SAN FRANCISCO -- Investors are loving semiconductor stocks these days. No wonder, considering the Philadelphia Semiconductor Index hit an all-time high of 509.98 Thursday. That's a growth of 179% since the index bottomed last October.
But is it headed for a fall? Chip stocks are, after all, cyclical. Just look at the chart: The SOX -- which measures the stocks of 16 leading chipmakers and chip-equipment companies -- peaked in summer 1995, bottomed summer 1996, peaked fall 1997 and bottomed last fall. Follow that pattern, and the index should peak sometime between now and October and not bottom again until next summer.
Hope for Chip Stocks
Long-time SOX watchers say don't bet on it. "The last few years have been a bit anomalous," says
analyst Gunnar Miller, who covers chip-equipment stocks. "The reality is that they tend to be three-year up, three-year down cycles if you go back historically."
The health of chip-equipment makers is a key indicator for the semiconductor industry. For a chipmaker like
to profit, it has to sell the latest and most expensive chips in large volume. It can't do that without the latest manufacturing and testing equipment.
The last true boom cycle started in 1993, when Intel introduced its first Pentium processor, and halted in 1995, says
Banc of America Securities
analyst Brett Hodess, who was one of the
first analysts to alert investors that the cycle was hitting bottom back in October.
Back in '93 and '94, Hodess says, chipmakers raced to build plants to meet rising demand. But once they had plenty of manufacturing capacity, excess equipment flooded the market just as PC demand slowed in '95. Demand for equipment wouldn't heat up again until the plants reached the end of their technology cycle -- generally three years -- and had to be retooled.
In mid-1996, the SOX began to bounce back. But rather than marking a broad-based recovery, the rise was triggered when a changeover in technology from a 0.35 micron process to a 0.25 micron process forced companies to buy lots of equipment, Hodess says.
Any impression that the industry was recovering from its slump was false. The utilization rate at chip factories was still a low 75%. So while chip companies like Intel were buying equipment to produce chips on the new process, consumers still weren't buying enough computers to keep the plants running at full speed.
In reality, the slump wouldn't end until around the end of 1998. Now, strong chip demand is filling the plants. Plant utilization rate is at about 90%. That compares with 70% in July of last year.
Meanwhile, companies are just beginning to buy equipment for another changeover in technology -- this time from 0.25 microns to 0.18 microns. Yet equipment orders are still low: Just $27 billion will likely be spent worldwide this year on equipment, compared with $45 billion during the peak year of 1995. And that's a sign, Miller says, that equipment orders have only started to rise.
"Most U.S. chip companies are spending below their depreciation levels," he says. "I'm a guy who, in '95 -- when we were at a peak -- felt it was time to sell the stocks. This does not feel like that to me. This feels more like something that's potentially sustainable."
There are other indicators of a continuing recovery. Communications chip companies such as
have been seeing
soaring demand for their products as people worldwide load up on cell phones, network computers and Internet gadgets.
Meanwhile, memory-chip prices are about as low as they can get. They dropped to about $4.10 for a 64-megabit synchronous DRAM chip, and that compares to $7.50 a year ago, when memory makers fretted that prices were too low. "They are the lowest they've ever been," says Steve Cullen, a memory analyst at market research firm
Cahner's In-Stat Group
. Memory makers tend to cut equipment spending when memory prices are low, and to beef up spending when the money pours in.
All that means the SOX stocks still have room to rise. "This summer and the second half of the year you will see improving fundamentals," Hodess says, adding that supply won't outstrip demand again until 2001 or 2002.