BRIEFING: NETWORKS
Routers: central theme of the next buildout
Today router companies have seen their sales evaporate, even though Internet traffic continues to grow at a torrid pace.
By David Lipschultz
April 5, 2002

A Different Rout
The industry has been waiting for a router company to solve this scalability problem. But there's little evidence it's likely to be solved any time soon by Cisco or Juniper. That's where startups like Caspian, Charlotte's Web Networks, Hyperchip, Pluris, and Procket Networks come into the picture. These well-funded startups are designing and building a new class of "super routers," with capacity that can be increased as the need arises, saving customers from having to buy additional routers so frequently.

Using parallel semiconductors to distribute the data, these super routers offer access to thousands of ports on a single router (compared with 16 ports on many current models). Aside from accommodating much more capacity, super core routers also help eliminate several "hops" along the network, which is where one port has to connect to another. The routers are able to do this partly because of advances in technologies like dense wavelength division multiplexing, ultralong-haul optics, and optical switches. When several of these technologies are used together, routers can extend data traffic farther along a network without being forced to hop. This creates a simple, single-tier network design, as opposed to the typically complex multitier network that has evolved as carriers haphazardly built their networks to handle explosive Internet traffic growth. The simpler design also yields a much faster and far less congested network.

One shortcoming of the single-tier model is that it requires a carrier do an extensive and costly infrastructure overhaul. Although critical, this is daunting for many carriers that don't want increased levels of data traffic to overwhelm their equipment. "The current architecture of the Internet's major backbones--designed in 1997 and 1998--has reached the end of its useful life," says Mr. Kamman.

The next generation of super core routers, however, not only might save carriers from costly network failures, it also might create an economic model that allows carriers to increase revenue while decreasing costs--something carriers are desperate to accomplish in this tough economy.

Under the current network architecture and the routers that support it, carriers are expected to lose money as they increase capacity to meet growing traffic demand, unless routers decrease in price by 60 percent (a very unlikely proposition). That's because, under the current revenue scheme, the marginal cost of buying a new router to accommodate traffic outweighs the marginal gain. Next-generation router technology is also expected to allow carriers to reduce the number of hops from three to one. Those efficiency gains would put their networks back on the track to profitability. "Carriers can add more revenue-generating traffic and spend less on core network overhead--a recipe for profits in a fixed-cost business," says Mr. Kamman.

Follow the Money
Other startups make the same case. And as we saw when Juniper was able to sneak out a next-generation product ahead of Cisco and garner a $75 billion market cap, the company that lays out the equipment with the best economics is likely to be in the catbird seat. To succeed, all a startup needs to do is win over a few carriers' business. The follow-on deals five years hence will be sales of core routers to enterprise businesses that will place them on the edge of their networks. "This is an opportunity where it's one or the other: you either make $5 billion or lose $200 million," says Jim McLean, a partner at ComVentures and one of the original investors in UUnet, the first big network provider for the Internet.

Still, don't count Juniper or Cisco out in this business, especially since Cisco has plenty of cash sitting on its balance sheet. Both may be slow getting to the next generation, but Mr. McLean says that many big companies like Cisco have a way of dealing with next-generation threats when they become real: "They may just wind up buying the one that shows the most promise."

David Lipschultz is a contributing editor for Red Herring.  Write to letters@redherring.com.