Call it the battle of the networks—the telecom networks, that is. In one corner are the incumbent equipment and service providers that have together spent upward of $300 billion building extensive proprietary phone networks. In the other corner are a crowd of upstarts hoping that the new technologies and superior economics of Internet protocol (IP) networks¹ will allow them to attack the incumbents’ position. Driving it all are a boom in voice and data traffic and the inexorable transition from a circuit- to a packet-switched telecommunications infrastructure.

The battle has already been joined in the datacom space (see text panel for a definition). Attackers have made considerable inroads into datacom during the past five years. Companies like UUNET (now part of WorldCom), BBN (now part of GTE), and Ascend/Cascade are using the Internet and its open standards to pick off horizontal slices of the telecom industry’s value chain such as hosting and redundancy. Now, the competition is spilling over from datacom into the much bigger and more lucrative telephony space as they attempt to hijack the voice traffic currently routed through proprietary networks. How far they succeed will largely depend on what incumbents do to protect their sizable investments in these networks.

We believe that the existing proprietary telecom network will retain much of its traffic and continue to generate solid returns for those incumbents that manage it well. But it remains to be seen whether incumbents or attackers will capture the lion’s share of the new value that will be created in datacom over the longer term.

Lessons drawn from other industries demonstrate that incumbents and attackers can both be winners in the battle for value. They also show that during the transition to open standards and products, the risk of making wrong bets on technologies or markets escalates—but then, so does the risk of making no bets at all. Failure to lock up positions in attractive emerging markets may compromise value capture over the entire life of these markets.

In short, this is exactly the right moment for telecom executives to take seriously both the transition to IP-based networks and the threat from datacom attackers. The coming battle will not be for the faint of heart. Value will migrate to the top two or three players in each industry segment. However, by looking at what has happened in other industries, especially computing, telecom players can gain insight into the business models that are likely to emerge and the ground rules for those who want to win.

An accelerating transition

Evidence is mounting that the transition from proprietary to IP-based networks is gathering speed, opening up a new battleground with new rules. Attackers are making substantial inroads into the datacom space in the first battle. The second will take place in the telephony market.

New battleground, new rules

Less than a decade ago, US customers were served by three autonomous proprietary networks. The phone network carried voice, the LAN/WAN carried data, and the broadcast network carried video (at that time, television). Each network was like a vertical pipe: a closed system in which a single vendor provided all of the hardware, software, and services to more or less captive end users.

That orderly world is becoming increasingly chaotic. Capitalizing on digitization, the Internet has enabled the vertical disintegration of the three pipes and created opportunities for many new entrants to participate in the telecom space. As a global network of networks, the Internet is a more economical means of transmitting voice, data, and video than the proprietary networks. Moreover, it has open standards for communications, applications, and services, making it possible for different companies to launch products that work together.

Deregulation has also played a part in this transition by ensuring that customers are no longer tied to a single provider. The 1996 US Telecom Act and the mandate for open competition in Europe by 1998 have created opportunities for attackers to compete in markets that used to be monopolies or oligopolies. As a result, customers are increasingly free to choose different components from different vendors and assemble their own solutions.

Battle 1: Invade at the periphery

The first point of attack has been on the periphery of the core telephony network, in the datacom market. Attackers have gained ground simply by providing offerings that complement existing telecom services. They are on track to capture the lion’s share of the $60 billion datacom market by 2001.

Attackers entered the datacom market by offering Internet services such as access, hosting, extranets, e-mail, and World Wide Web searching. Over 5,000 Internet service providers (ISPs) including UUNET, PSINet, and GTE/BBN are now advancing on fronts where they enjoy price advantages, such as fax, messaging, and EDI. Some incumbents are joining forces with these attackers to capitalize on the emerging market. Examples include GTE’s recent $600 million purchase of BBN to gain hosting and redundancy capabilities, WorldCom’s $14 billion purchase of MFS/UUNET, and Deutsche Telekom’s investment in VocalTec, one of the premier IP telephony software providers.

