Forecasters drop rose-colored glasses
and may have gotten too pessimistic

Just a few months ago, chip markets were booming, chip production equipment makers were booming, and analysts were talking about no more sharp business cycles for the chip business. A financial analyst and a columnist or two like me who began predicting a slowdown in the midst of a massive build up in capacity got their heads handed to them.

But this week the shoe was on the other foot at SEMI's Strategy Strategy Symposium in Pebble Beach. Everyone--analyst and industry manager alike--was trying to figure out when the downturn would bottom and business would turn up.

There was little consensus--business wasn't expected to get better until the second half at the earliest. And some folks even expected a second downturn would follow next year. "It's anybody's guess," laments Barry Mayer, CFO at Tokyo Electron America, U.S. subsidiary of equipment giant Tokyo Electron.

The chip industry is now entering into the seventh downturn in its history, notes analyst Bill McClean, president of IC Insights. He defines a downturn as a period when the chip industry falls below its historical annual growth rate of 17%. Now that's a positive spin.

McClean also believes the current downturn will be short lived--unlike the decline in the late '90s that lasted for more than three years. "This should be a short-lived down cycle," he says. "The industry should be back to normal growth rates by 2003."

That's not a universal feeling, however. "We're forecasting two soft years--including this year and next," warns analyst Risto Puhakka at VLSI Research.

The chip and equipment markets will both be "slow" for the first three quarters of 2001, although they both will rebound slightly by the fourth quarter of this year, says Klaus-Dieter Rinnen, chip gear analyst at Dataquest. But this will be short lived, he says, with both chip and capital equipment segments heading for another downturn in 2003.

(See Jan. 10 story.)

Chip business could end up
with negative growth this year

The chip industry's market researchers have switched dramatically from hot to cold and are looking now for more or less a flat year. But even that cautious outlook might be too bullish.

Out poured the latest forecasts this week at SEMI's Industry Strategy Symposium in Pebble Beach. The peak year of the current cycle was certainly last year, they concluded. What shot it down was a combination of global product growth slowing and a build up of excess capacity. "The global IC market will come down quite a bit from 2000," says analyst Bill McClean, president of IC Insights.

A big problem now is excess inventory, he says. "Buyers always err on the side of building inventory. They get fired for shutting down a line because there's not enough inventory. They get their hands slapped for having too much inventory," McClean comments.

The worldwide semiconductor market is forecast by McClean to grow 7% this year after a 36% surge in 2000, and he adds there is a greater risk that growth will fall short of that mark than exceed it.

In fact, McClean warns that the chip industry was twice as likely to show no growth this year as it was to grow 10% or more. He noted a discouraging sign already for this year--three companies, Fairchild, Microchip, and On Semiconductor--have already warned that the first quarter of this year will fall 5%-to-8% from the first quarter in 2000. First quarter results from the other chip companies will likely tell the tale for 2001, he believes.

(See Jan. 9 story.)

It's Ruiz's turn to call for
help from chip gear makers

It seems like every couple of the years or so there is a new outcry for chip companies to join forces with production equipment makers to solve the growing manufacturing problems as they move to more complex products.

Picking up the cry this year was Hector de J. Ruiz, COO of Advanced Micro Devices, who told SEMI's annual executive summit conference at Pebble Beach, Calif., that two industries must forge even tighter relationships with each other to bring technology to the next level, especially with the development of new and costly 300-mm fabs.

"Tool complexities and costs are increasing," warned Ruiz. "We must do some 'out-of-the-box' thinking to control these costs," he says. "It's a good opportunity for chip and equipment makers to innovate." He urged both IC and equipment vendors to work together to lower the cost of lithography, metrology, test, and other tools in state-of-the-art 300-mm fabs. "We must lower the costs of chip testability," he said. "We also have to measure devices that are plus or minus one nanosecond."

(See Jan. 8 story.)

Would you believe
wafer shortage is coming?

Despite the chip industry slowdown, there is growing worry that there'll be a shortage of blank production wafers beginning as early as this year.

