Comparing semiconductor industries in Taiwan and China
08/01/1998
Comparing semiconductor industries in Taiwan and China
George P. Koo, International Strategic Alliances Inc., Mountain View, California
Both Taiwan and China began to develop semiconductor industries at about the same time. From today`s view, Taiwan has been far more successful and is still growing dramatically. Despite China`s slow start, particularly related to the Chinese government`s approach to industry development, the necessary changes seem to be happening. Indicators show the potential of the Chinese semiconductor market and the need for expanded wafer fabrication.
In the early 1980s, the People`s Republic of China (PRC) imported its first semiconductor technology-the purchase of a turnkey, secondhand fab from the US that was installed in Wuxi, which is two hours west of Shanghai. That 3-in. wafer line, though soon obsolete, marked the beginning of a steady infusion of foreign technology to what is now known as Hua Jing Electronics, China`s largest wholly state-owned fab. In 1995, the company turned out 139 million discrete devices worth ~$34.3 million and 84.6 million ICs worth ~$68.9 million. (Accurate data about China`s semiconductor industry are slow to come out of the country.)
At about the same time, the Republic of China (Taiwan) began to develop its semiconductor industry with active government support. The government subsidized United Microelectronics Corp. (UMC) and Taiwan Semiconductor Manufacturing Co. (TSMC) just as China subsidized its chipmakers. Typical examples of the initial commercial output from Taiwan were simple sound chips for handheld games, serving a niche market at the time, but hardly impressive from a technical point of view.
Obviously, China and Taiwan were both late entering IC manufacturing. Both governments invested heavily and attempted to bring in considerable technology from the West. Of the two, Taiwan has so far been spectacularly more successful in its approach than has the mainland. Examining the different strategies adopted by each and the conditions that exist in each country today is helpful for other nations aspiring to join the elite semiconductor-manufacturing circle, a circle arguably more valuable and exclusive than even the world`s nuclear club.
The fabs
Today, China has six significant wafers fabs, ranging from established state-owned enterprises to joint ventures to one wholly owned fab-Motorola`s recent start up (Table 1). The latest fabs on this list-Motorola and joint venture Shanghai Hua Hong NEC Electronics-are barely under way. By contrast, Taiwan has 12 significant ongoing operations (Table 2), some with multiple fabs and multiple outside partners, plus at least two new start-ups announced for coming years (Advanced Semiconductor Manufacturing and Sincerity Semiconductor Manufacturing).
In a rough comparison, new fab capacity in China, even including Motorola and Hua Hong NEC in 1998, is coming on stream at one fifth the rate of new capacity in Taiwan. Another comparison is that while both countries started to develop semiconductor industries at about the same time, Taiwan caught up to the world`s leading edge by 1994. China is just reaching this level in 1998 with the 8-in. wafer, 0.35-?m geometry Hua Hong NEC fab.
There are six significant reasons behind the different rates of acceleration in the comparisons above:
China is just beginning to get rid of its inefficient, low-productivity system created by the former planned economy.
China still cannot resist the temptation of central planning for the development of its semiconductor industry.
Initially, Western nations would not sell China their most advanced process tools.
Almost from the outset, Taiwan adopted a strategy that made many of the world`s semiconductor companies de facto partners.
The concurrent emergence of Taiwan as one of world`s major producers of PCs and related products gave it a huge local market for its semiconductors, lessening the immediate need for Taiwanese fabs to compete globally.
Taiwan`s stock market has historically been supportive of high technology.
China`s electronics industry
An article in China Daily on Jan. 14, 1998, indicated that although there are 3500 electronics companies in China, its information technology industry is highly fragmented, inefficient, and a matter of serious concern. Most of these companies are small, including approximately 100 TV manufacturers, 200 manufacturers of video players, and 200 licensed PC manufacturers. Inaddition, China lacks local availability of key components and products.