All this activity signals the acceleration of the shift to IP-based networks. It also shows how a sizable market can come into being in a short time. A market that did not even exist just five years ago is now thought to be worth $8 billion in the United States alone. Though attackers (Internet- or IP-based players) account for just 30 percent of revenues from data transmitted, as against 70 percent for proprietary networks, their market capitalization to revenue ratio is roughly four times that of incumbents on average (Exhibit 1).

Of the expected over $60 billion datacom market (Exhibit 2), the Internet portion will grow at a much faster rate than the proprietary segment (about 58 percent as against 20 percent compound annual growth rate). That means attackers could capture two-thirds of the data traffic market by 2001. The cost advantages their IP-based networks give them over traditional phone networks will speed their growth.

IP-based networks are far more cost-efficient than traditional circuit-switched networks (Exhibit 3). Digital technology is roughly 10 times more efficient in terms of bandwidth than analog. Moreover, the compression of digital signals can offer a further bandwidth saving of up to 75 percent, depending on the data. A typical Internet WAN is at least twice as efficient as a traditional network when user needs include meshing, dial-up access, and international or trading partner connections. Admittedly, the IP-based solution does use more overhead than a circuit-based solution when transmissions have to go over a larger route structure, but as long as capacity exists, the cost advantages remain.

An added advantage is that IP-based networks with multi-point connections have built-in redundancy; calls basically route themselves. By contrast, most proprietary networks have a hub-and-spoke topology that requires costly point-to-point redundancy.

Moreover, instead of relying on telecom-focused vendors to develop solutions for proprietary networks, attackers can exploit technological advances from computing, semiconductors, and other industries that use the network.

The more attackers benefit from the constant flow of capital and talent into this promising market, the better equipped they become to address such issues as performance and security. Companies with the ability to solve IP network shortcomings will continue to start up. Major players and small technology houses are already forming partnerships, setting the stage for new levels of service. UUNET, for example, has revised the structure of its agreements with ISPs and is offering different prices for different services. From a security standpoint, encryption features such as routers and fire walls now abound in every device in the network.

Battle 2: Then challenge the core

Having established themselves on IP-based networks, attackers are now setting their sights on the lucrative telephony market. Voice service is its most attractive segment. At present, it is almost 20 times larger than the datacom market in bits and data volume, and enjoys substantial profit margins.

As in datacom, attackers will seek to capitalize on their price or performance advantage over proprietary networks. A voice-over-data network such as voice-over-frame relay is about 40 percent cheaper than a voice-over-voice network. It requires customers to use a PC with a phone capability instead of a telephone. Using the Internet instead of a frame-relay network is still cheaper. The large price umbrella provided by the existing voice service means there are substantial cost savings to be made for all users, especially for international calls.

Even with today’s not so great technology and service levels, many customers are already being lured away from proprietary voice networks by the promise of savings. For the world’s major telecom carriers, the threat of losing billions of dollars in revenue is too real to ignore. The migration of users from traditional voice networks to IP-based networks won’t happen overnight, but it will happen. Companies already on the attack and those to come are likely to accelerate the convergence of voice and data networks; for the views of two industry participants, see the text panel.

It is always risky to predict where attacks will come, but we expect that customers’ service and cost requirements will drive steady migration in seven product segments (Exhibit 4). The migration is already beginning in domestic and international fax traffic and international consumer calls, because IP networks meet the quality and cost needs of many consumers and a good portion of small to medium-sized businesses. Current estimates suggest that 8 to 12 percent of international traffic will be routed over the data network by 2001.

In an attempt to stem the tide, some incumbent telecom service companies are launching defensive programs. Deutsche Telekom, for example, has experimented with offering consumers international telephone calls via the Internet at approximately one-tenth of the price prevailing on its proprietary network.

