next question became: where in the US should they build? Looking at a map of the US, some of their Japanese counterparts pointed to the center of the map. So the code name for the project became “Kansas-yama”, which of course ultimately never came to pass.
The first issue to understand is the quantity of TVs that could be produced by a single fab. The largest LCD fab today is that Gen 10 fab in Sakai City, now called Sakai Display Products with ownership split between Terry Gou (his personal investment, rather than Foxconn), Sharp (now a subsidiary of Foxconn, Samsung (a 5% share), and a few other suppliers. Several Gen 10.5 fabs have been announced and/or are being built in China, by BOE, CSOT, HKC, and Foxconn. The Gen 10.5 fabs have a substrate size of 2940 x 3370mm, and therefore are capable of making 65” and 75” TV panels with high efficiency, with one substrate making 8x65” or 6x75”.
While Foxconn has not given any information about a US fab other than the investment amount ($7 billion) and one location under discussion (Pennsylvania), we can make some assumptions about the fab to understand the implications. The investment figure is similar to, but a bit larger than, the large fabs being built in China. These fabs have a design capacity of 120,000 substrates per month. An investment of $7 billion would get more than 120k capacity if it was a-Si, but it some or all of the capacity was a more advanced backplane technology such as oxide-TFT LCD or OLED, then the required investment would increase rapidly. So let’s assume that the proposed fab has 120k of capacity.
Such a fab, when fully ramped and running at full capacity, would be capable of making:
120k * 8 panels / substrate * 90% yield * 12 months = 10.4 million 65” panels per year
Or alternately, 7.8 million 75” panels, or a smaller number of larger panels (132” or 11 foot diagonal would also be made efficiently with a 2-cut) or 23 million 44” panels or whatever combination Foxconn would prefer. Let’s assume that they target the biggest sizes, and split capacity between 65”, 75”, and larger sizes in the ratio 50%/25%/25%. This would allow them to make 5.2 million 65”, 2 million 75”, and about 800,000 larger sizes (90”, 100”, up to 132”). Such a capacity for large-size TV panels would enable this fab to completely supply the US market for 60” and larger sizes. There were about 4 million 60”+ TVs sold in the US last year, and although the market for very large TVs has seen its growth slow over the last several years, when we consider the growth of 4K and even 8K programming, it is not implausible for this market to double in the 6-7 years before this fab would be installed and running at full capacity.
If a similar number of TVs were in Asia, what would be the shipping cost to get these to market in the USA? For 65” TVs, about 150 can be packed into a standard 40-foot shipping container. With the cost for shipment from Shenzhen to Los Angeles at about $1500 per standard container, that’s $10 to get to port. Add to that the inland truck shipment, assume an average of 2000 miles at $1.50 per mile is $3000, or another $20 per set to get to the average warehouse, for a total of $30 shipping. If the TV sets were made in Mexico from panels made in Asia, the ocean shipping can be substantially reduced by shipping LC cells and completing module assembly in Mexico, reducing the ocean shipping by a factor 3-4, but if assembly is in Tijuana there is still the truck shipment, so figure $23 per set in that case.
If LCDs were made at a plant in the USA, and that fab was strategically located in the industrial Midwest, the average shipping distance might be reduced to 1000 miles or maybe a bit less, giving a figure of $10 per 65” set, compared to $23 for a Mexican-assembled TV or $30 for an Asian-assembled one. Although Pennsylvania was mentioned by Foxconn, Ohio or Kentucky would be a bit more centrally located, and each of those states has some LCD resources to provide a prospective manufacturer.
The TV industry historically has struggled with low margins. Subscribers to our Quarterly Display Supply Chain Financial Health Report will know that even some of the top brands in the industry squeeze by with profit margins of just a few percent of sales; several brands in 2016 managed to set record-level profits but those operating profits still did not exceed 10% of sales. The savings on logistics of $13 or $20 per TV set would be 2-3% for a typical set today retailing at $1000 with a $700 wholesale price, and this percentage would increase as the set price fell over time (in the TV industry, prices always fall year-after-year), potentially making the difference between an unsuccessful business and a profitable one.
