MCI, Worldcom (UUNET), and Sprint were all major Tier 1 backbones, having been first movers in the explosive commercial backbone market. When MCI and Worldcom merged, a merger condition was that MCI had to spin off its backbone (MCI's backbone was acquired by Cable & Wireless). The merger of Worldcom and Sprint, however, was too much for antitrust review. The EU, DOJ, and the FCC all indicated that they would block the merger. On July 13, 2000, Worldcom and Sprint conceded, and filed to withdraw the merger application. On August 3, the withdrawal was approved.
In its complaint to block the merger, DOJ argued the following about consolidation in the backbone market:
- WorldCom's wholly owned subsidiary, UUNET, is by far the largest Tier 1 IBP by any relevant measure and is already approaching a dominant position in the Internet backbone market. Based upon a study conducted in February 2000, UUNET's share of all Internet traffic sent to or received from the customers of the 15 largest Internet backbones in the United States was 37%, more than twice the share of Sprint, the next-largest Tier 1 IBP, which had a 16% share. These 15 backbones represent approximately 95% of all U.S. dedicated Internet access revenues. UUNET's and Sprint's 53% combined share of Internet traffic is at least five times larger than that of the next-largest IBP. The Herfindahl-Hirschman Index ("HHI"), the standard measure of market concentration (defined and explained in Appendix A), indicates that this market is highly concentrated. The HHI in terms of traffic is approximately 1850; post-merger, the HHI will rise approximately 1150 points to approximately 3000. (Note: Throughout the Complaint, market share percentages have been rounded to the nearest whole number, but HHIs have been estimated using unrounded percentages in order to accurately reflect the concentration of the various markets.)
- The proposed merger threatens to destroy the competitive environment that has created a vibrant, innovative Internet by forming an entity that is larger than all other IBPs combined, and thereby has an overwhelmingly disproportionate size advantage over any other IBP.
- The proposed transaction would produce anticompetitive harm in at least two ways. First, it would substantially lessen competition by eliminating Sprint, the second-largest IBP in an already concentrated market, as a competitive constraint on the Internet backbone market. The elimination of this constraint will provide the combined entity with the incentive and ability to charge higher prices and provide lower quality of service for customers.
- Second, the combined entity ("UUNET/Sprint") will have the incentive and ability to impair the ability of its rivals to compete by, among other things, raising its rivals' costs and/or degrading the quality of its interconnections to its rivals. As a result of the merger, UUNET/Sprint's rivals will become increasingly dependent upon being connected to the combined entity, and the combined entity will exploit that advantage. Such behavior will likely enhance the market power of the combined firm, and ultimately facilitate a "tipping" of the Internet backbone market that will result in a monopoly.
- As is true in network industries generally, the value of Internet access to end users becomes greater as more and more end users can easily be reached through the Internet. The benefit that one end user derives from being able to communicate effectively with additional users is known as a "network externality."
- When the networks that constitute the Internet operate in a competitive market, this network externality creates powerful incentives for each individual network to seek and implement efficient interconnection arrangements with other networks. Efficient interconnection has many requirements, including the physical connection to exchange traffic and the effective implementation of cross-network protocols or standards. For example, providers in competitive network industries have strong incentives to cooperate in the development of new cross-network protocols or quality of service ("QoS") standards that would enable new services or applications to be used across interconnection points on multiple providers' networks. By securing efficient interconnection, an ISP or IBP makes its services more valuable to its existing and potential customers. End users can enjoy the benefits of network externalities regardless of which network they belong to so long as their cross-network communications are of similar quality to communications "on-net," or purely within their provider's network. Thus, a failure to secure efficient interconnection arrangements places any given network at a significant competitive disadvantage when such customers can turn to a competing network that is efficiently interconnected to other networks.
- The explosive growth of Internet traffic, which has been doubling in volume every three to four months, and the introduction of new applications that depend upon the transmission of large quantities of data, have made it necessary for IBPs to constantly increase the capacity, i.e., bandwidth, of their own networks, and of the facilities through which they interconnect with other networks. A network that upgrades bandwidth within its own network in an adequate and timely manner can maintain the quality of its customers' Internet experience with regard to communications that originate as well as terminate on that network. In order to maintain the quality of its customers' Internet experience with regard to communications that originate or terminate on another network, however, a network must constantly upgrade the capacity of its interconnections with other networks, as well as upgrade capacity within its own network.
- Any failure to keep pace with the growing demand for increased interconnection capacity -- or, worse yet, any degradation in the quality of existing interconnections with other networks -- would adversely affect the quality of an Internet user's experience regardless of the capacity and efficiency of an IBP's own network. Due to the Internet's growth rate, any failure to make adequate and timely upgrades of interconnection capacity is tantamount to a degradation of the quality of interconnection. When networks operate in competitive markets, they have mutual incentives to avoid such degradation.
