While technology has changed a great deal since Cogent CEO Dave Schaeffer founded the company in 1999, he has few, if any, regrets
During an investor conference last week, Schaeffer was asked what he would do differently in terms of building a network with a clean slate.
"The network we would build today would actually look almost identical to the network that we operate," he said.
Cogent benefited from buying distressed companies in the 2001 to 2004 time frame, but it also focused on using its own capital to extend its network only to locations that had high volumes of traffic.
"A common misconception of our businesses is that we somehow exist because of the acquisitions we did when assets became distressed," he said. "The Cogent business model, including the target locations and product set, as well as the network architecture, was actually outlined in a 1999 business plan, which we raised $500 million of capital for.
"We understood two fundamental points. First, that the internet would be the only network that would matter. That's proven to be correct, although it was a bit of a controversial call in 1999-2000. While there are walled garden networks, and closed infrastructure, virtually everyone is of the belief that all applications will eventually ride on a common internet infrastructure."
Schaeffer said the second fundamental point was that local area networks we're a natural monopoly, and therefore anyone who was not an incumbent started with a huge disadvantage.
"Our strategy was to connect to two types of endpoints; data centers, and large multi-tenant office buildings," he said. "We started by taking a map and putting pins into, and saying these are the locations we want to go to. From that, we then looked to buy dark fiber rather than build it to get as close to those end locations as possible, and to do so in a ring configuration for highest resiliency."
Cogent does build some small networking assets, such as from manholes into a building or up a building's risers, but it has stuck with its dark fiber plan.
From its inauspicious beginning of using push pins to identify its endpoints, Cogent has proliferated.
"We have bought dark fiber from 240 different suppliers around the world, and we have roughly 840 metropolitan rings in 204 markets," he said. "We serve 1,750 skyscrapers in North America and 1,200 carrier neutral data centers in 44 countries."
With over 57,400 route miles of intercity fiber and more than 33,600 metro fiber miles, Cogent offers its services in North America, Europe and Asia. The company provides its services to more than 202 major markets and interconnects with over 6,660 other networks. Schaeffer said Cogent is on its third generation of technology across four areas of its network; long haul transport, metro transport, edge routing and core routing.
"I don't think today, we would do anything different," Schaeffer said. "We continue to buy dark fiber every quarter. We continue to go into buildings that give us an adequate return on capital. So, we win two ways, one by lowering our capital costs by buying rather than building as much as we can, but still leveraging the advances in optronics that are available to a dark fiber operator.
"And, then the second is being extremely selective about the endpoints. We speculatively build in to the very best locations, and then bank on having a better value proposition to capture a very fragmented market with a superior product, whether it's superior quality of service, or in price. That strategy has allowed us to become profitable, to organically grow, and to do so in an industry that remains challenged."
While the acquisitions were a means to an end, Schaeffer said Cogent has long discarded that equipment into the scrap heap as it deployed newer generations of gear to "ride down the cost curve more efficiently."
"The fiber continues to serve us, and we have an average remaining life on our fiber agreements of about 16 years," he said. "We almost always have renewal options on those agreements. So we feel pretty comfortable with the architecture we have. The product set that rides on that is purely internet, which gives us, I think, a lower operating cost structure and a much lower cost per bit per mile than others have."
While Schaeffer doesn't see any major changes in regards to how Cogent has built its network, he does think other service providers would have done things differently given how services and applications have evolved since 1999.
"You would not replicate a cable plant with an HFC plant and a branch and tree architecture," he said. "You would not replicate a wireline telephony network with 25,000 COs (central offices) and 223 million pairs of hub and spoke copper. You would not replicate a mobile network that was based entirely on macro cells with low band frequency initially in 800 megahertz with 70,000 cell sites. You would build something very different.
"You would build fiber much deeper into markets. You would build fiber that is ring protected that can support native IP over DWDM and protected Layer 3. And, in theory, you would serve every location by subsidizing the less economically available locations with those that are more profitable, a utility like model. Because our government has insisted on facility based competition, because we have intermodal competition between legacy and new networks, we have seen no one in North America build a sustainable, profitable business building fiber to the premise. "