Advertiser and

Buyer's Guide
Buyers Guide

Chip Shots blog

Greatest Hits of 2005
Greatest Hits of 2005

Featured Series
Featured Series

Web Sightings

Media Kit

Comments? Suggestions? Send us your feedback.


Midyear Forecast Update

Ramp to 65 nm may offer sole relief after 2004 buying binge

The initial shift to the 65-nm process node later this year may provide the only relief for semiconductor equipment manufacturers facing a letdown after customers went on a spending spree in 2004.

"It's very much going to be a mixed year," asserts Dean Freeman, chief analyst for Gartner/Dataquest in San Jose. He points to a "precipitous drop" recently in average selling prices for both DRAMs and flash devices. Like starving shipwreck survivors, chipmakers binged on tool buys in 2004. The binge has resulted in a capacity glut, Freeman and other analysts say.

CAPEX REDUX: Capital spending comparisons show a doubling of Korea's share and a significant U.S. drop between 2000 (in parentheses) and 2005 (projected).


Driven by the industry's technology roadmap, the ramp-up to 65-nm processes expected to come later in 2005 should buck up the industry's fortunes. Sales of equipment crucial for the technological transition are expected to do relatively well compared with other systems.

"One of the things we've been saying all along is that the industry just didn't need to buy all that equipment and build all those fabs that it did last year," asserts Robert Castellano, president of The Information Network, a market research firm based in New Tripoli, PA. "One of the main problems now is that we combined what could have been two years of growth in the typical upcycle into one year."

The mixed-signal message is reinforced by early semiconductor revenue figures for 2005. The April edition of the McClean Report from IC Insights in Scottsdale, AZ, parses data from the Worldwide Semiconductor Trade Statistics (WSTS) to show a disappointing final quarter of 2004. Those results are offset by relatively healthy chip sales in January 2005.

Since 1996, the semiconductor market in January declined by an average of more than 20%, the report points out. However, revenues dropped only 10.5% between last December and January. This decline was the shallowest since the industry's 1995 boom year. Then, to muddy the waters further, in February global semiconductor sales declined by 0.5% from January's results, according to the report. This decrease is significant because it marks the "first February on record to show a semiconductor market decline from January's results." IC Insights cautions, however, that the weak sequential figures do not necessarily forebode a poor year for semiconductor growth.

Out of 10 total product categories, only two registered a market increase between January and February 2005. Sales of MPUs and display drivers grew 22.5% and a paltry 0.6%, respectively, according to the McClean Report. The remaining eight categories saw sequential declines as steep as 11.7% for digital signal processors. In other product categories, revenues for analog ICs dropped 5.5%, DRAMs decreased 4.1%, and MCUs declined 5.3%. By contrast, in 2004 only two of the 10 categories—MCUs and display drivers—saw sequential declines from January to February.

What do these figures portend for capital spending in 2005? "We're right in the middle of forecasting at this point," Gartner's Freeman says. "But what we're looking at for this year is what we said in April. Right now we think capital spending is going to decline about 7% for the year. That means fab equipment [revenues] will be down in the low teens to maybe low single digits at, say, 10%, plus or minus a couple of percent."

Castellano insists that the ghosts of 2004's binge will spook the industry this year. "Anything they bought in 2004 is going to haunt them in 2005. We're looking for on the order of a 9.5% drop in equipment sales for 2005, and most likely another 5% decrease in 2006, then 20% growth in 2007 and 27% growth in 2008."

The McClean Report asserts that the industry's "historically volatile" trends will continue through 2009. Following the big bust of 2001–2002, capital outlays grew 13% in 2003 and 56% in 2004. The latter year represents the industry's peak during the current business cycle. As a result, the research firm forecasts a 5% decrease in capital spending to $43.4 billion this year, followed by a further 7% decline in 2006 to $40.2 billion. IC Insights forecasts a return to growth beginning in 2007 and continuing through 2008. The April report shows total semiconductor industry capital expenditures reaching $66.1 billion in 2008 before decreasing 8% to $61.1 billion in 2009.

Revenues for front-end equipment grew 68% in 2004 over the previous year, according to Gartner. In 2005, however, the research firm forecasts a 12.2% decrease, followed by another decline of 4.3% in 2006.

SEMI reports that worldwide equipment billings reached $9.35 billion in 1Q05, a 2.3% increase over 1Q04 and 6.5% higher than 4Q04. In a press release announcing the figures, Stanley Myers, the trade association's president, says the January-through-March revenues "showed some positive momentum." SEMI also noted, however, that new orders of $7.25 billion in 1Q05 were 12% below bookings for the last three months of 2004 and 21% lower than the same quarter last year.

Myers did single out South Korea as a particular bright spot for billings. Chipmakers in the Asian nation spent $2.3 billion on equipment in 1Q05, a 165% increase over 4Q04 and a 61% increase over 1Q04.

Meanwhile, VLSI Research in Santa Clara, CA, reports equipment-sector growth of 8% in 1Q05 over the final quarter of 2004. However, bookings dropped approximately 3.5% in the first quarter compared with the first quarter of last year. The firm forecasts orders to grow in 2Q05 by single-digit rates and billings to dip below those in 1Q05. Global equipment revenues for 2005 are predicted to increase only 3.9% to $54 billion.

"As you start to look at it, so much of the market in an upturn like last year is driven by capacity buys," says Dan Hutcheson, CEO of VLSI Research. "When you have to go back to the technology buys, it's like going on a no-fat/no-sugar diet."