A lack of investments in new plants for these wafers could cause "severe shortages" of silicon substrates in 2001 and beyond, warns James Ellis, CEO of the U.S. subsidiary of Germany's Wacker Siltronic.

The world's supply of 300-mm wafer substrates is especially vulnerable to shortages this year, says Ellis, who was speaking to industry executives at SEMI's Industry Strategy Symposium. "I hope I'm wrong," he says, but "I don't think there's enough 300-mm capacity to serve the industry."

The problem is the lack of plant investments coupled with the frenetic activity in 300-mm fabs that will create worldwide shortages of 300-mm (12-inch) substrates, he says. "I think there are eleven 300-mm fabs going online in 2001," Ellis says. "In 2002, there will be seventeen 300-mm fabs," he figures. "But we still don't feel very secure about where the industry is headed down the road."

Wacker Siltronic's German parent is now conducting a feasibility study for construction of a new high-volume blank wafer plant, which would dramatically increase shipments of large-diameter silicon substrates to IC makers. But a decision to start work on the new facility is still not certain and would not likely happen this year if uncertainty continues over when volume 12-inch production begins at the new fabs.

There are some other troubling trends in this industry, Ellis says. A wave of consolidation in the blank silicon wafer business could limit the number of vendors that could afford investments in 300-mm substrate factories, he says. "In the future, I think only three or four companies will be able to produce 300-mm blank wafers."

(See Jan. 9 story.)

Varian's straw in the wind

Varian Semiconductor Equipment is redoing its business plan for this year after its chip-making customers started putting on the brakes on equipment deliveries.

The Gloucester, Mass., company says it has "recently received requests from a few customers to delay deliveries," confirms CEO Richard A. Aurelio. "These requests combined with general market uncertainty have caused us to update our guidance for the rest of fiscal 2001," he points out. "We now expect our annual revenue growth to be in line with the revised industry's growth projections of approximately 10% year over year." Varian is a leading supplier of ion-implantation systems.

(See Jan. 10 story.)

Here comes Japan's
third 300-mm fab

It's beginning to look like the Japanese semiconductor industry doesn't want to be left behind in the next big upturn. Mitsubishi Electric says it is going to build a new 300-mm wafer fab, making it the third 12-inch wafer fab to be announced for Japan.

Hitachi and NEC already have started their new generation fab. Mitsubishi is investing $1.7 billion to build a plant with a 8,000-square-meter cleanroom and a production capacity of 25,000 to 30,000 12-inch wafers per a month. The factory, which will be located at its main Kochi, Japan fab complex, will turn out mainly flash memories and system-on-chip (SoC) devices.

Production will begin in 2003 if reliable 300-mm production systems are available and the market is favorable, Mitsubishi says. Construction of the building will begin in the year beginning in April. Production equipment for the fab will cost about $1.7 billion.

Mitsubishi wanted Matsushita to join it in sharing the cost of the new 300-mm wafer fab. But Matsushita, which already is a partner with Mitsubishi on system-on-chip (SoC) process development, decided not to participate. "We've allied with Mitsubishi for development, but we have no intention to collaborate in semiconductor production at present," says Kunio Nakamura, president of Matsushita Electric Industrial.

(See Jan. 11 story.)

Motorola orders run
into 4th quarter wall

Yep, there has been a sea-state change in the semiconductor business. Take a look at Motorola's results. While big electronics company was able to ship 7% more chips in the fourth quarter than it did a year-ago, the wheels came off order rates.

The Schaumburg, Ill., company's semiconductor sales edged up to $1.9 billion, but chip orders took a swan dive, dropping 19% to $1.6 billion from the fourth quarter of 1999.

Motorola's fourth-quarter chip orders were down in all regions. And by markets, chip orders were significantly higher in imaging/entertainment, higher in networking/computing, lower in transportation, and significantly lower in wireless and standard products. Motorola Semiconductor's operating profits hit $158 million, nearly double the $80 million recorded in the fourth 1999 quarter.

(See Jan. 10 story.)

NEC Semiconductor
gains its independence

As expected, NEC is following the direction of several other giant systems companies and spinning off its semiconductor operations into a new company.

