2008 Economic Forecast: Fairly Unpredictable

Klaus Rinnen, Managing Vice President, Semiconductor Manufacturing, Gartner Dataquest

2008 will be a year when the industry's hopes and realities clash. Hopes are for a strengthening expansion in 2008, a year that should have history, speaking a lot of momentum, given the U.S. presidential election and Olympic Games. However, these hopes are confronted by the market realities of oversupply in the memory sector and the risk of slower end-user electronics sales caused by a rising U.S. recession risk and macroeconomic woes in many parts of the world.

So will 2008 bring hope or reality, growth or contraction? It will be both. And depending on the position in the semiconductor industry supply chain, segments and companies within will either enjoy market expansion or wrestle with contraction. For semiconductor sales, we hold to an industry expansion in 2008, albeit growth expectations decline. And for capital spending and equipment, we forecast a contraction. As the industry remedies its oversupply situation in the memory sector, spending will contract. Therefore, 2008 bears a varied picture where fortunes depend on segment and customer exposure.

Diving into a bit more of detail for semiconductor sales, we have revised downward our semiconductor market growth forecast for 2007 and 2008. For 2007, our forecast changed from 3.9% to 2.9% growth because of deterioration in DRAM and NAND flash memory market conditions compared with our 3Q07 update and because of generally weakening markets, which are still suffering from downward average selling price (ASP) pressure.

Although semiconductor supply chain inventory levels have been brought back under control throughout 2007, all eyes are on holiday season sales of electronic goods as the industry looks for early signs of potential demand-side weakness. This sensitivity to weakening demand has been heightened because the chances of a U.S. economic recession in 2008 are being talked up on the back of ongoing financial market uncertainty. At the very least, it now looks likely that global gross domestic product (GDP) growth will slow down in 2008, with the obvious impact on consumer spending on electronics and business IT budgets. While positive growth remains the most likely forecast scenario for next year, there is an increasing chance of a downside risk to the forecast, and we envisage virtually no chance of an upside scenario. Therefore, for 2008, our growth forecast has been revised downward from 8.2% last quarter to 6.2% now.

With respect to capital spending and the equipment market, we are forecasting a correction. Our latest 4Q year-end Q1 forecast has a contraction of ~10%, much dependent on the individual market segments. This decline is driven by a significant drop in DRAM-related investments of ~30%. While these are partially offset by continued strength in NAND-related investments, it should not be enough to compensate, especially considering that we do not expect much strength in foundry spending for 2008.

Overshadowing all are the, in recent months, steadily climbing prospects of a U.S. economic recession. IF — and we are not saying that it will happen — such an event would come to pass, we believe that our industry could fall deeper into a demand-driven downcycle. Semiconductor sales would contract and capital spending would further decline into the -20% range.

As we look to 2008, the year may become a nail biter. There is reason for hope and there is reason for caution. Can the industry continue its expansion streak? Yes, we think so, but the risks are increasing up and the expansion potential is declining.

Bill McClean, President, IC Insights

In 2007, 3.9% global gross domestic product (GDP) growth was slightly above the 30-year long-term average of 3.7%, while 2008 global GDP growth is expected to be slightly below average at 3.6%. In 2007, worldwide electronic system sales increased only 5%. However, driven by better computer system sales, total 2008 worldwide electronic system sales are forecast to increase 6%, 1 point up from 2007, and 1 point below the 30-year average annual growth rate (AAGR) of 7%.

In general, the "wild card" in the 2008 forecast continues to be the worldwide economy — the U.S. economy in particular. Most economists put the chance for a U.S. recession in 2008 at about 40–50%, with the highest chance of occurrence in the first half of the year. While much has been made about the "decoupling" of the United States with other countries' economies, we believe that a U.S. recession would still negatively affect the worldwide economy.

All of our forecasts assume the United States and worldwide economies will "muddle through" 2008 without encountering a recession. A couple of interesting points regarding U.S. recessions are shown below.

• The last recession in the United States was eight months in length, starting in March 2001. However, the official announcement documenting the beginning of the March 2001 recession, released by National Bureau of Economic Research (NBER), was not made public until Nov. 26, 2001, just four days before the recession officially ended. Thus, the American citizens were essentially out of the recession before they were even told it had started.
• In 2008, the United States will hold its presidential election. Of the past 10 U.S. election years, dating back to 1968, only one contained a recession. In 1980, the United States endured a six-month recession that occurred over the first half of the year. However, this recession had little, if any, impact on the semiconductor industry, with the worldwide semiconductor market growing by 27% in 1980.

