In The News
Engineering Times, © 1998, CMP Media Inc.
Chip Market Edges Forward
By Bill McClean
The worldwide electronics industry is composed of many different sub-industries. At the top of the food chain is the total electronic-systems segment with 1997 sales of $770 billion (Figure 1). With a still-healthy 7 percent growth rate forecast for 1998 and assuming an average annual increase of 7 percent to 8 percent after 1998, the electronic-systems market will exceed $1 trillion by 2001.
As a rule of thumb, the semiconductor market grows at about twice the annual rate of the systems industry (i.e., 17 percent cumulative annual growth rate (CAGR) for semiconductors and 9 percent CAGR for electronic systems). Because semiconductors are not standalone entities and must be incorporated into systems, the growth rate for the semiconductor market depends on more than just the growth rate of the systems market.
The driving force behind the historically high growth of the semiconductor industry as compared to the systems business is the increasing penetration of semiconductor value within electronic systems. Through 1997, the semiconductor value as a percentage of electronic-system sales peaked at 21.6 percent in the boom year of 1995. IC Insights forecasts that the average percent semiconductor figure will increase about 1.5 points per year over 1998's 17.9 percent level to reach 25.2 percent by 2002 (Figure 2).
While the correlation between the growth of a country's Gross Domestic Product (GDP) and the health and size of its semiconductor market is not absolute, there is some interdependence. It is very difficult to have a strong worldwide semiconductor market in the face of recessions. For example, over a 19-year period, (1978-1997), the CAGR was only 17 percent. Despite that CAGR, no single year during that period actually displayed a 17 percent growth rate. In fact, the closest any one year came to this "average" was in 1979, 1986, 1987 and 1992, all of which had growth rates that were 7 percentage points above or below the 17 percent figure. Thus, the semiconductor market over the last 19 years grew 10 percent or less, or 24 percent or greater in any one year. And 1998 is forecast to continue this trend -- unfortunately on the low end -- with an increase of only 8 percent.
The rest of the world (ROW) was the only major region to exhibit double-digit semiconductor sales growth in 1997 with 10 percent. Under normal circumstances, the forecast for the ROW semiconductor market for 1998 would be much higher than the worldwide average. However, with all of the current uncertainty surrounding the stability of many of the Asia Pacific economies, IC Insights forecasts that the 1998 ROW semiconductor market will grow only a few percentage points greater than the industry average.
Four Asian countries, led by Taiwan's 30 percent share, represented 82 percent of ROW IC consumption in 1997 (Figure 3). Although Taiwan and China are forecast to display slowing GDP growth in 1998, each country is still expected to post a better than 5 percent GDP increase (China 7 percent and Taiwan 5.7 percent).
One of the weakest and most uncertain countries in the ROW region is South Korea. However, it should be noted that it consumes only 15 percent of the ICs in the ROW segment, which equates only to approximately 3 percent of the worldwide IC industry. As long as a total economic collapse of the ROW region (which could possibly be triggered by the disintegration of the Japanese economy) is averted, the effects of the Asian flu on the worldwide semiconductor market should be moderate.
Volume and pricing
One of the surprising aspects of the 1997 IC market was the 22 percent surge in unit volume shipments, the highest exhibited since 1984, another boom year. Although average selling prices (ASPs) and unit volumes do not perfectly follow the elasticity-of-demand theory (lower ASPs spur higher unit volume and higher ASPs negatively affect unit shipments), historically there has been at least a loose correlation.
IC Insights forecasts that the IC unit growth rate for 1998 will return to a more normal, but still strong, level of 14 percent. If the abnormally weak increase of 2 percent in 1996 is combined with 1997's 22 percent surge, the average increase for this two-year period is 12 percent, the same as the 1993-to-1997 IC unit shipment CAGR.
Interestingly, the IC industry in 1996 and 1997 witnessed decreasing IC ASPs. IC Insights said that this was the first time in industry history that annual back-to-back IC ASP declines took place. With the ASP expected to decline another 6 percent in 1998, the industry appears to be heading for an unprecedented third year in a row of declining prices.
The real driving force behind the current IC pricing trends is capacity. The 1990s have thus far endured both extremes of the capacity spectrum. While 1993 to 1995 witnessed a paucity of capacity worldwide, the industry made a 180 degree turn beginning in 1996. The resulting overcapacity that began in January 1996 continues to drag down IC industry dollar volume growth in 1998.
