After the Blink: A Kinder, Gentler Downturn

Although opinions vary on how good a year 2005 will be, most agree that, if you took the time to blink in 2004, you probably missed the upturn. We asked industry analysts to give us their predictions for the coming year. Most are predicting that 2005 will be the beginning of the next downcycle, but it will certainly not be the cliff-diving experience we faced in 2001.
 
In-Stat
Frank Dickson, General Manager
 
The year 2004 is going to be known to many as the year we blinked. Many have been waiting for a return to the glory days of 2000 with "working" meetings in Maui and bonuses that take one's breath away; however, the peak has come and gone. Although it seems too soon for another downturn, the previous cycle peaked in the third quarter of 2000, and it has now been four years since that peak. The recovery from the last downturn seemed to take a long time to get started, but there were only five quarters of decline before the upturn started in the first quarter of 2002. It is the violent nature of the last downturn that makes it feel as though we barely blinked, and now the rally is a fading image in our rearview mirror.

So, it is time for a downturn, eh? Well, here it is, right on schedule. Since the last run-up was moderate, the current downturn should also be moderate. We don't expect to see another cycle like the one that peaked in 2000 for a long time, if ever again. That one resulted from a coincidence of a significant worldwide recession, bursting of the Internet bubble, the end of the Y2K demand surge, the end of the 3G hysteria, extensive excess inventory or excess capacity in many semiconductor end markets, and extreme overcapacity in the semiconductor industry itself, a combination that is unlikely to be repeated — which is a good thing.

The recovery in this cycle is unique from previous recoveries. It has been driven far more by unit shipment growth than by increases in average selling price (ASP). Although unit shipments fell slightly sequentially (by 0.7%) in the third quarter, they were 14% above the quarterly peak that had been reached four years earlier. ASP, on the other hand, was 12.5% lower in the third quarter of 2004 than it was in the third quarter of 2000. As a result, revenue in the third quarter of 2004 missed breaking the quarterly record, albeit by less than 1%. To be sure, ASP rose during the upturn as fab utilization increased, but the rise was neither as fast nor as far as the increase during the 1999-2000 upcycle. A 96.4% quarterly utilization rate peak in 2000 was accompanied by a $0.567 ASP peak, while the 95.4% peak in 2004 corresponded to a $0.488 ASP peak.

This is clearly a "bad news/good news" story. The bad news is that the upturn was not as strong as it might have been if the old ASP patterns had prevailed. The good news is that the downturn will be milder than it would have been in the past. In-Stat continues to believe that ASP is in a long-term decline and that short-term swings will become proportionally smaller as time progresses.

In-Stat expects that 2004 annual worldwide semiconductor revenue will be $211.4B, 27% higher than 2003's level. However, 2005 will be a down year with annual revenue sequentially off by 5.7% to $199.3B. In 2006 and 2007, we should see annual growth of 11.2% and 9.2%, respectively, before the next downturn occurs in 2009, when revenues are expected to fall by 2%. On average, unit growth during the forecast period will outpace revenue growth as ASPs continue to follow a long-term downward trend despite occasional small upticks during tighter markets.

2004 was a year of paradox for the semiconductor equipment market. While the market was ramping to 66% growth, the industry was waiting for the other shoe to fall. After four quarters of growth, the shoe dropped, with growth slowing significantly in Q4 of 2004, thus placing a significant damper on the outlook for 2005. At this time, the equipment industry is in a quandary: Why such a short ramp? Are the cycles shortening? Will they be steeper and faster? Will I be able to maintain positive net profits during this downcycle after reorganizing my manufacturing and supply chain? These are all difficult questions to answer when the crystal ball is cloudy.

So what is the outlook for 2005? At this juncture, with three years of positive growth in the semiconductor ranks and a boom year in the range of 26% growth in 2004, semiconductor growth is predicted to be flat in 2005. As a result of the 300 mm shells that were ready for equipment installs, a great deal of capacity has been added throughout the year. Capacity utilization peaked in Q2 at 98%, and will decline throughout 2005. About 8% additional capacity will be added in 2005 as a result of the equipment that has already been shipped or that is scheduled to ship in 2005. Capital expenditures are forecast to decline 13% in 2005, with wafer fab equipment expected to decline by 17%.