The effect of a U.S. recession on the worldwide semiconductor market would greatly depend on the magnitude and duration of the recession. Given the recent "decoupling" of the United States with many other countries' economies, a mild U.S. recession in 2008 would likely reduce our current forecast for worldwide semiconductor market growth from 10% to 5% (the lower end of our expected 5–15% range). A severe U.S. recession and possible global recession, however, could cause the worldwide semiconductor market to show a single-digit decline.

Although a U.S. recession would likely have a negative impact on the worldwide semiconductor market, there are still many positive factors affecting the semiconductor industry heading into 2008. One positive factor is that there isn't a huge semiconductor inventory stockpile that needs to be burned off in 2008. Also, IC capacity utilization is near 90%, and semiconductor industry capital spending plans are very conservative for next year.

Richard Gordon, Managing Vice President, Semiconductors Technology and Service Provider Research Gartner Dataquest

Global semiconductor sales are estimated at $270B in 2007, which represents a growth of 2.9% compared with market revenues of $263B in 2006. Although unit production of the major electronics system types, especially PCs, cell phones and consumer electronics, was healthy throughout the year, semiconductor device pricing remained under downward pressure, which held back semiconductor market growth.

The outlook for 2008 does not provide any reasons for optimism that semiconductor market conditions will improve. Most notably, DRAM market conditions deteriorated in 2H07 as leading vendors added significant manufacturing capacity. DRAM vendors have recently announced cutbacks in capital spending, but these actions have come too late and oversupply is expected to plague the market through much of 2008.

In the wider market, we are closely tracking holiday season sales of electronic goods, as the industry looks for early signs of potential demand-side weakness. This sensitivity to weakening demand has been heightened because the chances of a U.S. economic recession in 2008 are being talked up on the back of ongoing financial market uncertainty. At the very least, it now looks likely that global gross domestic product (GDP) growth will slow down in 2008, with the obvious knock-on effect on consumer spending on electronics and business IT budgets. While positive growth of 6.2% remains our most likely forecast for 2008, there is an increasing chance of a downside scenario.

Looking longer term and reflecting that fact that the semiconductor industry has become heavily influenced by price-sensitive end markets, we are forecasting a five-year compound annual growth rate (CAGR) through 2011 of just 4.8%, which suggests that there is little scope for wild swings in annual growth year-to-year within the forecast window. The year 2009 represents the peak growth year in our updated forecast, although at 8.5% annual growth, this can hardly be classified as a "boom." This stronger growth in 2009 is driven by a recovery in the DRAM market, but the negative influence of another DRAM market downturn on the total semiconductor market can be seen in 2010 when slower growth of 4.0% is forecast. In 2011, semiconductor market growth of just 2.6% is forecast as effects of the lingering DRAM downcycle are exacerbated by a forecast NAND flash market recession.

Moshe Handelsman, President, Advanced Forecasting

Growth of worldwide IC revenues in 2007 was noticeably slower than in the previous four years, indicating that the industry is at a peak of the IC cycle, as we originally alerted in February 2006 about underlying demand for ICs reaching a cycle peak in mid-2007.

By definition, a decline will follow. The question is how deep the ensuing decline will be in 2008.

IC revenues in the first half of 2007 were weak, negligibly above those in 2006 due, in part, to overheated shipments at the end of 2006. Based on the first 10 months of 2007, IC revenue growth is 5%, which is quite close to our December 2006 forecast of 7.5%, generated by a quantitative forecasting model that does not require retroactive modifications.

Year-over-year growth rates improved significantly from August to October, reaching 6–8%. However, this is also a point of concern because IC shipments are zooming high in contrast to the shrinking underlying demand for ICs. This discrepancy contributed in the past to exacerbation of industry declines. Early signs of the problem are already evident among memory vendors and the bookings status among end OEMs.

On the supply side, 2007 seemed to be a decent year for semiconductor equipment vendors. Billings of wafer processing equipment in the first three quarters delivered a solid 17% growth in comparison with the same period in 2006. Alas, bookings in 3Q07 declined 26% vs. 3Q06 with a certain impact on shipments in Q4. In sum, 2007 growth may end around 10%, not far from our original unchanged forecast of 8.9% given in January 2007, the most optimistic forecast at that time.