After speaking with a reporter from Bloomberg who indicated V-Technology’s stock rose over 180% last year, I was curious about how all the equipment stocks did in 2016. So, I put together Table 1 below, which shows the end of the month stock price throughout 2016 relative to 12/30/15. As shown, V-Technology did indeed have the best stock performance in 2016, up 185%. Second, was TES, up 156%, AP Systems up 130%, Coherent up 107% and Applied Materials up 70%. The average equipment stock rose 58%, impressive indeed. Most of the growth occurred from May when supply got tight and prices started rising. Tight supply/demand in most display markets and demand for capital intensive flexible OLED fabs boosted the capital spending outlook for 2017, 2018 and 2019. A rising tide floats all boats explains the 58% growth, but why are some companies outperforming and some companies underperforming. First, let’s look at the winners.
The winners should be companies who are seeing increased adoption for their process step, are gaining significant share or are expected to in the near future.
V-Technology’s trailing 4-quarter revenues are up 93% vs. the prior 4-quarter period. It has found a successful niche as the photo tool of choice for the photo alignment process. It also developed a high resolution (2um) system for the TFT backplane market for 6G LTPS fabs using its micro lens array approach which likely has shorter lead times and is less costly than more expensive systems from Nikon and Canon. It also continues to produce color filter exposure systems and inspection, repair and measurement systems. However, it has also developed products for the OLED and LTPS/oxide markets that could further accelerate its growth. According to the recent Bloomberg interview, V-Tech spent 5 years developing fine hybrid masks for use in vertical evaporation systems. The V-Tech. President, Shigeto Sugimoto, claims that if adopted these masks could bring in sales of as much as $441M, larger than its previous year’s revenues. The benefits of the mask include being thinner, lighter, less sag, ability to achieve smaller patterns (~5um vs. ~20um) and pattern pitch (2um vs. 5um) and higher pixels per inch (~700 vs. 570) than conventional masks. The resolution benefits will be important for VR/AR and as mobile phones continue to go higher in resolution. However, today’s evaporation systems use horizontal masks for mobile applications and it will require an equipment maker to develop a new system. There are rumors of certain companies working on this, and given the mask availability, it is even more likely now. V-Technology has also developed a partial laser annealing silicon (PLAS) system which offers oxide-like mobility’s of around 30cm2/Vs, but should be lower in cost and higher in yield than oxide TFTs. Multiple lasers are exposed through a mask over the targeted regions, selectively annealing the channel region giving better performance. The improved performance can be achieved without changing the TFT process flow, does not require an etch stop process like most oxide processes, keeps the mask count at 4-5 and does not have issues with photo stability like current oxide processes. By changing the number of laser shots or exposures, it can control the energy density and electron mobility. It can also be used in LTPS TFTs, which must often adopt more masks through a top gate TFT design with a lightly doped drain to reduce off leakage current. If this method is employed, V-Technology claims it can minimize off leakage concerns and enable bottom gate LTPS TFT devices with fewer masks. Bloomberg pointed out that a US hedge fund, Renaissance Technologies, took a $28M stake in the company and V-Tech now has a market cap of $690M. If either the Fine Hybrid Masks or Partial Laser Annealing Silicon system gains acceptance, its market cap is likely to continue to surge producing an excellent return for Renaissance and other investors.
Also enjoying rapid growth was Korea’s TES. TES saw its revenues bottom out in Q4’15 at just $8M. However, it has rebounded since, averaging $35M per quarter in 2016. It produces an atomic layer deposition tool for OLED thin film encapsulation, but the majority of its revenues are from its semiconductor deposition and etch tools.
