On Sunday, April 7th, the Robbinston Historical Society is holding an open house for the family at the Ridge Methodist Church. The entire town has been invited to come for coffee and sweets and bring along something for the family to re-equip their kitchen. Along with the dishes, toaster, and pans, folks will also bring one of their favorite recipes--the family also lost all of its recipes in the fire. My wife says we have never been as close to our neighbors any place we have ever lived as we are here in rural Maine--and that includes San Francisco, Marine del Rey, and La Mirada, Calif., Chicago, Port Washington, N.Y., Manhattan, or Washington, D.C.

Shanghai slated to become
China's new Silicon Valley

Chip makers already are beginning to get nervous about the rapid growth of the semiconductor industry in China, but they may have something to really worry about in another decade.

By 2015, the number of fabs in the Shanghai area alone will more than triple to 20 front ends, according to a new forecast made by a Chinese government ministry. Several of them will be 12-inch wafer fabs.

The six wafer fabs now in production in Shanghai account for more than half of China's IC production capacity, according to Xu Xiaotian Xu, semiconductor analyst for China's Ministry of Information Industry. That output will make a giant leap forward shortly when three new fabs go on stream in Shanghai--plants being built by Grace Semiconductor, Advanced Semiconductor, and Shanghai Belling.

"The China mainland IC industry will grow up within the next three-to-five years," declares Congjie Li, vice general manager of Via Electronics. "The base camp will be centered in Shanghai, with adjoining Jiangsu and Zhejiang provinces becoming the golden triangle for China's IC development," he says.

This growth is powered by the government as well as by strong local demand for consumer products, he says. But it won't all be smooth sailing. "Low design level and lack of IC designers are the bottlenecks for China's development," Li notes.

Growing at the same rate will be the local chip-production equipment and materials businesses. Already, Applied Materials--world's largest semiconductor equipment supplier--is looking for a Chinese partner to help it turn out materials and production equipment for the country's exploding chip manufacturing industry.

And next year, Photronics, a U.S. lithography photomask supplier, plans to build a plant in Shanghai's Zhangjiang Industrial Zone. This facility will develop equipment for 0.13-micron processes and 12-inch fabs in China.

(See April 3 story.)

IBM surprises with
higher ASIC sales . . .

By now, you undoubtedly know the global ASIC market fell off the cliff last year, plummeting nearly 45% from 2000 sales. But what you may not know is that the ASIC market leader zigged while the overall market zagged last year.

IBM Microelectronics was able to gain market share and even post increased revenues. The leading ASIC supplier chalked up $2.7 billion in ASIC revenue in 2001, more than double that of its nearest competitor, Agere Systems, according to Gartner Dataquest.

While most of IBM's ASIC rivals posted double-digit declines, the leader's ASIC revenues grew 1.2%. Now that's nothing short of amazing.

So what happened? IBM is just executing very well. "IBM is showing a focus on the ASIC business from a high level within the company that very few other vendors are showing right now," says Jordan Selburn, analyst at iSuppli. That focus also is evident in its technology, which has helped IBM win a number of high-performance communications designs away from chip rivals in the past few years, he adds.

While a lot of IBM's ASIC revenue comes from inside sales, IBM claims that its non-captive ASIC sales in 2001 exceeded 50% of ASIC revenue.

"IBM continues to bring a lot to the party--superior fab technologies, a superior breadth of IP, and strong industry alliances," points out Jerry Worchel, analyst at In-Stat Group. "This is especially impressive." He says, "when you consider they were late in coming to the party when they first entered the merchant market in the mid 1990s."

"Our customers have shifted toward system-on-a-chip designs with embedded controllers and more memory content based on either SRAM or DRAM," notes Tim Ravey, director of design services at IBM's ASIC Group. "They're trying to integrate a higher level of functionality on a single piece of silicon. And whether the application is based on copper technology, DRAM or SRAM, or our process technologies, we're doing things our competitors can't do, which the market share numbers demonstrate."

. . . as ASIC competitors
shrink from 100 to just 12

The ASIC market has changed big time in recent years. For one thing, there aren't many vendors left in this business. In the '90s, there were more than 100 ASIC suppliers competing in the '90s, but today there are only about a dozen, estimates In-stat analyst Jerry Worchel.

The number of ASIC design starts also has declined dramatically in recent years. And today, designs have become more complex while the cost of ASIC technologies has skyrocketed.

It takes big companies to compete in ASICs today. "There are fewer and fewer competitors at the high end of the ASIC market," says a spokesman for LSI Logic, which tied NEC for third place in Dataquest's ASIC ranking with $1.2 billion in revenues. "The reason is--as the market shifts to smaller feature sizes--there's more pressure for suppliers to offer a full suite of IP, develop flexible and robust design methodologies, and maintain the engineering resources to handle the real challenging designs."

