2002 Economic Forecast

As we head into 2002, the question on many people's minds is, When can we expect the upturn in the semiconductor industry? Semiconductor International asked industry analysts to give their views on what to expect in the year ahead — how the economy is likely to shape up, and what applications will help pull the industry through the downturn. For more predictions from key industry executives, check out the CEO Roundup.

For the short term, we still support our prediction from the beginning of 2001, that IC sales ($ shipments) will reach a bottom during Q401. We are relying on the same forecasting tool we used for predicting the 2H00 slowdown and eventual decline for 2001, more than one year before it started and without retroactive modifications.

Our forecasting methodology is quite unique to the semiconductor industry as we avoid the pitfalls of human opinion and extrapolation. By measuring a select set of economic factors over the past 15 years, Advanced Forecasting Inc. (AFI) has calculated quantitative-based forecasts to accurately identify the semiconductor industry's turning points. How these factors influence consumer behavior and the consumption of ICs during 2002 and into 2003 should be of particular interest to many companies now after having experienced a record (unexpected by most) decline of IC revenues during 2001.

Recently, we drew attention to the fact that data points from our IC Recovery Index have shown a much slower rate of descent. This is significant because, as with previous recessions such as 1996 and 1998, the position of these data points signals an approaching minimum point for the industry, confirming our forecast of April 2001.

The industry has reached the bottom of an extremely deep recession and, by definition, a recovery during 2002 will take place after hitting the minimum point. Of concern, however, is that the same factors that influenced our forecast for the industry to grow so rapidly in 1999 and 2000 are now very weak. During 2002 and 2003, the pace and the strength of the recovery will be negatively affected by these weak economic factors underlying demand for ICs and end products.

Which specific new applications (drivers) would eventually influence IC sales are secondary in importance to our models of underlying demand, since we are measuring the economic factors that allow the industry to grow or cause it to recede. Even "killer applications" or transition to new technology require favorable underlying demand to cause a period of strong growth for semiconductors.

The year ahead is shaping up to be one of transition for the semiconductor equipment industry — a state of affairs certainly common to the semiconductor industry (represented by both the capital equipment and chip sides) and its boom-and-bust cycles. However, gathering momentum as we move through 2002, the return to full recovery should be gradual and comfortingly free of any significant turbulence.

Last year saw the most severe downturn in the history of the semiconductor equipment industry. When the final numbers are compiled, it looks like the value of 2001 global sales will be almost 40% lower than the record-shattering $47.7B level achieved during 2000. Growth between 1999 and 2000 was an extraordinary 87% — about 5× the historical average annual rate of equipment sales increase — so a slowdown was in order. But nobody predicted this degree of market retrenchment; indeed, few at this time last year were forecasting anything more than a growth rate slowdown (not absolute decline).

But we were just starting to see the first glimpses of light as 2001 drew to a close. There were some tentative signs that orders — for both chips and equipment — were bottoming out during the final quarter of last year. Some companies announced plans to move forward with selective purchases of equipment.

Consequently, the equipment market should recover as we move through the four quarters of 2002. But it won't be until 2003 that semiconductor manufacturers as a group have any compelling reason to add to the industry's bloated level of production capacity. So purchases this year will continue to be almost exclusively technology-driven. Equipment that advances the relentless push from 200 to 300 mm fab production and the tools and process technology necessary to implement and enable the 0.13 and 0.10 µm device generation. Adoption of copper and low-k dielectric films will be part and parcel of this investment. Investments in new equipment and process technology this year will be driven primarily by the continued push on the part of chipmakers to achieve even better operating efficiencies and to further reduce total manufacturing costs.

As we enter 2002, it looks like the best we can hope for is something on the order of a "flat" sales year for semiconductor equipment. Following last year's estimated decline of about 37% in worldwide semiconductor equipment shipments, a much smaller loss on the order of 4% is likely during 2002. But this incorporates the expectation that shipments during the second half of this year will be worth 22.6% more than over the final six months of 2001; we're almost certain to see steep declines vs. the year-earlier totals during both the first and second quarters of 2002.

