Global Strategic Management/Final Exam

Questions

1. This question concerns Samsung Electronics’ semiconductor business. Compare

Samsung Electronics’ DRAM segment, including a value stick analysis, with the
corresponding segments of Hynix, Infineon, Micron, and SMIC. Does Samsung
Electronics have a competitive advantage in DRAM vis-à-vis these competitors? What
are the specific nature and sources of any competitive advantage or disadvantage? (40%)

2. This question concerns Samsung Electronics’ Central Asian fab (see page 11 and related
exhibits). (20%)

a. As a base case, determine the likely outcome from competition between the
Samsung Electronics and Hynix fabs. (5%)

b. How would the following changes to the scenario described in the case likely
alter or not alter your answer to question a., considering each change separately?
(You are not expected to create a new matrix for any of the following questions,
but considering how the matrix might change may or may not be helpful.) (10%)

i. The local electronics manufacturers agree a common product standard,

imposing minimum performance requirements and setting limits on the
design architectures of DRAM chips in the Central Asian market. (5%)

ii. Rather than being owned by Hynix, the other competing fab is owned by
Qinghai Electronics Industries, a Chinese semiconductor firm that has
licensed technology from Infineon to make older generation DRAM chips
that run less hot and use less power in calculators but, due to their
packaging, are awkward to install in electronic toys and medical devices.
Infineon is a minority investor in Qinghai. (5%)

c. Assess the prospects for the Samsung Electronics and Hynix fabs to do better
than the base case you calculated in a. (5%)

3. Evaluate Samsung Electronics’ corporate strategy (i.e., its business and geographic
scope). Please address the following in particular: (40%)

a. Samsung Electronics’ overall corporate strategy (i.e., its business segments) (15%)

b. The potential sale/divestiture of the Digital Appliances business (15%)

c. Whether and how Samsung Electronics is creating and/or capturing value by

operating internationally (10%)

1
Samsung Electronics & Its Semiconductor Memory Business
Samsung Group was founded in 1938 as a trading company by Byung Chull Lee. Over the
years, the group diversified into food processing, textiles, insurance, securities, and retail,
entering the electronics industry in the late 1960s and construction and shipbuilding in the mid-
1970s. In 1987, Samsung Group was separated into five business groups, which retained
significant cross-ownership and family ties: Shinsegae (retail), CJ (food, entertainment), Hansol
(paper, chemicals), JoongAng (printed media, hospitality), and Samsung Group, which
contained the remaining businesses. In 2004, Samsung Group was the largest family-controlled
business group in South Korea, with total sales of $135 billion, 337 overseas operations in 58
countries, and approximately 212,000 people employed worldwide. According to Interbrand,
“Samsung,” the brand, ranked 21st in the world in 2004 (ahead of Philips, Kodak, Sony, and
Panasonic) and was worth $12.6 billion, largely because of Samsung Electronics.

Samsung Electronics

Samsung Group’s flagship firm was Samsung Electronics (hereafter, just “Samsung”). Samsung
was established in 1969 to manufacture black-and-white TV sets. By 2003, the company had
over $50 billion in sales, $50 billion in assets, and 100,000 employees. Samsung alone was
responsible for 22% of Korea’s exports in 2004 and 23% of total market value on the Korea Stock
Exchange. Samsung’s chairman was Kun Hee Lee, the third son of Samsung Group’s founder.
The organizational structure of Samsung’s five operating divisions is shown in Exhibit 1;
divisional financial results are presented in Exhibit 2; and global market share numbers for
select products (along with those of leading competitors) are in Exhibit 3.

Digital media was the home of color TVs, audio and video equipment, and computers; this
business built the Samsung brand with global consumers. Starting as a contract producer, by
2000, Samsung had become a recognized supplier of high-quality, if not exciting, devices.
Because Samsung was the leading innovator in LCD panels (see below), its media and
telecommunications products enjoyed cutting-edge technology in this field. By 2004, it was
winning awards for designs and technologies related to LCD, plasma, DLP projection, and CRT.
Samsung led in global market share in TVs and DVD/VCRs (ahead of Philips and Sony) and
computer monitors and was second to Hewlett Packard in monochrome laser printers.

Telecommunications was a leading maker of cell phones. In 2003, Samsung pioneered the
antenna-less clamshell cell phone. More recently, it offered such innovations as a phone with a
screen that swiveled to a landscape view. More than 16 million of its BlueBlack slider phones,
introduced in 2004, would be sold by the end of 2005. Making use of Samsung’s in-house
expertise in displays and memory chips, the telecommunications business accelerated its
research and design efforts, trying to shorten the lifecycle of its products to six months. Aware
of the environmental implications, Samsung led in the establishment of recycling centers for cell
phones in Korea; the company also developed an environmentally friendly paint for cell
phones, for which it won an iF Material Award in 2004.

Digital appliances was slower moving than cell phones because of the longer life of products
like refrigerators, air conditioners, washing machines, microwave ovens, and vacuum cleaners.
Samsung, along with LG, were key players in Korea; Whirlpool (US) was its principal
competitor in many other markets (e.g., USA), with Haier (China) creeping up in the rankings.

