The changing future of the Asian foundry landscape
08/01/2003
By Aabid Husain, DongbuAnam Semiconductor USA, Santa Clara, California
As the semiconductor industry recovers from its worst ever global economic downturn, we should begin to see both fabless companies and IDMs increase their demand for foundry capacity. Significantly, the well-demonstrated and continuing cost efficiency of the fabless foundry model will force IDMs to shift their long-term manufacturing strategy away from cyclical expansion of their own wafer fabs toward adopting a "fab-lite" model. While the foundry industry has been dominated by the "Big Three" — namely TSMC, UMC, and Chartered — more than a dozen other foundries have entered the scene over the last five years, creating a much more competitive and diverse global foundry landscape.
 |
The most recent evidence of the shift was revealed in a Semico Research report that showed US-based IBM eclipsing Singapore-based Chartered Semiconductor as the world's third largest foundry (see table) [1]. Although the world's top two foundries, TSMC and UMC, currently dominate ~60% of the worldwide foundry market from their operations in Taiwan, they will be increasingly challenged by other Asian foundries that aspire to be ranked among the world's top-tier foundry service providers.
The profit-margin pressure imposed by emerging Asian foundries has caused TSMC and UMC to alter their business models in efforts to protect their customer relationships. With the implementation of its "virtual IDM" model, UMC has walked away from the "pure-play" foundry model. With this move, the world's second largest foundry is now investing in fabless companies and in some cases securing seats on their respective boards or assuming substantial operational control.
 Asian foundry models that will thrive in the future will provide end-to-end foundry solutions, solving many of the supply-chain issues that currently exist. |
null
Taking a different approach, TSMC is starting to lock in customers by offering a free pool of customized silicon-validated IP. But whether this approach reduces risk, or merely leads to a chip company finding itself virtually locked into a particular solution remains to be seen.
The winds of change now blowing across the Asian foundry landscape are not likely to abate anytime soon. As TSMC and UMC are increasingly challenged to defend their dominant market positions, the pathway for other Asian foundries that wish to achieve top-tier ranking is not without its own challenges.
While being a fast follower in technology is an absolute necessity, it is insufficient by itself to achieve success. Emerging Asian foundries with much smaller operations and fewer resources than TSMC and UMC must carefully manage their resources and investments to advance process technologies and expand manufacturing capacity. Equally important is gaining sufficient market visibility so that financial investments are timed well to take full advantage of an industry up cycle. For emerging Asian foundries that can meet these challenges, there are windows of opportunity that will enable them to see a clear path toward top-tier status.
Market growth opportunities
Most market researchers agree that the foundry segment will see dynamic growth in excess of 20% year-over-year. Amid this growth, TSMC and UMC — each serving in excess of 300 customers — are finding it increasingly difficult to continue offering "everything to everybody." Thus, there will be a growing opportunity for emerging foundries to differentiate themselves by offering specialized processes and focused services balanced by a more flexible business model.
Some emerging Asian foundries have already figured this out and are now providing a mix of foundry-compatible processes along with very well supported specialized processes that enable SoC designs to integrate functions such as embedded Flash storage, high-voltage power management, CMOS image sensing, and CMOS-rf signal processing. Additionally, those foundries based within well-developed semiconductor-manufacturing infrastructures, such as that found in South Korea, are well positioned to reduce the cost of logistics and improve supply chain management by offering a complete turnkey manufacturing solution from maskmaking to the drop-shipment of final tested units.
Emerging Asian foundries are offering extensive design and IP support as well as specialized value-added services such as packaging, assembly, and test through strategic partners (see figure). With high-growth market opportunities and large-volume drivers emerging outside of the mainstream CMOS processing, up-and-coming Asian foundries that execute well in this arena should be able to compete effectively with TSMC and UMC.
Beyond Taiwan
Against the backdrop of competition with the major foundries, it is worth taking a closer look at key Asian nations outside of Taiwan where there are foundries driving aggressive growth strategies, and their prospects for becoming top-tier suppliers in the years ahead.
China has made a major impact on the foundry landscape worldwide. The hype generated in financial circles about the huge untapped China market — coupled with the fear instituted by the Chinese Communist government regarding the requirement of local content in products sold into China — have made CEOs of chip companies sit up and take notice. The emergence of China-based foundries such as SMIC and Grace Semiconductor shows that China intends to be a serious player in the foundry industry. In fact, SMIC's aggressive technology development and wafer pricing continues to put pressure on the better-established foundries.
China's potential to rival Taiwan as a source of independent foundry service, however, will not be realized without overcoming some very daunting challenges. For starters, both the US and Japan have long-standing restrictions on the transfer of technology and equipment into China. Moreover, China lacks the technical talent and infrastructure that regions such as Japan, Korea, and Taiwan have developed over a long period. China therefore needs to import technical talent, which is not likely to stay more than a few years. Perhaps the most serious challenge facing China is the blatant piracy of intellectual property (IP). Fabless companies must be able to place proprietary designs with Chinese foundries without the danger of their IP being pirated. Indeed, until China can realign its culture to comply with trademark and copyright laws, IP piracy will continue to impede the efforts of Chinese foundries that are striving to attract innovative fabless and IDM customers. The degree of difficulty that China faces in meeting these challenges is proportional to the degree of opportunity there exists for other Asian foundries to emerge among the world's top suppliers along with TSMC and UMC.
