COMMENTARY >> TECH SAVVY
About That Coming Bandwidth Glut
By Jim Seymour
Special to TheStreet.com
7/13/99 3:28 PM ET
Some days, I think I could divide my email from TSC readers into just two (virtual) piles. Call one "Help!
Everyone says the sky is falling, and there's going to be a terrible bandwidth glut, and Qwest
(QWST:Nasdaq) and the other bandwidth resellers I know you like are going to collapse! What should I do?"
Call the other stack "Everything Else."
I had a great example of this over the past weekend. When I started writing this column Saturday morning, I
checked my email archives. I was genuinely surprised to see that I hadn't had a "bandwidth-glut fears" email
in all of 12 hours. But sure enough, Sunday's New York Times had a long piece on possible bandwidth
overcapacity, and by midafternoon, I already had 26 worried emails waiting in my inbox.
I'm not surprised that there's so much fear and confusion out there about a possible bandwidth glut. A
general misunderstanding of the evolving bandwidth market has led to unfortunate and confusing claims by
analysts, writers and others -- generally forecasting impending doom for the builders and aggregators of
bandwidth. Those claims fall in four areas:
There isn't going to be that much demand for more bandwidth. And even if there is, turning on only a
fraction of today's so-called "dark fiber" -- buried cable not yet in use -- can handle it just fine.
There's so much investment going into bandwidth expansion today that the industry will soon be
wildly overbuilt, and bandwidth will fall into the commodity-pricing swamp.
Wireless and satellite service is going to wipe out demand for bandwidth over any kind of wire: Before
long, everything goes wireless.
New technologies, principally Dense Wave-Division Multiplexing, or DWDM, are going to make data
transmission so much more efficient that we won't need much more bandwidth than we have today.
My take:
Demand. This, of course, is the big question: If demand does not continue to grow at least as fast as it has
over the past 12 months, bandwidth resellers are in trouble. Their cable plant -- their sunk-costs investment
in buried-fiber networks -- is not supportable at today's demand levels. If demand for bandwidth sags, so do
these companies.
So what's ahead? You know the answer to that already. We see the same forecasts for sharply rising
demand for data communications -- Web traffic doubling about every 90 days (and lately, even faster),
data-transmission demands so high that voice falls to a tiny fraction of total data-comm traffic. The specific
numbers aren't so important there as the trend line: Is demand growth going to stay that steep?
In a word, yes.
But it's going to get even hairier than that, to the great advantage of companies that today are building
bandwidth capacity for resale.
Many of the demand-doomsayers see demand for bandwidth expanding at an essentially linear rate -- in
other words, they see growth as being in the "More of the Same" category. In fact, demand isn't just going
to grow at its recent rate -- staggering as that is -- but faster still. Because increasingly, bandwidth is going
to be absorbed by new functions that just don't exist on current systems.
That's the point so many bandwidth doomsayers miss. It isn't just going to be more of the same-old
same-old. In fact, the availability of abundant, affordable bandwidth underlies the business plans of a huge
percentage of the hot companies of today and tomorrow. Yet the impact on the available pool of bandwidth
of all these new and generally unnoticed businesses' operations is largely ignored by the
bandwidth-doomsayers.
One example of an out-of-left-field biggie in bandwidth-demand: digital delivery of movies to the home. The
video-rental business is going to be squashed over the next decade by digital delivery of movies. No more
going to the never-quite-near-enough Blockbuster or Hollywood (HLYW:Nasdaq) store, searching the
shelves, then finding that the movie you want is out. Or if you get lucky, it is there, you take it home, view it,
then have to take it back.
Ordering rental videos over the Web for physical delivery doesn't work much better -- two-way shipping gets
too expensive and takes too long. And delivery from local video-rental stories isn't a bad interim answer, but
if you still have to take it back, you haven't solved the problem.
With online delivery directly to your TV for "live" viewing, (or to a digital-storage box attached to your TV set),
you'll be able to dial up, select the film you want, select the start time, click and boom -- it's there. The
movie studios are scared to death we'll capture and save these digitized films for later viewings as well -- of
course we will! -- and so are eager to find ways of making this a "live-viewing-only" market.
The content owners -- the studios -- hold the cards here, so at least at first, this will be mainly a
pay-per-view-style market, offering a finite number of viewings, or a finite viewing period, or both.
