Other Chip Makers Gunning To Snatch Technology Lead From Vitesse
======================================================================
From the on-line edition of SmartMoney Magazine
THE AVERAGE technology investor blanches these days at the dismal
developments in the personal computer market: Inventory adjustments,
pricing pressure on PCs, continual DRAM price implosions -- it's enough
to send even the hardiest hi-tech loyalist running for the hills to pan
for gold, or maybe airline stocks.
Yet there is at least one sector that has remained untouched by the
troubles infecting the rest of computerdom -- networking and
communications. Companies such as Cisco, Bay Networks, Lucent and other
giant firms seem to be sidestepping the mess by helping to build the
conduits that will transmit voice and data in the next millennium. While
forecasts for PCs are down this year, there's no sign of a shortage in
the network build-out within corporations or in the IT shops of the
Internet service providers and telecom firms constructing public
networks.
But rather than simply betting on the big four networkers -- Cisco,
3Com, Bay and Cabletron -- we thought it would be better to take a look
at some of the chip companies that make the components that Lucent and
its competitors put into their equipment. The idea is that as the
networking concerns rip each other apart over share of the data
communications market, the companies that supply the chips will continue
to benefit regardless of who wins the equipment war. What's more, these
companies will gain in overall importance to the Ciscos of the world as
more and more of the sophisticated networking functions that were once
solely the province of software become hard-wired into silicon. An added
bonus: The shares of many of these semiconductor companies were hit hard
by the Asian crisis and have yet to fully recover.
To test our theory, we screened for companies that serve the
communications equipment market and have significantly high yearly
growth. Only five companies from an initial 150 cleared our hurdles (see
recipe): Altera (ALTR), Level One Communications (LEVL), Lattice
Semiconductor (LSCC), VLSI Technology (VLSI) and Vitesse Semiconductor
(VTSS). While these companies are all in networking, they all specialize
in distinct areas of the semiconductor market: integrated circuits,
programmable logic devices and high-speed chips made from gallium
arsenide that are used for the fastest, most demanding communications
applications. For all these companies, success will be determined by how
quickly they can capture the latest hot networking functions in silicon.
Case in point: Level One, a maker of transceiver chips that go into
products for both data networking and telecommunications equipment, is
trying to move into more complex products. Level One has been a
highflyer over the past few years, with its stock price shooting up 36
percent between January and October of last year, before the Asia
meltdown made chip stocks dip generally last fall.
The challenge now for the company is to work its way up into the
higher-level functions of networking gear, where the real brains of
boxes from Cisco and others perform the routing and switching functions
that really constitute networking expertise.
That's not easy, since Cisco and Bay do most of their chip design
in-house for their most strategic products. But Level One is making
progress by partnering with firms like Galileo Technologies (GALTF) to
develop lower-cost versions of well-established but highly desirable
technologies such as Fast Ethernet switching. The challenge in the
coming years will be to develop commodity versions of what's now at the
high end, technologies such as gigabit-per-second packet switching.
VLSI Technology has a similar mandate. While VLSI started out
developing ASICs of various kinds, it has of late been captive to the
interests of Ericsson (ERICY), which has left VLSI overexposed to
volatility in the cell phone market. Recently, though, VLSI outlined a
transition strategy whereby it aims to broaden its customer base and
return to its roots in ASIC products. Through the use of so-called
Application Specific Standard Products (ASSPs), and by creating an
outsourcing network of subcontractors, VLSI believes it can bring in
more diverse assortments of custom chips, including more components that
go into networking boxes, like those developed by its competitor LSI
Logic.
Some of the company's strategic shifts are already bearing fruit. The
move away from personal computer products has boosted VLSI's gross
margin in the past year by decreasing the amount spent on inventory
write-downs for PC sales slumps. And the emphasis on communications
products has boosted its sales in Europe. In the last quarter, earnings
per share jumped to 45 cents from a loss of $1.08 in the December
quarter of 1996.
At the other end of the spectrum from VLSI, programmable logic
companies such as Altera and Lattice have enjoyed a more prominent
position in networking products where the latest software algorithms
have to be implemented in silicon. Programmable logic chips are generic
arrangements of circuits that can be reprogrammed by the customer either
by fusing new connections between transistors, or by using new software
stored in memory to alter the steps the chip executes. The danger for
these companies, according to analyst Scott Randall with Soundview
Financial, is that as portions of the networking market become
standardized, they start to fall to the ASIC programmers because their
solutions are cheaper.
The good news, says Randall, is that the skills developed by the two
in networking apply to the broader market for communications equipment,
including cellular base stations. That means that both Altera and
Lattice should be able to more than offset encroachments in one area of
their market by moving to areas that are more cutting-edge. At the same
time, Mark Edelstone with Morgan Stanley Dean Witter says that both
companies are poised to reap additional revenue in the coming year by
cannibalizing the low end of the ASIC market because they have finally
been able to put enough transistors on a chip to compete with so-called
"gate array" products. Should you buy in now? Randall says that the
coming slump in European sales for Altera will likely depress prices
this summer, so hold on to your money.
The last member of our group strikes us as the most technologically
adept. Vitesse, which makes transceivers for Gigabit Ethernet chips and
components for optical networking, has the difficult task of taking an
unlovely area of the chip market and making it seem efficient. Gallium
arsenide (GaAs) semiconductors, which are good for making high-speed
logic circuits as well as analog-capable chips, are traditionally hard
to manufacture in high-yields. However, Dataquest chip analyst Jeremy
Donavan says Vitesse has made tremendous strides in GaAs. "They've
certainly brought the cost down, and they're going to have the only
six-inch wafer fab in the world this year, which will give them
tremendous capacity."
Still, other chip vendors are licking their chops over dethroning
Vitesse and other gallium arsenide manufacturers by developing much
cheaper alternatives using standard silicon. While Vitesse's executives
say that such chips, known as BiCMOS, are still nowhere near commercial
production, Donavan says Applied Micro Circuits (AMCC) and Lucent (LU)
both have a chance to steal some of Vitesse's business in the low-end of
the telecommunications equipment market. "The move to smaller, meaning
0.18 geometries in silicon, is making the BiCMOS more viable," says
Donavan. "It would be foolish for Vitesse to be complacent about this
development."
For Vitesse, as for all of these companies, the challenge is to win
as much as possible of the business in new networking technologies.
That's where the premium for an innovative design is guaranteed to be
higher than the margin on commodity chips. The battle will be fought not
only among the chip companies but between software and silicon, as more
and more functionality, which used to run as software on generic
microprocessors, is designed into semiconductors themselves. Elias Moosa
of Robertson Stephens calls this phenomenon the "Silicon Cycle." And
some would even say we're approaching a stage where innovations in chips
will drive the nature of communications products. That's not to say that
superior software applications will be irrelevant. Far from it. But the
increasing importance of specialized chips and the companies that make
them is an undeniable fact in the age of the network.
Copyright (c) 1998 Dow Jones & Company, Inc.
All Rights Reserved. |