2
“Home appliance products are rooted in the lifestyles of each country,” explained Jeongmin
Kim, a manager in the digital appliances design group.” These appliances (frequently called
“white goods”) were also big and heavy, raising shipping costs; this suggested a need for either
decentralized manufacturing or premium positioning. Technically advanced and designed with
environmentally safe materials and energy efficiency, Samsung’s high-end refrigerators were
available across the globe in striking colors, such as cranberry, cobalt blue, black, and bronze,
which could be changed through an inexpensive replaceable panel system for individual
markets. Even so, Samsung was not a true luxury appliance brand like US-based Subzero.
Samsung did lead the market in side-by-side refrigerators in 32 countries, but its market share
for even this digital appliance product—its most successful appliance globally—lagged behind
other leading Samsung products in important markets like the USA; see Exhibit 4.

LCDs, the thin film transistor liquid crystal display business (TFT-LCD), launched in 1995, was
emblematic of Samsung’s success in aggressively focusing on digital, rather than analog,
technology. Samsung’s LCD technology went directly into the company’s own computer
monitors, notebook PCs, LCD TVs, and mobile phones, as well as to outside customers. By 2004,
Samsung’s reputation in LCD displays was such that Sony approached them to form a 50-50
joint venture to make seventh-generation LCD panels for TVs.

Semiconductors made its first memory chip in 1977, and by 1992, less than two decades later, it
was the world’s leading manufacturer. In 2004, it led in market share by a large margin in
memory chips like DRAM, SRAM, and Flash. The semiconductor business is discussed
extensively below.

Research & Development

As early as 1987, Samsung had determined that basic research was critical to its competitive
capability. R&D investment grew from $1.81 billion in 2001 to $5.34 billion in 2005—9.4 percent
of sales. In 2005 alone, Samsung registered 1,641 US patents, fifth most that year. With some
32,000 researchers (25 percent of Samsung’s workforce) in 16 research centers (6 in Korea),
Samsung had one of the largest R&D organizations in the world. The foreign locations allowed
Samsung to tap local knowledge bases, for example, in Silicon Valley, and attract engineers who
did not want to move to Korea. Even so, the heart of Samsung’s R&D was the Samsung
Advanced Institute of Technology, SAIT, opened in Giheung, near Suwon, South Korea in 1987.
With a mission to “make the world better,” it supported both current core businesses and future
possibilities. It sought to “bolster the synergy of the various Samsung units” and sponsored
specialized initiatives, such as an eco-group and an energy group. For instance, R&D and
production engineers in Samsung’s semiconductor business lived together in company-
provided housing. They shared meals and their worksites were near one another’s so that the
engineers could quickly solve design and process engineering problems together.

Manufacturing

Samsung operated 27 manufacturing plants in 13 countries. China was an interesting case

study: Samsung had opened or was opening several facilities there to manufacture LCD panels,
cell phones, home appliances, and other parts (e.g., cell phone cases; and semiconductor
packaging, which had a more labor-intensive process than making the chip itself—see below).
Most of Samsung’s competitors set up manufacturing in China to take advantage of its skilled,
low-cost labor and accordingly exported most of their output. Although some of Samsung’s

3
more labor-intensive work was directed at a global market, too, most of Samsung’s activities in
China were intended for the Chinese domestic market.

In general, Samsung’s manufacturing was characterized by flexibility. Said one Samsung

manager: “We can change our products any time we want. We have a human-oriented
production system. We get more productivity that way, and we can adjust inventories faster.
We don’t know which products consumers will buy, so we have to respond quickly.” Scale was
also important. Large, very expensive plants, especially for semiconductors, but also for LCD
panels and somewhat for televisions, were more efficient than smaller plants if and only if
Samsung could justify the production volume by selling across the globe.

Samsung prided itself on the reliability of its products. During the 1980s and early 1990s,
Chairman Lee had seen that his company was producing shoddy products. In a 1994 book
delivered to all employees, he explained that the Samsung Group had lost track of quality
because it had begun 55 years earlier selling commodities like sugar and textiles. Lee argued
that employees must now think of quality first. He launched mass burnings of tens of millions
of dollars’ worth of flawed products in an open field and admonished his employees about
quality control. Prior to 1995, the company had won one major industry award. Between 1995
and 2003, the company won awards for reliability and performance from most major customers.
Many customers, even rivals of one another, named Samsung their supplier of choice. For
instance, the company simultaneously developed a new Flash memory chip for Sony Ericsson
and a Flash memory chip customized for Nokia. Even so, some companies like cell phone giant
Motorola, expressed a reluctance to buy Samsung LCD screens or semiconductors, given that
Samsung was a rival of Motorola in cell phones.

Marketing

As an engineering-driven company for more than two decades, Samsung had not given priority
to either marketing or design. They were add-ons—the company’s engineers had assumed that
if their products were good enough, they would sell. In fact, most marketers had been
engineers. Perceptions began to change quickly with the First Design Revolution of 1996, as
Chairman Lee recognized the importance of developing a global brand strategy. First, Samsung
would focus on promoting a single brand for the firm, with emphasis on mobile phones and
digital TVs; second, marketing communications would be unified globally; third, the global
brand campaign would focus initially on already developed markets; and finally, Samsung
would emphasize sports marketing. In 1998, for example, Samsung was a major sponsor of the
Olympic Games, resulting in high exposure for Samsung mobile phones.

