SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Politics : Formerly About Applied Materials
AMAT 334.55+4.7%10:27 AM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Proud_Infidel who wrote (46637)5/11/2001 9:47:30 PM
From: Proud_Infidel  Read Replies (3) of 70976
 
Are IC suppliers heading for the 'Big Chill' with lower growth rates?

One troublesome theory says slowdown may be underway because of too much capacity, stalled chip content in systems
By J. Robert Lineback
Semiconductor Business News
(05/11/01 09:56 a.m. EST)

SCOTTSDALE, Ariz. --Something just isn't adding up in the fickle semiconductor markets, and that's causing one veteran chip analysts here to wonder if the industry is slowing down from its historical 17% annual growth rates of the past 30 years.

"This is not business as usual for this industry," observed analyst Bill McClean, president of IC Insights Inc. "Negative years [in terms of IC sales] used to be a very rare occurrence, but it's not so rare any more," he lamented.

McClean is still formulating his thoughts on what might be causing a six-year slowdown in overall IC growth, but he is beginning to finger two prime suspects:

A lower-than-expected ceiling on semiconductor content in systems; and,
Too much capital spending as a result of silicon foundry companies trying to convince major IC houses to outsource more of their production.
The alarm bells started going off when McClean began looking at growth rates since 1995. Pounded down by three "negative growth" years since then, worldwide IC revenues have not come close to the historic 17% compound annual growth rates of the 1970s, 1980s, and early 1990s, noted McClean, who plans to issue a report on the overall trends in June.

"The average growth rates of the 1970s and 1980s were made up of 12% unit volume growth and a 5% increase in average selling prices," explained the Scottsdale-based analyst, who has tracked the industry since 1980. "If we look back over the last five years we see a relatively normal 10-11% unit volume increase each year, but we also have a 2% drop in ASPs."

The drop in average selling prices "is the whole difference between the growth rates we have seen in the previous 25 years and the growth rates in the last five," he added. "The industry is just not allowing ASPs to get a foothold to increase."

Can't catch up

Average selling prices for integrated circuits hit their high point in the 1995 boom at $2.68 each, while in the troubled first quarter of 2001, ASPs for ICs were at a dismal $1.94, McClean reported. "Last year, the average was $2.04," he said, noting that even in a good year of industry growth, ASPs have not been able to catch up with the 1995 average.

"We cannot come up with a five-year forecast that gets us back to an IC ASP of $2.68 that we saw in 1995," McClean cautioned. "We may look back from 2005 and realize that we don't have an ASP that reached what it was 10 years ago! That is a fundamental change in the semiconductor industry."

McClean said there is nothing wrong with the market's consumption of IC units, but he see plenty of serious problems in pricing trends. He now worries that the IC industry might be bumping up against slower growth in semiconductor content (as a percentage of systems sales), while chip suppliers are getting clobbered by too much capacity.

IC content as a percentage of system revenues reached 24-25% in the booming fourth-quarter of 1995, McClean noted.

"Given a good 18% growth rate for ICs [in 2002 after an expected 9% drop in sales during 2001], and a 20% increase in 2003, we would get to around 24% in IC content in terms of a few years out--meaning we will be at the same IC content level in systems sales that we were at eight years ago.

"We may be bumping up against the limits, and while IC content might not completely flatten out in the next 10 years, it could go up only to 28% or 26% vs. 35%," he added, referring to the general belief that integrated circuits will top out at about one-third the value of systems later this decade.

He suggested that if IC content does not increase gradually to the mid-30 percent range later this decade, that will dampen off average growth rates from the industry's average of 17% since the 1970s. "Let's say we have 12% average annual growth in IC sales over the next 10 years [instead of 17%]," said McClean, playing the devil's advocate. "Our view is that the way this would work out is we'd see a 10-11% increase per year in unit volumes, and only a 1-to-2% increase in ASPs per year."

Flood of the fabs

One other major factor holding down average selling prices is wafer fab capacity--too much of it. Semiconductor capital spending rates have heated up and not cooled off since 1995. To understand how capital spending trends have shifted, McClean looks back at the downturn/recovery cycles of the mid-1980s and 1992.

Like the current slump of 2001, the 1984-86 recession was caused by an inventory buildup in ICs--back then for the first PCs vs. today's cell-phone and communications markets. During the inventory burn-off and recovery years of 1985 and 1986, industry-wide capital spending was down at 17% of semiconductor sales, noted McClean.

In the downturn of 1992, semiconductor capital spending was at 19% of chip revenues, he said. But then the level of investments in downturns suudenly shifted upwards in the late '90s.

"In 1998, the low point was 21% [capital expenditures as a percentage of IC sales]. Even in 2001--with a 20-to-25% cutback in capital spending from record levels in 2000--investments in terms of percentage of IC revenues will end up around 24-25%," McClean warned.

"If the IC suppliers get any whiff of double-digit growth again next year, they will probably start increasing their spending again," he said.

The foundry factor

The biggest change in semiconductor manufacturing during the past five years has been the strong emergence of the pure-play foundries in Asia, in particular Taiwan, which have set their sights on taking business from fab-operating chip companies--otherwise known as "integrated device manufacturers, or IDMs.

"You have the foundry companies trying to convince the IDMs to not build the next wafer fab, and all of them are counting on taking 30-to-50% of capacity from those integrated device manufacturers," McClean observed. "Companies like TSMC, UMC, Chartered--and now startups like Silterra and 1st Silicon [in Malaysia] and even Dongbu in Korea--are planning to take capacity from the IDMs.

"But the problem is that the 'word' has gotten down to the IDMs, which are not now planning to outsource 30-to-50% of their capacity," McClean noted.

In fact, it has gone the other way since the current downturn started. In the first quarter, major IDMs have shifted products back into their own wafer fabs from third-party foundries. That movement has driven foundry capacity utilization down below 50% in many cases in the second quarter of 2001, compared to what some analysts believe is 80-to-85% for the entire chip industry (counting the integrated device manufacturers).

The aggressive silicon foundries and IDMs are "kind of running into each other," McClean noted. And he believes this bump-and-grind will continue well into this decade until the IC industry works out the relationships between integrated device manufacturers and third-party foundries.

What should the IC industry expect? Well look no further than the still-troubled DRAM segment, which hasn't recovered from its plunge in the late 1990s. The DRAM business peaked in 1995 at $40 billion, "while this year it will be $21 billion," said McClean. "Six years later it will be half of what it was in 1995. Even if you put good growth in the forecast for DRAMs out to 2005, you might reach the market size of 1995.

"While there used to be 10 major DRAM manufacturers, there is now five. Is that good enough [of a consolidation]?" asked McClean. "It appears not, and may be we should have only two major DRAM manufacturers. The bottom line is that the industry hasn't cut back enough."

For the IC business--excluding DRAMs--the problem may be cutting back as well.

"I'm not happy about this scenario, believe me, but it could be that we could see growth rates of just 8-to-10% in the next five years," McClean said. "I'm really concerned about what happens with 300-mm fabs come online and companies try to make their last stand with those wafer fabs. It could get pretty ugly.

"A growth rate of 10% is pretty good for other industries, but for ICs it's a disaster," he added.
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext