From: Linda (email@example.com)
Date: Fri May 26 2000 - 11:27:21 PDT
[Some interesting thoughts from Andy Grove on Napster and its potential influence on network infrastructure.
Napster Is Clouding Grove's Crystal Ball
Forget consumer broadband, says the Intel chairman in
a FORTUNE interview. To see the future, check out
what's happening at Indiana University.
We always like it when Andy Grove comes
to town. He spends a few minutes talking
about Intel, and then, with the prodding
of a single question, he's off on the
challenges of the entire tech industry.
Arguably, no one thinks about the big
picture as well as the 63-year-old
chairman of Intel. On a recent morning,
looking tanned and fit in a white shirt and
blue-and-gray tie and sporting a cheap
digital watch, Grove pondered the
devices that will connect us to the
future, called broadband a pipe dream
(literally) for consumers, and talked about
why Napster may mean Intel is about to
waste $80 million a year.
Do you think wireless devices or
devices other than PCs will become
the main points of entry into the
Is an e-mail message or voice telephony
handled by packet switching an entry to
the Internet? If the answer to that is
yes, then my answer is yes.
Internet use is going to come in a big,
broad range of things, of which, I think,
only messaging--either voice or
textual--is likely to be a heavily formed
So are you down on broadband?
I have been making a very good living as
a prophet debunking consumer broadband
for five years, and I don't see any reason
Getting out to the consumer is a very
expensive undertaking. It's also a very
difficult technical undertaking. In the
United States, it is on the order of
a--and this is an illustrative
number--$100 billion infrastructure
investment. And while the tolerance of
the markets for investing $1 billion with
no foreseeable return has increased, the
markets have not reached the point of
tolerating investments of $100 billion that
have no foreseeable returns.
Twenty dollars a month gets you a basic
telephone connection to the Internet. To
pay for [broadband], you have to charge
something in excess of that. It's like a
first-class seat: It has to be more than a
few millimeters bigger than a coach seat.
There are plenty of businesses
desperate for bandwidth. You are
talking only about consumers, right?
Exactly. As all this is happening,
bandwidth is being deployed in very, very
major ways as backbone bandwidth and
as connections to businesses.
[Intel is] making a contribution to the
bandwidth economy. Recently we had a
strategic review organized by our CIO,
who just finished revamping our internal
network and wide-area network--a
project that we started in the mid-'90s.
It was a very expensive project. I forget
the number. It's a $100 million, $150
million kind of project. And the blueprints
[showed] this as a network that was
going to take us to the next century. And
it did. Just.
So now you're getting ready to pony
up just how much more money?
Eighty-million dollars a year for the next
five years. We are not getting another
application in there. We are just
expanding international capacity.
Do you see streaming video as
something to drive demand for more
It will. But I have a question about how
mainstream that is going to be. Now,
having said that, let me argue in a
completely opposite direction. One way
around two problems we are talking
about--the broadband access problem
and delivering the higher-quality
experience of streaming video--is going
to come in a very dramatically different
network architecture than what we are
dealing with today. You get glimpses of it
in the Akamais of the world and even
better glimpses of it in the architecture
that's behind Napster.
The University of Indiana has become the
most high-profile deployment place for
that. The Napster traffic is on campus.
It's a reversion of a large portion of
traffic back to some modern equivalent of
local-area networking rather than
wide-area networking. It doesn't have to
be music or video. It can be corporate
My worry about the building up of the
networking infrastructure that is costing
us $80 million a year is that we're going
to build that infrastructure just when the
usage pattern is going to a much more
peer-to-peer-based orientation. It shifts
the bandwidth requirement necessary to
go from us to the cloud [servers storing
data and applications] to bandwidth
necessary to go from my computer to his
computer, which may be in the cloud.
That may be the shortest way, or maybe
it's going to be a local-area network, or
maybe it's going to be some wireless
contraption. If the distances required for
these connections are reduced, then we
can solve the bandwidth problem without
solving it in a universal fashion.
There are some real trends of
decentralizing the cloud and bringing it
closer to the personal computer and even
involving the personal computer as nodes
in the network. This, of course, goes
back into the post-PC deal and throws all
that stuff [about the network computer]
out, because the device is too
So this is a wild card that can have a
profound effect on network build-out and
what has to play on the network.
Vol. 141, No. 11
May 29, 2000
This archive was generated by hypermail 2b29 : Fri May 26 2000 - 12:15:48 PDT