During the past year, Gartner Dataquest was forecasting a ramp, slowdown and ramp scenario; while this did not occur in 2004, it looks like the slowdown the industry is experiencing is a period of digestion. After four strong quarters of growth, the industry is now taking time to determine how much capacity is in place, if that capacity will meet the current demand, and what will be needed for the future. The current quarterly run rates look like the first three quarters of 2005 will have negative growth, with positive growth returning in Q4. The return in growth will begin the start of a growth period that should carry the industry into 2008.

During this period of slowness, firms should continue to work on improving their operations, looking closely at the core competency of the company and determining which operations will strengthen the bottom line. Firms should continue to look for alliances that will enable them to stretch their R&D dollars as well as position themselves for future growth for when the cycle turns up again.

In May of last year, VLSI Research began to warn of a slowdown as key indicators began to point down. It started with IC prices in the spot market. Starting in April, VLSI's Chip Price Performance Index (CPPI) began to drop at a Moore's Law clock rate of a halving every 12 months. By August, order activity for chip equipment was falling enough for me to label 2004 as another 2-for-1 deal — or two years of growth in one. The last time I did this was in 2000.

Nevertheless, the situation being faced in 2005 is very different than four years ago when the industry faced the worst downturn in history. For chipmakers, conditions are far better than they are for equipment suppliers. When this was written in late November, chipmakers were still experiencing increasing contract prices, increasing yields, increasing RPSI (revenue per square inch) and historically low inventories. While the general perception was that inventories were high, rising prices belied shortage conditions. Also, there has never been a downturn when RPSI was rising. It all points to a good year for chipmakers in 2005.

Yet, 2005 will have its difficulties. China has put a drag on electronics demand. Normally, this would not worry us. However, for the first time on record, an economy other than the United States is the primary driver of electronics demand. China became the most important regional economic driver in 2004, as it began to surpass all other regions in electronics consumption. Then, the controlled slowing of the Chinese Central Bank initiated in early 2004 slowed chip demand.

So, VLSI expects growth in worldwide electronics revenues to slow to about half the 12% rate achieved in 2004. Chips should grow in the 5-10% range. However, equipment growth will be flat at best and probably negative. The problem is in the combination of slowing IC growth and too much equipment. The result is that realized capacity utilization was down to 81% in October and unrealized capacity only cut another 3% from this figure. The low level of unrealized capacity is why we don't see a hard crash for 2005. While we have plenty to worry about for 2005, I suspect there will be more to worry about for 2006, so stay tuned.

It is interesting to note that, as of late 2004, there were no industry associations (e.g., the SIA, WSTS, etc.) or market research companies known to us that were predicting a higher IC market growth rate for 2005 than for 2004. Thus, it appears that nearly everyone believes 2004 was the peak growth-rate year for the IC industry in the current cycle.

Although each of the past IC industry cycles had their own particular characteristics (e.g., duration, magnitude of increase and decrease, etc.), all have exhibited a similar pattern. In every one of the past five industry cycles, once the peak growth-rate year was registered, the best IC industry growth rate the following year was only 11% (with 2001 exhibiting the worst decline, 33%).

Even more interesting is what the history of the IC industry cycles reveals about the second year (i.e., 2006) after the peak growth-rate year. In every past IC industry cycle, the IC market registered growth (typically low single-digit growth) in the second year after the peak year. If the current cycle follows the pattern of the past five cycles, 2006 will be a growth year and not a year when the IC market declines.

At the present time, most agree that the long-term average annual growth rate for the IC industry over the next 10 years is very likely to be 8-12%. When the IC market grows at a rate that is more than twice this average (e.g., 2004's 25%+ increase), there must be a period of time when the industry grows at a similar rate below the average. The phrase that comes to mind when describing the IC market adjustment period after a peak growth rate year is, "Pay me now or pay me later." But, rest assured, there will always be a payback.