The new company, as yet unnamed, will have annual sales of $1 billion and a capitalization of $100 million. It will be headquartered in Tamagawa, Japan. The spin-off should be completed by this fall, according to NEC. Officials say the move was made to allow the chip unit to better focus R&D and marketing efforts on specialty chips.

Within two years, the Japanese parent plans to make an initial public offering. Once the new company's shares are publicly traded, NEC expects it will be easier to acquire other optical and microwave chip operations through a stock exchange.

The new Japanese company claims to hold a 60% global market share in gallium-arsenide semiconductors in mobile and satellite communications, as well as a 35% share in silicon devices for the same markets.

(See Jan. 10 story.)

Now it's Vanguard that
wants to put up 12-inch fab

We've already predicted the world will have chips coming out of its ears due to the fast-growing crowd of global chip makers jumping on the 12-inch wafer bandwagon. But the new giant fabs keep coming.

One newcomer this week to the 12-inch world is Taiwan's Vanguard International Semiconductor, which is busy transforming itself from a DRAM maker to a foundry. It doesn't appear worried about the current chip slowdown or the number of chip makers rushing to put up the new generation of fab.

"We're not pessimistic about the industry's outlook, even though many people in the market are," declares vice president Robert Hsieh. "Upgrading to the 12-inch technology is one of the best ways to ensure we'll keep growing."

Because of the huge size of this new plant, Vanguard wants to team up in a joint venture with one or more chip makers to get the $2 billion to put it up. The company isn't saying who they are at this point, but Taipei press reports quote industry sources as saying the partners may include Altera and Silicon Storage Technology, a Sunnyvale, Calif., flash memory vendor.

Vanguard's 12-inch fab will turn out flash memory and other memory chips in Taiwan's Hsinchu science park, Hsieh says. "Everybody wants to get into the flash memory market," says Carol Yu, analyst at Taipei's Grand Cathay Securities.

(See Jan. 9 story.)

There's no way equipment
makers can keep sales up

Good old, feast-or-famine chip production equipment industry. Who would have believed that this business would have grown 82% last year? Not me. I kept waiting for this business to fall off the cliff, but it didn't do it in 2000.

Now 2001 is a different story, and this week the leading market research houses raced to lower their forecast for this year and tell industry executives in Pebble Beach, Calif., that the good times were over. Reasons for the chip gear makers' sudden change in direction were their customers' delaying tool orders and a sudden reduction in global capital spending.

Dataquest started off by chopping its forecast for equipment sales to a modest 10-to-11% growth for 2001. But it gets worse. Dataquest believes that there is a 50% chance that equipment sales could even drop this year--off as much as 6% from 2000. Just to show you how fast the market researchers changed, Dataquest projected 25-to-30% growth in semiconductor capital spending during 2001.

VLSI Research cut its forecast too, slicing it in half to just a 4.5% growth rate. This year the equipment business is moving towards a period of little or no growth, warns VLSI Research analyst Risto Puhakka.

"It's horrible out there," says Steve Jensen, vice president at Silicon Valley Group. Nearly every chip company has pushed out their orders for capital equipment, he says. "Even Intel has."

Most market observers believe the worse is yet to come. "It's going to get ugly," says analyst Steven Pelayo from Morgan Stanley Dean Witter. The semiconductor industry "is looking down a cliff, as opposed to a speed bump," adds analyst Timothy Arcuri of Deutsche Banc Alex Brown.

VLSI Research predicts that the chip-equipment industry will show only a 0.6% growth in 2002. "We're forecasting two soft years, including this year and next," Puhakka says. Then in another boom cycle, he predicts that equipment sales will surge 31.5% in 2003 and 54.6% in 2004. Then it falls off the cliff once again, he says, in 2005 when the equipment industry revenues will fall by nearly 17%. Gets you dizzy, huh?

(See Jan. 8 story.)

Another call for shortening cycle times

I wish I had a dollar for every time I've heard complaints about chip industry managers. As far back as the 1960s, so-called experts called industry management immature and that better management systems were badly needed by chip makers.

