Open Service for a Total Design Ecosystem
Twenty years ago TSMC broke the seal on the insular, monopoly-dominated semiconductor nexus. Now it is working on a system of total-process cooperation, to make sure its clients get it right the first time.
Open Service for a Total Design EcosystemBy Jimmy Hsiung
From CommonWealth Magazine (vol. 382 )
As one of Taiwan's heavyweight corporations, each and every move of Taiwan Semiconductor Manufacturing Company (TSMC) is closely followed and scrutinized.
Yet over the last two or three years, TSMC has undergone a stealthy transformation. Engineers of the shaggy variety commonly seen at IC design companies have gradually surfaced in large numbers at TSMC. Not clients, they are IC aces snapped up by TSMC from around the world.
This is TSMC's new platoon, a new team of designers working expressly on design and technology platform (DTP) technology, over and above the company's team of over 1300 research and development engineers renowned throughout the semiconductor realm. The company has invested in the DTP group, currently numbering over 500, to the tune of US$100 million a year.
'We're out to establish an IC design ecosystem,' reveals Rick L. Tsai, TSMC's president and chief operating officer. 'Clients can find whatever services they need within this ecosphere.'
TSMC, which jumped right into the semiconductor industry with an innovative business model, is moving further forward in an even more open fashion, working more closely with clients to create value. The new open cooperative model is sure to shock the industry again, just as TSMC's new alliance fearlessly took on powerful competitors IBM and Samsung.
Thus it's no wonder that, speaking of the open innovative business model he studies astutely, Henry Chesbrough, adjunct professor and executive director of UC Berkeley's Center for Open Innovation, affirmed that TSMC's is the best kind of open innovation, born right out of the open business model, and highly successful.
Two decades ago the semiconductor realm was a totally closed monopoly production system, with Intel and Texas Instruments handling every aspect from IC design to production to self-marketing using the integrated design and manufacturing (IDM) model. The first to puncture a hole in that order was none other than TSMC chairman Morris Chang.
Immediate Impact on Semiconductor Industry
Morris Chang pioneered the dedicated IC foundry model 20 years ago, carving out an area of OEM value creation from the vertically integrated IC manufacturing chain, not only generating substantial value, but helping shift the entire IC manufacturing industry into overdrive.
TSMC's production value in 2006 reached NT$317.4 billion, or 40 times its initial level. In a survey conducted this past February by market research company IC Insights, TSMC ranked sixth among all semiconductor companies worldwide. Moreover, according to a survey by the Nomura Research Institute, as of the end of 2006, TSMC's stock was worth US$53.49 billion, making it the most valuable Taiwanese company for the second straight year.
Reshaping the IC manufacturing value chain, TSMC single-handedly shepherded the emergence and rise of the worldwide fabless IC design industry.
With TSMC investing in prohibitively expensive wafer production facilities, IC design companies needed only to hand off finished IC designs to TSMC like batons in a relay race, and then market them to world-class customers. In this way, with TSMC's production capacity, even the smallest IC design houses could become partners with such top-tier corporations as Apple, Nokia, Sony and HP, giving them access to the entire worldwide IC design stage.
The production value of the IC design industry in Taiwan alone reached NT$323.4 billion in 2006, or approximately 20.3 percent of the US$49.7 billion industry worldwide. Five other Taiwanese IC design companies joined TSMC in the ranks of the world's top 20.
New 500-man Service Team
As the climate has changed, TSMC has had to adapt and in some instances take the lead by making changes of their own.
Celebrating TSMC's twentieth anniversary in late April of this year, Morris Chang noted that in the future, IC design houses and OEM fabs 'must go from the relay-style cooperative model to working together through the entire process, from design to production. The DTP is the key for TSMC to make that transition towards total-process cooperation.
Perusing TSMC's register of company vice presidents, one man stands out from the rest with his long ponytail, mustache and goatee: Dr. Fu-Chieh Hsu, vice president of TSMC's Design and Technology Platform.
Dr. Hsu is intimately familiar with the IC design business, having founded Monolithic System Technology Inc. (MoSys) in the United States and served as its chairman and CEO. VIA Technologies founders Wen-Chi Chen and Tzu-Mu Lin, along with Wayne Liang, founder and chairman of current industry darling MStar Semiconductor, were all Hsu's classmates in electrical engineering at National Taiwan University.
One day Hsu received a call from Hu Cheng-ming, former chief technology officer of TSMC and presently a professor at Berkeley, who told him, 'Rather than look for value for one particular chipset designer, why not help all IC design companies create value!' And thus on the other end of the line the TSMC 'design ecosystem' vision was germinated.
Kuo Wu, deputy director of TSMC's Product & Design Service Marketing Division, relates that not just Chieh-Fu Hsu, but many of the industry's top talent, including bosses and top executives at IC design companies, have drawn inspiration from TSMC. This has in turn helped catapult the company's DTP team from just over 100 people under the R&D structure to an independent juggernaut this year of more than 500.
Why have so many top dogs been inspired by the design ecology roadmap? They have identified a common trend.