Technology buys may prove to be the industry's bright spot in 2005 when the ramp-up to 65 nm begins later this year. "The drivers right now are in 65 nm," Hutcheson says. "That move is really kicking off this year, so that's going to be good for another couple of years; 65 nm means they need the high-numerical-aperture steppers to pull it off."

The VLSI executive notes that copper deposition tools—both for barrier metal and copper plating—also need upgrading. "It used to be you could lay the copper in there with a trowel. Not now. Now it's really difficult." The industry also will continue its "tremendous effort" to find a workable solution for high-k dielectrics, Hutcheson asserts.

He emphasizes that the expected slowdown should have little effect on the development of needed technology, such as immersion lithography. "Basically, the technology segments are going better; capacity is not doing as well.

Implant has been doing reasonably well. They're bringing in new material, they're trying to shrink linewidth; they've got new problems to solve."

Both Nikon and ASML have introduced immersion lithography systems, Hutcheson notes. ASML has a prototype tool at customer sites, while Nikon has chosen not to sell prototypes and is waiting until it has production-ready systems. The key to the different approaches is that a production immersion lithography system with a numerical aperture of 0.8 means the wet tool is "no better than the dry tool" at the moment. "If you really wanted to buy a cutting-edge system you'd buy dry today, you don't buy the wet."

There is no question that the technology works, Hutcheson says, because it has existed for more than 100 years in tools such as the oil-immersion microscope. Questions regarding cavitation and the interaction of water and photoresist are "really just an engineering problem." Other next-generation lithography solutions "were science problems, they weren't engineering problems."

Freeman agrees about the positive effects of the 65-nm push taking place in the second half of 2005, which he asserts "will give us the underlying support for the industry for the year." As for "200-mm-type purchases," the analyst foresees strict limits on any capacity additions.

The mixed-year expectations will show on the faces of participants at SEMI's annual mid-July equipment extravaganza in San Francisco, Freeman says. "Some people will come into Semicon West smiling because things are very good, and then you are going to see other firms trying to figure out how to make ends meet through the year."

The smiling equipment company reps will be the ones with firms "well positioned with respect to logic manufacturing. Firms that are well integrated with Intel will do well, because Intel's going to spend roughly $5 billion this year. Those firms are going to be in good shape because of the 65-nm ramp that's hitting late in the year."

Areas for potential growth include 193-nm steppers, associated metrology equipment for E-beam and thin-film measurements, as well as mask inspection tools. "Single-wafer spin-array processes look good, and ECD copper deposition probably won't drop as much as the rest of the marketplace," Freeman notes. "You've also got a new integrated technology that's emerging such as gate-stack technology where you do oxidation and nitridization in a cluster tool. That will see pretty strong growth with 65 nm."

Companies heavily involved in memory chips may fare less well, Freeman notes. Everyone's mood will depend to a great extent on the demand for memory products. "Right now, our assumption is that it's going to be declining."

Macroeconomic factors, particularly consumer attitudes, will play a big role in the industry's fortunes. "Consumer sentiment is fairly negative," Freeman points out. "You've got inflation worries. When all that stuff happens, people historically tend to hold onto their money. You're not going to see the consumer segment being driven very strongly during the second half of the year." Freeman foresees a reasonably strong Christmas rush, but he believes that the typically healthy back-to-school sales of electronic equipment "probably are not going to be as strong as they might have been had oil prices not topped $50 per barrel."

In addition to global oil prices, Castellano also points to the ripple effect of worldwide economic problems, such as a recession in Germany. "When you have pressures such as this, people are not going to be buying DVD recorders, laptops, and Palm Pilots that require these semiconductors."

Castellano says IC producers are smart for monitoring their inventory control "so that they're not shipping a lot of chips and making chip prices go down. At the same time, the capacity utilization is dropping because they're just not building as many chips. The tools are all in place, but they're trying to avoid a capacity utilization problem and an inventory problem by not building all the chips."

Capacity utilization will continue to drop, Castellano says. Even 300-mm foundries are using only 73% of their capacity at the moment. Total utilization for all manufacturers stood at only 86% at the end of 4Q04, down from 96% at the end of 2Q04, he says.

According to Castellano, in 2001 and 2002, chipmakers had $10 billion in excess inventory and "didn't need to build any more fabs for the next several years." Chipmakers believed forecasters' predictions and expected long queues at the purchasing desks. The semiconductor manufacturers asked: "'What if these guys are right? We're going to be waiting a year now for a lithography tool. Let's place all of our orders early in the year again as a hedge.' It turns out these analysts were right [about equipment sales growth]...but all the tools really weren't needed."

Castellano and Freeman both emphasize that the industry is focused like a laser on inventories right now. "They've turned off the spigot faster than they've historically done," Freeman says. "As a result, we didn't boom as strong as we potentially could have, but it doesn't look like we're going to bust as we have historically. We think we're going to have two slow years, and as demand begins to improve and capacity begins to tighten up, you'll begin to see an overall upturn."

Prospects for 2005 are not bleak, the analyst insists. "It's not gloomy. It's more of an 'is-the-glass-half-empty-or-is-the-glass-half-full' look at the overall market."—JC

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit

Questions/comments about MICRO Magazine? E-mail us at

© 2007 Tom Cheyney
All rights reserved.