In addition to networking and server applications, IBM is seeing more demand for its high-end technologies in "pervasive computing," such as PDAs, where low power and high integration are especially crucial, says Tim Ravey, director of design services at IBM's ASIC Group.

"We'll continue to proliferate our system-on-chip offerings by taking blocks of IP that are ready to be integrated," Ravey says. "We'll also continue to make investments in IP, which is integral for success in this space, especially for emerging applications."

(See April 3 story.)

An old problem reappears;
testers can't keep up with ICs

It seems like I've been hearing about this problem for decades and I have. IC testers continually have big trouble keeping up with fast-changing chip technology.

Latest to complain is Cisco Systems. A new class of communication chips for the next-generation of routers is pushing the limits for today's IC testers, warns the networking equipment giant, and new advanced chip testing technologies are needed.

Cisco is worried because it develops and uses cutting-edge network processors, switch fabric chips, and application-specific integrated circuits in its routers. Its high-end route for building Internet Protocol backbones, for example, runs at speeds over 10 gigabits per second.

"Routers are pushing technology to the limits," says Cisco engineer Matthias Kamm. "Routers are getting bigger and becoming more powerful, but power and heat are becoming real problems in systems design." He says the test challenges for devices in routers "have gone up dramatically."

As always, the cost of IC testing is a major issue, and this problem has prompted Cisco to use everything from build-in self-test (BIST) to boundary scan. Intel is taking a whack at the problem by pushing a technology called "distributed test." Cisco doesn't test its chips in-house, but uses a test subcontractor, Digital Testing Services.

(See April 3 story.)

Sony, Toshiba join IBM
in major R&D program

Some Japanese chip makers aren't going to wait for any "Japan Inc." solution for next-generation chip designs in consumer products.

In a major R&D effort, IBM, Toshiba, Sony, and Sony Computer Entertainment have expanded existing ties to co-develop advanced process technologies for 0.09- to 0.045-micron chip designs on 300-mm wafer substrates.

Their goal is to move silicon-on-insulator and other advanced semiconductor process technologies into cost-sensitive chips for consumer electronics gear. The companies expect to spend "several hundred million dollars" over the next four years to develop 90-nm (0.09-micron), 65-nm (0.065-micron), and 45-nm (0.045-micron) chips using SOI wafers, copper-metal interconnects, and low-k dielectrics.

As part of the deal, IBM will transfer its SOI technology to Japan's Sony and Toshiba and may become a foundry for the alliance. Toshiba will provide its system-on-a-chip (SoC) and manufacturing expertise and Sony will leverage the processes for its consumer devices.

The idea is for each company to be able to build these advanced chips in their own fabs, says Bijan Davari, vice president at IBM Microelectronics. He also expects this collaboration to lead to the development of high-performance, low-power chips for use in everything from digital consumer products to supercomputers.

IBM already is readying its 90-nm (0.09-micron) process for introduction "by the end of the year," Davari says. The group may end up developing a range of future devices for networking and consumer applications. "The PC is no longer the driving force in semiconductor innovation," declares John Kelly, IBM senior vice president. "Networking and consumer electronics applications are driving the evolution of a new semiconductor industry," he says.

(See April 2 story.)

Applied reportedly gives
reprieve to e-beam unit

Applied Materials reportedly came close to pulling the plug on its electronic-beam reticle mask-writer operation, but has now apparently given its Etec Systems subsidiary a reprieve.

Etec, world's largest supplier of e-beam tools, was in trouble because of product delays and lost opportunities in the market. Etec executives acknowledge they were under review, but insist they passed it with flying colors and that Applied will continue to back the e-beam operation.

Applied reportedly has given the e-beam operation a reprieve for about 18 months, but if it doesn't turn itself around by then, Applied will again review the group. Frank Abboud, general manager of Etec electron beam products, denies these reports and says there is no timetable and Applied is backing the e-beam group for the long haul.

Applied does have a major task ahead in turning around Etec's e-beam business, which has been losing market share for years. The company was more than a year late in delivering its next-generation, e-beam system--a delay that opened the door for rivals such as JEOL, Hitachi, Leica, and Toshiba.

(See April 1 story.)

Chalk up 2 more acquisitions
for Fairchild Semiconductor

Fairchild Semiconductor is still broadening its product lines by the acquisition route. This week it was I-Cube and Signal Processing Technologies that got sucked up by the rapidly growing Maine chip maker.

The 11-year-old I-Cube, based in Campbell, Calif., builds cross-point switches for Internet infrastructure, data communications, telecommunications, broadcast video, test equipment, and digital signal processing.

Fairchild purchased Signal Processing Technologies (SPT) in Colorado Springs, Colo., from Japan's Toko. The deal adds analog and mixed-signal products to Fairchild's line, including analog-to-digital converters, digital-to-analog converters, and comparators.

