The world's phone calls, faxes and e-mails zip through strands of glass no
thicker than a human hair, riding across countries and continents on
pulses of multicolored light.
The strands are bundled in cables that run beneath city streets, through
mountain passes and under the seas.
The cables were laid by a band of upstart companies that spent $50 billion
or more in the last few years to wire the planet. These massive networks
will serve the public for years to come, delivering the electronic goods
of the Digital Age. But the companies that built them are not celebrating.
Many are in financial ruin. The recent collapses of Global Crossing Ltd.
and other communications firms have roiled financial markets and cost
investors and employees tens of billions of dollars.
How did such a triumph of engineering leave so much corporate wreckage?
News reports of Global Crossing's meltdown have dwelt on accounting
sleight of hand and extravagant executive pay. But what actually drove the
company and others like it into the ground was an epic miscalculation.
These upstarts bet that if they built communications networks with far
more capacity, or bandwidth, than had ever been available before,
customers would rush to use them.
The network builders employed new technology that crammed much more data
onto each strand of glass. This enabled them to slash prices for
long-distance data transmission well below the rates charged by
established networks, such as those of AT&T Corp. and British Telecom,
that used older equipment.
The newcomers believed that the combination of low prices and abundant
bandwidth would unleash a frenzy of activity on the Internet. Consumers
and businesses would pay for all kinds of services that previously had
been too expensive. People would watch Web movie channels on their TV
sets. Doctors would diagnose illnesses via the Internet. Corporations
would hold video conferences with employees around the world.
The problem was that too many companies had the same dream, and they built
too many digital toll roads to the same destinations. The prices commanded
by long-distance networks did drop--but much more steeply than the
newcomers expected. And the demand for their services did rise--but not
nearly as much as they had banked on.
As a result, many of the upstarts couldn't bring in enough cash to pay
interest on the money they borrowed to lay all that cable.
Their plight is a textbook example of the boom-and-bust cycle of high-tech
capitalism. It illustrates how technological innovation, plowing
relentlessly forward, can make companies and then break them.
The financial outlook is not universally bleak--many network operators
remain healthy, and some regions are not overloaded with fiber. But on
many of the routes that drew the heaviest investment, such as those
between the United States and Europe, the bandwidth glut is likely to
remain for five years or more.
"People have laid huge amounts of fiber in the ground," said Internet
analyst Tony Marson of Probe Research Inc., "and there is a distinct
possibility that quite a lot of that will never actually see any traffic."
Explosion of Internet Traffic Fueled Demand
If any one person inspired the burst of network building, it would be an
English computer scientist named Tim Berners-Lee.
The expert in storing and retrieving data invented the World Wide Web in
1989 while working at a European nuclear research laboratory.
Before then, Internet users had to type arcane computer commands to search
for and view files on the network. Berners-Lee devised a way to present
documents, pictures and graphics on electronic pages that could be
retrieved with the click of a mouse.
The new technique transformed the Internet from a hard-to-use research
tool into a communications medium for the masses.
Two developments in the early 1990s aided that transformation. First, in
1992 Congress lifted the ban on commercial uses of the Net. Then, in 1993
and late 1994, the first easy-to-use browser programs were released,
simplifying the task of viewing or building a Web site.
Up to that point, Internet use had doubled every year or so. Afterward,
traffic exploded, increasing tenfold in 1995 and again in 1996, according
to researchers at AT&T Labs.
"People thought it could double every quarter forever," said analyst Paras
Bhargava of BMO Nesbitt Burns, a Canadian investment bank.
As people and businesses began buying, selling and chatting online by the
millions, it seemed that no amount of Internet bandwidth would be enough.
"All these [dot-com] companies were cropping up, it seemed weekly, and
there was no end to that in sight," said Glenn Jasper of Ciena Corp., a
telecom equipment manufacturer in Maryland. "So the conventional wisdom
was we've got to grow the capacity of our networks not for the traffic
that's out there now or even next week but for a year from now."
For years, those networks had been operated in the U.S. by a handful of
giant phone companies and abroad by government monopolies. These companies
relied on a small number of equipment suppliers, such as AT&T's Western
Electric subsidiary.
They lost their chokehold on the industry, however, just as Web traffic
was exploding. Governments around the world started prying open their
telecommunications markets to competition. At the same time, advancing
technology gave birth to a litter of new equipment suppliers that
specialized in fiber-optic gear.
Long frozen out of the telecommunications business, investors suddenly had
a chance in the mid-to-late 1990s to crash the party. Venture capitalists
opened their checkbooks to bankroll new networks and equipment companies.
Investors jumped on board as soon as shares were offered to the public.
"There was a lot of money available," said Todd Brooks, a general partner
at Mayfield Funds, a venture capital firm in Menlo Park. "You had
billion-dollar IPOs, and the gold rush mentality set it."
Before long, engineers were stringing glass around the globe--a "new
economy" version of the race to build railroads across America in the 19th
century.
Global Crossing and other companies tunneled under streets, carved
trenches and sent ships across the oceans, laying hundreds of thousands of
miles of fiber-optic cable. Many of the network builders were so sure of
the growth to come that they packed the cables with extra fibers that were
left inactive--"dark," in industry vernacular--for future use.
But all the while, technology was advancing in a way that would delay the
need for those extra fibers--and, paradoxically, lure more competitors
into the fray.
Fiber-optic networks use lasers to transmit light in split-second flashes.
Think of them as tiny, high-speed versions of the blinking semaphore
signals that ships use to communicate at sea.
Equipment makers improved that technology in two ways: by speeding up the
flashes of light and by using different colors to send multiple signals at
the same time over a single fiber. These innovations greatly expanded the
capacity of fiber-optic networks.
Statistics illustrate the magnitude of the change.
