TSMC Plays Solo

For 20 years TSMC has grown at a rapid and robust compound annual growth rate of 43% and in 2006 its yearly operating revenue broke through to NT$310 billion, close to US$10 billion. Market value for the company was NT$1.7 trillion (US$52 billion), 290 times its original value.

It was back in 1974 when six academics and government officials, including Li Kwoh-ting and Sun Yun-suan, met over a simple breakfast at a soya bean milk shop on Taipei's Linsen North Rd. and set the pace for the development of Taiwan's semiconductor industry. In 1985 Morris Chang, former group vice president at Texas Instruments and CEO at General Instrument Corporation, was recruited by the Industrial Technology Research Institute as its president. TSMC was born in 1987 under his leadership. Spurred on by TSMC's constant breaking of the mold and its bold innovations, the Taiwan semiconductor industry took off and from this point on the rules of the game for the industry worldwide were rewritten.

To celebrate its 20th anniversary, TSMC invited close to 1,000 customers and supplier reps from around the globe at the end of April this year. In addition to enjoying an exclusively sponsored performance of Beethoven's Symphony No. 9 ("Choral") by the London Symphony Orchestra, guests could attend a "technology forum" on the development of the semiconductor industry over the next 20 years.

Silver-haired TSMC chairman, Morris Chang, rewrote the rules of the game for the global semiconductor industry with his inimitable, daring innovations and management ability.

Innovation the key to success

The tough role of keynote speaker was filled by Harvard's Professor Michael E. Porter. A well-known business strategist, he is also an independent supervisor at TSMC. But the title of his address, "Redefining Health Care: Implications for Taiwan," seemed to have nothing to do with semiconductors or the IT industry.

In fact, Morris Chang had taken great pains to arrange this address. He felt Professor Porter, a longtime researcher on competition theory, had in recent years turned his attention to applied medical systems because he saw the costs of medical systems worldwide steadily rising, while the effectiveness of those systems deteriorated year after year. This state of affairs would inevitably affect the quality of life for the aging population of the future. In the US, for example, of the 139 heart transplant hospitals, why was the post-operative survival rate 0% in some and 100% in others? Moreover, commonly shared problems included quotas on resource allocation, deterioration of service, and continued high rates of medical error.

The overall reason: despite unprecedented modern medical technology, organization and thinking remain stuck in the 19th century. Professor Porter believes that in the future medical institutions should specialize in terms of specific diseases and that the patient should be the focus of attention. But this major reform requires the voluntary participation of top US corporations. This means first changing their habit of only looking at low-cost healthcare plans for their employees. Porter stressed the importance of value creation. If this new viewpoint of Professor Porter's could successfully be promoted, it would revolutionize global medical care. This innovative thinking jibes perfectly with the values that TSMC is aiming for.

"Integrity, commitment, innovation, customer partnership--this is our formula for success," says Chairman Morris Chang as he points out the four basic values that have guided TSMC over the last 20 years.

Sales breakdown by technology

Looking for a niche--service

When Morris Chang decided to enter the market with the innovative strategy of being a purely specialized contract manufacturer, almost everyone disagreed with him. Before, after global IC design companies had designed integrated circuits for various uses and of various capabilities, if you were not able to set up your own manufacturing plant you had to let contracts out to major integrated device manufacturers (IDM) for production. But because the IDMs themselves included all elements of the IC business, from design, manufacture, packaging and testing to brand sales, they regarded this type of contract work as just a secondary business.

As a result, TSMC saw a businesses opportunity and thus recommended itself to IC design companies. Better technology, lower defect rates, and more precise delivery times together with its production capacity all permitted a significant lowering of costs, enabling TSMC to offer its products at favorable prices.

Also when IC design companies made an order with an IDM they were most anxious about the danger of their designs being copied. Because TSMC did not develop or manufacture their own products but only provided a contract manufacturing service pure and simple, the question of trust uppermost in the client's mind was thus solved. Although the pure contract manufacturing model was soon copied by industry rivals at home and abroad, no one was able to make inroads against the success of TSMC.

"We are first and foremost a service enterprise, not a manufacturing one," is the article of faith that Morris Chang never forgets. After firmly establishing itself on the new path of specialized contract manufacturing, TSMC was not content with merely attaining technological excellence, but tasked itself with transformation from a manufacturing enterprise that concentrates its efforts on expanding its production capacity, into an enterprise that provides clients with the best and most comprehensive service.

