home Today's News Magazine Archives Vendor Guide 2001 Search isdmag.com

Editorial
Today's News
News Archives
On-line Articles
Current Issue
Magazine Archives
Subscribe to ISD


Directories:
Vendor Guide 2001
Advertiser Index
Event Calendar


Resources:
Resources and Seminars
Special Sections


Information:
2001 Media Kit
About isdmag.com
Writers Wanted!
Search isdmag.com
Contact Us





Focus Report: Design Libraries

By Peggy Aycinena
Posted  03/29/01, 03:22:07 PM EDT

After the leak was out-and the hype began-on inventor Dean Kamen's latest invention, Ginger or IT, the media was saturated with speculation on what the so-called revolutionary device might be. The details were fuzzy: It will cost under $2000 and make Kamen wealthier than Bill Gates. Some thought it might be a motorized scooter. Others pondered a jetpack. Still others wondered if it might be a non-polluting source of energy. One thing is certain-it's not a fab.

The semiconductor industry has evolved profoundly in the past decade-most notably with foundries and fabless semiconductor companies. Fortuitously, the two are linked. The foundries offer the capacity and fabless companies provide the demand. The success of one relies on the other. Each relies on its core competency. The advent of the fabless company and pure-play foundry, however, drew a lot of skepticism from the rest of the semiconductor industry.

The most famous critique came from Jerry Sanders III, CEO of AMD (Sunnyvale, CA), when he said several years ago "only real men have fabs." Nonetheless, by the late 1990s, fabless semiconductor companies, such as Xilinx (San Jose, CA) and Broadcom (Irvine, CA), saw their market caps reach the echelons of medium-sized integrated device manufacturers (IDMs).

"Over the past several years a change in the landscape has occurred, where foundries can now compete with large IDMs," said Bing Yeh, president and CEO of Silicon Storage Technology (Sunnyvale, CA), a Flash memory company which has licensed its Superflash technology to Taiwan Semiconductor Manufacturing Company (TSMC - Hsinchu, Taiwan). "The fabless companies can thrive in a market with companies which have fabs."

The fabless business model has proved to be a successful one. In December of 2000, the Fabless Semiconductor Association (FSA) honored thirty companies that doubled in revenue or net income in the previous year. FSA has a membership of 300 companies, including foundries, fabless semiconductor companies, and EDA vendors.

Dataquest forecasts a 25 percent compound annual growth rate for foundry services from 1999 to 2004. In addition, Dataquest predicts that foundries will produce half of all semiconductors by the year 2012. Today, the industry produces about 15 percent of all semiconductors. With an eye on the skyrocketing costs of building a state-of-the-art fab, the IDMs noticed the success. Therefore, in regards to foundries, the IDMs have reshaped their plans to include outsourcing as well.

"Even the IDMs with fabs need to think hard before investing in the cost of a new fab," said Peter Richmond, business development director of the system IC business unit at Toshiba America (New York, NY). Indeed, a trend that has developed in the industry is for IDMs to use the services of thefoundries for some of their mature products. "We recognize that today's foundries can offer expertise and cost benefits. Toshiba couldn't do without them."

Richmond sees the foundry industry as both partner and, to a lesser extent, competitor. For the first year of life for a Toshiba product, Toshiba uses only its fab. "We don't literally compete with foundries," Richmond said. "We tend to be a value-added service, such as embedded DRAM. Companies come to us because they can't get that from TSMC."

From criticism to praise

Early on, critics charged that the foundries lagged behind in process technology, offering processes, which were two to three generations behind the leading IDMs. Now, foundries deliver leading-edge technology. "In the past, foundries were trailing the IDMs," said Jim Ballingall, vice president of worldwide marketing for UMC (Hsinchu, Taiwan). "Now we are ahead of all IDMs."

"There is no technology loss when choosing a foundry over an IDM," said Sheldon Wu, senior director of field technical support for TSMC North America.

UMC has engaged in a joint-development program with IBM (East Fishkill, NY) and Infineon (Munich, Germany) to introduce Worldlogic standard 0.13-micron technology. In November of 2000, IBM, Infineon, and UMC announced that they have started building chips with what they call the most advanced 0.13-micron foundry process technology currently available. Ballingall also said that foundries lead the way in construction of 300-mm wafer fabs. The company has announced plans for three 300-mm fabs. UMC has delivered silicon from its 300-mm Trecenti fab, a joint venture with Hitachi in Japan, at 0.18-micron process technology. The third 300-mm fab will be built in Singapore, with equipment scheduled to move in by the third quarter 2002. The fab will cost UMC $3.6 billion and will implement the 0.13-micron and 0.10-micron Worldlogic platform.

In December 2000, TSMC reported that it delivered the industry's first 300-mm wafers. TSMC is currently constructing two 300-mm manufacturing facilities, Fab 12 in Hsinchu and Fab 14 in Tainan, which are expected to start production in the fourth quarter of 2001 and in early 2002, respectively. In addition, TSMC's third 300-mm fab is expected to start construction in the first quarter of 2001.

