Greetings from Down-East Maine--where all hell broke loose this week. Nothing like a Nor'easter to stir things up. Believe it or not, the wind was so strong that it not only blew down several of our 150-year-old pine trees but also ripped our 5-by-8-ft. U.S. battle flag into shreds. This new weekly column, successor to a weekly newsletter we wrote during the 1990s, analyzes and comments on the important developments of the past week in the chip world.
Welcome to the Taiwan Memorial Issue!
Just kidding, but it would seem so after reading all the news pouring out of Taiwan this week. In all honesty, there are at least two reasons for this tidal wave. Ace reporter Rob Lineback was on the island this week reporting the latest developments and Taiwan Semiconductor Manufacturing Co. was doing everything it could to take over as the world's largest producer of ICs.
TSMC wants to be chip leader -- and just might make it in 2001
The world's largest silicon foundry, better known as TSMC, told a Hsinchu briefing it is going to keep expanding at its current furious pace for at least the next couple of years. Reason for the huge investment is simply that TSMC wants to become the world's largest producer of chips next year. In order to claim the top spot, the Taiwan giant plans to invest $3.6 billion in 2001 to add wafer-processing capacity.
This would follow an astonishing increase of 167% in its capital spending this year. The big foundry is now well its way to spending $4.4 billion, an amount that would put TSMC close to what the capital-spending leader -- Intel -- is spending to expand its production this year. TSMC expects to fabricate 3.4 million 8-inch equivalent wafers this year -- or 6% of the wafers the entire global industry is expected to turn out.
This kind of increase is expected to help the Taiwanese company become the sixth largest chip maker in 2000. Toshiba looks to be the leading producer, followed by Hyundai, NEC, Intel, and Hitachi.
TSMC has even bigger production plans for 2001. It aims to crank out 4.7 million wafers -- a walloping 38% increase over this year, putting the Taiwanese company in a first-place tie with Toshiba and Hyundai as the world's largest supplier of chips. To help fund its huge expansion program, TSMC is now considering a new equity offering, which would be the first one since the foundry went public six years ago. (See March 29 story).
TSMC and UMC wage war of words over copper IC lead
Copper is hot! That was another big story coming out of Taiwan this week, where the world's two largest silicon foundries were trying to out-shout one another in their battle to get into production first with copper ICs, the next-generation of chips.
United Microelectronics, or UMC, says that Xilinx, its technology partner, is the first customer of a dedicated foundry to begin shipments of copper chips to system manufacturers. UMC is pushing hard to top its cross-town rival TSMC in copper-processing capacity this year. UMC expects to be processing more copper interconnects by the end of 2000 than TSMC is now projecting. Xilinx already has started delivering samples of the first copper-processed field-programmable gate arrays, which were fabricated with UMC's 0.18-micron technology using copper as the top two interconnect layers. They should be in volume production later this year. (See March 27 story).
Are Taiwan foundries catching up in process technologies?
The silicon foundries -- which until recently trailed the leading semiconductor giants in getting the latest production processes up and running -- are now pulling out all the stops to put new technology into production.
TSMC is a prime example. This week the foundry leader disclosed that it has already distributed the next generation of copper-based chips to five electronic design automation suppliers. TSMC is working closely with them to speed the availability of EDA tools and accelerate chip designs using its new 0.13-micron process, which is slated to go into initial production in the first quarter of next year. Volume production would begin on the chips with all-copper interconnects and low-k dielectric insulators in the third quarter of 2001.
Getting into production faster is part of the Taiwan foundry's strategy to close the technology gap between itself and leading semiconductor houses that operate their own wafer fabs. TSMC chairman Morris Chang says his company and perhaps its closest foundry rivals already are catching up with major chip companies such as Intel and IBM.
"If there is any distance between TSMC and the big integrated device makers, it is not very perceptible." Chang said. He predicts that both TSMC and its foundry archrival United Microelectronics will be producing devices at 0.13 micron before the end of the year. (See March 28 story).
TSMC's strategy to become to become No. 1 is driven by "Big Mac"
This week in Taiwan was also the scene of the formal dedication by TSMC of its huge Fab 6, which we believe to be the world's largest cleanroom for chip production. Nicknamed "Big Mac," the plant underscores TSMC's double-barreled approach for hiking its chip capacity over the next few years.