Under the campaign Samsung DigitAll, Everyone’s Invited, Samsung redefined itself as an
inclusive provider of digital products. In 2000, Samsung consolidated its advertising to execute
holistic marketing campaigns with a consistent look and feel across traditional media, outdoor
advertisements, point-of-purchase, and exhibitions. The company also promoted name
recognition beyond sports arenas; for a time, people could not walk anywhere in major Central
and Eastern European cities, such as Prague and Moscow, without seeing blue and white
Samsung banners fluttering from light poles and Samsung ads covering buses and trams. By
2003, Samsung’s marketing department had greatly expanded and was no longer dominated by
engineers but by people with marketing and other business experience.

Design

4
At roughly the same time as these branding initiatives, Samsung launched the first of several
design education initiatives. From 1993, a Design Membership program put 50 top Korean
university students each year through one- to three-year design training programs. At first, only
about half the graduates entered Samsung; by 2003, 80 percent joined the company. In 1995,
Samsung founded the Samsung Art and Design Institute (SADI) to introduce Western-style
undergraduate design education, focusing on fashion and communication design. After two
years in Seoul, SADI students spent two years at partner colleges abroad, such as Parsons,
Carnegie Mellon, or other American and British schools.

Since its founding in 2001, Samsung’s Corporate Design Center (CDC) had been the incubator
for numerous winners in the competitive markets of the company’s three consumer businesses.
Located in downtown Seoul, in a historic red marble skyscraper named after the city’s major
daily newspaper, the Joong Ang Ilbo, the CDC employed 600 people. The company’s 62 recent
design prizes included 3 from the US (IDEA), 35 from Germany (iF and Red Dot), 20 from Japan
(G-Mark), and 4 from China (Design for Asia Award, iF China). It ranked first (with 19 prizes)
among IDEA (Industrial Design Excellence Award) winners in the early 2000s. (Apple was
second with 17.) Kook-Hyun Chung, head of design, said: “The primary advantage of having
design together is that the designers can learn skills and ideas from each other.”

Young Jun Kim, vice president of Samsung’s Design Research Lab, elaborated, “Our marketing
research is focused on the current situation—market share, and so forth. Design research is
focused on user behavior and user experience.” Samsung designers visited people in their
homes to see how they used products. Samsung’s User-Centered Design Laboratory simulated
familiar living environments in which users tried Samsung products while being observed by
Samsung researchers. This led to user-friendly forms and features; in 2003, Samsung was
commended by the International Ergonomics Association for innovations in the workplace.

Design recognized the importance of international input. Samsung also had design centers in
Tokyo, San Francisco, London, Los Angeles, Shanghai, and Milan. These six design centers were
Samsung’s windows on global markets. Said Chung, “We use design centers overseas to learn
about lifestyle trends. The overseas centers also design their concepts for their own markets.”

Human Resources

In the past, Korean companies often hired employees because they came from the right high
school or region, but Samsung eliminated this practice. It was taboo at Samsung to ask a
coworker about their university or place of origin. Prospective employees were given an
aptitude test covering language, mathematics, reasoning, and space perception. Samsung also
broke with traditional seniority-based promotion, which was still widespread in Asia.
Employees were given evaluations on an A, B, C, and D scale every year, and only those who
earned two A’s within three years were eligible for promotion. As a result, younger, high-
potential, English-speaking managers were quickly promoted up the hierarchy. Among the
company’s senior-most executives, several had attained their current positions in their early 40s.

Samsung invested in employees’ business skills. The Regional Specialist Program placed high-
potential employees in a foreign country to learn the local language and culture for one year.
Upon their return, the Regional Specialists produced reports on their experiences that became
part of a codified knowledge database. Samsung sponsored hundreds of employees’ MBAs and
PhDs abroad.

5
Unlike most Korean companies, Samsung actively recruited foreign talent, including westerners
and members of the Korean diaspora who had long ago left Korea to live and work in the US
and Europe. These included Korean Americans like Chang Gyu Hwang, president of Samsung’s
Semiconductor Business; Oh Hyun Kwon, head of Samsung’s System LSI Division; Gregory
Lee, head of marketing, who had worked at Kellogg, Johnson & Johnson, and Procter &
Gamble; and Dae Je Chin, the previous president of Samsung’s Digital Media Business. In the
early 1990s, Chairman Lee created Samsung’s Global Strategy Group, primarily staffed by non-
Koreans, which served as an in-house strategic planning group. David Steel, the highest-
ranking Western manager at Samsung, joined the Global Strategy Group in 1997 and later
became Vice President of Business Development in the Digital Media Business.

Samsung claimed to invest more in its employees than competitors. When Lee became Samsung
Group chairman in 1987, he stated: “What do our salaried workers worry about as soon as they
open the door from their house to work? Probably over 90% will think about their family and
their own health, their children’s education, and their retirement.” Lee proposed that the
company would take care of 90% of their burden, allowing them to concentrate on innovation
and productivity. He also declared that the company would richly reward individuals for their
accomplishments, while not firing people for failure: “Take the example of a horse trainer: The
best ones never use a stick or whip, only carrots for reward. At Samsung, we reward
outstanding performance; we do not punish failure. This is my personal philosophy and belief.
We need punishment only for those who lack ethics, are unfair, tell lies, hold others back or
stand in the way of our unified march.”

