One of the Few
Arbinet is one of the few former Breakaway Solutions customers/ICG companies (the two were unfortunately almost synonymous) still around, and it is
thriving. Arbinet provides a platform for telcos to trade
IDD minutes amongst one another. The platform apparently is quite effective at lowering transaction costs, particularly now that they also offer credit to customers (from GMAC). They are growing very quickly and anticipate continued growth - 2002 volume was around 5.9 billion minutes, and they think they'll hit 9 billion this year. It's a great business model providing a clear value to customers - so much for disintermediation.
Like
the analyst at the ITU, Arbinet wants to see prices 'stabilize' ("nobody can make a business out of 1-cent minutes," says the CEO) and says that they appear to be stabilizing right now. Unfortunately for the telcos, the
cover article in the same issue of Telecom Asia reports on the expected 'second bandwidth glut' anticipated for 2004 as all of the major players emerge from Chapter 11 minus their debt obligations. The pricing trend downward is likely to continue.
A good deal of Arbinet's current business will be gone within five years - customers will transition to VoIP and long distance (and many IDD) rates will plummet to one cent per minute or below. Arbinet will need to quickly switch focus to both the countries where state-owned telcos are effectively resisting/preventing change (which they are doing with their increasing focus on Europe and Asia), and to IP transit-related services.
In the world of IP transit, Quality of Service is much more of an issue. Instead of providing a circuit connection from, say, New York to Shanghai, bandwidth providers just provide a connection in one point with a certain throughput (a T1, for example, is 1.5Mbps) and charge either based on the total amount of data transfer or percentage of bandwidth used (it is possible to buy point to point connections but much more expensive and increasingly uncommon). With a standard contract, there's no guarrantee that that speed of the connection to any other specific location will be acceptable. Part of the reason is because most traffic goes through the big exchanges such as
MAE-East where everyone interconnects with everyone else - this allows bandwidth providers to say that they don't have control over the packets for the entire trip and therefore can't be held responsible for QoS.
The end-to-end, packet-switching architecture of the Internet is its greatest strength - new applications don't have to get approval from any central authority, and any point can easily connect to any other point on the network. However, the lack of accountability from bandwidth providers is very frustrating for their customers. In response, number of new companies have been workign on this problem over the past few years. At Breakaway, we worked closely with
InterNAP. InterNAP avoids the big exchanges and instead establishes private peering points with all of the major networks. They then continuously test the throughput and latency of each connection, and send customer traffic down the fastest pipe.
As Arbinet transitions to IP transit services, it will need to beef up its QoS-related capabilities - bandwidth will never be as one dimensional a commodity as voice minutes. Partnering or merging with a firm such as InterNAP would create a very compelling value proposition.
Posted by Stephen Bronstein at March 15, 2003 05:26 PM