IC Insights believes that the sequential quarterly IC market growth rates in 2005 will follow the typical historical pattern of a "correction" year. Thus, IC Insights believes that the bulk of the IC market slowdown during 2005 is likely to take place in the first half of the year, with quarterly market improvement beginning in the second half of the year and carrying over into 2006. There is good news, however, in that 2006 could be a surprisingly strong year for the IC industry.

2004 was a great year by many counts. There was a nice rally in chip sales, which fed demand for semiconductor equipment. The growth (finally) generated some optimism for a very battered sector of the economy.

While the year could be considered good when measured against past performances, it also served to highlight the many long-term issues that still need to be addressed. Those issues being commoditization, too much manufacturing capacity and, last but not least, profitability.

There's little doubt that the first few quarters of 2005 will be a struggle for the industry. A chip inventory hangover — a product of over exuberant forecasts for demand by the end market — will be with us for a while. Yes, many things have changed over the past few years, but the industry still struggles to match end demand to production levels. It does appear as if we're doing better. Acknowledgement and reaction times are certainly faster.

In looking at the numbers that populate my spreadsheets, I get the sense that chip cycles have entered a phase where they will be much more frequent but not as amplified (top to bottom) as the ones we have witnessed during the past. Some of this is caused by the narrowing of the marketplace, and part is because of a maturation process.

It's been said by many that the chip and chip equipment industry has too many players and that a great number will die on the vine by the end of the decade. It's pretty clear that, to be profitable, size matters. I am expecting a great deal of consolidation, rationalization and reorganization during 2005. That process will be most evident in the capital equipment arena. We're already seeing it take hold in the ranks of the subsystem supplier. I would not be surprised to see this process move up the food chain as the year progresses. One could even go as far to say that 2005 will be a year where a lot of survival mechanisms kick into high gear.

Although the semiconductor industry has seen inventories climb and guidance lowered at key chip companies, we do not believe the current upswing in the business cycle is over. The weakness is temporary, in our opinion, as the supply chain readjusts to a rapid run-up in revenues created by purchasing managers worried about supplies. Overheated markets lead to rampant optimism and often trigger a downturn. The slowdown that occurred in the past quarter has a silver lining: It will be a factor in preventing the overheating in the industry similar to the peak that occurred during the 2000 technology boom-bust period.

As a result of these unfolding events, analysts have been bumping up forecasts for this year and downgrading next year. There is significant churning among analyst forecasts as day-to-day events in the news make the spin on the near-term outlook either optimistic or cloudy.

We are not as pessimistic as Wall Street appears to be at this time. We think it is reasonable to expect corporations and enterprises to return to the market, as IT spending was curtailed severely after the bubble burst in 2000. The pressure to replace older PCs in the installed enterprise base continues to increase. Wireless developments and the coming 3G cellular deployment cycle will fuel the chip industry in its next telecom phase. Consumer electronics will continue to converge with PCs with advances in displays, home PC/media networks, and the integration of theater PCs with never-before-seen features and ease of use.

This recovery has been marked by a state of cautious optimism among CEOs and, as a result, discretionary spending remains tight. The current business cycle upturn in the chip industry was tempered by an uncertain election year, high oil prices and uncertainties about the war on terrorism. The election issue has been resolved, and it is providing the markets a welcome boost. As the other problematic issues become better clarified, we believe semiconductor stocks will continue to make gains as investors grow in confidence and the industry gets past the current "soft patch" thinking.

If you blinked, you might have missed the semiconductor recovery. The semiconductor industry revival, having commenced in mid-2003 and continuing through the first half of 2004, now has fizzled in late 2004. The uptick of 2003-04 appears to have been one of the shorter recoveries in the history of the semiconductor industry.

