Scores of Net companies have collapsed in the past year, but Excite@Home was different. Rather than trying to sell bags of dog food online or creating content that appealed to a niche audience and even fewer advertisers, the Silicon Valley company had real assets: a high-speed network that generated monthly revenue from more than 4 million subscribers.
To be sure, the company was hobbled by the money-hemorrhaging portal side of the business, but the network was real and impervious to competition. Right?
Wrong. In the space of a few months, the value of Excite@Home, the world's largest high-speed Internet network, evaporated as the company misjudged the ease with which its cable partners could replicate its highly touted network.
Facing bankruptcy and dissolution, the cable Internet provider has seen its one-time cable partners all but abandon it, deciding to build their own networks instead of rescuing Excite@Home. In the most painful blow, controlling partner AT&T said it would move all of its 850,000 subscribers to a new network, constructed at breakneck speed, instead of buying Excite@Home's network.
Within hours of the announcement, Excite@Home declared it would cease operations in February. The shutdown is quite a turnaround from just a few weeks ago when the company--emboldened by its creditors--played hardball with AT&T on the theory that its network was worth a good $700 million more than Ma Bell was offering.
"Excite@Home's biggest mistake was believing that none of its partners could migrate to another network quickly," said one Wall Street analyst. "AT&T called their bluff."
To the surprise of even many inside the industry, calling that bluff took just a few weeks, as AT&T proved able to link its own data centers, network operations centers and cable lines into a data network that could substitute quickly for Excite@Home's own systems.
To investors and longtime supporters of the once high-flying Silicon Valley company, the Excite@Home meltdown has been bewildering. Many have even accused AT&T of deliberately devaluing the company. After all, unlike the legion of defunct dot-coms, Excite@Home actually owned a network, and received about $15 per month from some 4 million subscribers, they note.
But the company's dizzying fall from grace and the cable companies' decision to create their own networks almost overnight shows just how tenuous the battered company's grip on life was as its big cable partners quietly learned the Internet business for themselves.
When @Home started, "cable was an industry that was only capable of delivering one-way broadcast video," said Cynthia Brumfield, principal analyst for analyst group Broadband Intelligence. "Over the course of the last few years, bringing in (Internet technology) engineers has been a big priority for the cable industry."
The result? It's now taking the big cable companies just a few months to replicate what Excite@Home spent years building. It's too soon to say whether their own efforts will function smoothly, but it's clear that Excite@Home's partners viewed their sidekick as expendable.
What's a network, anyway?
Excite@Home's difficulties have always rested on the fact that much of its own "network"--indeed, the hardest pieces to re-create--were in fact made up largely of the cable companies' property.
A high-speed cable network piggybacks on top of the same network that a company like AT&T Broadband or Comcast uses to send video HBO or ESPN signals to subscribers' homes.
At the very end of the network is the cable modem, which reads the data traffic sent over the network and translates it into a form that PCs can understand. These modems have generally been owned by the cable companies themselves, which lease them to subscribers. Excite@Home simply certifies that they will work with the network.
The "last mile" connection between the customer's house and the first layer of serious data-routing machines is simply the regular cable network, upgraded to handle two-way data traffic. This too has been owned by the cable companies, not Excite@Home.
Farther up the line, the Silicon Valley company actually did own equipment. In many local cable "headends"--the centers where the local cable wires are connected to high-capacity fiber-optic cables--Excite@Home often placed servers that would store local versions of popular content, such as Yahoo or CNN Web pages, as well as data routers and switches that served a traffic-cop role, telling data where to go. In many areas this equipment was located instead in a regional data center, a little like the massive hosting facilities maintained by Exodus Communications and others. Excite@Home has 25 of these centers in North America, which also host the e-mail servers, news servers and domain name servers that provide basic Internet services for subscribers.
From these headends and data centers, data would travel across a high-speed local telephone link, or a network owned by one of the cable companies, on its way to the Internet. For carrying data over long distances, Excite@Home used a private AT&T network "backbone," which minimized the traffic jams that often snarl ordinary Internet traffic.
When Excite@Home began work in 1996, this kind of network was brand new. @Home's chief technology guru Milo Medin was touted as one of the few recognized experts in this area, helping to create the service on top of cable company networks that hadn't been built for data.
But the cable companies never had a comfortable relationship with the technology upstart, and relations grew more and more strained as Excite@Home's market value skyrocketed--even though much of the company's network belonged to the cable companies themselves. But until recently, the technical know-how to put together a competing network wasn't widespread enough to warrant creating a competitor. In addition, the companies' original contracts locked them into using Excite@Home service until June 2002.
All of that changed this year when Excite@Home's financial condition worsened and bankruptcy looked likely. The cable companies examined their options and found that creating their own networks wasn't so hard anymore.
Growing a network overnight
That's not to say that it's cheap or easy to replicate what took Excite@Home years to build. AT&T maintains that it would rather have paid the $307 million of its original bid than create a new version by itself. Cox says it will create its own for between $100 million and $150 million.
Excite@Home invested tens of millions of dollars in its network each year as customer growth accelerated. Technology advances have also seen it swap out early components for new equipment.
Last year, the company spent about $290 million on ongoing operations, about two-thirds of which was dedicated to network maintenance, connection costs, network equipment and customer service for subscribers. That figure has grown fairly steadily from 1997, when the company spent $22.5 million on operations.
At the time of its initial public offering in mid-1997, @Home said it had spent about $57 million on capital and operating expenses. That was for a very different environment, however. The company had just 5,000 subscribers, making it a hundred times smaller than what AT&T Broadband, Cox and Comcast have today.
But the cable companies aren't starting from scratch the way Excite@Home did.
AT&T Broadband is in the best position. It already has a network of regional data centers it uses for Web hosting. It has network operations centers it uses to monitor telephone and video traffic on the cable network. It owns its own cable and backbone connections.
The company has spent the last several months linking those assets into a network that can function much like Excite@Home's. That has meant adding data traffic routers and switches into the cable "head ends" and into its own data centers, installing e-mail and domain name servers around the network, obtaining rights to the Internet addresses that will identify its subscribers' computers, and retooling its video network command center to be able to also monitor the flow of data.
It has helped considerably that parent company AT&T is in the business of creating networks, data or otherwise, the company said.
"They do this on a daily basis for external clients," said Sarah Eder, an AT&T spokeswoman. "We're an internal client."
She declined to say how much the company was spending, or exactly what type of components had to be purchased. About 80 percent of AT&T's customers have already been moved to the new network, and the rest are expected to move over by Friday.
Why, then, did AT&T bid $307 million for Excite@Home's assets in the first place, if a new network could be re-created in the space of a few weeks or months? Eder says the agreement still would have been the company's "first choice," had Excite@Home not turned off service to AT&T subscribers last week.
The company could quickly have made up that $307 million, even as it built its own parallel network. The agreement would have included equipment that covered Cox and Comcast's territory. Those companies, which are farther behind in building their own networks, are each paying Excite@Home $160 million for the next three months of service. That's a recent development, but it is a sign that AT&T could have made back much of its outlay in a short period of time.
A Cox spokeswoman said that technology advances and the spread of know-how has simply made it easy to build a network that might even be more stable than Excite@Home's, which was plagued with e-mail and connectivity problems throughout its life.
Analysts say the speed with which the cable networks are re-creating Excite@Home's network is likely to have some nasty side effects. Speedy engineering often leads to "some things that aren't the most efficient," Brumfield noted.
In the end, it was the technology itself that overtook Excite@Home, making its network role as middleman less important, and less hard to replicate.
"The cable companies already own their own networks," Brumfield said. "What @Home brought to the table was really just an ISP."