Insights from computing

The transition made by the computing industry can offer useful insights into the possible outcome of the battle for the telephony market. Over the past 15 years, computing has moved from proprietary to open systems. In the original mainframe environment, single vendors provided hardware, software, and services in a complete proprietary solution. The advent of PC and server technologies represented the first wave of vertical disintegration. Their performance and cost advantages lured customers away from the mainframe providers.

A new segment was born: the networking industry that ties PCs and servers together. With it, a new breed of competitor emerged that typically focused on hardware, software, or services alone. Competition was no longer based on complete solutions.

Some differences...

Clearly, there are differences between telecom and computing. For one thing, the transition is occurring more slowly in telecom. Incumbents are more constrained by geography and have large fixed investments in network infrastructure. Historically, they have also had to play by the rules of regulation, while computer players have enjoyed market-based competition.

... many similarities

Despite these differences, the two industries have much in common:

• Migration from closed to open standards
• Emergence of new technologies (PC and Internet)
• New entrants with different business models (hardware, software, and service providers in computing; ISPs and computer industry players in telecom)
• Horizontal industry structures that allow companies to specialize in a single slice of the value chain
• Decentralized decision making in which customers can purchase products and services from "bottom-up" decentralized locations in their companies rather than through central agents. The computing analogy obviously cannot tell us how telecom will end up, but it can help us anticipate patterns that will evolve. In particular, it illustrates three trends in the flow of value and the basis of competition that should hold true for the telecom industry of the future.

Growing the pie. Attackers will not only seize value from incumbents, but also increase the overall size of the pie. If we look at the market value of a sample of computing players in 1980 (before the transition) and 1997 (after), we can get some idea of the possible value creation opportunity for new entrants (Exhibit 5). Hundreds of new computing companies occupy the space once shared by just a handful. At the same time, the value of the market has soared from $110 billion to almost $600 billion.

If patterns in computing hold good for telecom, value will be redistributed in at least three ways. Attackers will capture a portion of existing revenues from incumbents. They will also gain a disproportionate share of the new revenues that will be created as customers demand new services. Finally, value will transfer to customers as price umbrellas collapse.

Innovating from the outside in. Telecom’s potential is attracting talent, capital, and public attention. It will also encourage rapid innovation. We expect this innovation to be based on open standards, partly because many telecom attackers have their roots in computing and partly because that industry serves as a model of success for all players.

Before it migrates inward, innovation will begin at the telecom periphery: it is here that most of today’s open standards are developed, as there is freedom to experiment with new products and services in less critical applications. The PC phenomenon began with users and non-networked devices; the advent of LANs and, later, WANs allowed users and devices to be methodically networked. The development of datacom technologies in the telecom space is following a similar pattern (Exhibit 6).

Exploiting the power of share. Competition for share will take place within each horizontal slice. If telecom emulates computing, a disproportionate share of value will be captured by the top two or three players in each slice. As Exhibit 7 illustrates, the share leader in each computing industry slice captured value (in net income after tax) ranging from 1.14 to 3 times revenue. Compaq, for example, has an 11 percent market share in the PC manufacturing slice but has locked up 35 percent of the value. In printers and peripherals, Hewlett-Packard possesses roughly double the market share of its nearest rival and commands 69 percent of the value. Microsoft and Intel have both sought to achieve a dominant share in the slices they targeted; between them, they have captured more than 60 percent of the total value created in the PC market.

These leaders have at least two clear advantages over their rivals. First, share leadership enables them to develop scale and scope economies in such areas as cost spreading, learning, ability to reinforce a patent portfolio, and breadth of relationships with customers, suppliers, and partners. Second, it helps them seize opportunities to set technical and performance standards and construct winning industry webs.

Emerging business models

What do these insights from the computing industry mean for telecom? We think at least three business models will emerge as the industry shifts to open, distributed networks.

Model 1: Integrate across the layers

Companies following the integrator model aggregate product or service solutions across layers or slices of the value chain without owning assets in any of them. In computing, EDS pursues this model by assembling purchased hardware and off-the-shelf or custom-written software into a complete solution for customers. This is potentially the easiest model for incumbents to pull off, exploiting as it does their existing strengths in customer relationships and network management. However, because the scope for differentiation is limited, such a strategy may create less value than other options.