Unsurprisingly, the IC overcapacity environment can be traced back to the simple balance, or imbalance, of supply and demand. In this case, considering the 22 percent surge in unit volume and 98 percent increase in DRAM bit-volume use that occurred in 1997 (Figure 4), the surplus of IC capacity is not the result of lower demand. Thus, the focus must be on the supply side of the equation.
It is estimated that semiconductor capital spending must average 23 percent to 24 percent of total semiconductor sales for semiconductor producers to keep up with industry demand. The lackluster semiconductor industry from 1989 through 1993, when growth rates were 10 percent or less each year, caused many chip makers to slash capital spending. This frugality is what led to the capacity shortfall that began to show itself in 1994. It should be remembered that because of the 18-month-to-two-year startup and construction time needed for IC facilities to fully come online, the effects of over- or under-spending typically take some time to unfold.
As the IC industry began to gain momentum in 1994 and into 1995, producers attempted to make up for those spending cuts. By 1996, capital spending as a percent of semiconductor sales reached an all-time high of 32 percent. For 1998, the figure will recede to only about 26 percent (Figure 5).
Not surprisingly, the IC industry is rarely at a "normal" spending ratio for very long. It seems that the industry is always overreacting with its spending budgets one way or the other. What typically happens is that 15 IC producers each plan, and spend, to get a 10 percent share of a particular market (of course, this is a statistical impossibility). Siemens, Oki, Samsung and others do not check with each other about capital spending plans and adjust accordingly. Rather, each has its own strategic capacity and spending budget to carry out.
The inherent lag time in getting a new IC facility built and productive, coupled with the usual industry-wide over- or under-spending, has created the history of volatility in capacity. This in turn has led to the sometimes wildly fluctuating IC ASPs, which have helped create the famous -- or infamous -- boom-bust periods in the IC market.
What does the future hold for IC industry capacity? Besides the obvious answer of more volatility, it is interesting to look at the role the Taiwanese producers may play in the future IC capacity framework.
IC Insights estimates that worldwide MOS wafer capacity in 1997 was about 70 million 150-mm wafer equivalents. Moreover, at the end of 1997, Taiwanese IC producers had the equivalent capacity of approximately 10 million 150-mm wafers (with about 50 percent dedicated for foundry work). While this works out to a 14 percent share of worldwide MOS IC wafer capacity, the Taiwanese held only about a 4 percent share of 1997 MOS IC dollar volume (Figure 6).
The situation of the Taiwanese is such that the capacity exists for these suppliers to more than triple their MOS IC market share. However, since they are playing in a zero-sum game, for Taiwanese companies to gain 10 points of market share somebody has to lose 10 points. It is in this battle for market share that significant IC pricing pressures have been created.
For 1998, IC Insights forecasts that some of the severe pricing pressure of 1997 will be relieved. However, with capacity still relatively plentiful, IC ASPs are predicted to continue to slide, decreasing 6 percent in 1998 vs. the long-term average annual ASP increase of 5 percent.
One interesting facet of overcapacity is that time deteriorates the usefulness of some of this capacity. Because of the relentless move to finer feature sizes, increased densities and higher performance, capacity installed four or five years ago must be upgraded through new facilities and/or equipment in order to meet the technological demands of today.
Thus, the producer that intends to be successful over the long term must be dedicated to investing in new capacity even in the face of a slumping market. However, depending on the aggressiveness of these IC producers, or a small group of producers, continued spending during a downturn will usually keep the overcapacity-weak-ASP cycle intact. Of the four major regions, only Europe is forecast to have a higher GDP growth rate in 1998 than 1997.
With economic instability present throughout much of the Asia-Pacific region (especially in Malaysia, Thailand and South Korea) in 1998, most economists lowered their expectations of GDP growth for the individual countries in this area. Although China and Taiwan still possess relatively healthy economies, they have significant trade with the weaker Asian nations. It is this weaker trade situation that is expected to lower all the GDP growth rates of the region.
So with an overall weakening of almost all the major economies in 1998, why would the semiconductor industry show slightly better growth in 1998 than in 1997? Once again, the tie between GDP growth and the semiconductor industry is not that tight. For example, in 1996, when most of the major economies were enjoying relatively high GDP growth rates (e.g., 3.6 percent in Japan), the worldwide semiconductor market declined 9 percent.
For 1998, IC Insights believes that the slowing Asian economies will take some of the strength away from the ROW results. However, this region is once again expected to be the growth-rate leader for electronic system production as well as semiconductor consumption in 1998.
Copyright © 1998 IC Insights, Inc.