Our 2008 forecast is in line with last year's article: Our longer-term forecasting tool predicts the major turning points of underlying demand for ICs in revenues. It warns of a decline that will continue into 2008, but will be shallower than that of 2001. However, if the industry overheats during the last two months of 2007, as the October data indicates, then the decline will be steeper than that shown by the underlying demand forecast.

Laurie Balch, Principal Analyst, Gary Smith EDA

The EDA industry endured an extended growth slump through the first half of this decade, despite the substantially better performance of the semiconductor industry during the same period. Resurgence in IC device shipments and average selling prices (ASPs) unfortunately did not translate into increased investment in design software tools. The lack of next-generation EDA tools held back industry growth. Mercifully, as new tools have finally begun to emerge in earnest, EDA achieved its own recovery in 2006, and the positive trend has continued into this year, with growth forecast to reach ~10% for 2007.

Where the future for EDA lies depends on several factors. Of course, cyclical growth patterns in the overall semiconductor industry are a major contributor to the EDA forecast. With some uncertainty in the semiconductor industry picture for 2008 and 2009, EDA will likely experience reduced growth over the next couple of years. EDA growth is forecast to dip to ~8% in 2008 and 4% in 2009. However, spending on design tools won't simply dry up; the need for advanced design technology is just too great in this era of rapidly shrinking IC design features and burgeoning gate counts. Once semiconductor industry growth becomes less uncertain, EDA growth should head back into the 10% range. The five-year compound annual growth rate (CAGR) through 2011 is forecast to reach 8.2%.

Therefore, it is incumbent on the EDA industry to maintain its focus on fully developing this next generation of design tools. Although new tools have debuted in the past few years, many methodology transformations remain yet to be introduced. Design for manufacturing (DFM) issues will require that CAD tools be able to handle parallel processing and concurrent algorithms. These capabilities aren't minor improvements to today's CAD tools; they will necessitate a complete rewrite of today's CAD tools. That's a job that could easily take EDA vendors three years to accomplish, and IC designers are already anxious to begin implementing new CAD technologies.

The front end of the IC design flow is also currently in the midst of an inflection point. The evolution of the electronic system-level (ESL) design methodology is a bit thornier of a problem to address. Because an ESL-based methodology involved bridging between IC designers, system designers, and embedded software designer — groups with traditionally different design needs and tool budgets — developing a full-fledged ESL flow is no small task. The high cost of software development is even starting to cause pricing pressure on functional verification tools, as users need to slash spending on downstream tools to put more resources toward ESL tools. But creating a full suite of ESL technologies will be a necessary step for EDA to keep adequately serving electronics designers through future semiconductor generations.

Michael Cowan, Independent S/C Industry Analyst

A new semiconductor sales forecasting model called the Cowan LRA Model has been developed to facilitate the determination of future global sales of the semiconductor industry on a monthly basis. The results of the latest sales estimates for 2007 and 2008 derived from this approach are to be presented, but first, an overview of this modeling technique will be discussed.

The Cowan LRA (linear regression analysis) Model, which forecasts global semiconductor sales, is a mathematically based model that features statistical analysis carried out on the past 23+ years of historical, monthly global semiconductor sales that are collected by the World Semiconductor Trade Statistics organization and published by the Semiconductor Industry Association (SIA). It is a dynamic, mathematically pure view of near-term worldwide semiconductor sales looking forward over the next five quarters. The model is devoid of economic assumptions or emotional biases. It exploits linear regression analysis applied to the "appropriately transformed" actual monthly sales numbers, thereby "rendering" the global semiconductor sales data highly linear and, therefore, very amenable to linear regression statistical analysis techniques. The resulting numerical transformation of the past 23+ years of the monthly actual sales numbers — from 1984 to 2007 YTD — that is invoked is not a complicated mathematical expression but very straight forward and "makes sense physically," yielding extremely high linear regression correlation coefficients approaching 0.97 and greater. In exercising the model each month, a total of five distinct sets of linear regression parameters (of the format y=mx+b) are employed to calculate the resulting global semiconductor sales forecast estimates for each of the five quarters associated with the model's forecast horizon — namely, 4Q07, 1Q08, 2Q08, 3Q08 and 4Q08 and, thus, 2007 and 2008.