AP Systems and Coherent are big beneficiaries of the rapid growth in LTPS TFTs and OLEDs and its stock prices are up 130% and 107% respectively. Coherent produces the laser and optical subsystems found in excimer laser annealing (ELA) systems produced by AP Systems and others that are used to convert a-Si backplanes to LTPS backplanes for use in LTPS LCDs and LTPS OLEDs. LTPS backplanes dominate the smartphone market and are moving to tablets and notebooks as well. With LTPS capacity rapidly growing as panel suppliers replace their LCD capacity with OLEDs, these two companies are major beneficiaries. In addition, for flexible OLEDs, a laser lift off step is required and these two companies have developed products for that step as well. AP Systems also produces glass encapsulation equipment for OLEDs and one drop fill equipment for LCDs. In addition, it produces rapid thermal process and bump sputter packaging equipment for semiconductor fabs. Its revenues are only up 3% over the past 4 quarters vs. the previous 4 quarters, but investors are expecting to see rapid growth in revenues and profits in 2017 on ELA installations in Korea. Coherent has enjoyed 7% growth in revenues over the past 4 quarters to $857M vs. the previous 4 quarters of $802M, but its backlog has more than doubled over this period to $890M indicative of strong revenue growth to come. Its FPD backlog alone is up 282% to $382M. Coherent supplies lasers to numerous different verticals with microelectronics only accounting for 58% of revenues.
Applied Materials also beat the average, up 70%. It is seeing strong growth in both semiconductors, where it is the leading equipment supplier, and in displays. In its fiscal Q4’16 ending on October 30th, its semiconductor revenues were up 42% Y/Y while its display revenues were up 92% Y/Y. Its backlog stood at over $4.5B with over $1.5B in displays. On the display side, AMAT is benefitting from the trend toward flexible OLEDs, which expand AMAT’s TAM due to additional CVD steps in the backplane and in the OLED frontplane for thin film encapsulation. It has also introduced a new eBeam review tool, which has gained rapid acceptance. AMAT is benefitting from strong investment in both semiconductors and displays and earned record operating income and margins in fiscal Q4’16. We discuss the outlook for its fiscal Q1’17 earnings in the latest issue of The Display Supply Chain Monitor.
The biggest loser was Canon whose stock price fell 10% in 2016 and Nikon and ULVAC whose stock prices rose just 12% and 4% respectively. These companies are more diversified and less able to reflect the rapid growth in displays on their overall growth. Canon has benefited from its subsidiary Tokki who is the leader in OLED evaporation systems. Canon has reported that its Industrial Equipment Group, which includes Tokki, is expected to see 62% growth in revenues to $1.27B in 2016. However, Canon’s FPD litho business is losing share, only growing from 34 systems in 2015 to 38 systems in 2016. Nikon’s FPD litho business, on the other hand, is expected to double from 46 units in fiscal 2016 ending in March 2016 to 92 units in fiscal 2017 ending in March 2017. Nikon’s systems have been better suited to the higher resolution (1.5um) requirements for smaller substrate LTPS systems at 5.5G and 6G. In fact, Nikon’s 6G and smaller tool shipments from Q1’16 – Q3’16 are outpacing its 7G+ shipments by nearly a 6:1 margin. However, Canon has recently developed a new higher resolution lithography system for smaller substrates and is rumored to have won significant business with Samsung due to bundling the litho tools with the Tokki evaporation systems. With Tokki having limited capacity for its evaporation tools, Canon has a lot of leverage which should result in more exposure tool sales to Samsung. So, 2017 should be a very good year for Canon’s FPD equipment business. However, it still does not have a 10.5G litho system and will be missing a big opportunity if it doesn’t quickly come to market with a 10.5G litho tool. Its bigger problem is that office products and imaging systems account for 85% of its revenues and declined 20% and 18% in yen Y/Y last quarter. It recently reduced its 2016 outlook by 4.5%, implying a 12% Y/Y decline. Nikon is also having some challenges despite rapid growth in lithography. It recently lowered its revenue outlook from 820B to 800B yen. It also lowered its net income forecast by 6% or 12B yen to 18.2B yen. FPD litho equipment are now expected to account for 92% of its operating income despite account for less than 25% of its revenues. Like Canon, it is suffering from weakness in digital camera sales and is also losing money in semiconductor lithography due to high R&D costs and disappointing sales of its ArF immersion scanner business. It has undergone a significant company restructuring effort to improve growth.