Japan has a well-established semiconductor-manufacturing infrastructure and a culture that complies with trademark/copyright laws. Although Japan was one of the very first countries to provide foundry services, this region's semiconductor industry continues to be dominated by major IDMs that have their own wafer fabs. Indeed, Japanese IDMs have preferred to license their technologies to pure-play foundries instead of offering foundry services in a significant manner. There is little evidence of this trend changing.
With its slumping economy over the last decade, Japan has not been a fertile environment for wafer fab expansions. As a result, the country's IDMs have increasingly tapped other Asian foundries to manufacture wafers. Going forward, Japan's IDMs could play an indirect role in shaping the Asian foundry landscape, depending on their willingness to license their IP and processing technologies to other Asian foundries and placing substantial business with them.
South Korea offers many attributes that make it an optimal region for the next top-tier Asian foundry to emerge. For starters, its semiconductor-manufacturing infrastructure rivals those that have been established in Japan and Taiwan over a long period of time. As one of the most technologically literate regions in all of Asia, South Korea also offers a vast and stable pool of technical talent. Moreover, it has a very respectable track record in complying with trademark/copyright laws, and boasts an excellent track record in providing the industry's most comprehensive range of packaging and testing services.
Significantly, South Korea is a high-growth market for consumer electronics products, including cell phones, TV sets, and PDAs. This fact is often overlooked by fabless and IDM companies in evaluating foundry partners. With South Koreans currently purchasing more than 10 million cell phones annually, and South Korean manufacturers exporting cell phones that account for ~20% of the worldwide market, it should come as no surprise that the South Korean government is encouraging localization of silicon content in products purchased by its citizens.
While South Korea has long hosted foundries such as those operated by Hynix and Anam Semiconductor, major investments made by the Dongbu Group have changed the Korean foundry landscape. These investments were first made to create Dongbu Electronics, and then subsequently to gain control of Anam Semiconductor, thereby paving the way for the two foundries to join forces, showing that there is an opportunity for a synergistic union of two foundries to achieve top-tier player ranking. Indeed, DongbuAnam Semiconductor is today ranked among the top five foundries worldwide, and is the world's fourth largest in the pure-play category.
Singapore has been host to Chartered Semiconductor, which until recently was ranked as the world's third largest foundry. Chartered has slipped to the fourth position behind IBM (not a pure-play foundry) as it continues to be buffeted by certain adverse factors that appear to be related specifically to being located in Singapore. First, there is the difficulty of retaining imported talent for more than a few years. Because the technical talent pool in Singapore continues to be limited, the growth of Chartered or the establishment of other foundries in Singapore will be constrained. Another factor is the cost of doing business in Singapore. Establishing a new foundry in Singapore is like establishing a new foundry in Manhattan; it's very expensive. For the foreseeable future, it seems that Chartered will remain as the only major fab in Singapore.
Malaysia is a varied region, much of it still third world. As such, it probably faces more challenges in being an attractive location for building a top-tier foundry and has a long way to catch up with the robust semiconductor-manufacturing infrastructures that are well established in other Asian locales. Moreover, Malaysia suffers the same malaise as Singapore and China, in that it has to import talent as well as having the challenges of building a vibrant semiconductor-manufacturing infrastructure in a quasi-third world economic environment. The future for Malaysian foundries will likely be limited growth, as they will face formidable challenges in competing with the top-tier three or four Asian foundries in the future. Consolidation of foundries in and around Malaysia seems to be the only survival choice.
Conclusion
As the foundry landscape changes and companies compete to capture top-tier status, there are unpredictable variables that could have a significant impact on who wins and who loses. Such variables should be weighed during the foundry selection process. The earthquakes, fire, and water shortages that hit Taiwan in the past have adversely affected the output of foundries based there, as well as the complete supply chain of the electronics industry. More recently, we have witnessed how the SARS outbreak has slowed the wheels of commerce in China and certain other regions of Asia. Accordingly, the successful Asian foundries of the future will have contingency plans in place to mitigate any adverse affects caused by such variables.
Regardless of the variables, one thing is certain: the semiconductor industry (particularly the foundry segment) will continue to mature, following in the footsteps of the automotive and aviation industries. Within this maturation process, it is expected that diversification will take place technologically, geographically, as well as in the range of services offered. The increasingly diverse foundry landscape will free both fabless and IDM companies from their historical dependency on the major Taiwanese foundries, thereby providing a broader scope of strategic foundry options for achieving long-term success. The foundries based in Asia will continue to play a key role in making it happen.
Reference
- "Foundry Suppliers: The Competition Heats Up," Semico Research Corp. report MA102-1-03, April 2003.
Aabid Husain is director of worldwide marketing at DongbuAnam Semiconductor USA, 2953 Bunker Hill, Suite 206, Santa Clara, CA 95054; ph 408/330-0337, e-mail [email protected].