The average salary at Samsung was $44,000, versus comparable figures for competing
semiconductor firms Micron, Infineon, Hynix, and SMIC of $54,000, $72,400, $24,600, and
$10,800, respectively. Beyond base salaries, Project Incentives rewarded project members with
cash bonuses at the end of a successful project and ranged from a few thousand dollars to more
than $1 million for a project team. Productivity Incentives rewarded employees for performance
at the division level but could be modified at the department or team level on the basis of its
performance and contribution. Productivity Incentives paid up to 300% of annual base salary. A
Profit Sharing program rewarded each member of a division, paying up to 50% of annual base
salary depending on divisional performance as measured by economic value added (EVA).

That said, Samsung had a well-deserved reputation for hard work. To wit, in the mid-1980s
Samsung built its first large fab (semiconductor manufacturing plant). Building fabs was
difficult and time consuming because the production-related machinery was highly sensitive to
dust and electronic shock. At the time, the construction period for a new fab was normally 18
months, but Samsung did it in just six. Construction crews worked shifts covering all 168 hours
of the week in a harsh Korean winter, even completing a four-kilometer-long road in just a
single day. Engineers said their weekly schedule consisted of Monday-Tuesday-Wednesday-
Thursday-Friday-Friday-Friday. This culture permeated the firm.

The Semiconductor Memory Industry and Samsung’s Memory Business

Memory Industry

In 2000, the semiconductor industry enjoyed $200 billion in sales and had grown by an average
of 16% per year since 1960. Semiconductor chips were classified into two types: Logic chips
were used for processing and control, and memory chips stored information. Intel dominated

6
the market for logic chips because of its control of the personal computer central processing unit
(CPU) market. Logic chips were generally higher value-added than memory, but lower-end
logic chips were used in devices like calculators and toys. Memory chips were further classified
into DRAM (Dynamic Random Access Memory), SRAM (Static RAM), and Flash.

Memory chips accounted for $33.7 billion in sales worldwide in 2003, a little over half in DRAM.
Historically, DRAMs were used mainly in PCs, but the share of DRAMs going to PCs declined
from 80% to 67% between 1990 and 2003. Telecommunications and consumer electronics were
growing consumers of DRAMs. TVs, set-top boxes, and game devices such as Playstation
represented 7% of the market. Products ranged from so-called larger chips, i.e., “frontier
products” (e.g., 512Mbit DRAM) at the cutting edge of technology to smaller “legacy products”
(e.g., 64Mbit DRAM) that makers of less advanced electronics products continued to use.
Within each product generation, there also existed “specialty products” (e.g., DDR2 SDRAM,
Rambus DRAM) using customized architectures for niche markets. Prices for large frontier
chips were higher than for chips in the “sweet spot” of the market (at the time of the case,
256Mbit DRAM), but stayed high for only a couple of years, before plunging rapidly as the
market moved on. Even so, legacy product lines with smaller sizes could be transformed into
profitable niche chips for products like calculators, toys, or watches. See Exhibit 5 for market
prices of DRAM chips of different types and sizes.

SRAM and Flash memory accounted for 10% and 32% of industry sales, respectively. SRAM
was a type of buffer memory that facilitated computer processing and mobile phone
functionality. Flash memory, which was the hot-growth area, was used heavily in digital
cameras and mobile phones. Flash memory could store data when a device was turned off,
unlike DRAMs, which lost their data.

The memory industry contained powerful suppliers and price-conscious customers. With each
generation of semiconductor equipment, the technology grew more complex and the number of
suppliers became more concentrated. Only two or three main players—Applied Materials,
Tokyo Electron, and ASML—dominated key segments of the equipment market. Suppliers of
memory raw materials to memory chip makers provided discounts of up to 5% for high-
volume. Customers were far more fragmented, with no single chip user controlling more than
20% of the global PC market. Memory represented 4%–12% of material costs for a PC maker,
and 4%–7% of material costs for a mobile phone producer. Because rivalry between PC
producers was intense and they had to face price-conscious consumers, PC makers negotiated
hard on price. However, defective memory could destroy a computer’s value, so PC makers
would pay upwards of a 1% price premium to a reliable supplier.

Semiconductor Production Process

After designers had made a blueprint of a semiconductor based on its intended function, the
physical shape of the chip was transferred to a mask used to create identical chips. Separately, a
cylindrical silicon ingot was shaped to the desired diameter (in early 2000s, up to 12 inches),
and the ingot was further cut into wafers that were extremely thin (250–350 microns, thinner
than a human hair). Next, companies like Samsung took the wafers and produced memory
chips through a series of thermal, metallurgical, and chemical processing steps. In this process,
billions of electronic circuits were defined within numerous individual chips (also called “dice”)

7
on the wafer. The result was a matrix of rectangular chips on the wafer. Finally, the wafers were
sawed into individual chips. Throughout the process, chips were tested for reliability.

One of the main goals of a memory chip producer was to generate as many individual chips in
one production step as possible while minimizing defective chips. To do so, producers made
design and process improvements that would allow more electronic circuits to fit on ever-
smaller chip sizes as well as ensure more uniformity in the manufacturing process. More
specifically, production yields depended on the number of good chips that could be cut out of a
wafer, which in turn hinged on the size of the wafer and the precision (fineness) of the design
rule used to etch the circuits on the wafer (see Exhibit 6). Over time, wafers had gotten larger,
and design rules had shrunk. Although any firm with sufficient capital could in principle buy
the latest manufacturing equipment, mastering its use—i.e., keeping the defect rate down—
varied considerably among memory chip makers; and even the best firms faced a steep learning
curve when using the latest equipment. Insiders likened this aspect of the industry to playing
classical music: Anyone with enough money could buy a Stradivarius violin, but the ability to
make beautiful music with such an instrument varied considerably among musicians.

