Three years ago when I worked at AT&T Laboratories, I wrote an essay called "The Rise of the Stupid Network," which analyzed how new technology would reshape the telecommunications industry. In the old days, I explained, the phone company exercised total control over its network, and it made the network more valuable by adding features like 800 service and Caller ID. Today the Internet makes it possible for individuals who don't own wires and switches to create powerful network applications that the phone company never did, like e-mail, Web browsers, and audio on demand. In the Internet era, I argued, the business of the phone company would be different. New phone companies would succeed by delivering raw connectivity, the most bits for the buck.
In hindsight, it's not surprising that AT&T wasn't able to hear my message. Indeed, one of the reasons AT&T is floundering today is that it is stuck on the idea that network ownership can command premium value. AT&T has plenty of company too. The world's telephone and cable companies see the "last mile" between their networks and their customers as the last scarcity, the best place to extract a premium from ownership of the network. They're rushing to offer high-speed Internet connections like DSL and cable-modem services. Both services are big improvements over dial-up modems; a typical DSL connection is ten times faster than the best regular modem, and cable modems can be up to 100 times faster. Demand for the high-speed services is growing swiftly; by the end of the year around five million U.S. customers will use them, half of whom are hooking up this year.
But DSL and cable-modem services are too little, too late. They're retrofits of almost-obsolete copper networks. Cable companies developed cable-modem technology because it works with the networks they've already built. Telephone companies adopted DSL for the same reason. Yet both technologies have serious problems. Long strands of copper act like antennas, especially when transporting signals at the frequencies necessary to carry data at high speeds. Electromagnetic interference from adjacent wires--crosstalk--is a huge stumbling block to universal deployment of DSL. Compensating for crosstalk is a signal-processing burden that makes deployment spotty and costly. DSL demands that $150-an-hour technicians search like Diogenes for a circuit that is clean enough to carry the signal. Technicians must untangle--literally, not metaphorically--an undocumented legacy of twisted copper wires dating back to Alexander Graham Bell. That is why DSL often takes several customer-frustrating weeks to install. Two DSL users groups are suing their phone companies over such problems.
Cable-modem service has an advantage over DSL. Cable networks are simpler, and they can carry more information because cable was designed for broadcasting video signals. But that advantage comes with a price--500 or more households share each high-capacity connection. And as a direct consequence of the shared design, cable companies must upgrade the network of an entire neighborhood with two-way amplifiers before the first house can get the service. Cable modems have reliability problems too. As I was writing this article, for example, my cable modem inexplicably slowed to a crawl for an entire afternoon.
Today's network owners may be able to preserve their old business models for a time with cable modems and DSL. But eventually these technologies will be superseded by the convergence of two radically faster, simpler, cheaper technologies--Ethernet and fiber optics. Ethernet is a simple, well-established computer-networking technology invented in the mid-1970s at Xerox PARC, birthplace of the graphical computer interface and the mouse. Ethernet is widely used to connect computers in offices. The first installations couldn't span distances longer than a few hundred feet, but those limitations loosened as the technology improved. Today Ethernet networks can have links of 50 kilometers or longer. Ethernet got faster too. It began at DSL-like speeds but soon jumped to ten million bits per second, then to 100 million bits, and now to a billion bits per second. A ten-billion-bit version of Ethernet is in trial, and stories of 100-billion-bit-per-second Ethernet are beginning to filter out of Silicon Valley labs.
Fiber-optic communications technology is advancing as explosively as Ethernet. The peak-hour telephone traffic of the U.S. can be squeezed onto a single glass fiber using off-the-shelf equipment. Technology in the lab today could provide ten times that capacity. Furthermore, fiber-optic cables hold as many as 1,000 fibers. Such cables are about as thick as the cables of copper telephone wires that hang from every pole. With a 1,000-fiber cable coming down the street, there could be one fiber for every house. The mythical last-mile bottleneck--the last bastion of the old business model--would be shattered.
The faster versions of Ethernet work hand in glove with fiber optics. Ethernet over fiber can connect homes to the Internet at speeds hundreds of times faster than DSL or cable modems and tens of thousands of times faster than dial-up modems. Lest you think that building a new network would be too expensive, the gear needed for 100-megabit Ethernet service costs about the same as the equipment needed for broadband service over telephone or cable networks--several hundred dollars a connection, according to the Dell'Oro Group, a Silicon Valley market research firm. Companies in places with fiber-friendly policies are already installing fiber-optic Ethernet service at (or even below) the cost of retrofitting old networks for DSL or cable-modem service. In Canada, for example, the phone company must allow competitors (and even some customers) to install fiber on its poles or rights of way. Such places have determined that plentiful bandwidth is as critical to their economy as running water, sewers, roads, and electricity.
A fiber-optic Ethernet system would be cheaper to run than DSL or cable-modem networks. Fiber optics are inherently reliable. Since glass is not an antenna, the components are simpler. Ethernet has huge economies of scale too; there are over 250 million Ethernet links in the U.S., vs. four million DSL and cable-modem connections. According to Dell'Oro, Ethernet could reduce back-office data-equipment costs by 90% compared with traditional telephone technology designed for voice.
The fiber-optic future will arrive first in Montreal, Stockholm, and several fiber-friendly U.S. cities, such as Palo Alto. In Stockholm residential customers can buy a ten-million-bit-per-second Internet connection for about $25 a month. In Palo Alto the service (still in trial) costs $70 a month. When the rest of the network matures, customers with the new Internet service will be able to switch from Website to Website as fast as changing TV channels. They'll be able to make high-quality phone calls and get TV-quality video over the Internet--and even broadcast their own video.
Don't expect old-style phone companies to bring us this kind of Internet service. They will not allow their cash cow to be slaughtered. And don't expect cable-TV companies (like AT&T and Time Warner, the parent of FORTUNE's publisher) to bring that kind of Internet service either. The owners of the old video entertainment business will not willingly provide the infrastructure to destroy it.
But new network service providers that are not tied to past technologies and obsolete business models--like Cogent and Telseon and Yipes in the U.S. (disclaimer: I'm on the advisory board of Yipes), like VDN in Canada and Bredbandsbolaget in Sweden--will see the opportunity in the new "deliver the bits" technology. In the U.S. the economies of fiber-friendly cities will leap ahead, adding another dimension to the digital divide. Old industries that we love to hate--cable companies, telephone companies, even the movie and record industries--could either fade to insignificance or be changed forever. The communications revolution is just beginning.
DAVID S. ISENBERG is the founder of Isen.com, a Westfield, N.J., telecommunications analysis firm. He served as a Distinguished Member of Technical Staff at AT&T for 12 years.
DSL AND CABLE-MODEM SERVICES ARE TOO LITTLE, TOO LATE--RETROFITS OF OBSOLETE NETWORKS.
PLACES WITH FIBER-FRIENDLY POLICIES THINK BANDWIDTH IS AS CRITICAL TO THE ECONOMY AS ELECTRICITY.
Copyright © 2000, Time Inc., all rights reserved.You may now print or save this document.