Web host keeps going and going ...

Antonio Salerno, CEO of Web host Conxion, discusses the demand for reliable, continuous uptime in the enterprise market and Conxion's chances of becoming a $1 billion company.
Written by Max Smetannikov, Contributor

Some Web hosters are just now finding their footing in the enterprise market, but there are players that have been pitching to Fortune 500 companies since the beginning of dot-com history. Conxion might not be as well known as its Silicon Valley neighbor Exodus Communications, but that could be due to a lack of high-profile outages and spectacular customer bankruptcies. In fact, on Aug. 21, Conxion celebrated its 2,000th day of continuous uptime. With that kind of reliability, it's little surprise that the shop has customers ranging from Microsoft to Visa International. Interactive Week's Max Smetannikov talked to CEO Antonio Salerno about the chances of Conxion becoming a $1 billion company during his lifetime.

What's the elevator-speech description of Conxion?

We have three different kinds of customers. The first kind--and this is not any kind of an official name--are the bad boys of the Internet. These are the companies that have extraordinarily high requirements for security, bandwidth and uptime. They tend to be failsafe customers. These are companies like Microsoft, Intel, Visa and Oracle. In this same first category we have a number of network customers. They buy network transport and backbone, and they bill the other networks that buy bandwidth and fiber from us. Their requirements are high bandwidth and security. Almost all of these customers take a very good look at their vendors before they sign multi-year agreements.

The second kind is vendors, resellers and [original equipment manufacturers]. They range from Fortune 10 companies to boutiques that specialize in things like security or integration of a special application. Some of them we have announced, such as Totality, but there is one OEM that we can't talk about that is extraordinarily large. There is a medium size one by the name of Equant, which is now part of Global One. So we have [between] four to five dozen resellers.

The final kind of customers just seek a full services Web hosting firm, customers like Seagate Technology, LSI Logic, Veritas [Software] and Siebel [Systems]. They need a solid company that is a specialist. They may not need a gigabit all the time, but they want to be sure we know what we are doing with servers.

Do you see more competition, now that giants like Exodus are switching to serve enterprise market?

We found that companies that are trying to transition to fully managed complex hosting but have 80 percent of their business as colo[cation], really - at least according to the customers we talk to - don't get it. I can give you an example. Take a look at a recent introduction of something called Conxion Enterprise XI. This new service allows us to go into customer's own data center as if it were one of our own data centers, and conversely make our data centers look to our customers as if they were their data centers. Therefore our customer doesn't lose its investment in people, machine room, servers, applications. They get to use all that investment and extend it into the Internet. We have been doing this for over three years. What happened is that almost all of our largest customers have had this architecture that we use, and everybody just loved this thing. This updates your strategy with the infrastructure you already have in place. It was a natural. We just figured out how to market it and put a name on it.

For someone like Colo.com or Intira, someone coming from power and pipe business, it would be very hard for them to even conceptualize this, let alone write a software program to make it happen. We wrote software called HotRoute. And then they would need to have the network to allow it to happen.

How do you see people like IBM and Accenture incorporating more Web elements in their outsourcing?

Obviously IBM has an outsourcing partner of some sort, and I would imagine they and their partner are constantly revising and updating services that they currently offer. Since we have a lot of OEMs, we constantly are updating all services that we have.

Your take on MSPs?

With Totality, we have been working together for three months, and we had half a dozen sales meetings. We see Totality making progress, getting traction in the market. We do see companies like Loudcloud as competition. One thing we were noting is that while Loudcloud probably does reasonably good work, their cost structure is so flawed that one of two things is going to happen. They either will price themselves out of the market, or they are going to have to buy the business and their expenses will be so large they will not be able to keep going. We believe they probably will capture some revenue, but will they price it in a way that they have a real business?

The reality is, they can't control their network costs or data center costs. I know people argue they are not significant, but in reality they are. And take a look at companies that have poor network costs. Some analysts argued that the major reason why Digital Island and Digex had to be sold was because they were not able to control their network costs, in terms of fiber and peering. Those costs can be extraordinarily high. If you are hosting with somebody who can't control these costs, they have to charge you extra because they can't keep these costs down. [If] they get a customer who needs high bandwidth or something else that is not mainstream, they are in the position where they can't get much margin out of it. So they try to go beyond breakeven on services. But most customers have an occasional need for services, but an all-the-time need for bread and butter of the industry.

I made this argument against PSINet and so many others for so long. It didn't appear to me that, with the debt load and their cost structure, they had a company that could survive. They might take some revenue out of the market for a year or two. But at some point, they run out of gas. With the current economic circumstance and inability to raise cash by our competitors, we know that time is on our side. They can't win a pricing war. And we've got a reputation for uptime and managed services.

Your take on data-center and bandwidth glut?

They are both true, they are out there. We ran our business with that expectation for the last several years. Bandwidth comes down to the cost of large local loops and cost of long-haul fiber. And we know that one of our major competitors announced in 1999 a deal that they were very proud of. This deal had them paying a certain amount per [fiber] mile, and in fact they at that point had signed a deal that was 100 percent more expensive than the deal that we signed in 1997. And now that price for us has been cut by a factor of 60. That's hard to believe. Our cost of doing business is dropping remarkably. We benefit from the bandwidth glut.

The cost you charge the customer per megabit of Web hosting has not dropped nearly as much as the cost of the backbone fiber. You are talking orders of magnitude.

As far as data-center glut, we built more distributed, more power-dense data centers. Our data centers don't have anywhere the square footage of large major competitors, but the revenue potential is remarkably similar. If you take a look at Exodus, it derives about $1,000 per square foot, per year, and they claim to have 5 million square feet. So if you factor that number into their revenue, you wind up with something like $240 per square foot, per year. That's ridiculous, that's less than what you get from a 7-Eleven. If you take a more careful look at that and factor in only useful space like machine rooms, you still come up with something like $1,000 per square foot, per year.

But what the analysts want to know is what is your revenue potential based on the capital investment that you already made. In our case, the revenue potential of our data centers, as they are in place and debt free, is $1 billion. There are a couple of ways to get to this number. It depends on old servers and new servers. Old servers, you can get only five in a rack. The newer servers, you can get 20 in a rack. But you also have to have hard disks, switches, and firewalls. The reality is that you can't get to 20 either. The real number is ten. Based on ten servers per rack, and overhead like switches and network equipment, we are at about $1 billion. While we've got the same revenue potential, we have no debt. We are not worried about our data centers being in play, we can expand our data centers as necessary. But, if you take these power-and-pipes facilities that have cages in them, you've got to tear them down to expand them.

Editorial standards