Unfortunately, this brief resurgence follows a decade-long trend of declining average growth in the semiconductor industry. The rolling five-year compound annual growth rate (CAGR) for the global semiconductor industry has been stuck in single digits for the past six years. And while 2004 seemed like it might have been the springboard for a return to double-digit average growth, optimism has given way to caution, with most forecasters now offering rather weak growth scenarios for the next two years.

The semiconductor industry is undergoing a much-needed correction at this time as it strives to reduce bloated inventories. However, this period of rectification is very likely to transform into a semiconductor overcapacity situation, as sales growth to equipment markets begins to slow.

iSuppli's semiconductor revenue growth forecast for 2005 has been trimmed to the mid-single-digit range, with a 4.7% increase in revenue expected for the year. Electronic equipment production, which drives semiconductor sales, is expected to increase by 6.2% in 2005, down from 10.1% in 2004. While this growth is slower than in 2004, an increase of >6% still is quite respectable.

The general trend of increasing semiconductor content in electronic products generally means that chip-sales growth is higher than end-equipment revenue increases on an annual basis. However, the needed inventory correction, coupled with a return to semiconductor ASP erosion, will constrain semiconductor revenue growth so that it is in line with the increase in end-equipment sales.

iSuppli continues to believe that 2006 will represent the nadir of the present semiconductor cycle. The second half of 2005 will bring some modest growth, with sales rising by 4% compared with the first six months of the year. However, this seasonal uptick will not lead to a recovery because the slowing market, led by a DRAM revenue downturn, will limit worldwide semiconductor industry growth to only 2.2% in 2006.

Semiconductor shipments in 2005 are likely to surprise on the upside, since most forecasts were made during the weakest months of 2004 when sales growth was stalled and manufacturers had little pricing power. That weak period was misinterpreted as the end of the market expansion cycle rather than recognized as an inventory cycle within the market expansion cycle.

The 29% shipments growth in 2004 was the sum of a quick 17% jump in sales early in 2004, followed by essentially no change in sales during the six months through October and probably a month or two longer. Over ordering early in 2004 caused an inventory surplus, which was aggravated by the loss of consumer and business purchasing power when oil prices soared, U.S. economic policy turned from stimulation to restraint, and China bluntly restricted investment to slow an overheating economy. Semiconductor shipments were less than chip consumption in assembled products.

The inventory surplus ended in the United States at the end of the summer, but it lingered into the fall in Europe because of the appreciation of the euro, and in Asia, where earlier over ordering was most severe.

Expect semiconductor shipments to resume growing at the end of 2004. The underlying economic environment is unusually strong. World economic growth in 2004-05 will be the strongest two-year period in more than 20 years. This will increase 2005 shipments by 15% unless the 2004 overexpansion of semiconductor capacity ignites excessive price cutting.

2004 has proven to be a very strong growth year for the entire semiconductor industry, including the material and equipment segments of the industry. Strong double-digit growth has been observed in most segments of the materials markets. At $35B, 2004 will end up being the second-highest revenue year in history for semiconductor capital equipment market tracked by SEMI.

While the industry has experienced strong growth, recent data trends and industry reports have shown that the semiconductor industry is slowing down. The recently issued SEMI Capital Equipment Consensus Forecast reflects this trend, as the surveyed SEMI member equipment manufacturers are now estimating a 5% decline in the total equipment market in 2005, with the assembly and packaging segment expected to decline ~15%. A moderate recovery of 3% growth in the equipment market is expected in 2006.

The SEMI Silicon Manufacturers Group is forecasting low to mid-single-digit growth in silicon wafer area shipments over the next two years, though stronger growth in 300 mm wafer shipments will occur over this timeframe as new 300 mm fabs ramp production. Production will also increase using silicon-on-insulator (SOI) and strained silicon wafers as semiconductor manufacturers increase production of sub-100 nm process technologies. These advanced substrates, along with other new and advanced materials, are critical for the next generation of devices outlined by the International Technology Roadmap for Semiconductors.