At the end of 1995, wholly foreign-owned enterprises accounted for only 0.4% of the 1.7 million work force in China`s electronic industry, but more than 4% of the total revenue for the industry. Foreign-invested joint-venture enterprises accounted for 4.9% of the work force and 18.2% of revenue. On the other hand, state-owned enterprises constituted 80.9% of the work force, but contributed only 52.2% of the revenue. Average productivity of the wholly foreign-owned enterprises, as measured by sales/employee, is in fact 15 times higher than that of state-owned enterprises. Perhaps China`s biggest hurdle is to resolve the problem of too many people doing too little.
 |
Within China`s electronics industry, 33 enterprises under various forms of ownership manufacture semiconductors. Comparing the performance of foreign-invested enterprises with state-owned enterprises clearly shows the lack of competitiveness that plagues the latter: For example, a calculation based on published data shows state-owned Hua Yue Microelectronics at ?76,000 in sales/person and joint-venture Shanghai Belling at ?800,000/person [1]. Other data about these two manufacturers show Shanghai-Belling`s sales at ?426 million with an average unit value of ?15.23 and Hua Yue sales at ?88 million with an average unit value of ?2.75 [1]. Although timely production and sales statistics about China`s semiconductor industry are often difficult to find, it seems reasonable to conclude that state-owned enterprises are at least an order of magnitude behind in productivity, and make less valuable devices.
 |
Hua Yue Microelectronics, owned by China Electronics Corp., is in Shaoxing, south of Shanghai, and manufacturers bipolar devices using 3- and 4-in. wafers with 3- and 5-?m geometries. While it is the second largest state-owned enterprise making ICs, it is not listed in Table 1 of significant semiconductor fabs in China because it has not benefited from the level of technology infusion enjoyed by Hua Jing. Other smaller fabs of Chinese origin are even less competitive than Hua Yue.
Significantly, Hua Yue has been searching for a partner for more than three years. Indeed, finding willing partners is another part of the problem that China`s semiconductor industry faces. (Unverified rumors at press time indicated that Hua Yue may have found an outside partner.)
China`s partnering efforts
In a recent government reorganization led by Premier Zhu Rongji, China`s Ministry of Electronics Industry has been merged into the Ministry of Information Industry, along with the former Ministry of Post and Telecommunication. Those electronics companies not strong enough to "go public" in either the Hong Kong or China stock exchanges and unable to find a willing foreign partner have been put under the umbrella of China Electronics Corp. (CEC), a holding company with distinct Chinese characteristics.
CEC`s mandate is to reorganize and consolidate-and if necessary shut down-the many disparate entities, in order to create viable enterprises. At this juncture, China`s electronic industry is very much in transition.
The partnering process in China, irrespective of industry, shows the natural inclination of the country`s central planners to leapfrog to the latest and best technology, hold foreign partners` "feet to the fire" on achieving technical targets, and insist that partners export a portion of their output as China`s investment payback. Striking an alliance with a foreign partner is an inherently complicated process, made more difficult when expectations of Chinese officials far exceed what Western companies find acceptable. This gap is particularly obvious in the case of semiconductors because the technology is tightly held and evolving rapidly, and the investment needed to play in this arena is breathtakingly large. Under these conditions, the potential reward frequently falls short of the risk from a Western point of view.
IBM, Rockwell, Siemens, and Texas Instruments (TI) are among companies that looked hard at the Hua Hong deal but ultimately bowed out because of irreconcilable differences. The Chinese side wants control of decisions on which products to make, but insists that the foreign partner give a technology guarantee and a commitment to export a fixed percentage of output.
NEC Electronics is now dealing with the dilemma in the Shanghai Hua Hong joint venture. The Chinese side of the venture wants to make ASICs, while NEC thinks that increasing the work force by starting with memory chips is a more prudent course. Furthermore, NEC privately complains that there is really no accurate read on China`s current market for semiconductors. Compared to other commodities, ICs are easy to sneak into China-escaping the attention of customs-and thus accurate statistics on China`s consumption are not available to serve as a reliable guide to the true nature of ASIC demand.
Certainly there is a lesson in China`s 1994 state-funded Project 908. Instead of a joint venture, AT&T agreed to a technology transfer that included worker training and instructions on chip design. Hua Jing then went on to establish design centers around China that have not generated much business. Apparently the Hua Jing design staff has not yet learned the importance of working closely with customers and coming up with designs that suit a customer`s requirements. Along with the planners at the Ministry of Electronics Industry, they do not understand that a semiconductor manufacturer cannot make "standard" ASICs and then find customers, a business practice more familiar to those in more traditional industries but contrary to the nature of semiconductor manufacturing.