Major Memory Competitors

Many semiconductor manufacturers produced memory chips. Other than Samsung, the leading
firms in the early 2000s were focused firms. Exhibit 7 presents comparative financials for select
firms, including Samsung.

Elpida Memory—Japan’s only remaining DRAM producer—was established as a joint venture

between NEC and Hitachi in December 1999. In the next three years, Elpida suffered losses due
to a DRAM market decline and a decision not to invest in new products and production
capacity as the market recovered. Subsequently, Elpida decided to focus on memory products
for mobile devices and consumer electronics, so it could try to sell primarily to Japanese
customers who had, until then, bought memory chips from Samsung and Micron. In 2004,
Elpida started construction of its second 12-inch wafer fab next to its current manufacturing
facility in Hiroshima. The cost of the new fab was $4.5 billion, and Elpida partially financed it
with a $100 million investment from Intel, along with a public equity issue.

Hynix Semiconductor was founded in 1983 in South Korea as Hyundai Electronics. It changed
its name in 2001 while separating from the financially distressed Hyundai Group. In the early
1990s, Hynix enjoyed some of the same advantages as Samsung, but it lost the technological
lead. Moreover, Hynix had trouble timing its capital investments. In 1996, when the DRAM
market began experiencing a cyclical decline, Samsung maintained minimum capital
expenditures needed to smooth business operations, while Hynix dramatically increased its
capital investments into the downturn. Hynix lost more ground to Samsung in 1999 when the
market began to expand. Samsung significantly increased its investment, while Hynix reduced
its investments because of uncertainty surrounding its merger with LG Semiconductor, part of
LG Group, which the Korean government had insisted on. Hynix’s ultimate acquisition of LG
Semiconductor loaded it with enormous debt, which together with a cyclical industry downturn
forced Hynix almost to collapse in 2001–2002. A multibillion-dollar bailout allowed the
company to survive. Still, Hynix was forced to lay off 30% of its workforce and sell all non-core
operations such as mobile phones, LCD panels, and televisions. Thus, Hynix went from
resembling Samsung’s diverse business portfolio to being a much more focused firm.

8
Infineon Technologies, a German firm, was spun off from Siemens in 1999. Siemens had been
in the semiconductor industry since its beginnings. Throughout its history, Siemens’
semiconductor unit formed alliances with competitors to reduce investment risk and shorten
time-to-market. As a result, the company stayed near the front of the pack in the industry. In
recent years, Infineon entered a product purchase and capacity agreement with Taiwan-based
DRAM manufacturer Winbond, under which Infineon agreed to license its 0.11um DRAM
technology to Winbond in exchange for the output using that technology. Infineon also formed
a joint venture with Taiwan-based Nanya Technology to build a new plant in Taiwan.

Micron Technology, based in Boise, Idaho, was founded in 1978. It sold its first DRAM product
manufactured at its own facility in 1982 and went public in 1984. Micron was the sole U.S.
producer remaining in the memory segment, and it had grown primarily through acquisitions.
In 1998, Micron purchased the memory chip business of Texas Instruments, including plants in
Texas, Italy, Japan, and Singapore. Subsequently, Micron purchased Dominion Semiconductor,
a unit of Toshiba located in Virginia. Over its 26-year existence, Micron had encountered
numerous periods of severe financial distress. Starting in the late 1990s, Micron exited many of
its non-DRAM memory businesses and reduced its workforce by 10%. As of 2003, Micron was
focused almost entirely on DRAM (96% of sales). In 2003, Micron received a $500 million
investment from Intel to invest in next-generation DRAM technology.

Nanya Technology, based in Taiwan, was the fifth-largest DRAM manufacturer. It had two
fabs. In 1998, Nanya purchased DRAM technology from IBM Corporation. In December 2002,
Nanya and Infineon launched joint developments for next-generation process technology. They
formed a joint venture named Inotera and invested a total of $2.2 billion toward a fab near
Taipei. Inotera began producing 256Mbit DRAM in June 2004.

Semiconductor Manufacturing International Corp. (SMIC), established in 2000 and

headquartered in Shanghai, was China’s largest foundry, manufacturing logic and memory
chips including DRAM. Foundries did not design chips as Samsung did, but, rather, took
designs from other firms. In 2003, Infineon agreed to license technology to SMIC in exchange for
purchasing rights to much of the output. SMIC also made a similar agreement with Elpida and
purchased a $1 billion Chinese production facility from Motorola in 2003. Through this deal,
Motorola took a minority stake in SMIC and agreed to license technology to SMIC in exchange
for exclusive use of the resulting output. SMIC’s revenue had increased from $50.3 million in
2002 to $365.8 million in 2003. In March 2004, the company completed a dual listing on the New
York and Hong Kong stock exchanges.