Eventually, as the studios learn not to fear digital delivery -- just as they had to learn not to fear
videocassette rentals, which now provide about half of Hollywood's revenues -- you'll be able to save films to
a hard disk, effectively buying them instead of just renting them over the Net. (We'll see third-party systems
that bypass limits on this emerge very early in this market, of course; when I say you'll eventually be able to
save digital movies directly to a disk for later repeated replay, I mean the time when you'll be able to do so
without a little black box that defeats the anti-save technology inevitably larded onto the early
digital-download business by Hollywood Inc. I have hopes -- but only small ones -- that the demonstrated
success of such current products as the TiVO and Replay digital-download-and-save boxes will convince
Hollywood studios that allowing customers to keep downloaded copies of movies isn't such an evil thing,
after all. Then again, this is Hollywood, the land of paranoia and denial.)
Making downloadable, live-viewing digital movies work takes a little less than a 7 megabits-per-second
Net-access service, easily achievable with both ADSL and cable modems -- though with both cranked-up to
speeds not usually delivered to Net-access subscribers by either today. (And, we should note, not possible
within the limits of the dumbed-down G.Lite flavor of ADSL, which maxes out at 1.5 Mbs, now being pushed
by most RBOCs and CLECs.)
Imagine the cumulative bandwidth demand of, say, 25 million films cascading down the wire every night, and
you begin to get a sense of just how big a bandwidth sucking sound this will be.
That shot-in-the-dark guess is based on the Census Bureau's 1996 estimate of 91.9 million to 94.2 million
U.S. households -- an estimate subsequently widely criticized as lowball, with the real number now
generally accepted at about 100 million. Of course, next year's census will increase that number. But even
using the 1996 estimate, I think one household in four watching a movie on a typical night is about right,
once Net access to pay-per-view "live download" movies is widespread. With the continuing sharp decline of
network television as a source of early hours G-rated home entertainment, and with the powerful
convenience of watching whatever movie you want, whenever you want, I don't think 25 million households
"consuming" films on a given night is high at all.
This direct-delivery-to-the-home market for movies has been forecast for a long time, but there hasn't been a
means in place to deliver all those millions of concurrent, high-bandwidth video streams. Over the next few
years, that kind of fat pipe into American homes will be commonplace.
Bandwidth demand isn't going anywhere but up. Period.
Tomorrow I'll tackle the other three questions. But you can guess where my answers are headed...
About That Coming Bandwidth Glut, Part 2
By Jim Seymour
Special to TheStreet.com
7/14/99 2:53 PM ET
Yesterday we looked at the popular and scary question of whether we're about to see a profit-destroying
bandwidth glut, a surfeit of available capacity that will severely hurt the dominant bandwidth vendors. Four
themes emerged:
Will demand for bandwidth really continue to grow as fast as it is growing today?
Will the small avalanche of investment dollars pouring into the leading bandwidth resellers today all
but guarantee an overbuilt market?
Will wireless crush the demand for bandwidth over buried fiber lines?
And will techniques for cramming more data down existing lines do in bandwidth vendors, which have
continued to increase capacity?
We examined the biggest of all those questions, the one underlying this growing fear among investors that
the bandwidth business won't be so hot much longer: What about demand? As I said yesterday, I think the
growth of demand for bandwidth is not only going to continue at today's torrid pace, but that growth rate will
actually increase, as new technologies and markets, absolutely dependent on ample bandwidth at
reasonable prices, emerge.
The issue, I'm convinced, isn't going to be a bandwidth glut, but just the opposite: whether we can build-out
total U.S. bandwidth resources fast enough to avoid hurting these new, emerging industries which will rely
so profoundly on cheap and abundant bandwidth to deliver their products and services.
On to the other big three questions:
Investment. Yes, investment in bandwidth-for-sale is soaring. Four companies now dominate this area, and
all are attracting new capital: Qwest (QWST:Nasdaq), which I am long, Level 3 Communications
(LVLT:Nasdaq), Williams Communications (WMB:NYSE) and IXC Communications (IIXC:Nasdaq).
QWST has gotten distracted by its unfortunate bid for U S West (USW:NYSE) and Frontier (FRO:NYSE);
CEO Joe Nacchio has decided he wants Qwest to play more in the value-added-services area than as a
pure play in buried fiber for rent. That's understandable, but wrong-headed: focus, Joe, focus. But the other
three companies, particularly LVLT, are forging ahead as single-minded providers of bandwidth for the major
telcos and private company networks.
LVLT looks so interesting right now because with QWST's apparent refocusing -- the company would say
"added focus" -- on the value-added-services market, Level 3 emerges as the primary pure-play investment
in IP bandwidth. And as a technology leader, buying the best-of-breed solutions it needs to push its network
to the front of the quality-of-service ratings.