"I-Cube's technology will add a further dimension to Fairchild's interface portfolio, providing the 'switch' in opportunities for switch fabric interconnect," said Kirk Pond, president, CEO and chairman of Fairchild Semiconductor.

I-Cube's patented architecture enables its switches to serve multiple end applications, says Fairchild CEO Kirk Pond. This fits well with Fairchild's multi-market strategy of providing high performance building block products.

Fairchild intends to combine its own manufacturing processes with SPT's analog and mixed-signal technologies and go after emerging high-end consumer products in video as well as medical ultrasound, professional broadcast, and infrared imaging. The mixed-signal ICs will also be aimed at cellular phone infrastructure, satellite communications, and instrumentation such as laboratory equipment, oscilloscopes, and test systems for memories.

The SPT purchase will accelerate Fairchild's expansion into analog ICs, says Pond. "In five short years, Fairchild has developed its analog products business from zero to more than $300 million annually," he points out.

(See April 1 story.)

Mitsubishi, Hitachi venture
may add NEC, Matsushita

This R&D joint venture still looks increasingly to me like the "Japan Inc." of 2002. The new system-on-a-chip effort launched by Mitsubishi Electric and Hitachi in March, eventually might be opened up to other partners.

New partners could be added after the combined SOC company starts operations next year, says Tamotsu Nomakuchi, new president of Mitsubishi. They could include NEC and Matsushita Electric, he says.

The Mitsubishi president also says the new SOC joint venture with Hitachi might cooperate in the development of chips with other Japanese companies even if they don't join the new joint venture.

(See April 4 story.)

Infineon could be getting
anxious to grow DRAM biz . . .

Infineon Technologies is working harder than ever to grow its DRAM business. And it is looking more than ever to Taiwan companies for help.

The German chip maker now confirms it is meeting with Nanya Technology to discuss a possible DRAM alliance, a move that would help Infineon to expand its memory capacity so that it could better compete with its larger competitors.

Observers in Taiwan believe the two companies will set up a 50-50 joint venture to build one or even two 300-mm wafer DRAM fabs. It could be a done deal by June, they say.

Infineon clearly wants to be one of the major surviving DRAM players. "The current consolidation in the DRAM industry will determine which companies are strong players when the market recovers," says Jan du Preez, president of Infineon Technologies North America. "Infineon intends to remain a top-tier competitor, based on its commitment to DRAM technology development, ability to produce both volume and specialty products in the most cost-effective way, and a stable financial base."

"Infineon is getting more anxious than anybody else in the industry," comments Helen Huang, analyst at ABN AMRO. "It has to do something to expand market share and increase pricing power, and Taiwan is the right place to go because it's attractive as a partner."

The Nanya talks are only the latest in a big Infineon push to line up additional DRAM capacity, especially in Taiwan. Within the past month, Infineon expanded its DRAM outsourcing relationship with Mosel Vitelic. It already is getting 48% of the output from ProMOS Technologies, their joint venture, and is starting to ramp output at their 300-mm wafer fab.

Infineon also has just formed an alliance with Winbond Electronics, which is licensing Infineon's advanced 0.11-micron technology. In exchange, Winbond will provide Infineon with exclusive access to standard DRAM chips that it will make using this technology.

"Infineon's strategy is to expand market share through extended alliances" with Taiwan, explains Andrew Lu, semiconductor analyst at Salomon Smith Barney in Taipei. Lu believes that Infineon's strategy to barter process technology for capacity will pay off for the company. "As long as Infineon continues to sell its technology to Taiwanese companies and get more capacity from Taiwan, the company will be able to compete."

(See April 1 story.)

. . . as German IC maker's strategy
gets crucial to survival in DRAMs

Infineon's strategy could be crucial to survival in the rapidly consolidating DRAM industry, analysts believe. If Micron Technology is successful in acquiring Hynix Semiconductor's DRAM fabs, it will have a 35% market share, making it the No.1 DRAM manufacturer, according to ABN AMRO in Taipei. That would pose serious threats to Micron's biggest rivals, Samsung Electronics and Infineon.

"Infineon has to expand capacity, since there will be only three to five big players that can stay in this competitive market, while the rest of them will be forced out when the next downturn comes around 2004," says Rick Hsu, analyst at Nomura Securities in Taipei.

Infineon's strategy in outsourcing a vast quantity of added DRAMs is to shoot for a 20% global market share, says Sherry Garber, analyst with Semico Research in Phoenix. "This is all about the ongoing quickening pace of DRAM industry consolidation," she adds.

"First-tier OEMs and major motherboard vendors predominantly want to deal with only the biggest and most reliable DRAM suppliers," Garber notes. "These are the memory vendors who will get long-term supply agreements and contract prices, which are more lucrative than slugging it out in the spot market."

