In The News
Business News, © 1998, CMP Media Inc.
Manufacturing Capacity is Industry's Driving Force
By Bill McClean
The current demand for integrated circuits is growing faster today than it has in more than a decade. Yet average selling prices (ASPs) are declining for the second year in a row.
What this seeming contradiction shows is that the real driving force in the IC industry today is capacity. Never mind deep submicron processing, billion-transistor memories, or system-on-a-chip ASICs. It is capacity, not demand, that dictates IC pricing. And an overcapacity situation that began two years ago has been drawing down IC industry dollar volume growth in one of its typical boom-bust cycles.
This overcapacity had a dramatic effect in the marketplace for the year just ended. IC unit volumes are estimated to have increased by 23% in 1997, but the growth in worldwide IC revenues was held to just 5% by too much capacity.
The culprit was the steep slide in IC ASPs, which was brought on by excess wafer fabrication capacity. Price cuts were even greater in 1997 than they were in 1996 because the overcapacity situation had gotten worse. For 1997, IC ASPs declined by 14%. This steep fall followed a 10% drop in ASPs in 1996. Also contributing to the lower average selling prices (when they are calculated in U.S. dollars) were the falling exchange rates in Japan and Europe.
In 1998, average selling prices will continue to be a bit of a drag on industry growth.
This is probably the first time in the history of the IC industry that IC ASPs have declined two years in a row. While historical data from the Semiconductor Industry Association show ASPs declining in the 1984-1985 period, they did not include Japanese production. Combining Japanese data with the SIA statistics, however, would most likely have showed an increase in average prices in 1984.
IC Insights sees prices coming up only 2% in 1998, but that's a big improvement considering that the industry was still on the down slope of pricing in the fourth quarter. The ASPs for ICs in all of 1997 is estimated to be $2.05, but the average price in the fourth quarter pricing was $1.95.
IC prices will have to rise quite a bit from where they were in December to get the industry's ASP up to $2.10 in 1998. But they should firm gradually throughout the year and then continue to improve in 1999. The 2% increase in ASP should be enough to help global revenues show a 17% growth rate in 1998 because of higher unit volumes.
This optimistic outlook is predicated on rationality returning to capital spending outlays. And this change in attitude is now beginning to happen. As a result, the IC industry appears to be heading into a period of "normalcy" in average selling prices. The current outlook for the next five years is that the industry will show the 7% annual increase in ASPs that it has recorded over the past 20 years. This has come primarily from the growing functionality of ICs.
Overcapacity will still exist in some market segments, but it won't be nearly as prevalent as it was in 1997. Helping to bring more of the market into balance are the troubled economies in Asia, coupled with the continuing bad market conditions in DRAMs. This environment will lead to cutbacks in capital spending, which in turn will bring back normal ASP increases over the next six years.
This slowdown will be similar to the situation in the late 1980s and early 1990s when the Japanese began to curtail their capital spending after their economy took a dive. This situation eventually contributed to a major shortage in memories, which was then followed by high ASPs and phenomenal growth in DRAM revenues until 1996.
There was a belief in early 1997 that the industry would never see price increases again based on the huge increases in capital spending levels worldwide. But historically, at some point in time, companies will pull back. And that's what is beginning to happen now.
Instead of being triggered by Japan's problems, the cutback this time around will be caused by a South Korean economy that is in trouble. Add that to poor DRAM prices and Korean suppliers can no longer keep spending as if there is no concern for tomorrow. Reality is setting in.
The duration of these slumps plays a key factor in reducing the ranks of suppliers. No one seems to care after just one bad year. But after two bad years, some companies are scared out of the business. In the late 1980s, the DRAM makers had three bad years in a row and a number of companies got out of the DRAM market. That led to the shortage of DRAMs in the early 1990s.
Some companies are getting a little scared, now that there has been two bad years in a row for DRAMs. But that kind of attitude should help firm up pricing and make for a more stable market in 1998.
IC Insights is predicting a 14% increase in DRAM ASPs, based on the assumption that the market will be impacted by a continued pullback in capital spending by Korean, Japanese, and other memory suppliers.
But the modest improvements expected in the 1998 IC market will not be sufficient for some companies to continue in markets depressed by oversupply. The problem is that DRAM pricing will not recover sufficiently to bring that market anywhere near the revenue levels of two years ago. IC Insights estimates that DRAM revenues dropped 20% in 1997 after falling 38% in 1996. So a 14% increase in 1998 ASPs will not support a lot of new capacity. This will be perceived by some suppliers as a third bad year in a row and will lead some suppliers to pull back in memories.
Back in 1994-1995, DRAM plans had been based on much higher revenue levels. With the business still far below those levels, IC companies are not as interested now in this segment. The DRAM market won't be coming back to the 1995 levels until 2001.
Falling ASPs in 1997 stole growth from more than just the DRAM market. Overcapacity in flash memory devices pushed down their ASPs, which fell 33% in 1997. Total flash revenues, however, increased 5% because a whopping 56% more units were sold.
Both DRAM and flash memory ASPs are expected to fall by 7% in 1998. That's not good news. They should be increasing 5% to 10% annually because of the continuing increase in the number of bits on each chip.
The ASPs for analog ICs also dropped 7% in 1997, but a 25% increase in unit volume enabled analog revenues to grow by 16%. And while the ASPs for microcontrollers also fell 11% in 1997, revenues still rose by 12% because unit shipments soared 26%.
History just may be repeating itself. Current conditions could be setting up a shortage of leading-edge capacity in 1999, which could lead to severe shortages and a repeat of market conditions of the early 1990s.
The problem with this industry has been the herd mentality in capital spending. Spending for semiconductor production equipment as a percent of semiconductor sales must average 15% to 16% in order for producers to keep up with industry demand. The lackluster IC industry from 1989 through 1993 (when growth was 10% or less each year) caused many producers to slash their capital spending budgets.
The frugal spending on capital expansion during this period is what led to the capacity shortfall that showed up in 1994. The effects of over- or under-spending take time to unfold because of the 18-month to two-year startup and construction time needed to get a fab on-line.
As the IC industry started gaining momentum in 1994 and 1995, IC producers began to attempt to make up for the spending cuts of 1989-1993. As a result, capital spending for semiconductor production equipment in 1996 reached an all-time high of 20% of semiconductor sales. For 1998, this figure will only recede to about 17%, which is still in the "overspending" range.
It seems that the industry is always overreacting with its capital spending budgets one way or the other. Unlike OPEC, which counts how many barrels of oil leave port each day, no one polices the IC industry's capacity and spending. What typically happens is that 15 IC producers all plan, and spend, with the goal of getting a 10% market share, which of course is impossible.
The inherent lag time in getting a new fab built, coupled with the usual industrywide over- or under-spending, causes these widely fluctuating IC ASPs. That's what will continue to cause the IC industry's boom-bust cycles.
Copyright © 1998 IC Insights, Inc.