When all the numbers are in, 2004 will prove to be one of best years ever for the worldwide semiconductor industry. Global sales are now projected to reach $214B, surpassing the previous record of $204B in 2004, with a year-on-year growth rate of 28%.

Going forward, the picture is somewhat murky. After a very strong second quarter, concerns about a number of issues — high energy prices, end-market demand, inventory, capacity and fundamental market changes — dampened sales growth in the succeeding quarters.

While some industry analysts have predicted that 2005 will be much like 2001 — a severe decline in sales after a record year — there are significant differences in underlying conditions. Major end markets are likely to see slower growth rather than actual declines. For example, PC unit growth is projected to slow from 14% to 10%. The handset market will slow from 30% unit growth (670 million units in 2004) to 8% in the face of a slowdown in new subscriber growth and a saturation point in upgrades from 2G and 2.5/3G devices. Digital camera sales now exceed film camera sales, but unit growth is likely to slow from 24% to a more modest 6%.

The inventory situation is also very different from 2001. Inventory in the supply chain began to build in the second quarter, reaching roughly $1B at the end of the third quarter. In contrast, there was $15B in excess inventory at the end of 2000. The entire electronics industry reacted much more quickly to inventory concerns in 2004, and there is now evidence that excess inventory in some sectors has been depleted.

Capital spending has also been more restrained in the present cycle. Currently, capital spending is running at 70% of 2000 levels at $46B this year, compared with $65B in 2000. Again, the industry has reacted much more quickly than in the past to modulate capacity expansion in the face of slower growth in demand.

Meanwhile, we are seeing a fundamental shift toward consumers as a primary driver of semiconductor sales. If we define "consumer products" as products purchased by individual consumers with their own money, consumers ultimately drive roughly half of all semiconductor sales. In this changed marketplace, underlying economic conditions have a greater influence on chip sales. For example, increased energy prices over the past 12 months have taken $656 out of the $2700 the average U.S. household has available for discretionary purchases.

Taking all the factors into account, total semiconductor sales in 2005 are likely to be essentially flat at the record level of 2004. There are encouraging signs that the industry has learned from past experience and that future swings in this very cyclical industry may be more moderate.

Despite the current slowdown, 2004 was a great year in terms of growth rate in ICs, silicon wafers and semiconductor equipment. Shipments of worldwide ICs in revenues during the first 10 months of 2004 totaled $145.2B (based on a three-month moving average), a 32% increase from the same period in 2003. Taking into consideration the current decline, growth of IC sales for 2004 is likely to end at ~30%.

Shipments of semiconductor equipment worldwide during the first nine months of 2004 reached $27.3B, a 79% increase from the same period in 2003. Because of the current slowdown, semiconductor equipment sales for 2004 are likely to end at close to 70%+.

Advanced Forecasting's Materials and Equipment-Components Turning Points Forecast alerted as early as March 2004 that a slowdown was pending for the second half of the year. The impact of this slowdown is currently felt throughout the industry. In spite of it, sales of silicon wafers during the first nine months of 2004 yielded a 27.2% increase over the same period a year earlier.

Advanced Forecasting's quantitative forecast model for the IC cycle, which has accurately predicted each of the major turning points in the industry for nearly two decades and is not modified retroactively, shows 2005's year-over-year growth to be below 10%, substantially lower than that of 2004. We expect that the current softness in IC shipments will continue into the first quarter of 2005. However, the decline is expected to be mild, in contrast to the 2001 recession, caused in part by the cautious behavior of industry leaders that stems from the consequences of that recession.

Growth in semiconductor equipment in 2005 will also be substantially below that of 2004. Our forecast for the materials and equipment-components market is expected to remain flat through the first quarter of 2005 as well.

The 2001 recession continues to have far-reaching effects on the industry today. Before that recession, boom periods drove companies to extrapolate the past, disregarding the fact that previous periods of decline could occur again. This attitude caused excess capacity, inventories, and the inevitable recessions. The 2001 recession changed this behavior, with companies now looking for any hint of a slowdown and decreasing the lag time in response.