By David Gross
Ethernet exchanges have rapidly become the flavor-of-the-month in telecom services. Just about every provider offering some kind of data center interconnection, business Ethernet, or corporate VoIP service has announced they are offering an Ethernet Exchange platform, or has built a connection to one. A remarkably quick development for a service that didn't really launch until late last year.
Turning Ethernet into Frame Relay
While Ethernet Exchange services are fairly new, the interconnection technology that supports them has been under development for about five years. The need for Ethernet Exchanges arose out of carrier frustration with the nonstandard connection guidelines for handing off Ethernet traffic to other service providers. This problem kept growing as many Ethernet customers were using the technology as a private data service to connect buildings, not as an Internet access service as 2001-era Ethernet providers like Yipes and Telseon were advocating.
As businesses started using Ethernet to connect locations beyond metro areas, their traffic had to be sent through multiple carrier networks, which created all kind of custom configuration requests, in addition to QoS and management issues. Essentially, Ethernet had become a shared network private data service like Frame Relay, except without an NNI (Network-to-Network-Interface) reference to make carrier handoffs simple to configure and reliable to operate.
The MEF (Metro Ethernet Forum) took up this issue around 2005, and got to work on an NNI standard to allow carriers to hand over Ethernet traffic like they had been doing for years with Frame Relay/ATM traffic. The President of the MEF, Nan Chen, also saw this as a business opportunity, and started CENX, or Carrier Ethernet Network Exchange, which launched about a year ago. The benefit of this service is to eliminate the requirement that carriers connect to each other at multiple points throughout their networks, and instead just build a single link (or a handful of links) to the exchange.
CENX vs. Equinix vs. Telx/Neutral Tandem
At the same time that CENX launched, Equinix (EQIX) developed a competing service based on the new MEF NNI specification. Equinix already had all the major service providers sitting in its IBX centers, a big advantage for establishing this kind of service. Another co-lo provider, Telx, entered the market in June, through a partnership with Neutral Tandem (TNDM) which already provided a third-party interconnection service for cell phone carriers and VoIP carriers who needed to interconnect voice traffic with long distance networks. Neutral Tandem co-founder Ron Gavillet is also the co-founder of CENX.
Competitively, the data center co-lo providers are now fighting a startup, one that has been very successful with large carriers, announcing last week that Cox Business would be coming into its exchange, in addition to Verizon (VZ), Level 3 (LVLT), and China Telecom who are already there. Equinix also has Level 3 as a customer in addition to AboveNet (ABVT), all three of whom are components of our DataCenterStocks.com Services Index.
CENX makes the claim that it's completely neutral, and does not compete with its carrier customers for corporate accounts like Equinix and Telx do. While this might be technically true, competing with customers has never held back Verizon Wholesale or the wholesale telecom market in general. Carriers have long given business to their competitors in order to provide greater network reach for their other customers. Additionally, Nan Chen is also still President of the MEF, which represents both carriers and vendors, so there are conflicts on all sides - never a surprise in wholesale telecom. A greater advantage for CENX than its claim to be conflict-free is focus, which has been the key to success for most data center service providers. Digital Realty (DLR), Equinix, Rackspace (RAX), and Akamai (AKAM) have the highest market caps in this sector, and also the narrowest product lines - they don't mix co-lo with managed services, or CDNs with bandwidth sales.
Interconnection Margins Are Typically Higher than Bandwidth Margins
CENX has an odd "let's keep things quiet" strategy that extends to who has actually financed them. That said, margins on interconnection services are typically higher than traditional bandwidth services because there is no need to spend capital on laterals to office buildings, and sales costs are reasonably low because there is no need to reach out and get a message out to small or mid-size business. Additionally, sales cycles are often shorter than they are for long-term bandwidth or IRU contracts. Neutral Tandem's financials reflect this, with net margins around 20%, gross margins in the high 60s, and a Revenue/PP&E ratio over 3 - all much higher than those of its carrier customers.
Wall Street fell out of love with Neutral Tandem when revenue flattened at around $45 million a quarter, which occurred even when VoIP and cell phone minutes were still increasing last year. Similarly, CENX could face similar revenue constraints as traffic grows, but customer counts level off. Nonetheless, Equinix and Telx will have to decide how seriously they want to compete in this sector. While losing out out to CENX wouldn't derail their businesses, the number of larger carriers going with the startup could soon establish CENX as the clear leader in Ethernet Exchanges.