ULVAC only experienced 4% stock price growth in 2016. Its trailing 4-quarter revenues were up 8% over the previous period. However, it is only forecasting 1% revenue and operating income growth for the fiscal year ending June 2017, making it less interesting to investors. FPD equipment accounts for 40% - 45% of ULVAC’s revenues with semiconductor equipment, components, industrial equipment and materials accounting for the remainder. FPD equipment is its largest category and returned to Q/Q growth in $US in Q3’16 after being down in Q1’16 and Q2’16. ULVAC leads the PVD category but has a strong competitor in Applied Materials. In addition, the PVD market does not benefit to the extent of other markets from the rapid growth in LTPS and OLEDs.
With FPD equipment spending expected to generate new records in 2017 and 2018, we expect to see equipment supplier stock prices continue to outperform.
For more data and insight on all FPD equipment producers as well as publicly traded companies in the display supply chain including analysis and comparisons across all financials parameters, please see our Quarterly Display Supply Chain Financial Health Report.
Table 1: 2016 FPD Equipment Stock Price Growth
LG Display and Samsung Display reported Q4'16 earnings on Tuesday, January 24th and each company exceeded expectations.
As indicated in our Quarterly Display Supply Chain Financial Health Report, LG Display had double-digit Q/Q revenue growth to $6.8B as shown in Figure 1, exceeding Samsung Display for the first time in Q4’15 on a 16% sequential increase in ASPs more than offsetting a 1% decline in shipments on area basis. LG Display also enjoyed triple digit Q/Q earnings growth to $711M, the highest in at least 3 years. On a Y/Y basis, revenues and earnings were positive after 5 consecutive quarters of Y/Y declines.
LG Display's results were much better than guidance/consensus as prices rose faster than anticipated:
Shipments were down due to its P8 8.5G capacity conversion from a-Si to oxide backplanes for OLEDs. When asked what kind of capacity loss there is as a result of this conversion, their CFO indicated 41K LCD capacity becomes 26K OLED capacity due to the higher masks in the oxide process. The utilization of their producible capacity on an area basis remained flat at 82%.
On an area basis, ASPs rose 16% Q/Q and 2% Y/Y to $642. ASPs were the highest since Q1'15 and were up Y/Y for the first time since Q2'15. Supporting the area increase in ASPs was a jump in mobile shipments, which rose from 27% of Q3'16 revenues to 31% of Q4'16 revenues on strong iPhone demand.
LG Display also generated a record $1.2B cash from operations and their estimated free cash flow was $389M, their highest since Q2'15. Capex was $870M and over $3B for the year, split 50/50 between LCDs and OLEDs.
LG guided for panel shipments on an area basis to fall by 5% in Q1'17 due to its continued 8.5G a-Si to oxide fab conversion for OLED TV panels as well as allocating production capacity to R&D, increased demand for larger sizes and fewer days of operation in Q1. It expects blended ASPs to continue to rise in Q1'17 on panel price increases, tight supply/demand, low inventory levels and the trend toward larger panel sizes. No meaningful supply growth and demand for larger sizes will reduce panel output.
Regarding OLEDs, LG indicated they intend to strengthen its leadership in OLED TV panels by increasing volumes and focusing on picture quality, design flexibility and embedded sound systems. LG will have 60K 8.5G OLED TV panel capacity in 2H’17. It expects to ship 1.5-1.8M OLED TV panels in 2017. Yields have already hit 80% in OLED TV panels. LG indicated it took 10 years for LCDs to hit 80% yields while OLEDs achieved 80% yields in 2 years, although we believe it has been 4 years. 65”+ OLED TV panels were 30% of 2016 output, but are expected to reach 40% of 2017 shipments. It introduced Crystal Sound OLED (CSO) technology at CES. LG Display management commented that this technology can only be used in OLEDs, where an exciter on the speaker side of the OLED panel enables control of the sound on all 4 quadrants which was developed by LG Display. It solves the speaker problem for set makers, they indicated. LG also indicated it will start flexible OLED production on its 6G line in Q3’17.