Samsung’s Memory Business

History. Korea’s semiconductor industry started wafer production when a small start-up called
Korea Semiconductor Company began operating in 1974. Without strong financing and
proprietary technology, the start-up quickly ran into financial difficulties. Samsung’s Kun Hee
Lee decided to purchase Korea Semiconductor Company using his personal savings and merge
it with Samsung, then a producer of low-end consumer electronics. At the time, Samsung relied
on labor-intensive assembly lines, importing semiconductors and other advanced products
from abroad. The first semiconductor developed by the merged firm was the “watch chip,”
used in wristwatches. Then-president of South Korea, Jung Hee Park, was so proud of the

9
company’s accomplishment that he had his name printed on many of the watches. President
Park would personally give the watches to visiting foreign dignitaries.

Kun Hee Lee wanted to get into DRAMs, the high-growth memory segment in the 1980s and
1990s. So, from 1983 to 1985, as the global semiconductor market went into a recession and Intel
exited the DRAM business, Samsung allocated more than $100 million to DRAM development,
even though at the time, it cost $1.30 to produce a single 64K DRAM chip versus a market price
below $1.00. As the capital requirements for a fab increased during the late 1980s and early
1990s, Japanese competitors struggled to make the investments necessary to compete in
emerging generations of chips, but Samsung poured in many millions of dollars.

Technology development. To design and produce its first 64K DRAMs in the 1980s, Samsung
had to license DRAM technology and manufacturing know-how from Micron. To develop
frontier technology for the next generation of DRAM, Samsung created what was, at the time,
an unusual internal competition across global R&D sites. The company hired one team
composed primarily of Korean Americans with extensive job experience in the semiconductor
industry and located that team in California. At the same time, Samsung set up a team in South
Korea, also headed by two Korean Americans with extensive industry experience. The teams
were told to be cooperative, but each was to come up with its own solution. The team in
California won the competition for designing 256K DRAM, but in the following generation of
1Mbit technology, the team in Korea won. In subsequent years, the company set up competing
product development teams throughout its operations.

Beginning with 4Mbit DRAM in the late 1980s, semiconductor firms faced a critical decision
about how to fit four million cells onto a tiny chip. Each cell, a location to store information,
consisted of a transistor and a capacitor. Two ideas were debated: “Stacking” involved
replacing what had been a one-level construction with an apartment building-like structure of
cells stacked on one another. “Trenching” involved creating floors below the surface of the chip.
Both technologies had pros and cons, with IBM, Toshiba, and NEC using trenching and
Matsushita, Fujitsu, and Hitachi adopting stacking. Chairman Lee made the ultimate decision:
he chose stacking, because if a problem was discovered in a trench-style chip, one could not
look inside to see what was wrong; everything was covered and hidden from view. In contrast,
stacking was simple and modular, making it easier to see and fix mistakes.

IBM, Toshiba, and NEC subsequently discovered problems with trenching, but had already
made multibillion-dollar commitments to the technology and had created design routines that
worked only with trenching. When the companies tried to switch to stacking, they lost years of
development time. In the meantime, Hitachi became number one in the industry, and Samsung
began to catch up with Hitachi.

As of the early 1990s, Samsung had joined the industry’s top echelon. Samsung still wanted to
be number one, so the company’s senior managers invested $1 billion towards mastering the
latest, larger 8-inch wafer sizes for DRAM chips. No one else in the industry was willing to take
the risk of investing in 8-inch mass production so early. The decision paid off. Samsung gained
number one market share in the DRAM industry in 1992 and maintained its leadership over the
following 13 years. This leadership held up during market peaks and lows.

Product mix. As of 2003, Samsung offered over 1,200 different variations of DRAM products.
Given that DRAM products were conventionally thought of as commodities, the ability to

10
produce 1,200 different varieties was unprecedented in the memory industry; only Hynix came
somewhat close. Exhibits 8 & 9 compare leading memory semiconductor firms’ DRAM
production volume by density (size) and type of chip. [The data in these two exhibits is stated in
terms of 256Mbit equivalent chips (current sweet spot of the market), i.e., a 512Mbit chip is
equivalent to two 256Mbit chips and a 64Mbit chip is equivalent to ¼ of a 256Mbit chip.
Restating data this way was common in this industry. If an exhibit herein presents data in terms
of 256Mbit equivalent, the exhibit clearly indicates this normalization.]

Samsung led the industry at finding profitable uses for fabs that made older “legacy” chips,
with Hynix being the #2 competitor. The result was that Hynix and Samsung were in
competition in many regional markets. Typically, each firm would engage in some product
customization for buyers (although the designs were not that high tech) and synchronize
supply with buyers’ manufacturing schedules. For example, both Samsung and Hynix had each
dedicated an older fab to making 64MBit chips for electronics manufacturers in Central Asia
(e.g., Kazakhstan). In these markets, Samsung had to think carefully about how to set prices to
maximize its profits given market dynamics and Hynix’s likely prices. Exhibit 10 provides a
pricing analysis of the Central Asian market for 64Mbit chips from the Global Strategy Group.

Design and production. Unlike its competitors, Samsung tried to create new uses for DRAMs.
Over time, Samsung had launched new DRAM products with product-specific applications in
laptops and personal game players, for example. Many of these applications shared a common
core design, as did even two seemingly different architectures, DDR DRAM and Rambus
DRAM. The difference was that Rambus had a high-speed interface, which Samsung connected
to the DRAM core design. Samsung sought to customize its products around a core design.

