Barron's article Pt.2
Cover Story, Part 2
Cover Story, Part 1
Barrett maintains that selling server chips is not that different from selling PC
chips for PCs. While they are manufactured by a variety of firms, including
IBM, Hewlett-Packard and Dell Computer, the key element is the processor, a
business Intel knows well. To win business from established server chip
makers, Intel will be peddling a range of offerings starting with its Pentium III
processors and the blazingly fast Pentium Xeon processor that was rolled out in
June '98. Next year, Intel plans to launch its 64-bit Intel chip, code-named
Merced. And to further kick-start its way into the Internet business, Intel
quadrupled its research and development spending to an astonishing $8 billion
last year.
A second and no less important leg of Intel's Internet strategy is designing and
manufacturing chips for networking devices, the powerful computers that
control traffic over the Web. Much like a traffic cop, networking computers
analyze the differences between voice and data, determine whether data are
coming or going, assign importance and priorities to traffic and control just what
data get sent when and to whom.
Intel chips are already being bought by the likes of Cisco Systems, Lucent and
Nortel, the major communications vendors, but the Santa Clara company wants
a bigger slice of the pie. Toward that end, over the past two years, Intel has
spent roughly $8 billion to invest in or buy outright eight different companies,
many in some way key to the networking push. Especially important was the
acquisition of StrongARM, a novel chip design that can be used in high-end
switchers and routers from Digital Equipment and the $2.7 billion purchase just
this year of Level One, a maker of chips that combine analog and digital
technologies in local and wide-area networks. A key element of Intel's new
strategy is not just to sell the chips into this expanding market but also to try to
lay down some basic guidelines on how network chips interact with one
another. Last month, Intel unveiled its new Internet Exchange Architecture (or
IEA), in effect a blueprint for networking based on a new Intel family of
products.
What Intel is in effect attempting to do is replay the PC model in networking.
Just as there were many different standards for PCs in the early 1980s, not all
of which were compatible with one another, there are many standards for
networking today, and few are compatible with each other. If Intel has its way,
the standards would merge, and Intel would be the biggest beneficiary.
"If you step back and look at the big picture, you find that nature rather abhors
narrow vertical solutions to anything," says Barrett. "Nature likes horizontal
solutions with standard building blocks and open interfaces. Being a success in
this market is not dissimilar to being a success in the PC market. You don't just
create one device. You create an architecture, a processor plus a way of doing
things. You're throwing out a complete solution. That's the strength of our
push."
Intel's most exciting offer ing in this market so far is the IXP1200 network
processor, one of a new breed of interchangeable programmable chips from In
tel that will allow network system designers to add functions to chips even after
they are installed with software. Competitive chips from the likes of Motorola,
AMD and Texas Instruments are pre-programmed for their particular function
and not changeable once installed.
However, the Intel Inside concept may not give the company the same kick in
this market that it gets from PCs. Says Yankee Group's Howard Anderson, "If
you knew that Domino Sugar was in Coke, would that make you buy it instead
of Pepsi?" In other words, explains Anderson, the guts of the machine are less
important to customers than what they actually do. What's more, he says, "The
industry is deadly afraid of letting Intel be the power in networking that they
are in PCs."
Intel is also keeping a close eye on the interactive market-everything from
Palm Pilots and TV-set Internet access boxes. In one sense, this growing
market represents a threat to Intel's core PC business. After all, if people
migrate toward special-purpose PCs, and away from Intel-based PCs, Intel's
PC processor business could be even further eroded. Barrett, however,
chooses to see it as an opportunity for Intel's lowestend processors, including a
variant of its StrongARM chip designed to run in low-powered units on AA
batteries. "When I read stories about how Intel has missed out on the appliance
wave," he says, "I wonder if those people ever took high school math. Sure, the
chip we make for such a device may sell for only $20. But if you are talking
about a market for 100 million such devices, that's still 100 million times $20, or
$2 billion. It may not be huge, but that's still between 5% and 10% of our
current microprocessor business."
A growing awareness of these heady prospects helps explain just why Intel
stock has raced ahead to the present 76 from a low of about 50 in May,
outstripping both the S&P 500 and the benchmark semiconductor index by a
wide margin. Yet there is a very real danger that, in its enthusiasm, the market
has gone too far, too fast.
Let us look first at the numbers. This year, with six months in the bag and
third-quarter numbers, due out within weeks, set to be upbeat, Street analysts
are shooting for earnings as high as $2.40 a share, with a consensus
somewhere around $2.30 -- an advance of 29% over the very depressed 1998
return. In 2000, at least one bull figures Intel could earn $2.90 a share, but most
are a lot closer to $2.70, up just 17% over 1999. On the basis of those
consensus numbers, the stock is trading at multiples of 33 for 1999 and 28 for
2000, roughly twice the underlying rate of growth in earnings.
By comparison, the S&P 500 index trades at roughly 1.5 times 1999 growth
estimates. This is significant, because Intel has long tracked the S&P pretty
closely on this measure. But now, just as the company faces the challenge of
remaking itself with all the attendant risks that this entails, its shares have raced
ahead to the point that they trade at a substantial premium.
And while Intel seems to have a clear strategy going forward, there is much
that could go wrong. What if, for example, the competition for server chips
from the likes of Sun, HewlettPackard and IBM proves fiercer than expected?
And what if Intel has a tough time trying to dictate networking standards and
that new Internet architecture fails to catch on as quickly as hoped? And what
if Intel's new Merced chip, already delayed for a year, faces even further
delays next year? And what if the rollout of broadband Internet services to
American homes takes longer to accomplish than the phone companies and
cable companies, which are racing to get there first, anticipate?
None of these factors are under Intel's control any more than the cannibalizing
of the PC market by sub-$1,000 PCs was. Says Greg Blatnik, vice president at
Zona Research: "Intel is performing a delicate balancing act in these new
markets. In the past it had the advantage of owning the market, of being a
benevolent dictator. But that's changed. They now face new competition and
the difficulties of establishing new relationships with customers. I think they'll
succeed but there is certainly a risk that things will not work out as planned."
Factoring in the current risk, the company's shares should probably trade at a
P/E equal to 1.5 times the company's 17% pace of earnings growth. As noted
earlier, that's somewhere south of 60. No matter how good Intel management
and how bullish the prospects, anything above that is all fluff.
|