Well, it came around again this week at the Pebble Beach Industry Strategy Symposium. The sharp cyclical swings that have always plagued the industry could be mitigated if it could tighten the flow of information across the supply chain, declares Kevin Canty, planning vice president for Applied Materials.

Due to the way the industry has always done business, "minor disturbances in end demand can translate into huge disturbances at our suppliers," he points out. This "bullwhip effect" has traditionally been accepted as an inevitable part of doing business in the industry, he notes.

Cycle times are lengthened simply because of delays in receiving information. "Our suppliers are waiting for a signal from us, and we're waiting for a signal from our customers. There's more wait time in the cycle than actual physical assembly time," he says. Speeding this up could have "a tremendously favorable impact on the efficiency and predictability of the supply chain," Canty predicts.

While supply chain management will never make the industry immune to cyclicality, Canty figures that "if we're able to streamline the information flow, hopefully we can mitigate the consequences of the boom/bust cycle our industry has lived with for many years."

E-business is one way to improve the flow of information, but "it's not enough to have e-business working on transactions and forecasting. The real advantage is in collaboration," Canty says.

(See Jan. 9 story.)

Broadcom keeps on buying
up chip companies . . .

Neither a lower stock price nor a business slowdown is stopping Broadcom's mad dash to buy up chip companies and turn the Irvine, Calif., supplier into a semiconductor giant. This week it gobbled up another chip company.

The company is swapping nearly $1 billion in its stock for ServerWorks, a high-speed input/output chip supplier, with the aim of significantly increasing the performance of its products for networked servers and systems connected to the Internet.

The addition of ServerWorks' integrated-circuit portfolio and technologies, Broadcom says, will enable Broadcom to deploy chip products for 10-gigabit-per-second designs such as InfiniBand, Ethernet, and Fibre Channel.

The six-year-old ServerWorks has shipped four generations of system-I/O solutions for computers based on Intel microprocessors. More than 10 million input-output devices already have been shipped by the Santa Clara, Calif., company.

(See Jan. 8 story.)

. . . with latest deal spawning
new high-speed chip set

Acquisition of ServerWorks by Broadcom will lead quickly to a new generation high-speed chip set that will obsolete the upcoming PCI-X I/O bus, claims Yossi Cohen, Broadcom marketing chief.

Even the 8-gigabit-per-second speed of PCI-X, he says, will become a choke point for enterprise and network servers in 2002 when four different I/O systems hit the 10-gigabyte market--Ethernet, fiber channel, OC192 Sonet, and Infiniband. "We must replace the separate PCI-X bus with a direct I/O connection through the chip set to support all the new 10-gigabyte systems," he says.

ServerWorks will use Broadcom technology and resources to complete development of this chip set, says David Pulling, ServerWorks vice president. The new chip set is expected to be available in 2002.

(See Jan. 9 story.)

AMD is planning
second 300-mm fab

Like almost all the big players today, Advanced Micro Devices is in the planning stages of a next-generation, 300-mm wafer fab. Aimed at microprocessor production, the new fab has an anticipated start-up date of around 2004, according to CEO Hector Ruiz.

The facility, to be called Fab 35, will likely be a joint venture in order to share the risks and costs of the fab, he says. Location is not yet set. Total cost could hit $4 billion. AMD already has a 300-mm flash memory fab in the works in a joint venture with Fujitsu. Called Joint Venture 4, this fab is located in Aizu-Wakamatsu, Japan and was originally expected to cost about $1.5 billion. It is scheduled to begin production in the second half of this year.

And a day after AMD disclosed it was looking for partners to build the 300-mm wafer fabs, reports in Taiwan indicated that TSMC and UMC were leading contenders for the AMD partnership. No confirmation yet.

(See Jan. 8 story.)

Taiwan foundries still think
their sales will jump this year

This is a bit hard to believe, but Taiwan's major foundries claim they are weathering the first quarter blues and the fallout from the PC slowdown and will ship a lot more product than they did last year.

Both United Microelectronics and Taiwan Semiconductor Manufacturing are predicting higher first-quarter revenues than they had in the first quarter last year. This is due primarily, they say, to mergers, expansions, and a process mix that relies less upon the prevailing technologies of a year ago--0.35- and 0.25-micron.