Comparison with Taiwan
In 1996, Taiwan`s total IC production was worth $5.4 billion, roughly ten times more than China`s. Today`s success began in the 1980s when, to alleviate the periodically agonizing shortage of DRAMs, Acer-Taiwan`s leading PC maker-struck a deal with TI to build the first DRAM fab in Taiwan by offering attractive terms for the joint venture. At about the same time, Taiwan`s government enticed Dr. Morris Chang to Taiwan to head TSMC, a joint venture foundry with Philips Semiconductor holding a minority stake. Chang was a former senior executive of TI`s semiconductor operation. He was able to recruit a group of American and Taiwanese expatriates to the venture.
The really significant difference compared to semiconductor industry evolution in China is that the strategy of both operations was to concentrate on mastering and improving the manufacturing process with the help of foreign partners and customers. They did not necessarily overtake Japan and Korea on yield and labor productivity, but found ways to maximize use of manufacturing equipment and thus reduced the unit cost of capital.
In retrospect, the TSMC business model-to concentrate solely on being a foundry for the international semiconductor community-was pivotal in keeping Taiwan up-to-date in technology and know-how. Virtually all the fabless semiconductor companies and even some companies with their own fabs are willing to do business with a foundry service that publicly renounces any intention of competing with their customers by developing its own proprietary products. Thus, helping TSMC by providing advice and know-how to fine-tune the manufacturing process and improve yields became fully compatible with the customers` self-interest. As a result, beginning in 1984 with 3.0-?m technology, when the world standard was ~1.8 ?m, Taiwan caught up to the world standard at 0.5-?m in 1994.
Sales for TSMC in 1997 were projected to reach $1.38 billion with a net income of $547.7 million. In 1996 the company had net income of $462.7 million on sales of $1.24 billion. Gross margin was over 40%. The TSMC model became the dominant business model in Taiwan.
UMC was the first merchant semiconductor company in Taiwan. The company has now adopted the TSMC model by forming several new foundries. But UMC took the model one step further by asking selected customers to join as investors in these new fabs in exchange for guaranteed allocation of production capacity. Virtually all the newly announced fabs have stated the intention of becoming foundries for others.
Gradually, Taiwan began to develop its own technology base. Consider, for example, that in 1992 UMC received only eight patents, but by 1992 UMC filed and received 447 domestic and foreign patents-a giant leap in just four years. The value of this level of technology is that UMC has been able to get equity for its technology in joint ventures.
Taiwan can now claim more than 20 semiconductor wafer fabs, 72 fabless design houses, three wafer suppliers, 23 assembly services, 11 testing and burn-in services, two maskmakers, six chemical suppliers, and nine leadframe suppliers. While Taiwan`s IC revenue still accounts for less than 5% of the world total, it is the fastest-growing because IC imports still exceed domestic production and are driven in part by Taiwan`s prowess in PC manufacturing.
Taiwan`s PC industry
Thanks to Western high-tech investments in the 1960s from such companies as General Instruments, Philips, RCA, and TI, a plethora of printed circuit board stuffing and related contract manufacturing services came into being to serve and support foreign investments. Because of the competitive nature intrinsic to OEM manufacturing, the participants had to be extremely cost conscious, while maintaining satisfactory quality. The survivors quickly became adept at preparing cost-plus quotations as a function of the desired number of quality "sigmas" in a customer`s product specification. In the 1980s, shortly after the dawning of the PC revolution, Taiwan was ready to fill this need when American PC companies began to look for lower-cost manufacturing.
Since then, Taiwan has emerged as a major source of PCs and the leading source of PC related hardware and peripherals (Table 3). Taiwan`s key to success lies in its start in contract manufacturing with thin margins, and its need to find ways to sharpen manufacturing processes continuously. According to 1996 data, multinational companies such as H-P, IBM, and Intel today make $12 billion worth of OEM purchases from Taiwan annually. Many of these purchases, along with Taiwan exports under their own brands, contain ICs. So far, the semiconductor demand from Taiwan`s PC industry continues to exceed by far the local capability to supply it, thus creating the drive for new fab capacity even in the face of a somewhat gloomy world market that is not looking for new capacity.
