In the past, the semiconductor foundry market seemed immune to the cycles that the rest of the industry experienced. Dedicated foundry providers reported higher growth rates than the overall semiconductor industry and did not suffer as much during the slowdowns. The downturn in 2001 was different.

The foundries logged in a more severe decline in demand for wafer processing services than the industry overall in 2001. As the foundries become a bigger player in the market, dedicated foundries will have to learn how to manage through those swings in order to continued the success of the foundry model.

New help could be on the way. During March, Semico launched the "Semico Inflection Point Indicator" (IPI), a new forecasting tool designed to track the next inflection point in the semiconductor industry. Semico uses the IPI in conjunction with its Tracker Index, which was developed to provide a short-term outlook for sales in the following three months. The Tracker Index was designed in the late 1990s to provide additional data that was lost when industry trade groups (such as WSTS and SIA) stopped reporting bookings.

Since then, Semico has been searching for a set of variables that would provide a leading long-term indicator for the direction of the semiconductor industry. The IPI does exactly that, accurately predicting the direction of semiconductor sales six to eight months prior to the inflection point. The IPI reveals significant changes in sales trends and is a prelude to the strength, stability, or weakness in the industry. Semico's IPI tool has proven to be a leading indicator to the trends, anticipating directional changes, and additionally providing a platform for long-term forecasting direction.

The IPI is a composite of various statistics, incorporating semiconductor sales, end market data, and inventories, combined with first derivative analysis. Semico has taken this model and tested it over the past 25 years of industry historical data. The IPI has proven to be an outstanding leading inflection point indicator.

Data from 1984 and 1986 shows how the IPI works. In the period between May and December 1984, revenue shipments were on an upward trend, as indicated by a black line in the IPI chart (see 1984-19986 chart). During that timeframe, the PC market was just evolving, and as a result, semiconductor manufacturers were anticipating a booming market. Based on this data, it would have been easy to deduce that the market was on a positive course.

However, a look at the IPI, as depicted by the blue bars, shows a definite decline in the indicator's rate of change during this same period -- an early sign that the market would be heading down in the following months. Applying Semico's IPI to this historical data, it is clear that this new forecasting tool would have accurately predicted the 1985 downturn.

Similarly, in the period between September 1985 and January 1986, the inflection point indicator began to improve -- an early indicator for the next inflection point. As shown, the start of the upturn began six months after the IPI improved, in February 1986.

Historical data between 1995 and 1997 reveals a similar leading correlation between the inflection point indicator and future sales. The inflection point indicator began to show weakness in July 1995 through January 1996, indicative of the coming downturn that occurred in the first half of 1996 (see 1995-1997 chart). Likewise, the positive movement that occurred in the second half of 1996 was a preliminary sign that the market would enter into a period of positive growth in early 1997.

Perhaps most telling is how the inflection point indicator provided an accurate assessment of the sharp downturn of 2001. Total worldwide semiconductor revenue shipments had been escalating between March 1999 and December 2000. In September 2000, the market still appeared to be on the rise. Sales were up, and manufacturers were adding capacity to address the shortages that had plagued the market in the previous 12 months.

However, once again applying Semico's IPI to historical data, it is apparent that the indicator showed significant weakness in early 2000 and a dramatic downward movement in September 2000 before the market hit a nosedive (see 1998-2002 chart). The IPI would have been an accurate warning sign for the impending slowdown.

By applying the Semico IPI, we can forecast the next industry inflection point. In August 2001, the inflection point indicator began to improve and turned positive in October. The market hit bottom during this time. Applying the IPI model in which the rate of change is an indicator of long-term sales, Semico forecasts the worldwide semiconductor market to show positive growth in 2002. We expect worldwide revenue shipments to reach $163.3 billion, a 17.5% increase.

The foundries have been one of the first to enjoy an uptick in demand. The major dedicated foundries are already reporting higher capacity utilization rates and expected increases in capital expenditures. Semico has had one of the more optimistic forecasts for 2002 and we are sticking by it. The foundries should experience almost 40% growth in wafer demand in 2002.

(See the last commentary on foundry trends.)