It is emphasized that each month's actual global sales number released by the SIA is a "lagging indicator" because it is published a full month after the fact. The Cowan LRA Model, however, "turns" this lagging monthly sales into a "leading indicator" by virtue of its near-term forecasting capability looking out over the next five quarters. This is the "beauty" of the model and, therefore, makes it dynamic in the sense that it can be run each month using the most recent actual global S/C sales number published by the SIA. It can rigorously "track" the near-term sales forecast outlook of the global semiconductor industry on a real-time basis. Consequently, the model's monthly sales forecast does not "sit still" but evolves each month because conditions change rapidly and unexpectedly in the semiconductor industry, and market forecasts are hard pressed to keep up with these changes. How can industry management be sure that a forecast issued two, three or more months ago is still relevant to what's happening in today's market?

Given this model overview, I present the latest monthly run of the Cowan LRA Model's global sales forecast estimates covering the next five quarters — namely, from 4Q07 through 4Q08, respectively. These latest forecast sales are based on October 2007's actual (non-averaged) global sales of $21.680B as posted (Dec. 3, 2007) on the SIA's website. These sales forecast estimates are: 4Q07=$68.632B and 2007=$257,425B; 1Q08=$65.367B, 2Q08=$65.325B, 3Q08=$71.418B, 4Q08=$76.282B and 2008=$278.393B. For the two yearly sales forecasts, the year-over-year sales growth expectations are 2007=3.92% and 2008=8.15%, indicating a modest 2× recovery in 2008's sales growth expectation compared with 2007's projected growth relative to 2006.

Carl Johnson, Executive Director of Research Infrastructure

Transitions in the semiconductor and semiconductor capital equipment industry are relatively easy to predict. Getting the timing right? Well, that is another issue. Short term, we can focus on the macroeconomic disruptions that we read in the daily press.

Is consumer spending (finally) going to roll over? Are the challenges of staying in step with Moore's Law going to hamstring companies that do not have the financial muscle to stay in step? Will the oft-predicted consolidation among devicemakers, equipment suppliers and the supply chain pick up speed?

These are only a few of the questions industry forecasters ask each and every year. We come back to the forecasting bench late in the year to gaze into the crystal ball only to see that things have not changed all that much. These issues are clearly visible, plausible and, in many respects, inevitable. They just take time to happen.

Next year is going to be a lumpy year. IC unit volumes should continue to be strong, but average selling prices (ASPs) will be under pressure. This has become a recurring theme. At best, a flat year for semiconductor revenues is in the cards. I'm expecting early year weakness in IC demand caused by the macroeconomic dislocations in the United States and other parts of the world. These are primarily debt-related issues, and they are not going to disappear in just a few quarters.

It is decision time for the devicemaker. Do I outsource to foundries or do I push ahead with leading-edge R&D? If you do not have a big bankroll and a well-defined roadmap, you probably have to step back and take advantage of the capacity available in older, more understood process technologies. Sure, there are a few companies that will step right along with the leading edge. Intel, Samsung and Toshiba come to mind immediately. Others are dropping back — the action at Texas Instruments and Advanced Micro Devices speaks volumes. The foundries know this, and that is why they are reluctant to increase their capital spending. For most of 2008, they will continue to be very cautious on the spending front.

Despite the slowing capital spending front, IC production will continue to roll forward. There's a silver lining here — increasing unit volumes is good for the materials companies. I expect the semiconductor materials segment, barring a significant drop in wafer starts, to do very well in 2008.

The stage is set for weakness on the capital equipment front. The narrowing customer base is going to make it difficult for small, non-diversified players to grow. I'm anticipating that capital spending in '08 will be down at least 10%, with the brunt of the decline occurring in the first half of the year. Those that have diversified into emerging, faster-growing markets will do much better than those with a large exposure to IC production.

How about the oft-mentioned consolidation? Several late-in-the-year stories have suggested that the smaller players in chip and chip equipment industry consolidate or else they will die. To me, I think 2007 provided a reality check to many business owners. In many conversations, owners seemed much more open to the suggestion of merging, acquiring and partnering with industry cohorts. Expect more action on this front — the games are just beginning.