No Recovery in Sight for Semiconductor Shares
By ERIC J. SAVITZ
So here's the scary thing about semiconductor stocks: Despite their
dramatic losses since early 2000, the companies still trade at
sky-high multiples of sales and earnings.
Corporate information-technology spending is barely growing. Telecom
spending barely exists. And that's trouble, since most chips still go
into computing and communications gear. True, the economy is growing
again, which eventually will mean higher corporate profits; and as
profits grow, so should tech spending. But growth won't be nearly as
robust as it was in the bubble years. And what investors sometimes
miss is that while stock prices have fallen hard, sales and profits
have fallen harder.
The valuation conundrum was the dominant theme at the annual
semiconductor outlook panel held in Palo Alto last Wednesday by the
Churchill Club, a Silicon Valley speakers' forum. This year's panel
featured two affirmed bears -- Dan Niles of Lehman Brothers and Drew
Peck, a former SG Cowen analyst who recently moved to private
investing with Crimson Ventures -- and a cautious bull, Morgan
Stanley's Mark Edelstone.
While the three had some distinct differences of opinion on the
outlook for chip stocks, the heart of the evening's discussion
revolved around their shared opinion that semiconductor stock prices
remain far above historic norms. The question was whether the stocks
would fall back in line, as Peck and Niles contend, or if the
companies can "grow into" their current valuations, as Edelstone
argues.
Niles is certain the stocks are headed further south. For many chip
companies, he says, expectations for the 2003 recovery have become far
too optimistic given the reduced expectations for their customers. He
notes that analysts expect Dell Computer to increase 2003 revenues by
about 9%, down from compounded growth of 43% in the period from 1995
to 2000. But Intel, which grew 16% a year compounded over that same
span, is expected to show 15% growth in 2003, even as growth in its
primary market, personal computers, is slowing. Niles suggests an
explanation: Intel estimates are too high.
The pattern is repeated with the communications and data storage
sectors. While growth has come down dramatically at companies such as
Cisco Systems, EMC and Nokia, Niles notes that revenue forecasts for
2003 for their component suppliers, such as Xilinx, Altera and LSI
Logic, have the chip companies expanding as fast or faster than they
did in the 1995-2000 period. He thinks estimates for those companies
also will prove unrealistic.
Overall, Niles says, semiconductor demand from here is likely to track
end-market demand for personal computers, cellphones and other
electronic gear. That sounds obvious, except that the three analysts
agreed that the industry in recent months has benefited from
"inventory replenishment," as some hardware companies restocked
depleted warehouses. Niles, who also follows hardware stocks, warns
that there are few signs that a robust PC-replacement cycle is about
to begin.
"Hopes and reality are about to collide, and it won't be that pretty,"
he says. Niles expects industry-wide revenues to be flat to down 5%
this year. That's a gloomier view than the slight uptick most analysts
expect. (An even more bullish view can be expected on Wednesday, when
the Semiconductor Industry Association issues its mid-year forecast.)
Drew Peck is equally pessimistic. Noting that valuations in the chip
sector have been rising almost continuously since 1994 "based on a
belief in a pot of gold at the end of the rainbow," Peck offered a
particularly venomous view of Intel, displaying a chart comparing
Intel's valuation to its growth rate over time. Labeled "The Jaws of
Death," the chart showed the growth in Intel's valuation opening up a
big gap above growth in revenue -- the gap being the jaws that he
contends will close via a fall in Intel's stock price.
By contrast, Edelstone says the industry has bottomed, noting that
first-quarter revenues were sequentially higher for most chip makers
and orders are increasing, driving up backlog for many companies.
Capacity utilization rose five to 10 percentage points to more than
70% in the first quarter for many chip factories, he notes, improving
both margins and pricing power.
Edelstone concedes that "valuations are still too high" and suggests
"fundamentals need to catch up to stock prices." But he thinks we're
due for a near-term rally driven by second-quarter earnings, which he
believes will meet or beat expectations. Edelstone expects the sector
to "move sideways for a lengthy period of time," perhaps 12 to 18
months. Chip-industry revenues, which last year fell 32%, are unlikely
to return to bubble-era levels until 2005, he says.
Part of the ritual of the Churchill Club's annual panel involves a
stock-picking contest that includes the three panelists and the
moderator, Bill Tai, a venture capitalist and former chip analyst who
recently joined Charles River Ventures.
Peck highlighted the same stock he picked last year: Agere, a spinoff
from Lucent Technologies that sells a broad array of chips for
consumer and telecommunications applications. "Agere has an enormous
amount of intellectual property -- 6,000-plus Bell Labs patents, some
of them very fundamental," he says. "There is intrinsic value."
Not surprisingly, Peck also offered a pair of shorts: graphics
chipmaker NVidia and Broadcom, the communications chip firm. Both, he
said, will be hurt by high stock valuations and rising competition.
But if Peck is deeply bearish about most chip stocks, he's a raging
bull on companies that make chips for consumer-electronics devices
such as DVD players, digital cameras and flat-panel televisions. To
illustrate his point, he showed slides comparing the innards of a VCR
with a DVD player, noting that the DVD players features far more
sophisticated circuitry, including chips designed by U.S. companies
and made in Asian fabs. Peck expects that combination -- U.S. design
talent and Asian manufacturing -- to dominate the industry in coming
years. In an interview after the panel, Peck cited Zoran, Oak
Technology and, in particular, Cirrus Logic as well positioned to take
advantage of this trend.
Edelstone named Silicon Laboratories his top pick, noting that the
company is showing 15% to 25% sequential quarterly growth and is on
track for a tenfold increase in revenue over the next five years. The
Austin, Texas, company makes mixed-signal chips for wireless, wireline
and optical communications devices.
Niles repeated his pick from last year, On Semiconductor, a
Phoenix-based spinoff from Motorola that makes power-management chips.
And Tai's long recommendation was Xicor, which has been changing the
focus of its business from speclialized memory chips to mixed-signal
chips used in communications gear and flat-panel displays.
online.wsj.com
stolen from Lee's Marketswing site <g> |