The essence of these two mindsets differs: manufacturing aims at low cost, service aims at high value. In the past, for example, TSMC was famous for its exacting demands, such as operating at full production capacity to avoid wasting money by having equipment standing idle. It insisted on operating at 100% of nominal daily production capacity, and sometimes pushed output to as high as 120%. But following market changes and differing demands from customers, today TSMC is willing to bring that rate down to 85-90% to give customers room to make changes in the fabrication process, or to hold additional capacity free to meet unscheduled customer demand.

Virtual fabrication

In the age of 10x competition where seconds mean winning or losing, TSMC came up with the concept ten years ago of a "virtual fab" to assist customers smoothly to enter into the advanced mass production process. It also continued to develop various "e-foundry" measures to raise the transparency of the production process, providing customers more, better and faster service, thus also steadily increasing its own value-added quality.

In 2000, for example, TSMC developed the Design Sphere Access design platform and then began working with analogous Electronic Design Automation tool companies to shorten the period needed for customers to overcome technological barriers. In addition, it improved customer ordering, change-orders and production processes. On the web TSMC set up an electronic library with a common design module for specific customers so they need not do a design from scratch but could shorten the time spent on making the design and bringing the product to market. At the same time TSMC generously freed up its own patents for its customers' use, solving the vexing patent infringement problem customers faced.

On last year's list of "Taiwan's Top 100 Patent Holders" TSMC was the new winner with a total of 441 patents. Because there is a high degree of commonality of design elements in different IC products TSMC uses "technology modularization" to make it convenient for customers to rapidly put together different designs for different demands. The "virtual fab" provides customers with a one-stop shop for IC design. TSMC's computer system can automatically run a verification and respond with all the information pertaining to product delivery. Even at great distances, customers can get a comprehensive understanding and exercise oversight of the progress of their contracted product via the computer. There are even specially assigned staff to handle customer complaints so the customer feels the whole experience is as convenient as if the wafer plant were his own.

To serve the customer, TSMC also vertically integrates the latter stages of testing, packaging, packing and delivery to offer so-called "one-stop turnkey service." The customer need only provide the production order and then TSMC gets things ready and takes care of delivering the wafers to the downstream customer. This greatly reduces costs and shortens delivery time. For manufacturers under the pressure of time-sensitive competition in an era of marginal profits, this an enticement they can't refuse.

TSMC's Global Locations

Working with partners

TSMC, which attaches great importance to partner relationships, has joined with third-party intellectual property providers (i.e., CPU and the associated intellectual property) and over 100 domestic design teams to support IC design companies. TSMC has enabled these companies, originally small scale and local in nature, to compete with IDM companies a hundred times their size and to become competitive in the global industry. Because TSMC and these companies regard each other as good and trusted partners, TSMC has been able to enjoy high profits as the companies have grown. "Our partners' success is our success! We welcome our partners to be even more successful than us." At the 20th anniversary celebration Morris Chang did not forget to point out the beneficial result of jointly creating with partners and mutually sharing in the profit.

The most important value for TSMC is "innovation." Morris Chang expects his colleagues to make innovation and change their normal work pattern. He has even set up an Innovation Award and Customer Partnering Award. Each year he distributes award money in line with profits and the Innovation Award program has given out a record NT$10 million.

But innovation alone cannot guarantee success. You need cross-departmental managers who have a comprehensive understanding of operations to integrate and utilize resources. In the past, TSMC only had three people at senior levels, including the chairman of the board and the president, who were able to do this kind of integrated policymaking. But in the last three years some ten "strategic initiatives" based on market segments have been introduced in order pass on these capabilities to other senior managers. This pro-active policymaking mechanism is also an industry first.

Pushing the nanoscale envelope

The service projects discussed above constitute TSMC's competitive advantage but as a specialized contract manufacturer, it is the manufacturing technology that represents their "basic skill." With the dual demands of both increasing wafer sizes (from six, to eight, to 12 and now to 18 inches) and decreasing space between circuits (from 1.5 microns to 45 nanometers), semiconductor factories around the world have made enormous investments in a never-ending technological race.

For TSMC 2000 was an important year of manufacturing breakthroughs. Before 2000, TSMC was at the technology catch-up stage, but in this year, after declining to work with IBM in developing a project for a 0.13-micron copper process, TSMC decided to move in the direction of developing its own technology and from that point on opened up a gap between itself and its competitors.