The reverse brain drain
One very significant key to Taiwan`s success has been the so-called reverse brain drain, the return of Taiwanese expatriates to staff and run the country`s wafer fabs. According to Robert Tsao, Chairman of United Microelectronics, 82 companies in the Hsinchu Science Park, or about 40% of the total, have been started by entrepreneurs returning from the US [2]. Thanks to its strong capital market, Taiwan has been able to offer returning entrepreneurs stock option packages that are more attractive than a career in the US.
This trend, which began in the mid-1980s when Taiwan`s government created and modeled Hsinchu Science Park after Silicon Valley and encouraged the formation of venture capital, is now well known. For Taiwan`s semiconductor industry, early ventures such as TI-Acer and TSMC provided the leakage of personnel needed to seed other ventures and complement the returning Taiwan expatriates. The voracious demand for ICs from the PC industry continues to driveinvestments toward new fabs and attract the needed skilled personnel to Taiwan. Now, even with as many as 20 fabs, Taiwan is still a major net importer of semiconductors.
Can China follow this model?
China`s PC sales in 1997 reached nearly 3 million units, and China has become the second largest market in Asia next to Japan. China is also the world`s second-largest consumer of cellular phones and a major market for pagers, smart cards, and consumer electronics. Since its economy is still developing rapidly, the need for ICs is vast and will continue to grow.
Taiwan`s PC makers have been moving their operations offshore to reduce cost, and China is a favorite destination. This adds interest to developing a world-class semiconductor industry there. Thus, China represents a huge long-term demand for semiconductors, which holds multinationals` unwavering interest.
Taiwan has sent more than 350,000 students overseas for education and training. The mainland has sent slightly fewer and is expected to overtake Taiwan soon. However, China has to accomplish more structural reform before financial incentive packages can be created to attract a reverse brain drain similar to Taiwan`s. Its capital market needs more discipline, and listed companies need to offer more transparency to investors. Its currency needs to be fully convertible to encourage the participation of foreign institutional investors. The government also needs to take its hands off this industry and let market forces decide who, when, where, how much, and other basic issues.
The Beijing government is committed to letting market forces rule its economy. Historically, the government has not been successful in steering this industry to a world-class level. Eventually, the central planners will have to let go.
At present, China`s stock market does not offer the path to liquidity that professional venture capital investors require, and the formidable capital investment needed for a fab cannot yet be made without the heavy-handed presence of the government. When all the conditions mentioned above are met, China can be expected to become another semiconductor powerhouse.
 |
Motorola`s new fab near Beijing, using 0.35-?m processing, could be a sign of a loosening attitude by export-control officials in Washington. If so, US semiconductor equipment manufacturers will finally be able to compete on a level playing field in China. Other Western nations have long recognized the end of the Cold War and departed from US export-restriction policy by removing the limits on equipment for sale into China. NEC`s early partnerships with Shougang and Hua Hong are clear indications of a very different attitude toward China on the part of the Japanese government. Of course, NEC and Japanese equipment makers do not have to deal with the US Congress.
The economic pressure is simply too great for China not to join the electronics elite. Tony Liu, vice president of ASMC, reviewed China`s prospects in a recent presentation [2]. He placed China`s IC market today at $5 billion/year, of which domestic production accounted for only $400 million. By 2000, the market will be at $15 billion and only 20% of the market will be satisfied by domestic production. This suggests ample room for new entities. Liu worked in Silicon Valley before joining ASMC in Shanghai. In a way, he is a harbinger of China`s future in the semiconductor industry.
References
1. 1996 China Electronics Industry Yearbook (published in Chinese).
2. Presented May 1997 at the annual conference sponsored by the Monte Jade Science and Technology Association in Silicon Valley.
GEORGE KOO is the founder of International Strategic Alliances and has been advising American companies on doing business in Asia since 1978. He is an instructor for SEMI`s course on China. International Strategic Alliances, 1265 Montecito Ave., Suite 109, Mountain View, CA 94043; ph 650/969-1671, fax 650/969-1673, e-mail [email protected].