In 1994, the entire global communications network could transmit about 1
trillion pieces of data a second, said economist, author and technology
pundit George Gilder.
Today, a single fiber strand has more than 1 1/2 times that capacity if it
uses the best optical equipment on the market.
In the U.S., 10 of the largest networks had a total of about 40 such fiber
strands in service in 2000, according to a study by Probe Research, based
in Cedar Knolls, N.J. The networks also had 570 dark fibers waiting in
reserve. And since then, two emerging national data networks, Touch
America Inc. and Velocita Corp., have added more than 100 fibers to the
total.
The increase in bandwidth is even more dramatic between the U.S. and the
rest of the world. For example, the capacity of networks linking the
United States and Europe has multiplied nearly 80 times since 1997, said
Richard Elliott, co-founder of the Band-X technology research group in
London.
By the end of 2002, capacity is expected to nearly double again--and
that's just counting the fibers that are in service, not those left dark
to accommodate hoped-for growth.
Marketing to Businesses
A catalyst in this explosion of capacity was Global Crossing. Unlike AT&T
and other established long-distance companies, Global Crossing showed
little interest in consumers' phone calls. Instead, company executives
wanted to sell bandwidth wholesale to other long-distance companies and
corporations, which would use it for their own communications needs.
Global Crossing's founder was financier Gary Winnick, a onetime furniture
salesman and investment banker who worked alongside junk-bond king Michael
Milken in Beverly Hills. Trumpeting the opportunities presented by
telecommunications deregulation and fiber optics, Winnick raised $750
million in 70 days in 1997 for the first leg of his network: an 8,700-mile
cable from the United States to Britain, Germany and the Netherlands.
No single company had ever built an undersea cable with private investors'
money before. But when Global Crossing quickly found buyers for all that
new capacity, "any doubts about the need were quieted," said Elliott of
Band-X.
Winnick soon had plenty of company on the fund-raising circuit. A host of
other entrepreneurs dazzled investors with charts showing the skyrocketing
growth of the Internet and the plummeting cost of doing business.
For example, James Q. Crowe, chief executive of Colorado-based Level 3,
boasted that his company's state-of-the-art fiber network would undercut
its older rivals' prices by 15% to 20%. Crowe raised a reported $6.5
billion before Level 3 had activated its first strand of fiber.
"There were a lot of companies sort of going at this in parallel," said
Dave Passmore, research director at Burton Group, a network analysis firm.
"They all got in when they viewed this as an unexploited market."
Ron Kline, an analyst for the telecommunications research firm RHK Inc. in
South San Francisco, said it wasn't necessarily wrong for a new carrier to
think it could win the battle for customers. "The problem was there were
too many people thinking about it."
The greatest advantage went to the carrier with the newest technology and
highest capacity. It spent less to push data through its network than its
competitors did, which meant it could charge lower prices.
So companies kept building networks even as the supply of bandwidth grew
well beyond demand. And as technology kept improving, the upstarts soon
had to compete with newer, more advanced players. No network could hold on
to its advantage for long.
"In some bizarre movie about the telecom industry, you would have guys
from the carriers going out and killing guys in the labs to prevent them
from coming up with new technologies," said Ron Banaszek of TFS Telecom, a
Swiss consulting and investment firm for communications and energy
companies.
Today, about 16 advanced transcontinental fiber networks are competing in
the U.S. long-distance business, said Larry Roberts of Caspian Networks in
San Jose, which supplies communications equipment. That's three times as
many as there were two years ago.
The increase in capacity and competition drove prices to the floor.
Wholesalers such as Global Crossing typically sell companies a certain
amount of capacity from one city to the next--for example, enough to
transmit 155 million pieces of data a second from New York to London. A
bank with offices in those two cities might use that capacity to connect
its computers.
In 1997, that capacity cost about $14 million upfront, plus annual fees of
$250,000 to $380,000, said Elliott of Band-X. Today, the same bandwidth
could be bought for $350,000 upfront and $15,000 a year.
Industry Forecasts Were Too High
So what happened to the burgeoning demand that was supposed to be the
industry's lifeblood?
The extreme growth rates in Internet traffic seen in 1995 and 1996 were
just a blip, reflecting the advent of Web browsers, said Andrew Odlyzko,
director of the Digital Technology Center at the University of Minnesota.
Since then, he said, the amount of data flowing over the Net has reverted
to its previous rate of increase, roughly doubling every year.
That's a lot of bits to move--but not nearly enough to fill the networks
built during the boom.
One reason demand failed to mushroom as expected was the shortage of
bandwidth in local fiber-optic networks. Before consumers start
downloading symphonies or watching pay-per-view events online, they need a
high-speed connection to the Internet. But in the U.S., fewer than 10% of
all homes have one.
There may be a data fire hose running from coast to coast, but the typical
consumer is still connecting through a straw.
Many consumers are unwilling to pay the extra cost of a high-speed line
because, in their view, the Internet is not compelling or important enough
to justify it. The entertainment companies that could make the Net more
appealing to consumers, including most movie studios and TV networks, are
staying on the sidelines until more homes have high-speed connections.
Some analysts and equipment makers argue that demand is growing faster
than the prevailing estimates indicate, increasing 2 1/2 to three times a
year on the main U.S. Internet pipelines. They argue that networks are
getting so jammed in some areas that long-distance companies will be
ordering more within a few months to a year.
But even those growth estimates fall well short of the giddy projections
of a few years ago. And the situation may get worse before it gets better,
if Bankruptcy Court allows Global Crossing and other insolvent carriers to
write off their debt and stay in service.
"That will launch a whole 'nother round of price wars that will cause pain
for everybody in the industry," said Russ McGuire, chief strategy officer
for telecommunications consultant TeleChoice Inc. "It will get worse for
everyone, and in the end, Global Crossing will still go away."