In that, Level 3 is following the path laid out earlier by Qwest, with even more speed and intensity. Building
partnerships, for example: I counted at least six joint agreements for LVLT during June alone, including the
Big One, a deal with Lucent (LU:NYSE) to adopt Lucent's Softswitch technology and products to bring to
LVLT's IP network the reliability we're accustomed to getting with older, circuit-switched networks. Combine
the speed and economies of IP with the quality and reliability of circuit-switched systems, and you've got
telecom Nirvana. That's where LVLT is headed.
(Softswitch, which transparently translates calls back and forth between the older circuit-switched networks
and more modern packet-switched networks, is a key to building IP datacomm networks that can be used
as easily as today's telephones. Customers, both business and residential, simply won't put up with today's
Keystone Kops implementations of voice-over-IP, or "VOIP" service, dialing long prefix strings and putting up
with voice quality that sounds like something from a World War II-era movie about submarines at war.)
Because demand is going to continue to grow at such a steep pace, the additional capital flowing into the
bandwidth business is not a problem for investors, for whom the real risk is not participating in this immense
market.
Wireless. As I've said often here, yes, I do think everything will eventually go wireless. Well, not everything,
actually: There are too many security problems in an all-wireless system, and the available over-the-air
bandwidth -- the parts of the spectrum available for the various wireless services -- is limited. Noise, reach
and other problems will also restrict use of wireless communications. But yes, absolutely, the trend is
toward wireless communications.
But hardly ever pure wireless. Because it's just too cheap and easy -- and so much faster -- to complete
most of those wireless data exchanges, data or voice, over fast buried-fiber landlines.
Think of datacom as consisting of layered, inter-connected services. Wireless is emerging as the
highest-perceived-value, topmost layer in this stratified model, the sexy service everyone wants. But believe
me, the back room supporting that wireless layer is going to remain full of connections to terrestrial fiber
networks -- the kind of networks Qwest, Williams, IXC and especially, Level 3, are now building across
America.
Multiplexing. Multiplexing is a way of wringing more capacity out of a network, wired or wireless. It allows
each fiber strand or each frequency in wireless to carry more data, making the system more economically
efficient. Multiplexing over fiber networks is red hot today; wireless multiplexing will be similarly hot in a year
or two, as various forms of data compression and multiplexing help stretch that limited spectrum's capacity
to carry our voice messages and data transmissions.
The hottest of the hot in fiber multiplexing technologies is called DWDM, for Dense Wave Division
Multiplexing. Think of DWDM as a way of sending many different colors of light down each strand of fiber
simultaneously, each color carrying separate data streams. With DWDM, you can increase the carrying
capacity of a given buried-fiber plant, far more cheaply than by laying additional fiber strands.
Again, sure DWDM is going to be an issue in pricing bandwidth. But I'm convinced the demand for telecom
capacity will more than absorb the capacity increases made possible by multiplexing.
Indeed, I see the intelligent use of multiplexing technologies as one of the two safety valves the bandwidth
industry has to meet exploding demand with the finite (but substantial) development capital available to it.
The other, to go back to the beginning of this argument, is the amount of "dark fiber" presently owned by
each of the bandwidth vendors. Dark fiber refers to those glass strands not yet turned on -- lit, put to use --
in a bundle of buried fiber cable. Each of the major bandwidth companies has a substantial amount of dark
fiber at present. Level 3 is probably the leader here, for it has been more aggressive than the others in laying
extra capacity as it buries cable.
In the end, your assessment of the likelihood of a bandwidth glut comes back to your estimate of the growth
in demand likely over the next twenty years. If you're a pessimist -- like IBM (IBM:NYSE) execs in the early
'50s, who thought that the world market for computers might, just might, be as great as 12 units -- then yep,
it's easy to understand that you're going to see ahead a bandwidth market soon saturated.
I fall in the other camp. We have always approached bandwidth as an expensive, capacity-constrained
commodity, something to be preserved, to be used efficiently, even parsimoniously. I think the decade
ahead of us will instead be a time of essentially unlimited bandwidth.
"Bandwidth will be free," say some of the bright minds -- not because today's nascent bandwidth
aggregators and builders will be so tapped out by a market flooded with unneeded capacity, but because we
will be using bandwidth in such great gobs than we will begin to think of individual units of bandwidth as
almost cost-free, priced way below mills-per-something or any other tiny pricing unit you prefer.
That's the kind of business I want to own -- one so integrated into the threads of worldwide business and
society that it has become an essential but taken-for-granted part of the environment. And keeps churning
out profits, as volume grows and unit prices fall ... and revenue keeps growing. |
