By Tom Smith
On June 9, the major trade association for chipmakers, the Semiconductor Industry Assn. (SIA), released one of its semiannual updates to its World Semiconductor Forecast. While we view these releases, in some aspects, as lagging behind many research outfits in calling the near-term pace of the industry -- largely because they involve an extensive and formal forecasting process -- we think the SIA forecasts contain plenty of interesting data that hint at where the industry is going. They're well worth some study.
Starting with the top line, the SIA expects global industry sales to rise from $166.4 billion in 2003 to $214 billion in 2004, a 28.6% jump that would take global sales above the prior cycle peak of $204.4 billion in 2000. In its November, 2003, forecast, the SIA projected 19.4% growth for 2004, which reminds us of how much assumptions about the chip industry can change in six months. For 2005, the SIA expects 4.2% growth, to $223 billion, ahead of an anticipated dip of 0.8% in 2006, and a return to growth at 11.7% in 2007.
As the table below of forecasts from some other research outfits and chipmaker Micrel shows, 2004 is plainly a boom year. Bulls think 2005 should be strong, too, ahead of an expected downturn in 2006. However, bears think the downturn hits in 2005.
Annual Semiconductor Industry Estimated Sales Growth (%)
Source: SIA World Semiconductor Forecast 2004-2007 (June 2004), VLSI Research (June 14, 2004), Gartner (June 2, 2004), Semico Research (Feb. 2, 2004), In-Stat/MDR (Apr. 21, 2004), Micrel (Apr. 22, 2004), Standard & Poor's Equity Research Services.
With all this in mind, we at S&P Equity Research Services are positive on the semiconductor industry. We believe that prolonged underinvestment in plants in the last downturn implies a longer time for supply to be built out to overtake demand. Also, more end markets are appearing for chips, so the overall industry is less liable to swoon all at once if one industry fades.
Where should bulls graze in the semiconductor pastures? One strategy is to look to the SIA's 2003-07 compound annual growth rates for select chip categories and aim at the faster growers. A standout category, in our view, is digital signal processors (DSPs), which are estimated to grow from $6.2 billion in 2003 to $8.0 billion in 2004, a 31% jump. From 2003 to 2007, DSPs are expected to grow at a 17% annual rate. Other expected fast growers in this time frame are optoelectronics, at 15.5%, and flash EEPROM (electrically erasable and programmable read-only memory), at 15.1%.
In the DSP market, our top choice is Analog Devices (ADI ; 5 STARS, or buy; recent price: $48). We also like Texas Instruments (TXN ; 4 STARS, or accumulate; recent price: $25) and Maxim Integrated Products (MXIM ; 4 STARS; $50).
These three companies, along with Linear Technology (LLTC ; 4 STARS; $38) and Micrel (MCRL ; 4 STARS; $13), are also players in analog chip markets, which the SIA expects to grow about in line with the overall chip industry in 2004, at a 28% pace. The SIA sees analog chip sales increasing 12.4% in the 2003-07 period, which is above the estimated 10.4% overall pace for the chip industry.
The SIA forecasts that the analog category will rise from $26.8 billion in 2003 to $34.3 billion in 2004. This means analog sales would overtake the microprocessor category, which is expected to rise to $32.3 billion in 2004 from $27.4 billion in 2003. By 2007, the SIA still sees analog ahead, at $42.7 billion, vs. $37 billion for microprocessors.
Of course, the triumph of analog chips (and the complementary DSP chips and flash memories that often go with them) over microprocessors is related to the rise of cell phones and other equipment that sense real-world signals that are linear (analog) in nature, such as sound waves, and convert them to digital format for easy storage and transmission. It might sound contradictory, but digitizing real-world signals such as voice, light, temperature, pressure, acceleration, and rotation requires lots of analog chips.
Prospects for industry heavyweight Intel (INTC ; 5 STARS; $28) would seem to be dimmed by the SIA's estimate for microprocessor sales to trail the industry in 2003-07. However, sales aren't the whole story -- profitability matters, too. In our view, Intel's new generation of 300mm wafer-fabrication plants has already brought up gross margins above 60%, a good performance compared to prior cycle booms, and that's even before getting to the seasonally stronger sales we project for the second half of 2004 (see BW Online, 6/15/04, "Intel Turns Up the Juice"). Also, Intel aims to diversify its product line to derive about half of sales from computing markets and half from communications. It's already a big player in flash memories, a category growing faster than the overall industry.
WINNERS AND LOSERS.
Missing in action is dynamic random access memory (DRAM), which the SIA expects to rise by 55.8% in 2004, to $26.0 billion, but then is forecasted to be more volatile than the industry as a whole and fall to $22.6 billion in 2007. The DRAM area is particularly subject to commoditization and average-selling-price fluctuations, in contrast to the analog areas where design differentiation is key and pricing is relatively steady. We rank the big American DRAM maker, Micron Technology (MU ; $14) a hold.
In the spirit of the upcoming Olympic games, we'd award DSP chips with the gold medal, analog would get silver, and flash memory would get the bronze. The runners-up would be processors, followed by DRAM.
Note: Tom Smith has no stock ownership or financial interest in any of the companies in his coverage area. He's a registered representative of Standard & Poor's Securities, Inc. (SPSI). Affiliates of SPSI received non-investment banking compensation from Analog Devices, Texas Instruments, and Linear Technology during the past 12 months. Price charts and required disclosures for all STARS-ranked companies can be found at www.spsecurities.com
Smith follows semiconductor companies for Standard & Poor's Equity Research Services