Samsung’s main fabs were at a single site just south of Seoul, South Korea. In contrast, its
competitors’ facilities were scattered across the globe. (Hynix’s fabs were split between two
Korean locations, a legacy of its acquisition of LG’s fabs.) With the benefit of collocation and
scale of fab investment, Samsung was estimated to have saved an average of 12% on fab
construction costs. The site was in the mountains on flattened land, where the air was clean and
free of dust. In its fabs, Samsung produced multiple product architectures on each production
line. Process engineers had reputedly figured out how to modify its production equipment for
all kinds of contingencies. Exhibit 11 compares semiconductor firms by production technology
used to make DRAM. Exhibit 12 compares DRAM operating results for these firms.

Pressure to Focus

For all its success, Samsung had critics. One theme was that it was more diversified than many
of its leading competitors. Investors and bankers had urged Samsung for years to divest the
Digital Appliances business. As Joon Mahn Lee of Good Morning Securities in Seoul put it:
“The brand Samsung connotes high quality electronic goods to the global consumer. Home
appliances are inherently low-tech, even if a Samsung fridge tells you to buy more milk.”
Others thought that Samsung held onto the Digital Appliances business out of sentimental
attachment, given the longstanding prominence of the products in Korean households.
Nonetheless, Samsung leadership believed in the synergies. Asserted Tae-Hoon Chung, VP of
business development at Samsung, “Our brand means high quality without extravagance. Once
consumers try and like a Samsung TV or cell phone, they will try a refrigerator or air
conditioner and appreciate its good value.”

11
Exhibit 1 Samsung Electronics Organizational Structure

Exhibit 2 Samsung Electronics Financial Results by Division1

Digital Finance &

₩ millions Digital Media Appliances Telecommunications Semiconductors LCD Panels Other Intracompany Total Group
Revenue 29,814,761 8,482,065 24,547,998 38,107,278 15,254,706 5,139,998 (56,529,350) 64,817,456
Internal 13,892,798 3,545,530 9,244,846 19,550,409 10,295,767 ‐ (56,529,350) ‐
External 15,921,963 4,936,535 15,303,152 18,556,869 4,958,939 5,139,998 ‐ 64,817,456

Operating Income 567,424 (49,160) 2,817,809 4,582,004 (14,763) (1,584,879) (22,208) 6,296,227
Assets 8,877,694 3,688,652 9,825,081 27,829,190 8,453,571 20,416,998 (11,049,445) 68,041,741

1 At the time of the case, there were about 1,050 Korean Won (₩) to US$1.

12
Exhibit 3 Global Market Share of Samsung and Two Leading Competitors for Select Products

Exhibit 4 Market Share % of Select Samsung Electronics Consumer Products in Select US Cities

Market Share

Side‐by‐Side
City Cell Phones LCD TVs Refrigerators
Phoenix, AZ 9.3 10.3 0.0
Seattle, WA 11.9 14.1 1.5
New York, NY 15.4 17.8 4.2
Oakland, CA 7.7 7.8 0.0
Chicago, IL 11.5 15.3 1.8
Boston, MA 13.2 16.2 2.5
Miami, FL 12.5 15.2 2.7
Salt Lake City, UT 8.1 12.7 0.0
Detroit, MI 11.0 14.4 1.6
Dallas, TX 13.3 16.8 3.0
Los Angeles, CA 18.4 22.8 6.8
Washington, DC 12.6 16.3 2.2
Atlanta, GA 13.3 17.2 3.3
Philadelphia, PA 10.6 12.9 1.4

13
Exhibit 5 Average Industry Selling Prices (US$) for Different Individual DRAM chips in 2003

Density Average SDRAM DDR SDRAM DDR2 SDRAM Rambus DRAM

4Mbit 0.97
16Mbit 1.19
64Mbit 1.88
128Mbit 2.75 4.87 4.82
256Mbit 4.68 4.67 4.55 8.67 8.45
512Mbit 21.70
1Gbit 83.57

Exhibit 6 Comparison of 8-inch Wafer vs. 12-inch Wafer2

2 Net die per wafer is based on a 0.13m design rule.

14
Exhibit 7 Financial Comparison of Memory Semiconductor Firms in 20033

Samsung
Electronics Hynix Infineon Micron
₩ millions ₩ millions € millions US$ millions
Sales 64,817,456 3,620,426 6,152 3,091
Cost of Good Sold 42,252,493 3,175,802 4,614 3,112
R&D 2,020,431 351,000 1,089 656
Sales, General & Administrative 14,248,305 353,069 793 509
6,296,227 (259,445) (344) (1,187)

Net Interest Expense (76,226) 225,616 52 18

Other Non‐Operating Expense 410,206 1,259,941 (45) 68
Net Income 5,962,247 (1,745,002) (351) (1,273)

Assets 68,041,741 6,906,052 10,805 7,158

Exhibit 8 DRAM Production Volume by Density (Size) in 2003

Production Volume (million unit, 256Mbit equiv.)

Samsung Micron Infineon Hynix
16Mbit 1.3 0.1% 1.0 0.2% 0.0 0.0% 10.0 1.9%
64Mbit 16.4 1.8% 29.7 4.4% 0.0 0.0% 33.6 6.4%
128Mbit 151.6 16.9% 88.1 13.1% 43.7 8.2% 96.8 18.6%
256Mbit 695.8 77.6% 540.1 80.3% 479.5 89.6% 374.2 71.8%
512Mbit 30.4 3.4% 13.7 2.0% 11.5 2.1% 6.8 1.3%
1Gbit 1.0 0.1% 0.1 0.0% 0.6 0.1% 0.0 0.0%
Total 896.4 100.0% 672.8 100.0% 535.3 100.0% 521.5 100.0%

Exhibit 9 DRAM Production Volume by Product Line in 2003

Production Volume (million unit, 256Mbit equiv.)