Samsung Display's revenues rose 5% in Korean Won, but only 2% in $US due to the impact of the stronger dollar. Revenues were up 14% Y/Y in $US, the first Y/Y increase since Q3'15. We believe LCDs drove the increase, up 9% Q/Q, with OLEDs down 4% Q/Q on modest unit growth offset by shift in product mix towards smaller, rigid panels resulting from the Galaxy Note 7 recall. LCD results benefited from higher prices due to tight supply/demand conditions and a focus on high-end products such as UHD and larger screens. However, TV panel shipments were down high single digits Q/Q and down mid-single digits on an area basis as they closed their L7-1 line optimized for 40" and 46" panels. Operating income rose 27% Q.Q and 352% Y/Y to $1.2B, the highest in at least 3 years. Operating margins rose from 14% to 18%, well ahead of LG at 11%. Operating income was also driven by LCDs, up 483% Q/Q, but down Y/Y. As shown in the Figure below, LCD operating margins jumped from 3% to 15%. OLEDs continued to account for a majority of Samsung's operating income, however, accounting for a 63% share in Q4'16. OLED operating income was down 13%, likely impacted by the Galaxy Note 7 recall potentially leading to some cost write-offs as well as a shift in product mix towards smaller, rigid panels.
For Q1'17, Samsung guided to focusing on higher earnings in OLEDs through more shipments of new products and effectively managing their product mix. In LCDs, they indicated supply/demand will remain tight despite seasonal weakness due to reductions in supply. Samsung will continue to focus their LCD business on expanding sales of high value added products (UHD, larger panels, curved panels) and improving cost competitiveness.
All of this and other data for over 60 companies in the display supply chain is available in Excel Pivot Tables and Powerpoint slides within 48 hours after earnings results. For more information, see our Quarterly Display Supply Chain Financial Health Report.
Figure 1: LG Display and Samsung Display Revenues and Growth
Figure 2: LG Display and Samsung Display LCD and OLED Operating Margins
In the latest issue of DSCC's Weekly Display Supply Chain Monitor, we discussed the outlook for 10.5G fabs. Here is an excerpt of that article dealing with the constraint of lithography systems which is expected to slow down the 10.5G ramps for the late comers.
With BOE, CSOT, Foxconn/Sharp, HKC and LG Display all having announced or expected to build 10.5G fabs, there is a tremendous amount of 10.5G capacity expected to be installed from mid-2017 to mid-2021. As shown in the table below, it could amount to as much as 600K 10.5G substrate per month input capacity. This works out to demand for 76-84 10.5G litho or exposure tools. However, there is only 1 litho tool supplier, Nikon, selling 10.5G systems as Canon has so far opted not to produce their tools at 10.5G as they anticipated a small market after losing the Sharp 10G business years ago. Due to the multiple, large projection lenses used in the Nikon 10.5G system and the precise lens matching and uniformity requirements, Nikon is only able to produce one of these systems per month. Thus, Nikon will only be able to produce approximately 48 systems over this 4-year period, with priority going to the companies who already submitted POs to the company. BOE has reportedly issued POs for 13 units with CSOT booking 8 units through the end of 2018. Nikon’s capacity constraints will limit how much and how fast Foxconn, HKC and LG can add capacity. In addition, it may also delay implementation of oxide and OLED capacity as oxide requires more litho tools due to more masks.
10.5G Fab Timing and Capacity
It is likely that Nikon will eventually invest to boost capacity to 1.5 – 2.0 units per month and Canon may be encouraged to scale its tool to 10.5G. However, scaling up its tool from 8.7G to 10.5G can take at least 12 months if not longer and invites additional risk. Meanwhile, we hear Canon is pushing its 8.7G tool which has been scaled up to 2300 x 2700mm. A number of companies are considering converting their 8.5G capacity to 8.7G or building a new 8.7G fab. With Sharp de-committing on 65” and other panels to Samsung, it is expected that LG will or has already signed an agreement with Samsung to build panels. We believe it will start shipping 43" and larger to Samsung as early as June. It will also help LG justify building its 10.5G fab sooner than later. In addition, the Sharp decision is also expected to cause Samsung Display to delay the termination of another of its older lines in order to secure production of 75" panels.
10.5G litho tools are not the only tools expected to be constrained in 2017. At the China OLED Summit on January 12-13, it was said that ELA lead times had been extended to 20-24 months and Tokki’s evaporation lead times were 16-18 months.
