P7 (Analysis 4-Korea & Taiwan) (Auto-Saved)
P7 (Analysis 4-Korea & Taiwan) (Auto-Saved)
P7 (Analysis 4-Korea & Taiwan) (Auto-Saved)
Entrepreneurial vision
“The Korean electronics industry cannot
survive in the world market if it only
continues assembling parts like today. In
order to stabilize the electronics industry,
the development of a semiconductor industry
is indispensable. Moreover, the
semiconductor industry is the basis of the
every other industry” (Lee Byung-Chul)
Samsung Electronics
Sending all of those involved in writing
the negative report to the development
team of 16K DRAMs
“It was probably Lee Byung-Chul’ implicit
order that they should devote their energy
and courage to oppose the top management’s
decision to developing DRAMs” (Moon Sang-
Young, quoted in the Chosun Ilbo, 20 Aug.
2015)
Money-devouring machine
in the beginning
The cumulative deficits
- about US$227 million by 1986
- far more than the whole Samsung Group’s
profit in 1986, $136 million
- impossible to sustain them without intra-
group support
Keeping a high level of
investment
Maintaining high-level of investment
- over 50% investment-to-sales ratio all
through the 1980s
The group-level strategic support
“the total sales of Samsung group was $21.1
billion in 1987 (Mody 1990: 295, table 3). And
those of Hitachi and Toshiba were $23. 6
billion and $20.2 billion respectively (KEIPA
1988: 173; 175, $ = 123.5 at the end of 1987).
Dynamic catch-up strategy
Narrowing the gap as soon as possible
Employing a dynamic catch-up strategy
- 1.5 years gap in 4M DRAM (1988)
- nearly same in 16M DRAM (1989)
Overtaking in cost and quality
Aggressive technological choice
Investing in the state-of-the-art technologies with a
long-term future, although they were not mature
and risky
(1) Bigger size wafers
- 6 inch over 5 inch in 64K DRAM
- 8 inch over 6 inch in 4M DRAM in 1989
(2) Trench vs. stack method
- inevitable transition from 16M DRAM
Overtaking in cost and quality
Establishing ‘ramp-up’ capability
Inter-generational application of technologies
Deepening toward higher-end DRAMS
Diversification and the early lead in mobile
technologies
The increase in technology standards
→ turning into “first-mover advantages”
Development and ramp-up
The best competitive strategy in DRAMs?
- Too early development ahead of market is
costly
- Time gap between development and mass
production
- Waiting for the development of the next
generation chip closest to the market creation
and ramping up production as fast as possible
Ramp-up capability
Samsung official’s comment:
“We estimate that we overtook Toshiba in
ramping-up capability in 16M DRAM in 1991.” (A
Samsung official)
Further strengthened during the leadership
period
Samsung’s competitive advantage about which
its “competitors feel most curious and are most
inquisitive”
Ramp-up capability
(1) Fuller integration between development and
production
- directing engineers from every stage, from design
to mass-production, to take part in the entire
development process together
- better at problem-solving and information-sharing
(2) “Samsung-style TF(task force)”
- “different from Japanese-style TF”
- mastering ‘parallel problem-solving capability’ as
against ‘serial problem-solving capability
Ramp-up capability
(3) Testing yield rates with pilot production lines
- Solving many potential production problems at the
development stage
- attained a near ‘golden yield rate (80%)’ in the initial
mass production of the 12-inch line in 2001
(4) Intra-firm knowledge transfer system
- a comprehensive database
- an internal practice that about half the engineers
working in a new production line should have worked in
the previous generation production line
- ‘Maintenance Prevention (MP) Information Sharing
Conference
Inter-generational application
of technologies
Breaking the match between circuit density and
design rule from the early 1990s
- 64M DRAM: 0.35µm → 0.25µm → 0.18µm → 0.13µm
- Shrinking chip size huge reduction in cost and
improvement in stability and quality
eg. 254M DRAM: 0.18µm → 0.13µm (50%) → 0.11µm
(59%/30%)
→ Making DRAMs more easily available for mobile
devices
Inter-generational application
of technologies
A formidable entry barrier
- Vital to survive the life-cycle downturns
- Japanese competitors considerably
weakened from accumulated losses during
16M & 64M DRAM life cycles while Samsung
maintained positive profits
- Pre-empting new entrants
- Re-enforcing first-mover advantages
Deepening
Xilinx + UMC
- ‘the “virtual IDM” relationship’ in 1995
- two companies under one roof
“Xilinx moved all their new production to UMC …
Xilinx and UMC worked together to develop the
process technology, create test chips, and so on. In
fact, Xilinx had a whole floor of one of UMC’s
buildings for their own employees.” (Nenni &
MacLellan 2019: 60)
- enabling technologies: FPGAs(field programmable
gate arrays), EDA (electronic design automation) …
TSMC and UMC
Not much demand in the beginning
“There was very little business from fabless companies. In fact, all
the designs from fabless semiconductor companies together could
not have filled a single fab in 1995.”
The growth with a close connection with the U.S. design houses
- by 1996, handling 40% of the output of the US fabless companies
“Most Taiwanese foundries involve joint ventures with American
fabless firms (Langlois & Steinmueller 1999: 57)
Morris Chang’s invention?
“Chang wasn’t trying to reinvent the semiconductor industry when
he started TSMC. He did it because government officials in Taiwan
wanted him to start a semiconductor business, and with Taiwan
weak in both design and marketing, he didn’t see any other way.”
(Perry 2011)
TSMC and UMC
The exponential increase in the cost of fab and the
rapid growth of the fabless-foundry alliance
- The cost of fab: $1bn (1998) → $2bn (2002, Intel) →
$4bn (2006, Samsung) → $10bn (2013, Intel)
- IDM becomes a game among the largest players
“… when a fab cost $3 billion to build, a company would
face a depreciation cost of roughly $1 billion per year,
meaning that they need to have a running semiconductor
business of perhaps $5 billion, around the size of AMD,
who was the only competitor to Intel in the x86
microprocessor business.”
TSMC and UMC, “pure foundry players” at the right
time of the industry transformation
Symbiosis and competition
between ‘pure players’ and IDMs
IDMs’ need for moving into the foundry business
- the foundry itself is a profitable business
- an option to increase utilization of their expansive
fabs
TSMC vs. Samsung vs. Intel?