La Tecnologia è Mobile

Race Against Pride & Cost

For those less familiar with the acrimonious history between South Korea and Taiwan, the bitter-sweet attitude of Taiwanese tech industry towards its Korean archrival can be best understood by analyzing Samsung Electronics, the global consumer electronics powerhouse that generates over $140 billion of revenue in 2011, or roughly the total revenues of TSMC, HTC and Hon Hai Precision, the three pillars of the Taiwanese Tech Pride. 

A casual look at Samsung's financials suggests it has four major segments: memory, display, handset and TVs. Samsung is the largest and the most profitable DRAM, NAND Flash and TFT-LCD makers, the largest digital TV maker, the largest handset maker second only to Nokia, whose volume is shrinking by the minutes. A deeper look, however, tells more of a story of industrial migration of the past 20 years and how different industrial mindsets lead to different path dependency and outcome.

Take memory chips for example. Technological advancements and cost-down efforts by Taiwanese and Koreans over the past twenty years have forced the Americans, Europeans and Japanese out of the market. At each downturn, Samsung quickly increased capex, handsomely spent on R&D, and fiercely compete on price. The goal is not to make contrarian capital allocation decision, but to drive competitors out of the market with concentrated force.

The economics is simple: a 1Gb DRAM chip costs $0.5 to make, $0.4 to package, and $0.4 to finance the 50nm immersion equipments that render the production in Taiwan viable. The going price is $1 and en route to $0.6 within two years. Only Samsung and Hynix can survive on this cost curve. A more telling story is that, when DRAM price staged a short-lived rebound in 2009, the Koreans were buying advanced semiconductor production equipments to effectuate process upgrade, while Taiwanese were busy restructuring their overdrawn bank credit lines and shareholder loans to pay technology licensing fees to Americans and Japanese.

The result of this bifurcation, is to empower Samsung to control over essential patents and semiconductor manufacturing processes in DRAM, NAND Flash and SSD, thereby making a respectable 23% of EBIT margin for DRAM and NAND Flash in 2011, or about $5 billion of operating profits. On the other hand, Taiwanese DRAM makers, five of them in total, have been responsible for almost 100% of the $20 billion of global losses in memory sector since 2007. Nice little pair trade to have there.

Small & Fast beat Large & Fat

The history of display industry has a similar flavor. The first generation of large liquid crystal display was developed by NEC in 1990 and attracted many Japanese firms to follow suit. By 1995, the 3rd generation technology encouraged Samsung to enter the market and begin fostering its own supply chain. Japanese branded display makers, facing heightened costs and reduced risk appetite, ceded the thrown of global top TV brands to Samsung and LG following the Korean dumping amid Asian Financial Crisis in 1998.

Unable to tolerate Koreans to take the top seats,  Taiwanese came in with government-backed funding again, transferring Japanese IP, ramping-up capex on large-size display, thinking it might be able to ride the coat tail of Sony, Sharp and Toshiba and be the shadow manufacturing partner of global consumer brands.

Unlike Samsung and LG, none of the Taiwanese display firms can control upstream raw materials and core technology, nor do they have the political freedom to penetrate Mainland Chinese market and build profitable brand equity. In display, as in memory, or in any other hi-tech industry, speed to invest in efficiency upgrade is paramount: if you cannot keep up with others, you are bound to lose. In 2011, Samsung lost only $500 million of EBIT in display, thanks to its small and midsize panel growth initiative and Galaxy product lines, while the five Taiwanese display makers, loaded with debt, lost nearly $3 billion. This stark contrast has made many Taiwanese tech bosses publicly ask government to come up with a better, comprehensive policy package (or backstop facility) to sustain the Taiwanese competitiveness, while taking orders from Samsung and enjoy good rally of their stock prices.

A perfect example is the Taiwanese LED industry. Backtesting these LED stock price performances to the announcement of receiving Samsung order suggests strong correlation and has been a well-known trade to make a killing. The real killer, however, remains firmly in Korean hands. It has been the history of Samsung to ping competitors' strength and weakness by giving them orders with demanding product specifications. For example, Samsung would give large LED orders to Taiwanese makers on one hand, and buy or build advanced MOCVD, the core LED production technology, on the other. Once Taiwanese firms become dependent, they would serve as convenient virtual fabs for Samsung to optimize its manufacturing efficiency.

Miniaturization & Militarization

To say that Samsung is benefiting from global brand and currency manipulation to beat opponents is a gross misunderstanding of its competitive mindset. The common feature of this ongoing annihilation war against Taiwanese tech industry is Samsung's will and ability to mount capex offensive through economic and product cycles to drive favorable changes of consumer behavior. More importantly, Samsung has demonstrated its ability to establish dominance in both upstream and downstream parts of the supply chain: its memory, panel and battery products account for over 40% of key electronics component costs globally, its $20 billion worth of brand is no. 1 in China. With digital convergence continues, Samsung can leverage, if not militarize, its global design, manufacturing, and logistic resources to rapidly introduce affordable and fashionable consumer gadgets while beating Taiwanese tech firms left and right. The key question then becomes: How long can TSMC sustain? 