- Similarly, when operating in competitive markets, networks have incentives to negotiate reasonable prices for interconnection arrangements. An IBP that sells transit to another network will have incentives to charge reasonable prices for that service in order to prevent a transit customer from taking its business to a rival IBP. Furthermore, two networks will have incentives to enter into peering arrangements when, for each, the cost of terminating the other's traffic is roughly comparable to the benefit of having its own traffic terminated by the other, taking into account, among other factors, whether the networks have comparable traffic levels, similar geographic scope, and a roughly comparable input/output ratio at each interconnection point. As long as there are a sufficient number of Tier 1 IBPs of roughly comparable size, there exist sufficient incentives for all Tier 1 IBPs to peer privately with each other at the necessary capacity levels. In turn, this enhances both Internet connectivity and competition among Tier 1 IBPs. Nevertheless, an IBP makes peering decisions on a discretionary basis, and may refuse to peer or may terminate a peering relationship with any other IBP on short notice or without cause if it determines that doing so is in its self-interest.
- When a single network grows to a point at which it controls a substantial share of the total Internet end user base and its size greatly exceeds that of any other network, network externalities may cause a reversal of its previous incentives to achieve efficient interconnection arrangements with its rival networks. In this context, degrading the quality or increasing the price of interconnection with smaller networks can create advantages for the largest network in attracting customers to its network. Customers recognize that they can communicate more effectively with a larger number of other end users if they are on the largest network, and this effect feeds upon itself and becomes more powerful as larger numbers of customers choose the largest network. This effect has been described as "tipping" the market. Once the market begins to "tip," connecting to the dominant network becomes even more important to competitors. This, in turn, enables the dominant network to further raise its rivals' costs, thereby accelerating the tipping effect. As a result of an increase in their costs, rivals may not be able to compete on a long-term basis and may exit the market. If rivals decide to pass on these costs, users of connectivity will respond by selecting the dominant network as their provider. Ultimately, once rivals have been eliminated or reduced to "customer status," the dominant network can raise prices to users of its own network beyond competitive levels. Once this occurs, restoring the market to a competitive state often requires extraordinary means, including some form of government regulation.
- If the merger is allowed to proceed, the Defendants will be in a commanding position vis-à-vis all of their Tier 1 IBP rivals. With a majority of all Internet traffic on its own network, UUNET/Sprint and its customers will derive relatively less benefit from being efficiently connected to smaller networks than will the customers of these smaller networks derive from being efficiently connected to UUNET/Sprint. Whereas in a competitive environment Tier 1 IBPs have roughly equal incentives to peer with each other, the merged entity will be so large relative to any other IBP that its interest in providing others efficient and mutually beneficial access to its network will diminish. Because other Tier 1 IBPs will have a relatively greater need to be connected to UUNET/Sprint, in the absence of a peering relationship, they will be forced to purchase transit services from UUNET/Sprint to maintain adequate interconnection capacity.
- Whereas in a competitive environment Tier 1 IBPs have incentives to charge reasonable prices for transit, the merged entity will be so large relative to other IBPs that its interest in providing reasonable prices or terms for transit service will diminish. Ultimately, there is a significant risk that, as a result of the merger, the combined entity will be able to "tip" the Internet backbone services market and raise prices for all dedicated access services.
- The proposed transaction will substantially enhance the risk that UUNET/Sprint will have the power to engage in anticompetitive behavior. Such behavior may involve refusing to peer with other Tier 1 IBPs for interconnection, and either failing to augment (e.g., by denying, withholding, or "slow-rolling" requested upgrades) or otherwise degrading the quality of interconnection capacity between peers.
- The Defendants already require both their transit customers and peers to enter into strict nondisclosure agreements ("NDAs") as a condition of doing business. The NDAs prohibit these customers and peers from disclosing the nature or existence of the interconnection agreements and, in the case of customers, the prices charged. By enforcing secrecy, these NDAs will enhance the Defendants' ability to price discriminate (i.e., charge different prices) among their customers and to grant or deny peering on an arbitrary basis.
- Another way in which a combined UUNET/Sprint will be able to limit rivals' abilities to compete will be by refusing to cooperate with other Tier 1 IBPs in implementing interconnection arrangements required for the development of new Internet-based services, such as voice over Internet protocol ("VoIP"), video conferencing, live video transmission, or Internet protocol virtual private networks ("IP/VPNs"). These new services are becoming increasingly important to Internet users and require specialized arrangements for effective transmission across two or more backbone networks. For example, cross-network QoS standards that are required for two individual networks to share in providing certain Internet-based services have not yet been adopted on an industry-wide basis. UUNET/Sprint will be able to take advantage of its size to enhance its market power by implementing a QoS standard "on net" while refusing to cooperate in the implementation of cross-network QoS standards. Because UUNET/Sprint will have such a large percentage of traffic on net, customers seeking to use these services over as much of the Internet as possible will have little choice but to migrate to or select it as their provider. UUNET/Sprint will also have the incentive and ability to exploit its unmatched scale and scope to control the development of these new services so that only its own customers will have access to them.