Partnerships and alliances

Establishing partnerships with systems companies, design services houses, EDA vendors, library suppliers, and intellectual property (IP) providers has spurred the success of the foundries as well. Last year, UMC instituted its ASIC Plus program, where partners manage the entire chip development process for the customers who don't have the needed infrastructure or expertise.

Also last year, TSMC unveiled its Design Service Alliance. The organization encompasses three service areas that make up the IC design process: third-party libraries, IP, and IC design implementation services to help designers deliver system on a chip. TSMC is also involved in a joint-venture company, called Systems-on-Silicon Manufacturing Company (SSMC), with Philips Semiconductor (Eindhoven, The Netherlands) and Singapore Economic Development Board Investments (EDBI), which is building a fab in Singapore.

In January of this year, Conexant (Newport Beach, CA) and TSMC announced that the two companies have entered into a cross-licensing wafer supply and technology agreement. Under this agreement, TSMC will license advanced specialty RF process technology IP from Conexant to be used for bipolar and silicon germanium bipolar complementary metal oxide semiconductor (SiGe, BiCMOS) applications, and will supply foundry capacity for these technologies to Conexant.

For its part, Chartered Semiconductor Manufacturing (Singapore) has a joint venture R&D arrangement with Agere Systems (formerly Lucent Microelectronics of Allentown, PA), which will co-develop advanced process technologies. The company is also working with Ericsson Microelectronics (Stockholm) to develop RF-CMOS and BiCMOS technologies to support wireless communications applications, including Bluetooth.

Chartered is focusing on system-level technologies in the communications markets, said Kevin Meyer, vice president of worldwide business development for Chartered. Although continuing to develop leading-edge technologies, Meyer said that the industry's insatiable demand on transitioning to deeper submicron process technologies has given way to a strategic supply issue. Meyer said that every company needs an assured strategic supply built upon a foundry relationship.

The services provided by foundries also help companies to differentiate from the competition. "Methodologies play a big role in the semiconductor industry," said John Wright, manager of digital technology development at AMI Semiconductor (San Diego, CA), where according to FSA, 7 percent of the fabless industry goes for foundry services. "AMI really shines with its methodologies, compared against some of the bigger guys."

The success of the foundry model has inspired some newcomers into the competition as well. In Malaysia, 1st Silicon and Silterra aim to give the big three-TSMC, UMC, and Chartered-a run for their money. 1st Silicon is ramping up at its 200-mm wafer fab, which will implement 0.25-micron process technology. It will be able to process 30,000 wafers per month. Its second fab will use 0.18-micron technology on 200-mm wafers.

"You don't have to have all your eggs in one basket, meaning California and Taiwan," said Claudio Loddo, CEO of 1st Silicon. "Our industry always has a surprise ahead of us, no matter how well we plan," he said, referring to the power crisis in California. "The solution is to be able to obtain products all the time from anywhere in the world." Loddo said that Malaysia has the infrastructure in place-reliable and available power and an abundance of water.

Like their now successful predecessors, some skeptics give the start-ups little chance of success. Critics see the cost barrier of building a leading-edge fab to be formidable and they doubt whether the start-ups can ensure the demand so that their fabs are always running -a key to success.

"I think it's very hard for new (foundry) companies to compete against established foundries," said Ballingall. "The capital barrier is one factor. To be competitive, you need a 300-mm fab and you need a strong technology partner."

Loddo said he believes that there is a market for 1st Silicon's services. The company aims to establish long-term partnerships and co-development agreements. Its first technology partner is Sharp.

"The opportunity is still there," Loddo said. "The war is declared again by the big boys to capture customers ignored during the good days."

Slowdown

Loddo said that the success of the foundry industry could be sensed in the atmosphere at Semicon West.

At Semicon West 1998, it was a depressed environment. In 1999, there was a bit of a sparkle. By 2000, people were celebrating the return of the heady days, he said.

The year 2000 began with a crush of demand for foundry services. TSMC ran at full tilt through much of the first half of the year, effectively eclipsing theoretical capacity utilization, said Dan Sheldon, a public relations manager for the company.

By the end of the year, however, signs began to appear that a slowdown in the semiconductor industry was imminent. It caught a lot of people by surprise.

"The inventory issue is affecting end-market demand," Meyer said. "The foundry industry won't be immune. We expect it to pick up in the third or fourth quarter."

Market conditions may have forced the foundries to rethink their strategy. In late January 2001, Chartered announced plans to cut back capacity to counter weaker growth in the communications market. In addition, Meyer said that the company's next fab, Fab #7, which was originally slated for 200-mm wafer production, will be outfitted for 300-mm wafer production-giving it a much longer life.

In a report by Reuters, Salomon Smith Barney analyst Andrew Lu said in a research note, he expected TSMC's capacity utilization rate to fall to 85 percent in the first quarter of this year and 79 percent in the second quarter. Although not nearing the capacity utilization rates of the first half of 2000, TSMC said demand is still very strong for 2001.

"The semiconductor business has softened dramatically," said Ken Hurley, U.S. president of Nanya Technologies (San Jose, CA), a supplier of DRAM. "This has caused companies to be conservative on fab construction, which we think will contribuin the next couple of years. Nanya will continue to invest in new fabs."