A recent news article in Taipei may point to a possible crack in TSMC seemingly impregnable position. It is reported that last November, a 18-inch wafer foundry was being built in SUNY Albany, New York and top engineers from every major global semiconductor firms were present to test their equipment prototypes. The purpose of this joint project is not only to test the next-generation technology for 10nm geometry, 3nm  away from the physical limit of miniaturization, but also to size up competitors' readiness. TSMC, the global leader in wafer making, approached this project as an industrial espionage battleground, for a very logical reason. After dominating memory that are easier to make, it is only a matter of time before Samsung takes on logic chips and challenge TSMC head-on. The most advanced 18" foundry could cost over $5 billion to build and requires over 120 different types of equipments and over 900 manufacturing processes that requires absolute precision. If Samsung could exert influence over several processes, it could sabotage TSMC's cost structure. After losing several key executives to Samsung in recent years, TSMC fully understood the gravity of the situation.

When the testing results were announced, TSMC was shocked: of all the qualified equipments for the follow-up 18" foundry R&D, one key equipment for making mobile chipset was of Korean origin. This piece of Korean technology blocks a crucial bottleneck TSMC cannot bypass. This not only implies Samsung's continuing investments in South Korean semiconductor equipment makers have cultivated several firms that can compete against Intel, but also spells a dim future for TSMC: what if Samsung announced that all mobile device orders must be made with Korean equipment? A global shipment volume of 300 million units and growing can buyout loyalty and patriotism much more easily than you think.

In the future, there are probably only a handful of semiconductor equipment makers who can afford to invest in the 18" geometry. With Moore's Law approaching its theoretical limit, the equipment cost might eat up scale benefits of large wafer size. More importantly, such a future requires a device maker to produce antenna, micro machinery and analog chips more efficiently. If Samsung can break into 18" equipment making, it can quickly build a more sustainable integrated supply chain and sell retrofitted legacy equipments to other competitors who are late to the R&D game. Intel also reportedly has the ability to fit the most advanced process technology to old equipments. The cost saving can be enormous.

TSMC, because of its traditional insistence to focus on core business, never ventured into upstream equipment R&D seriously. Even with experienced engineers, it would still take two to three years to design next-generation equipments. In the world of semiconductor arms race, such speed gap almost spells defeat. No wonder TSMC immediately sent engineers to every major equipment suppliers to develop bespoke technology for TSMC. Samsung, on the other hand, announced record high wafer capex of US$7 billion in 2012. If Samsung spending continues, it is possible that TSMC, without compromising its NT$3 cash dividend per share to shareholders, will access debt capital market for ammunition. How wide should a TSMC credit trade? Would that event sound the overture of yet another DRAM-style Götterdämmerung ? Only time would tell.

Will HTC be the next?

The origin of HTC's great rally which peaked in April 2011 can be traced back to a eureka moment on a business trip: Cher Wang, Founder and Chairwoman of HTC, feeling tired ot packing her bulky laptop and cellphone en route to airport, once lamented how wonderful it would be if she could browse the web, check emails and do conference call in a single, light-weight device. That lament triggered HTC to focus on hand-held computer and smartphone in 1997, and became the largest OEM of PDA by 2002. Knowing the future of wafer-thin margin of a contract manufacturer too well, in 2006 HTC shifted to branding and innovation in mobile devices, thus embarking on a path to challenge its customers head-on, a path seldom chosen by fellow Taiwanese technology firms. Again, looking at the HTC share price versus Acer, Asustek, Quanta etc., the contrast cannot be starker.

HTC is arguably the only brand that can mount a credible threat to Samsung in its ongoing battle against Apple and Nokia. However, it still has a fatal weakness versus Samsung: HTC cannot make TV. In the era of digital convergence, the power of TV as a distribution channel for digital contents cannot be underestimated. Just imagine a world where you can voice-control your TV from your smartphone to play iTune contents while playing AngryBirds. Samsung is the only company in the world that can make money in memory, display, handset and TV simultaneously. Without a strong platform supported by patents and branding, HTC's only remaining leverage is its Mainland China business, where it only shipped less than 10 million handsets in 2010.

Given the size of Chinese market, such volume clearly suggests growth ahead. It is the intention of Beijing to foster its own telecommunication standards and leverage its market access to dictate the terms to foreign firms. It is also politically expedient to treat HTC as a native Chinese firm with global ambition and design a new set of rules to facilitate its Mainland growth. Given Cher Wang's all-out support for the pro-unification ticket in the 2012 Taiwanese Presidential Election, this publication smells value in HTC's depressed valuation. When will Beijing come to the rescue? We shall see.