The 0.13-micron generation of semiconductors is the most technologically difficult to date. Almost every major manufacturer devotes tremendous efforts to R&D but there has been much delay in reaching the mass-production stage. This is because of two new technologies, the copper damascene process and low-k dielectric materials. They are the new challenges for the development of semiconductor fabrication.

Although at the time IBM was the company that had been developing copper-based chip technologies the longest, TSMC thought it would take at least another year for the technology to be perfected and transferred. When this happened, TSMC would probably just be a No. 2 in the industry, always struggling to keep up with the industry leader.

After TSMC made the decision to develop this technology itself, in order to speed up the work, the company not only recruited a team that included component expert Carlos Diaz, lithography specialist Burn J. Lin and Denny D. Deng, author of basic research on bipolar transistors, as well as American semiconductor R&D experts, it developed a common fabrication process and, focusing on the individual customer, created tailor-made applied processes. This strategy led to the rapid rise of TSMC's R&D capacity to become the second company in the world that successfully moved to 0.13-micron mass production, just behind Intel.

Competition never ends

But a momentary success is not a permanent guarantee. TSMC's top competitor, United Microelectronics Corporation (UMC) is an example. Although UMC gave up on the 0.13-micron (130-nanometer) process, it has taken the fight to a higher level, the 90-nanometer process. At the same time Chartered Semiconductor of Singapore and IBM have begun cooperation and have gone even further to develop a 65-nanometer process. Added to this is a newly emerging Taiwanese company, Semiconductor Manufacturing International Corporation (SMIC), and even in Germany, Ireland and Malaysia dark horses are stepping up--all possible competitors for TSMC.

"Competition will not fade in this business, it will get more intense," says Dr. Rick Tsai, CEO of TSMC. Fortunately, following its successful experience with 0.13-micron development, TSMC in 2004 was the first to develop "immersion lithography" based on the ideas of Burn J. Lin, one of Morris Chang's star employees. It is estimated that the 45-nanometer mass-production stage will formally be reached this September and people are able to look forward to developing a 32-nanometer process. In terms of technology, TSMC can now hold its own with US-based Intel and other major companies.

Breaking through bottlenecks

At the 20th-anniversary summit forum, Morris Chang predicted that in the coming ten years the growth rate of the global semiconductor industry would slow down to only 5-10% per year. He said that without a "breakthrough" material in the offing, the wafer industry would hit a technological bottleneck.

Facing the challenge of slower overall industry growth, Chang thinks that IC design companies and contract wafer manufacturers must move from the current "relay-race" model of cooperation between upstream and downstream entities to an intimate relationship over the whole course, setting up even deeper and closer partnerships.

Further, there is the increasing investment in advanced semiconductor processes. During the era of the eight-inch wafer, it cost only US$1.2 billion to build a factory. When the scale went to the 12-inch wafer, the price of construction jumped to US$3 billion. For the next generation, that of the 18-inch wafer, costs will rise to some US$12-15 billion. In the future the barriers to entry to the contract wafer manufacturing industry will be more difficult to overcome and the number of those entering the field will decline.

Another aspect is that product design costs for IC design companies have already greatly increased from less than US$10 million at the 0.13-micron level to more than US$45 million at the 65-nanometer level. This increase in cost may force many large vendors to abandon their own independent design work and to cooperate on design with TSMC and other contract wafer manufacturers. This also represents an opportunity.

For the future, Morris Chang thinks that while in the past TSMC concentrated on contract manufacturing of logic chips, in the future it will produce a diversified product line that will include embedded DRAM as well as flash memory, niche memory, mixed-signal chips, high-pressure process technology and image sensors, all virgin markets for TSMC. Chang believes that just by strengthening the partner relationship with customers, in the next ten years TSMC's compound annual revenue growth rate will beat the industry average of 5-10%.

At the end of the 20th anniversary forum, CEO Rick Tsai summed it all up: IC products are uniquely capable of being used in every branch of industry and totally changing the lives of all mankind. "We must continue to advance to higher and higher levels and move forward to the top of the value chain, in step with everyone else."

What is certain is that the future semiconductor world will rest on the three legs of a tripod: Intel for CPUs, Samsung for memory and TSMC for contract wafer fabrication. Will the next movement in this symphony see TSMC capable of overawing their competitors and taking the field? It's something worth looking forward to.