But clearly things should be looking up as we enter the early days of next year. A veritable "mini-boom" should yield shipments growth of about 31% during 2003. A far cry from the 87% gain realized during 2000, yes, but it would be a less unnerving world for semiconductor equipment makers if market cycles began to more closely resemble those experienced by the rest of the manufacturing sector.

After the dismal year that the semiconductor industry had in 2001, can 2002 be any worse? The surprising answer to that may be yes. Although the turnaround seems to have begun, it is expected to proceed slowly with quarter-to-quarter growth averaging 5-7% in 2002. Because of the steep drop in the first few quarters of 2001, that growth will not be enough to bring total 2002 revenue above the 2001 level. We at In-Stat are forecasting a 5.7% decline from $138.3B in 2001 to $130.5B in 2002.

The turnaround is expected to occur in two phases. Since the steep drop in 2001 was in part due to customers realizing that they had accumulated excess inventories (finished products as well as raw material semiconductors), the first phase of the turnaround will come when inventories are depleted. In general, this has happened in the computer industry, has not fully happened in the wired communications industry, and the wireless industry is somewhere in the middle.

The second phase of the turnaround will occur when demand for end products picks up. Since the IT infrastructure, which absorbs so much of the semiconductor sales, was overbuilt in 1999-2000, this is unlikely to occur until sometime after the general economy picks up and IT user demand catches up with installed capacity.

The 2001 worldwide semiconductor market declined at almost twice the rate of the previous record decline exhibited in 1985
(-17%). For 2002, IC Insights forecasts that the worldwide semiconductor market will be essentially flat with 2001. However, it should be noted that given the low semiconductor market level at the end of 2001, an essentially flat 2002 requires a significant sequential quarterly rebound in the semiconductor industry.

Although the annualized 2002 semiconductor market is expected to be about flat with 2001, on a quarterly basis, the 2002 global semiconductor market is forecast to show a V-shaped recovery. IC Insights believes that overcapacity will still be a problem throughout the IC industry in 2002. IC average selling prices (ASPs) are expected to show steady gains in the final three quarters of 2002, but on an annualized basis, still end up below 2001 figures.

IC Insights believes that excess IC inventory issues will not have a significant impact on the 2002 semiconductor market. The 2002 semiconductor market is likely to be a direct reflection of worldwide economic health and, more precisely, electronic system sales volume. Thus, IC Insights believes that semiconductor unit and dollar volume growth will rebound beginning in 2Q02 and pick up momentum in the second half of the year.

There is no doubt that we have seen the turn in the semiconductor business. It's easy for me to say that in the capital equipment business we have seen a few signs of stabilization. We're certainly closer to the bottom than we are to the top. I expect that equipment orders and shipments will not rise much until late in the first quarter or early in the second quarter of next year.

Assuming the industry follows this script, the second quarter question will then center on the amplitude and duration of the cycle. The progress of 300 mm projects and the developing presence in countries like China will be the most important factors when measuring the life of the equipment cycle.

The turn in the chip industry is more than welcome but I harbor some concerns about the big picture. My biggest issue rests with this seemingly self-perpetuating, always productivity-enhancing, profitless prosperity. We don't have to look very far to see that the ability to generate revenue is redistributed. Unfortunately, the ultimate consequence of globalization and redistribution is equalization of cost and, by equation, the equalization of income. To offset the negative equalizing effects of globalization, the proverbial pie must expand.

When I look at the growth rate of semiconductor revenues over the past decade vs. the long-term historical growth rate, I see the chip business evolving into a mirror image of the growth in end markets (that means a compound annual growth rate that is slower than the 17-18% quoted so frequently). Is there a message here? The message is really rather simple: The world pie may not be getting big fast enough because the changes are simply happening too fast for all of us to adapt in time.

It is difficult to predict the short-term future of the chip industry following the tech wreck of 2000-01 and the industry's pullback from sky-high valuations. The often-asked question these days is, "Are we at the bottom yet?" and "If not, when?"

After watching major cycles surge and wane in my 30-year career in the chip industry, I can make the following assertions with confidence.