The display industry, and the electronics device industry which it supplies, has established an efficient global supply chain that has generated increasingly powerful and beautiful products at ever-lower prices. This global supply approach, though, faces its greatest political challenge with the presidency of Donald Trump. The new president, during his election campaign and since the election, has repeatedly called for imposing a 35% import tariff on goods from Mexico, and as much as a 45% tariff on goods from China. Since 95% of TV imports into the US come from those two countries, such a move, if implemented, would profoundly disrupt the industry, Even before any additional tariffs are implemented, there are a few companies working to implement some US manufacturing.
America’s biggest retailer Walmart has driven some of this push, as part of a general shift across all industries. Decades ago, Walmart contributed to the growth of China’s assembly industry by shifting sourcing of electronics and other goods from US suppliers to China importers. In a case where I personally participated in hearings, Walmart was found guilty of importing dumped CRT TVs from Chinese suppliers in the waning days of the tube industry. As labor costs in China increased, though, and perhaps recognizing the shifting political landscape, Walmart started in 2013 to push for increased US manufacturing, saying it wants to spend $250 billion on US-made products by 2023.
Today Walmart sells one of the few electronics products assembled in the US, from Winnsboro, South Carolina-based Element Electronics. Element offers a line of TVs in sizes from 19” to 65”, selling only at Walmart, and an important part of their value proposition is that the product is assembled in the USA.
During and after the CES show, I spoke with a former Philips colleague who now works for a contract manufacturer similar to Element. In fact, Lotus International, in Canton, MI, assembled Element TVs several years ago under contract, before Element set up their own facility in South Carolina. This discussion gave some insight on the economics of a decision to shift to US assembly.
My colleague indicated that other retailers are looking to add a US-based product to their portfolio, and a number of manufacturers who currently lack North American operations are likely to start contracting with Lotus this year. It’s worth noting that these products don’t claim to be “made” in the USA, instead they use the term “assembled”. Only the final assembly occurs in the US, all the components are imported in “kits” which contain all the materials (display, boards, mechanical parts, even the packaging and user manuals) for the final product.
If everything is imported, why bother with assembly in the US? One reason is the import duty advantage. TVs imported from outside North America are subject to a 3.9% duty (Harmonized Tariff code 8528.72.64), and while TVs imported into the US from Mexico are duty-free under NAFTA, the Mexican government requires a higher value-added content than simple kit assembly to qualify as a Mexican-made TV. Of course, if NAFTA is renegotiated as promised by President Trump, all bets are off. TV “kits” consisting of all the parts of a TV but incomplete or unfinished (Harmonized Tariff code 8528.72.04) can be imported to the US duty free from any country under most-favored-nation rules.
A second reason for US assembly, according to Lotus, is savings on returns and damage. A ½% difference in return rate can more than pay for the labor cost of US assembly, and final inspection and testing on a US assembly line, coupled with shorter supply and shipping times, improves reliability.
Although an increase in US assembly is likely, a significant increase in US supply to the electronics industry is much less so. Even components like the mechanical structure of the TV set (bezel and back) benefit from the huge economies of scale of the global industry. Although there are certainly capable suppliers in the US to produce these plastic parts, the tooling costs for large extruded components make it economical to have a single global source unless volumes reach into the hundreds of thousands, which would apply only to the top 2-3 brands in the US, all of whom currently have operations in Mexico.
This week as the presidency of Donald Trump has begun we have seen new reports that Foxconn is studying the possibility of building an LCD plant in the US, in conjunction with its now-subsidiary Sharp. It’s not the first time this idea has been floated, and it won’t be the last. Given the clear statements made by President Trump during the election campaign about favoring US manufacturers, it would be wise for any company with substantial sales in US end-markets to be reviewing their manufacturing plans. Nevertheless, although there might be some advantages to making LCD in the US, there are huge (yuge!) barriers to overcome before this would become a reality.
Foxconn chairman Terry Gou was quoted on Sunday, January 22 as saying that the investment under consideration would amount to more than $7 billion. The investment could be supported by Japan’s Softbank Group Corporation, whose head Masayoshi Son is a close friend of Gou’s. Gou was quoted as saying “There is such a plan, but it is not a promise, it is a wish” that would depend on investment conditions and negotiations at both state and federal levels.