The consumer-oriented age is upon us, the market has become highly changeable, and the product life cycle has shrunk. With consumer demands for smaller, thinner and lighter products, products are becoming increasingly complex, leaving no margin for error at any stage in the decision-making process along the product value chain.
Not only has product design become more difficult, but time has also gotten compressed. 'The product lifecycle for cellular phones has gone from 18 months to just three months in the last five years,' observes Dr. Ho-Ming Tong, general manager and chief R&D officer of the ASE Group, a package-testing partner of TSMC. 'You've got to get it right in one shot or you can't keep up,' he adds. Moreover, with system products dropping an average of 15 percent in price each year, it is up to R&D to find ways to reduce costs.
'There's only so much money and so much time. If you waste it, it's gone for good,' says Tsai, empathizing with the pressures of his clients.
With its clients' worlds getting more and more compressed and charged with tension, TSMC must maximize its own strengths to simplify its clients' challenges and make sure they make a winning move right away.
Last year products made with TSMC's 0.13-micron advanced manufacturing process accounted for 49 percent of overall sales. Presently, the company's production line has already advanced to where 90 nanometers is the norm. 'We'll start volume production of 45-nanometer products in the fourth quarter of this year,' adds Tsai.
One nanometer is about one ten-thousandth the width of a single strand of hair. The figure of 90 nanometers refers to the width of the circuit node. 'To say nothing of the increased difficulty and complexity of IC manufacturing, if the IC designer doesn't get assistance from the fab and the design service company, it's tough to get it right the first time,' says K.C. Shih of Global Unichip.
Kuo Wu chooses an interesting metaphor, saying, 'TSMC is like a vendor that specializes in developing film.' In the past, when the technology was simpler, TSMC only acted as the film developer. But now that manufacturing technology has become increasingly precise, TSMC must act as the photographer as well, even giving advice on shot selection and approach, so as to avoid 'missing the shot or getting the film stuck in the camera.'
Unlike IBM and Samsung's joint manufacturing technology, TSMC works with upstream designers and intellectual property strategists to provide technology and IP to clients, thus engaging in total-process cooperation with clients. First developing the concept of the design ecology three years ago, TSMC spun off its dedicated design team last year.
Rick L. Tsai stresses that the sum total of technological development constitutes a contract chipmaker's most basic competitiveness, making cooperation essential. 'But whom do you work with, and to what extent'? That, he says, is the key point to judge.
'TSMC will never use Samsung and IBM's cooperative model,' he asserts. 'It doesn't allow enough autonomy.'
Especially in the OEM chip manufacturing business, where production technology is everything, 'If you lose your technical autonomy you can forget about the rest, because that's the core value with which we service clients,' says Tsai, driving the point home.
With manufacturing technology as the core competency for contract chipset manufacturers, TSMC has always relied largely on an autonomous development approach, yet the company's cooperative arrangements can nonetheless be traced. For instance, Tsai mentions TSMC's close long-term technical collaboration with NXP. Still, he stresses that this European-based cooperative arrangement is focused mainly on fundamental scientific research in manufacturing technology. 'There's no cooperation in 65- or 45-nanometer production, but there is a lot in the 22 nm area,' he relates.
Although TSMC has stressed independence and autonomy in R&D from day one, Tsai is quick to point out that the company is by no means working in the dark behind closed doors.
To be sure, with one generation of chipset manufacturing processes requiring a minimum investment of US$1.5 billion, TSMC is always open to cooperation, says Tsai.
Robbins Yeh, country director for Synopsys, a global leader in EDA (electronic design automation) tools, confirms that TSMC approached both EDA and SIP (silicon intellectual property) vendors when it was looking to establish its IC design ecosystem for clients.
'Going it alone, that's too slow,' relates Tsai. But it is important to have control over the specs of technological development, autonomy in the development schedule, and even more autonomy in the essential area of intellectual property. These are basic factors for any contract chipmaker, stresses Tsai.
Cooperative Models Unlimited
In the latter packaging stage of chipset production, ASE's Tong reveals that his company has quite a few technicians stationed within TSMC working expressly on such new techniques as system-level packaging.
'We need to get everything right and to offer value,' stresses Tsai. The design ecosystem is an extremely valuable investment for TSMC. As for the latter production stages, 'Others have already provided value; TSMC isn't going to get involved,' he reveals. Expanding, he offers that since this is a link in the industry value chain, even if it is not a major focus it is still vital to know everything there is to know about the technology and trends in this area.
TSMC has picked up another cooperative model in recent years, that being strategic reinvestment. Whether due to client demand or to gain better familiarity with the chipset package testing field, in recent years TSMC, which at one time stressed attention to core business, has made reinvestments in packaging companies VisEra Technologies and XinTec.
From external partnerships to internal cooperation and strategic reinvestments, one can trace both the constants and the changes in TSMC's development. To sum those up, the company's approach could be described as a highly variable open operational model under an unchanging overall innovative business model.
Morris Chang has established TSMC's business model of innovation and long-term profitability over the past 20 years. Now under the model of total-process cooperation, TSMC is preparing to take its first step towards the next 20 years of innovation.
Translated from the Chinese by David Toman