Monday July 10, 8:26 am Eastern Time
Forbes.com
Chips Are Up, Chips Are Down
By Arik Hesseldahl
It's like a sudden rainstorm during a garden party. But once the rain stops, the party starts up again.
Here, the party is the semiconductor sector, and the rain is a downgrade in outlook from Salomon Smith Barney analyst Jonathan Joseph, who detected what he called ``first mover'' evidence that the recently strong semiconductor industry is nearing a downturn that could be more apparent in six to nine months.
Joseph named four companies, Advanced Micro Devices (NYSE: AMD - news), National Semiconductor (NYSE: NSM - news), Texas Instruments (NYSE: TXN - news) and Silicon Storage Technology (Nasdaq: SSTI - news), he was concerned about and cut his ratings on all four.
The results were immediate. The Philadelphia Semiconductor Index lost more than 9% of its value on July 5. AMD fell $9; National lost more than $8; TI more than $3; Silicon Storage fell more than $15.
The sector recovered the following day as other analysts rushed to defend the chip sector. The Philadelphia Semiconductor Index regained more than 3%. AMD picked up $2; TI another $2; National and Silicon Storage lost more ground.
But the sudden volatility in the chip sector is an indication for concern in an industry that is historically one of the most cyclical. As semiconductor stock prices have added value in recent months on surging demand, those who have watched the industry for a long time know that a downturn is on its way. The question is when.
Doug Andrey, director of finance at the Semiconductor Industry Association, based in San Jose, Calif., says that typically, the chip industry has run in boom-and-bust cycles that last four years--sometimes a little longer, sometimes a little shorter. The current upward cycle, Andrey says, started the second half of 1999 and is expected to last until 2002. That's when the SIA says the growth rate in chip sales, currently running at 31%, will slow to 14%, followed by an even slower 12% in 2003.
The main factor, according to Andrey, is that chip companies, many of which are beset by shortages prompted by increased demand they didn't expect, are currently building new factories, or ``fabs,'' that will start churning out more chips in the next 12 to 24 months. Once that manufacturing capacity roars to life, the industry will find itself in a situation of oversupply, waiting for demand to catch up.
It's happened before. In 1984, worldwide semiconductor revenue surged to more than $26 billion, up from $17.8 billion the previous year. In 1985 they fell to $21.7 billion. By 1989, global chip sales reached $49.6 billion and inched up slightly in 1990 to $50 billion.
So, when's the next downturn? Most analysts agree that late 2002 is about right, though manufacturing capacity won't be the only determining factor. While 13 new fabs are under construction right now, that pales in comparison to the building boom of 1994 through 1996, when chip companies built 113 fabs, according to market research firm Dataquest, San Jose, Calif. The increase in fab capacity ultimately led to conditions of oversupply that reached its worst in 1998, and demand has only in the past year caught up.
But in addition to new fabs, the industry is making a transition to using bigger wafers in the chip manufacturing process. Wafers are the thin round discs of silicon on which several chips are created. Once completed, the chips are cut from the waferlike square slices from the round pizza. The current standard wafer size is about 8 inches in diameter, but the bigger wafers are 12 inches.
Most of the new fabs under construction, planned for 2001 and 2002 completion, will use the larger wafers. Bigger wafers mean more chips per wafer, and more chips per wafer could contribute to the next period of market oversupply.
With supply conditions constrained now, but loosening in the next two years, the question about the chip sector's future is one of demand. And forecasting demand has not been easy. Santa Clara, Calif.-based Intel (Nasdaq: INTC - news) in particular has blamed poor forecasting of demand for its ongoing shortage of PC processors.
Much of the recent demand for semiconductor products, especially flash memory, has been fueled by new products like PalmPilots, MP3 players and digital cameras, but mostly by the increasing popularity of cell phones. The growth of the Internet's infrastructure, including routers, switches and hubs, has also pushed chip demand upward.
``If all this manufacturing capacity comes online and the demand for all these products continues to go through the roof, then we may not come to overcapacity,'' says Jim Feldhan, president of Semico Research, a Scottsdale, Ariz.-based market research firm.
New products aside, the personal computer continues to be the biggest market for semiconductors. But PC demand has shown signs of reaching maturity. International Data Corp. forecasts only a 9% growth in worldwide PC revenue for 2000 from 1999, half the rate of growth from 1998 to 1999. But that demand could surge again, as new markets open up.
``In spite of all the growth we have been seeing, we may be underestimating demand,'' says Manoj Nadkarni, publisher of ChipInvestor.com, an online investment newsletter focusing on the semiconductors market. ``If you take into account the possibilities of India and China alone, the numbers are simply mind-boggling.''
Nadkarni says that chip companies that don't have their own fabs tend to suffer less during industry downturns. Examples include Altera (Nasdaq: ALTR - news) and Xilinx (Nasdaq: XLNX - news). Chip companies like Vitesse Semiconductors (Nasdaq: VTSS - news) and Triquint Semiconductor (Nasdaq: TQNT - news) are partially isolated from the ups and downs of the rest of the industry because they make specialized communications chips using a material called gallium arsenide rather than silicon.
Whenever the downturn comes, it's not likely to be as severe as in previous years. The main reason is the mixture of products that consume chips. Should demand in one segment plummet, others may not.
``We don't see an end to the cycle, but rather a flattening of the amplitudes,'' says SIA's Andrey. ``The extremes between the high and low points will be reduced.''
But it never hurts to pack an umbrella.
Go to www.forbes.com to see all of our latest stories. |