As a key part of his “America First” campaign theme, the 45th president has repeatedly called for imposing a 35% import tariff on goods from Mexico, and as much as a 45% tariff on goods from China. In response to criticism from President Donald Trump, automakers in particular have announced a wave of investments and a pull-back from production in Mexico. For example:
It’s no accident that there have been almost no investments in new LCD fabs outside of China for the last 4 years. All the investments have been in China because local governments there have been willing to partner with display companies, even those with little track record such as HKC. In China, local government investment can supply as much as ½ of the capital required for a new LCD plant, so any investment outside of China becomes unattractive unless it gets a similar level of government support.
Could such government support be secured in the US? There have been some very large government subsidies for manufacturing. For example:
Nevertheless, if we imagine that with some combination of business opportunity, government incentives, and creative financing a company would decide to invest in a new display plant in the US, then we encounter the next hurdle: the display supply chain. For a period of 25 years, display suppliers have built an infrastructure and expertise in Asia, and while importing components and materials is possible (and for certain materials like liquid crystal, importing the raw chemicals is likely the most economical solution), a local supply base is essential for making a productive fab.
Although some of the cutting edge display technology involves flexible displays on plastic substrates, the big investments in Gen 10.5 fabs still will require glass, and even plastic substrate production in smaller size is executed using a carrier glass. Up to now, every Gen 10 or Gen 10.5 fab built has had a glass plant co-located, because Gen 10.5 glass is too big to ship economically. So any big LCD investment in the USA would need a corresponding big investment in display glass: the Corning investment in Hefei, China associated with BOE’s Gen 10.5 fab amounts to $1.3 billion, with 2/3 of that money coming from government and commercial incentives.
The most likely supplier of display glass to a US fab, of course, would be Corning. Corning has an existing plant in Harrodsburg, Kentucky with several fusion draw lines. That plant made the first volume production display glass in the LCD industry, and was Corning’s only site for melting LCD glass into the 1990s. It currently manufactures Gorilla glass cover glass, and this plant has been cited by Apple in defense of claims that the iPhone does not support US jobs.
While Corning would be the most likely choice, it might not be the only alternative. Corning’s competitor Asahi Glass Company has substantial operations in automotive and architectural glass in the USA, and recognizes more than 10% of its revenues in the Americas. While the borosilicate chemistry of display glass is profoundly different than the soda lime used in automotive and architectural applications, Asahi uses the float process for forming all its glass, so process expertise would not necessarily be lacking in the US.
Glass is not the only material in a display, but most other materials in an LCD cell (or for that matter in an OLED stack) are much more easily transported. I recall a discussion with an analyst from Merck who indicated that the entire supply of LC materials for the industry for a year would fit in my office. Although this was a few years ago, and we now may be up to several offices, it doesn’t change the economics. Merck makes core chemicals mainly in Germany, but has mixing operations in LCD-manufacturing countries, a sort of final-assembly step for that product.
Beyond the capital-intensive LC cell, though, is yet another challenge to US display manufacturing. Module assembly is the most labor-intensive step, and for this reason most module assembly today, even for LC cells made in Taiwan, Korea, or Japan, occurs in China. So a US-based display maker would look to find a cheap source of labor for module assembly. Under the Trump administration, Mexico is not an option, so module assembly costs are likely to be higher. Likewise, the next step in production, the assembly into electronic devices, is often co-located with module assembly because it is also labor intensive, and therefore it also would add cost.
All this suggests that the challenges to making displays and devices in the US are daunting, making a development like a big Foxconn LCD plant in the US unlikely. Even if those challenges could be resolved, Foxconn would run into another barrier that would delay implementation. With the current queue of Gen 10.5 fabs standing at five (BOE, CSOT, LGD, Foxconn, and HKC), implementation of yet another fab would be limited by equipment vendor capacity. DSCC sources suggest that Gen 10.5 exposure equipment from Nikon is likely booked with these fabs until 2020 or even later. If you’re looking to buy a 65” TV with a US-made panel, you’re likely to need to wait until the 2nd Trump administration.
While we’re not likely to see large-scale US display manufacturing soon, the impact of President Trump’s “America First” approach will see itself in some moves on device assembly, the subject of my next article.