How does this all affect the start-ups? Loddo said he believes that the slowdown may prove to be a benefit to his foundry. "Everybody, big or small, feels it," Loddo said. "We are fortunate because we are in ramp-up and qualification. [The decrease in demand for equipment] allows us to get the equipment delivered."

Loddo said that the success of the foundry industry could be sensed in the atmosphere at Semicon West.

At Semicon West 1998, it was a depressed environment. In 1999, there was a bit of a sparkle. By 2000, people were celebrating the return of the heady days, he said.

The year 2000 began with a crush of demand for foundry services. TSMC ran at full tilt through much of the first half of the year, effectively eclipsing theoretical capacity utilization, said Dan Sheldon, a public relations manager for the company.

By the end of the year, however, signs began to appear that a slowdown in the semiconductor industry was imminent. It caught a lot of people by surprise.

"The inventory issue is affecting end-market demand," Meyer said. "The foundry industry won't be immune. We expect it to pick up in the third or fourth quarter."

Market conditions may have forced the foundries to rethink their strategy. In late January 2001, Chartered announced plans to cut back capacity to counter weaker growth in the communications market. In addition, Meyer said that the company's next fab, Fab #7, which was originally slated for 200-mm wafer production, will be outfitted for 300-mm wafer production-giving it a much longer life.

In a report by Reuters, Salomon Smith Barney analyst Andrew Lu said in a research note, he expected TSMC's capacity utilization rate to fall to 85 percent in the first quarter of this year and 79 percent in the second quarter. Although not nearing the capacity utilization rates of the first half of 2000, TSMC said demand is still very strong for 2001.

"The semiconductor business has softened dramatically," said Ken Hurley, U.S. president of Nanya Technologies (San Jose, CA), a supplier of DRAM. "This has caused companies to be conservative on fab construction, which we think will contribute to a supply problem in the next couple of years. Nanya will continue to invest in new fabs."

How does this all affect the start-ups? Loddo said he believes that the slowdown may prove to be a benefit to his foundry. "Everybody, big or small, feels it," Loddo said. "We are fortunate because we are in ramp-up and qualification. [The decrease in demand for equipment] allows us to get the equipment delivered."

Loddo said that the success of the foundry industry could be sensed in the atmosphere at Semicon West.

At Semicon West 1998, it was a depressed environment. In 1999, there was a bit of a sparkle. By 2000, people were celebrating the return of the heady days, he said.

The year 2000 began with a crush of demand for foundry services. TSMC ran at full tilt through much of the first half of the year, effectively eclipsing theoretical capacity utilization, said Dan Sheldon, a public relations manager for the company.

By the end of the year, however, signs began to appear that a slowdown in the semiconductor industry was imminent. It caught a lot of people by surprise.

"The inventory issue is affecting end-market demand," Meyer said. "The foundry industry won't be immune. We expect it to pick up in the third or fourth quarter."

Market conditions may have forced the foundries to rethink their strategy. In late January 2001, Chartered announced plans to cut back capacity to counter weaker growth in the communications market. In addition, Meyer said that the company's next fab, Fab #7, which was originally slated for 200-mm wafer production, will be outfitted for 300-mm wafer production-giving it a much longer life.

In a report by Reuters, Salomon Smith Barney analyst Andrew Lu said in a research note, he expected TSMC's capacity utilization rate to fall to 85 percent in the first quarter of this year and 79 percent in the second quarter. Although not nearing the capacity utilization rates of the first half of 2000, TSMC said demand is still very strong for 2001.

"The semiconductor business has softened dramatically," said Ken Hurley, U.S. president of Nanya Technologies (San Jose, CA), a supplier of DRAM. "This has caused companies to be conservative on fab construction, which we think will contribute to a supply problem in the next couple of years. Nanya will continue to invest in new fabs."

How does this all affect the start-ups? Loddo said he believes that the slowdown may prove to be a benefit to his foundry. "Everybody, big or small, feels it," Loddo said. "We are fortunate because we are in ramp-up and qualification. [The decrease in demand for equipment] allows us to get the equipment delivered."

Loddo said that the success of the foundry industry could be sensed in the atmosphere at Semicon West.

The year 2000 began with a crush of demand for foundry services. TSMC ran at full tilt through much of the first half of the year, effectively eclipsing theoretical capacity utilization, said Dan Sheldon, a public relations manager for the company.

By the end of the year, however, signs began to appear that a slowdown in the semiconductor industry was imminent. It caught a lot of people by surprise.

"The inventory issue is affecting end-market demand," Meyer said. "The foundry industry won't be immune. We expect it to pick up in the third or fourth quarter."

Market conditions may have forced the foundries to rethink their strategy. In late January 2001, Chartered announced plans to cut back capacity to counter weaker growth in the communications market. In addition, Meyer said that the company's next fab, Fab #7, which was originally slated for 200-mm wafer production, will be outfitted for 300-mm wafer production-giving it a much longer life.

In a report by Reuters, Salomon Smith Barney analyst Andrew Lu said in a research note, he expected TSMC's capacity utilization