At the same time that the foundry is cranking out higher volumes from its current 8-inch wafer fabs, it is gearing up to switch over to the new generation of 300-mm wafers. Fab 6 "is simultaneously our last 8-inch wafer fab and our first 12-inch fab," declares TSMC chairman Morris Chang.
Fab 6's 8-inch line began production in January with the goal of 50,000 wafers starts a month by 2001. The plant is immense. When fully loaded in 2001, it will contain nearly 1,000 sets of manufacturing tools and employ more than 2,300 workers. In July, TSMC will start equipping its first 12-inch pilot production line at Fab 6. By November 2001, the pilot line will be capable of running 4,500 12-inch wafers a month with 0.18-micron technology.
All the 300-mm production gear now appears to be ready for delivery in the summer, but TSMC has had a scheduling problem with the 300-mm i-line scanners ordered from ASM Lithography. TSMC says its 300-mm schedule puts it slightly ahead of UMC, which will begin processing 12-inch wafers with 0.18-micron technology in early 2001 at its joint-venture fab in Japan. (See March 30 story).
Here's a neat idea, but are we getting too many e-services sites?
A new Silicon Valley startup is gambling on a bold, new approach to one-stop shopping on the Internet for the entire IC design and manufacturing supply chain.
SiliconX already is field-testing an early version of its new web site and expects to launch a full-fledged version in time for the Design Automation Conference in June. While there are many e-commerce sites already supporting electronic design, most of them -- firms like ChipCenter, PartMiner, and WebPRN -- are aimed primarily at providing components and services for board designers, says Judy Owen, CEO of the startup.
There is very little focus at these other companies on IC design, she maintains. SiliconX is different, she claims, because it wants to be the one site to offer it all. In order to provide an infrastructure for e-commerce, SiliconX is now building a services mall that will link users with design services, design automation tools, intellectual property, silicon foundries, and packaging and test firms. While it will derive some revenue from advertising, SiliconX expects to make money by taking a share of the transactions made on its site.
The company also is developing a resource locator, which will use intelligent search engines to link users to tools, services, technology, and manufacturers. That would certainly change product development -- if it works. (See March 29 story).
Coming soon -- a NEC semiconductor spinoff?
Spinning off its semiconductor business seems to be the "in" thing for a conglomerate to do these days. Siemens was the latest to follow this path with Siemens Semiconductor (now Infineon) and it raised nearly $6 billion in an initial public offering this winter. But frankly we were surprised this week to see Japan's NEC take some initial steps in this direction.
NEC will be restructured on April 1 into three new internal companies -- Electron Devices, Networks, and IT Solutions. While NEC's president, Koji Nishigaki, says there are no plans now for any IPOs or spinoffs, "it may be something that we will want to do in the future." NEC wants more time to develop its three new business units before considering any IPOs, he adds. The three new companies will have greater autonomy, Nishigaki says, and will be responsible for their own R&D, capital investments, as well as their own procurement and purchasing. NEC Electron Devices will include semiconductors, flat panel displays, passive components, and hybrid chips.
NEC expects to take a $2.2 billion one-time charge to cover the restructuring, but this would still leave it with net profit of $190 million for the fiscal year ending Mar. 31. (See March 28 story).
Analog chips more important to TI than some people think
Some folks will answer digital signal processors when asked what chips are behind the splendid growth of Texas Instruments these days. But that wouldn't be giving enough credit to analog parts, which have become an integral part of the Internet and multimedia area.
"Of all the semiconductor components, analog is the category to watch," says Mary Ann Olsson, Dataquest analyst who has just completed a study on analog chips for the market researcher. And for the third year in a row, Dataquest has ranked TI No. 1 in the analog market. In fact, TI's analog business grew faster than the overall market last year, she says, hitting 23.3% compared to 18.8% growth for the total market.
The Dallas chip maker owned 10.7% of the $26.2 billion global analog market, according to Dataquest. TI maintained its No. 1 position, Olsson says, because of its strength in traditional categories such as power management and its ability to position analog technology and software expertise to target new application-specific markets.
TI beat out ST Micro in 1997 for the analog lead, but the European chip maker kept pace in 1999 with TI by growing 22.2% and moving up to second place with an 8.6% market share. Philips Semiconductors slipped from second to third place, and held 6.9% of the market. Infineon, which grew faster than TI last year, moved up a notch to No. 4, while National knocked Analog Devices out of No. 5. (See March 30 story).