The chip industry is real and is very unlike the dot-coms, which sprang up rapidly and bombed with equal intensity. Chips are the crude oil that fuel the electronics industry and the Internet age. They have had, and will continue to have, a profound impact on society worldwide. My comments also apply to equipment and materials suppliers, as well as those in the EDA business. The semiconductor industry is not a fad and will not go away. While that sounds obvious, I point it out because high-tech has been closely associated with the Internet dot-com boom, which was little more than an investment craze built largely on smoke and mirrors. Consumers went for it lock, stock and barrel because they could easily understand the benefits. The average consumer does not have the same understanding of the semiconductor industry, even with Intel's widely recognized marketing campaigns.

Secondly, the semiconductor business cycle is not dead. Back in 1999 and early 2000, some very smart industry spokespersons proclaimed that the business cycle is no longer as relevant to the industry as it was in the past. However, as long as there is overbuilding of capacity; as long as there are technologies that explode on the market like the Internet; as long as commodity chips are priced like airplane seats, there will be a business cycle.

In our view, the upturn will not be quick, given the slowing economy in the United States and weakness in Japan and Asia. InsideChips sees 2002 as a long, arduous turn-around year and 2003 as a return to normal growth rates.

2002 will be a year of two halves for the semiconductor industry — the first half seeing negative (but improving) year-on-year growth rates and depressed earnings, the second seeing real-sell through demand growth, improving margins, and a resumption of capital spending based on our economic forecasts. Semiconductor revenue declined 48% year-on-year in August 2001 driven by inventory draw-downs along the supply chain as much as by weaker end market demand. The three months of data since August back up the assertion that the industry is now past its rate-of-decline trough and is heading toward historical average monthly sequential growth rates. Modeling such rates for 2002 leads to flat calendar revenue growth projections for the coming year. However, this hides a sharp underlying improvement in growth rates during the second half of 2002.

Global semiconductor supply increases have been stopped by the 40-45% reduction in capital spending anticipated for 2001 and 20-25% reduction we forecast for 2002. Capital spending as a percentage of global revenue dropped to 25% during 2001, down from the peak 2000 level of 29% but still well above the trough level of 20% established during past downturns. We estimate that the industry will hit the 20% capital spending/revenue level during 2002. This will lead to a second-half recovery in spending. Assuming semiconductor equipment sales in the first half of 2002 are a bit below the fourth-quarter 2001 run rate, second-half 2002 equipment sales would need to grow more than 15% quarter over quarter for the year to be down 20-25%.

We expect leading-edge capacity utilization to increase through 2002 as demand returns and inventory reduction runs its course. Sub-0.2 µm technology was at 79% utilization in the most recent reported third quarter of 2001. With signs of accelerating demand for advanced photomasks occurring, it's likely that new product introductions will continue to tighten the leading-edge capacity utilization through the first half of 2002. However, 0.25 and 0.35 µm fab capacity utilization is in the low 50% range. Thus, we expect the first wave of investment to be upgrades to existing fabs to 0.13 µm.

Since the transistor does not change much in the move to 0.13 µm, and lithography is extendable from 0.18 to 0.13 µm with adding new masks, we believe that copper interconnect and dielectric-related production equipment and yield management equipment will be the earliest segments to see a recovery. Materials should also see earlier improvement in business trends as fab use improves before capacity-oriented buying resumes.

The yearlong tech crunch seems to be coming to a close as inventories run down and production cuts come in line with final demand. New orders are bottoming or showing signs of life. Finally, the 11 Federal Reserve rate cuts this year are opening the capital markets to allow new equity and debt capital to find their way into more companies. Companies with new capital can start to spend the capital to fix up or expand their businesses.

The road to semiconductor demand recovery is finally being paved. We should start to see a normal seasonal demand pick up in the final quarter of 2001. The seasonally slow first quarter should not be so weak because of low inventories. Signs of a real return to growth probably won't start until we are well into the second quarter.

Initially, demand growth will probably start slowly. World economic growth will lag the United States, given that U.S. demand has been the engine of world economic growth since 1998. Also working against a sharp demand recovery is the apparent lack of new killer applications. Outside of PC upgrades (e.g., more memory for XP, flat panels, and CD burners), new games and replacement wireless phones, it's hard to see the equivalent of the 1990s drivers as the wireless phone, PC and Internet.

For more predictions from key industry executives, check out the CEO Roundup.