(See April 1 story.)

Despite downturn, EDA
kept growing last year . . .

Here's one big semiconductor-related market that actually grew last year despite the downturn in the economy. Electronic design automation generated $4 billion in global revenues, a 6% rise over 2000, according to the EDA Consortium's market statistics service.

EDA even grew in what overall was a lousy fourth quarter, hitting a record $1.046 billion in revenue vs. $1.035 billion in the year-ago period. In fact, the global EDA industry has generated more than $900 million in quarterly revenue for seven straight quarters.

Excluding revenues for semiconductor intellectual property, consulting and other services, EDA product and maintenance revenue jumped 10% in the fourth quarter to $935 million vs. year-ago sales. This accounted for 89% of the industry's total revenue. For all of 2001, EDA product and maintenance revenue rose 12% to $3.4 billion.

The IC layout sector set the pace for revenue growth, rising 24% in the fourth quarter to a record $309 million. For the year, IC layout accounted for $1.1 billion in revenues, up 37% over the previous year.

EDA's largest tool category, computer-aided engineering (CAE), generated $530 million in fourth quarter revenues, 5% more than a year-ago. For all of 2001, CAE revenue hit $1.9 billion, 4% higher than the previous year.

Revenues for printed-circuit board and multi-chip module layout totaled $96 million in the last quarter of 2001. For the full year, these revenues amounted to $373 million, down 4% from 2000.

The EDA industry's semiconductor intellectual property revenue amounted to $27 million in the fourth quarter, and $109 million for the year, down 15% from 2000.

. . .with EDA in U.S. and
Europe growing the fastest

EDA revenue in North America last year hit a record $2.3 billion, or 57% of worldwide business. Western Europe's EDA business grew 16% in 2001, a record total that was 20% of the global market.

EDA revenues from Japan dipped 8% from 2000 to end up at $641 million for 2001. That was 16% of the global market, according to the EDA Consortium's market statistics service. The rest-of-the-world accounted for $277 million last year, or 7% of the global market.

As usual, Unix seats beat out Windows seats in total EDA software revenue. Unix platforms generated $2.9 billion, 86% of all EDA software revenue. Revenue for software that runs on Windows-based platforms was $488 million for 2001.

(See April 1 story.)

Motorola predicts it
will lose money in '02

Our reporters were plowing through SEC documents that Motorola filed this week, and the news is not good--though not unexpected.

Motorola does not expect to be profitable in 2002. The company tells the SEC it believes it can be profitable before special charges despite anticipated lower sales. It expects to be able to do this because of a reduced cost structure from its corporate reorganization, improved manufacturing margins, increased sales of wireless handsets, and the expected recovery in the worldwide semiconductor industry.

Motorola says that its new semiconductor strategy, adopted in response to last year's downturn, "is to focus our internal manufacturing capacity on leading edge and specialty technologies, while supplementing our internal manufacturing capacity with joint venture manufacturing facilities and purchases of products from outside vendors, primarily foundries."

The new business model focuses on providing silicon-to-software solutions to the wireless communications, networking, and transportation markets, it says, because Motorola believes these are growth markets. The maturing PC market provides little opportunity for the company now.

The new business model also calls for the company to license more intellectual property to increase royalty revenues over the next several years, the company says.

Motorola reports that its top ten end-customers, which accounted for about half its revenue in 2001, were Apple, Bosch, Delphi, Hewlett Packard, Lucent, Motorola, Qualcomm, Siemens, Sony, and Visteon.

(See April 3 story.)

TI joins crowd trying
to speed up IC delivery

Maybe it was PC makers like Dell that got the chip industry to do something about speeding up their deliveries. Texas Instruments got on board this week when it started talking about its new strategy to cut delivery times on high-volume, standard logic devices.

TI says it is cutting the industry's eight-week turnaround cycles down to two weeks. The chip company says it has made "significant investments" in inventory in order to improve availability--irrespective of current market conditions. TI products targeted under this new delivery strategy include "building-block logic" ICs that make up 80% of the company's logic unit volume.

The Dallas-based company was following the lead of other logic chip suppliers, such as rival Fairchild Semiconductor, which have launched efforts to speed up delivery of standard products. Fairchild launched a new inventory program called Direct Ship Plus that it says will streamline the supply chain and minimize the warehousing of products. It says customers will be able to place a "pull order" from anywhere and receive the product in three days or less.

To speed up delivery, the Maine-based company also has built a $13 million, fully-automated logistics warehouse in South Korea. Robots and automated storage retrieval systems will be able to handle up to 150,000 product shipments to Northeast Asian customers every quarter.

(See April 4 story.)

We want your feedback, comments, criticisms, or questions. E-mail us at bhenkel@aol.com.

(Click here for last week's Semiconductor Alert!.)