Still no level playing field for U.S. chip makers in China
There goes the U.S. semiconductor industry again -- competing overseas with one hand tied behind its back. This time it's China, that humongous market that chip makers the world over have long coveted.
The Japanese chip makers are moving in while U.S. chip vendors watch helplessly as U.S. export controls prohibit them from building state-of-the-art fabs. But nothing like that seems to bother the Japanese. NEC's China joint venture will begin producing 128-megabit DRAMs at its Shanghai fab later this year. This represents the most advanced chip to be produced in China, according to NEC. The fab is now upgrading its 0.35-micron processing equipment to 0.25-micron to produce the new DRAMs, the first time design rules have reached this level in China.
U.S. export controls still prevent Motorola from equipping its fab in Tianjin, China, with similar 0.25-micron process equipment. Japan is largely free to approve the export of advanced semiconductor equipment. NEC Hua Hong Semiconductor's Shanghai fab will double its capacity this year, expanding from 10,000 wafers to 20,000 wafers a month. DRAMs will account for 80% of these wafers, while the remaining 20% will be made up of logic chips for handheld cell phones and PDAs. (See March 29 story).
Memory chips responsible for record semi business last year
Dataquest came out with its final figures this week for the 1999 chip business and no surprises. But it was nice to read strong numbers for a change.
The market was powered by the return of the memory business, which had lagged badly in the previous year. Memory revenues jumped a big 41% in 1999. As a result, the global semiconductor market grew 22% over 1998, posting revenues of $168.6 billion. Those revenues also broke the previous high of $151.3 billion recorded back in 1995, the market researcher notes. The Americas continued to lead all regional segments, accounting for nearly one-third of global revenue, but Asia-Pacific, with 25% of the total market, made the strongest regional comeback from 1998 with an increase of nearly 33%.
Like most analysts, Dataquest is looking for another good year in 2000. Continued expansion of the communications network, including the Internet, will provide growth for such product segments as digital signal processors, application-specific standard products, and optical semiconductors, according to Dataquest.
The market researcher also came up with its top 10 list. The rankings did not change all that much, the one exception being Motorola, which showed a drop of nearly 10% in revenue due to the divestment of its discrete-semiconductor business. That dropped Motorola from third to sixth place in the global rankings. The other two U.S. companies on the list held their own: No. 1 Intel grew 17.7%, slower than the entire industry's 21.6% growth, and No. 5 Texas Instruments grew slightly faster with a 22.3% increase. Samsung sales jumped 50%, moving it from sixth to fourth largest thanks to memories roaring back.
The three European companies in the top 10 once again were clustered in the last three spots: Infineon was up 33.6% to $5.22 billion, STMicroelectronics grew 20.9% to $5.08 billion, and Philips was flat at $5.07 billion. The three Japanese vendors included NEC in second place, Toshiba in third place, and Hitachi in seventh place. (See March 29 story).
Printing plastic transistors on press may be getting closer
We always like to keep tabs on the latest technology coming out of Bell Labs. Why? Because this is where it all began for the chip business, that's why. Monitoring what's happening at the Lucent Technologies lab should help from getting blindsided by new technology.
So check out what's happening in the plastic transistor department. Latest news: Researchers have come up with a new material the lab says could be the key ingredient for making plastic transistors a reality. Now, a commercial product isn't going to happen tomorrow, but this R&D could end up creating an alternative to silicon chip fabrication.
The missing link until now has been suitable material for making n-type circuits. The right material for p-type plastic transistors was already identified. Now Bell Labs says that it has found a new material, called F15, that can be dissolved in a solvent and then used to print electronic features on a plastic sheet. This method of making transistors would be similar to ink-on-paper printing and would be much less costly to than the standard silicon process, declares Bell Labs researcher Howard Katz.
"Developing an n-type plastic transistor that's easy to manufacture was considered a Holy Grail," he notes. "It's difficult enough to get electrons to move through organic materials, but it's even rarer for n-type organic semiconductors to be functional under everyday conditions."
Such plastic circuits with both n-type and p-type transistors would be highly useful in a variety of high-volume applications such as flexible displays, more durable smart cards, and other low-cost electronics applications. (See March 29 story).
If you have any comments or questions, don't hesitate to E-mail us at email@example.com. Have a great week!
(See previous week's Semiconductor Alert! March 20-24).