During the heyday of network construction in the late 1990s, a huge amount of broadband capacity was added to existing infrastructure. Service providers leasing bandwidth on certain routes had excess capacity on other routes, so swaps between carriers became common. These swaps took many forms: partnering on network builds, selling fiber, exchanging indefeasible rights of use, and initiating long-term leases.
As prices began crashing during the second half of 2000, carriers looked elsewhere to generate revenue. They found that leasing wavelengths required less up-front capital because it eliminated the SONET layer and the equipment associated with it. It also reduced operating expenses since the network management became the responsibility of the buyer. It looked like a win/win solution, so the wavelength market took off.
Today, the future of the wavelength services market is clouded by industry consolidation. There are fewer carriers, which means fewer customers for wavelengths. Many of the surviving carriers have acquired additional capacity through acquisitions and mergers or through purchasing assets at fire-sale prices. These facilities have been or will be rationalized and integrated into their infrastructure, reducing the demand for purchasing capacity from other carriers. However, the need for bandwidth at peak times still necessitates the availability of these services (see Table).
The underlying technology for wavelength services is WDM, naturally. WDM technology has advanced very rapidly, and equipment costs have been subject to tremendous reductions-which has had a profound effect on network service providers. During this period of drastic change, it was not uncommon to see DWDM equipment costs decline by 50% over 18 months. The more established network service providers were hard-pressed when new competitors, seeking to compete on the basis of service pricing, could purchase network gear at exceedingly lower prices.
Putting equipment price erosion to one side, two distinct wavelength services are being offered. One service consists of an unprotected wavelength between two points. If there is a service disruption, the service provider offers no instantaneous recovery as part of its service. It is incumbent upon the buyer to add more equipment, network protocol layers, and alternative routes to make the system robust. If the buyer is a typical IXC or wireless operator, it is likely they already have a SONET network and can readily integrate the unprotected wavelengths into a robust network architecture. Unprotected bandwidth is typically available in increments of 2.5 or 10 Gbits/sec, since the underlying network was designed at these speeds (OC-48 and OC-192, respectively).
The other wavelength service entails a SONET interface, which is typically available either as protected or unprotected. With the SONET interface, the wavelength service is indistinguishable from a traditional private line, albeit one at high speed. Similar to the true wavelength offer described above, these services are typically available at 2.5 and 10 Gbits/sec. For the facilities-based carrier, selling such service involves adding equipment to an existing network at the hubs where the service is requested. Delivery tends to be closer to weeks than days, and pricing is customized for each situation depending on length of contract, number of circuits, and various other factors.
Wavelength services, a subset of private-line revenues, took a battering over the past two years due to price erosion-and industry consolidation will further minimize the demand for such services. One contributing factor to the price erosion was the tremendous overcapacity that exists in the long-haul network. Another factor was the type of customer buying these circuits since wavelength sales are predominately carrier-to-carrier. Carriers have the expertise to shop for low-priced routes and build in the redundancy at the network level; moreover, they will switch vendors more readily than an enterprise customer.
Several factors have affected the direction of wavelength services. Most obvious is the bandwidth glut and the subsequent industry consolidation. The major players have always planned to own their own facilities; leasing wavelengths was a short-term strategy until their network build-out was completed. As the consolidation becomes reality, Insight Research expects that some wavelength leases will no longer be necessary. Many of the smaller carriers have merged with others, been acquired by a larger carrier, or had their assets sold off.
However, Insight also believes there will continue to be niche players. Service providers, such as ASPs and wireless providers, may require large amounts of bandwidth but may not want to invest in facilities. Instead, they will focus their investment on their core competency. For example, ASPs or wireless providers will need to terminate voice and data traffic. These types of customers will continue to purchase wavelength services as their traffic grows.
Thus, our research suggests this market will grow enough to replace the demand lost from the consolidated mega carriers. The result will be a relatively flat market during the forecast period of 2005-10.
Bob Rosenberg is president of Insight Research (Boonton, NJ, www.insight-corp.com).