Samsung Micron Infineon Hynix
SDRAM 206.1 23.0% 191.8 28.5% 79.1 14.8% 117.5 22.5%
DDR SDRAM 585.3 65.3% 475.6 70.7% 437.8 81.8% 401.4 77.0%
DDR2 SDRAM 40.4 4.5% 0.0 0.0% 1.7 0.3% 0.0 0.0%
Rambus DRAM 37.9 4.2% 0.0 0.0% 2.4 0.4% 0.0 0.0%
Other DRAM 25.9 2.9% 5.4 0.8% 14.3 2.7% 1.3 0.2%
Total 896.4 100.0% 672.8 100.0% 535.3 100.0% 521.5 100.0%

Note: SMIC’s production volume was 68.2 million units in 256Mbit equivalent. A breakdown by
size and product line was not available.

3 At the time of the case, there were about 1,050 Korean Won (₩) and 1.25 Euros (€) to US$1.

15
Exhibit 10 Average Selling Price (US$) for 64MBit Chips and Cash (Marginal) Profit for Samsung’s and Hynix’s fabs serving Central
Asian market4

Samsung Electronics Price Level

3.2 ‐ ‐ ‐ ‐ ‐ 0.8 5.5 10.2 14.9 19.6 24.3 29.0
3.9 8.9 13.9 18.9 23.9 28.5 31.0 32.7 33.7 34.0 33.6 32.4
3.0 ‐ ‐ ‐ ‐ 0.7 5.1 9.4 13.7 18.1 22.4 26.7 31.1
3.9 8.9 13.9 18.9 23.6 26.4 28.5 29.9 30.5 30.4 29.6 28.1
2.8 ‐ ‐ ‐ 0.7 4.6 8.6 12.6 16.5 20.5 24.5 28.5 32.4
3.9 8.9 13.9 18.6 21.8 24.3 26.0 27.1 27.3 26.9 25.7 23.8
2.6 ‐ ‐ 0.6 4.2 7.8 11.4 15.0 18.6 22.2 25.9 29.5 33.1
3.9 8.9 13.7 17.3 20.1 22.2 23.6 24.2 24.1 23.3 21.8 19.5
2.4 ‐ 0.5 3.8 7.0 10.3 13.5 16.8 20.0 23.2 26.5 29.7 33.0
3.9 8.8 12.7 15.9 18.4 20.1 21.1 21.4 20.9 19.7 17.8 15.2
2.2 0.5 3.4 6.2 9.1 12.0 14.9 17.7 20.6 23.5 26.4 29.3 32.1
3.8 8.1 11.7 14.5 16.6 18.0 18.6 18.5 17.7 16.2 13.9 10.9
2.0 2.9 5.4 8.0 10.5 13.0 15.5 18.0 20.5 23.0 25.6 28.1 30.6
3.5 7.5 10.7 13.1 14.9 15.9 16.2 15.7 14.5 12.6 10.0 6.6
1.8 4.7 6.8 8.9 11.1 13.2 15.4 17.5 19.7 21.8 24.0 26.1 28.3
3.3 6.8 9.7 11.8 13.1 13.8 13.7 12.9 11.3 9.0 6.0 2.3
1.6 5.6 7.4 9.2 11.0 12.8 14.6 16.3 18.1 19.9 21.7 23.5 24.4
3.0 6.2 8.6 10.4 11.4 11.7 11.2 10.0 8.1 5.5 2.1 ‐
1.4 5.9 7.3 8.7 10.2 11.6 13.0 14.4 15.8 17.3 18.7 19.4 19.4
2.7 5.5 7.6 9.0 9.6 9.6 8.7 7.2 4.9 1.9 ‐ ‐
1.2 5.4 6.5 7.6 8.6 9.7 10.7 11.8 12.8 13.9 14.4 14.4 14.4
2.4 4.9 6.6 7.6 7.9 7.4 6.3 4.4 1.7 ‐ ‐ ‐
1.0 4.2 4.9 5.6 6.3 7.0 7.7 8.4 9.1 9.4 9.4 9.4 9.4
2.1 4.2 5.6 6.2 6.2 5.3 3.8 1.5 ‐ ‐ ‐ ‐
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2
Hynix Semiconductor Price Level

4 In each box, the upper figure is Samsung Electronics’ profit, and the lower figure is Hynix Semiconductor’s profit.

16
Exhibit 11 Leading DRAM Competitors’ Production by Technology5

Exhibit 12 Comparison of DRAM Operating Results for Select Competitors in 2003

$ Millions Samsung Micron Infineon Hynix SMIC

Revenue 5,092 3,317 2,703 2,592 302
Fully loaded costs 3,863 4,447 2,687 2,780 330
Raw materials 1,058 1,299 846 1,006 125
Labor 484 632 407 266 16
Depreciation 1,210 1,265 803 772 111
R&D 538 383 380 302 55
SG&A 583 861 246 433 23
Operating Profit 1,228 (1,130) 11 (188) (28)

5 Yield rate is for 0.11m design rule applied to making 256Mbit DRAM chip.

17