UMC is sticking by its December projection of 40% sales growth, despite experiencing a utilization dip of about 10% so far this year. TSMC expects first-quarter revenues to be "much higher" than a year ago.

And unlike many other chip makers caught in today's downturn, TSMC and UMC, as well as Winbond and Macronix, are still proceeding with their expansion plans. Within the next three years, the four Taiwanese foundries expect to ramp up a total of seven 300-mm fabs using 0.18-micron and finer processes. That's a lot of production.

UMC still expects to produce about 3.4 million wafers in 2001, with more than half of those using the 0.18- and 0.25-micron processes. The Taiwanese companies believe that any inventory buildups will be burned off during the first quarter, sending them back up to full capacity utilization.

"Our feeling is that the cycle is not coming to an end yet," says UMC spokesman Alex Hinnawi. "I think people are overreacting, especially if you look at the valuation of the stock. People are seeing gloom and doom . . . but it's just not there."

If the Taiwan foundries are correct here, then they're going to gain some market share. But I don't think they are.

(See Jan. 10 story.)

Asat gambles by expanding
to move out of second tier

They usually say the best time to build a new factory or buy stock is when the market outlook isn't looking good. Well then, Asat Holdings might be doing things right. Despite a slowdown in its IC-packaging and test business, the Hong Kong company has just broken ground on a 100,000-sq.-ft., packaging plant in nearby Shenzhen, China.

Production is set to begin in November or December on several types of new and advanced packages, such as BGA, chip-scale, and leaded devices. This will be the company fourth factory--it currently has two plans in Hong Kong and one in France. The new plant might propel Asat into the top ranks of companies competing in this market--analysts now consider Asat to be a second-tier provider of back-end chip services.

Asat currently is gaining ground on the leaders. Propelled by such customers as Analog Devices, Broadcom, Conexant, Infineon, Lucent, and Vitesse, Asat reported sales jumped 90% to $312 million last year. This year, Asat figures its sales will hit between $400-and-$500 million. But it is focusing on telecom products and is not going head-to-head with the leader, Amkor, in the DRAM packaging business.

(See Jan. 11 story.)

But is a shakeout coming
in IC packaging and test?

Now may not be the best time after all to expand in the IC-packaging and test business. Despite the fast growth of these markets, a major shakeout is going to develop.

That pessimistic outlook comes from president John Boruch, president of Amkor, the world's largest IC-packaging and test house.

The huge investment needed now to compete in this business, combined the slowdown in the IC industry make the small- to mid-sized providers of IC-packaging and test services especially vulnerable to a possible consolidation, says Boruch. He told executives at SEMI's Industry Strategy Symposium this week that even some of the larger IC-packaging houses may not survive in the long term.

Of the 50 companies in the IC-packaging and test business today, 10 of them are large-scale companies, Boruch says. "But we think there will be only three to four large-scale companies that will survive."

Third-party packaging, however, is still expected to grow strongly over the next few years, since the vast majority of IC-packaging is still done in-house by integrated device manufacturers (IDMs). Only 26% of all chip production is now finished by third-party IC-packaging houses, but this percentage is expected to climb to 38% by 2003, Boruch estimates.

(See Jan. 11 story.)

NEC going strong in China

NEC is deadly serious about the Chinese market. For some time, now, it has expected significant growth in China's semiconductor industry. It now figures that this market hit $7.73 billion last year. So it is expanding its operations there.

It is spending $300 million to expand a joint-venture wafer fab in Shanghai, raising its output from 20,000 to 30,000 eight-inch wafers per month. The 200-mm fab was set up in 1999 as a joint venture with Shanghai Hua Hong to build chips for Chinese mobile communications. The expansion should be done this year.

NEC also operates a second, joint-venture plant in Beijing. Called Shugang NEC Electronics, this seven-year-old fab has a capacity of 13,000s wafer a month and produces driver ICs for liquid crystal displays as well as remote control devices for a variety of applications.

(See Jan. 11 story.)

If you have any comments or questions, don't hesitate to E-mail us at bhenkel@aol.com. Have a great weekend!

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