At first glance, this might seem the logical choice for telco service providers. They are accustomed to a service business model, and typically own their customer relationships. Recent moves by large players seem to be following this path. AT&T’s service subsidiary, AT&T Solutions, began offering integration services in 1995, the same year that MCI completed its $1 billion purchase of SHL Systemhouse. Sprint recently purchased Paranet to enter the computer network services business.

From the customer’s point of view, independence is one of the main attractions of integration service companies. Customers like to have unbiased advisors assemble their "best of breed" solutions. The question is whether these companies can gain any synergy from running a telecom network and acting as an integrator.

Model 2: Create multi-layered solutions

This model involves leveraging a position in a layer by integrating your own products or services with those of others. Ownership of a particular asset or technology provides a platform on which players build as they try to control more than one slice or layer of the value chain.

In computing, Hewlett-Packard pursued this model. It aimed not only to be a leading player in peripherals such as printers, but also to leverage its customer base and manufacturing skills by entering PC hardware, semiconductors, and services. Intel’s forward integration into motherboard manufacturing and PC assembly represents another example.

In the IP-networked world, companies adopting this model typically own assets or intellectual property in a layer as well as customer relationships. GTE’s purchase of BBN represents an example of an incumbent strategy. The acquisition will boost GTE’s ability to run its existing network, which it will upgrade with a $2 billion investment. It will also be able to retain control of customers in the near term, let other players put value-added services on top of the network, and transition the GTE customer base to these new services over time. In addition, BBN will help GTE not only control the network, but also edge out into hosting and backup/redundancy services.

Model 3: Dominate a horizontal slice

This model is the hardest for incumbents to pull off, but it also offers the biggest rewards. In this model, it is market position that determines value creation. The best strategies thus focus on early entry to establish a position that will lead to sustained leadership.

Successful exponents of this model are typically nimble attackers that use intellectual property assets as competitive weapons to collapse price umbrellas in existing markets. Attackers pursuing this strategy in IP networks include a host of security providers such as VeriSign. Qwest is also adopting this model in an attempt to become one of the premier network operators.

For incumbents, this model poses a major challenge. Indeed, no computing incumbent has succeeded at it—not even IBM, the originator of the PC. Incumbents have difficulty managing multiple business models nimbly. They have legacy assets to protect, and find it hard to sell to customers in only one layer of the value chain when they have traditionally competed in several. Moreover, they tend not to be attracted to markets that they perceive as smaller than those in which they already compete.

Managing through the transition

If these are the forms competition will take, how can incumbents or attackers get from where they are today to their chosen model for the future? How can they capture value as they navigate the transition from proprietary networks to an IP-based world? Two cases from other industries provide lessons that may prove helpful to telecom equipment suppliers and telco service providers.

Lessons for incumbents: Computing players’ experience

As incumbents in the computing market during the advent of PC-based open systems, IBM, DEC, HP, and Data General provide examples of the models we identified. Each pursued most of the elements of the model 1 integrator role by migrating their service and support businesses toward professional services. Each also implemented model 2 to varying degrees by using existing product or asset positions in a single layer to create multi-layered solutions. DEC, IBM, and HP all capitalized on existing positions in semiconductors (and in printers in HP’s case) to migrate their customer bases to the open portion of the industry in the PC and server hardware markets.

From a tactical standpoint, each attempted to accomplish the migration by introducing hardware based on open systems architecture—a radical departure from existing proprietary systems. Examples include HP’s migration to HP-UX operating systems and PA-RISC microprocessors in servers; IBM ensuring PC and server compatibility not only with its operating systems (such as OS/2), but also with Microsoft’s NT and Windows products in PCs and servers; and DEC’s moves to offer NT-based hardware and forge a service partnership with Microsoft. While continuing to support existing platforms, each company has worked to shift its installed customer base to new platforms over time using a combination of migration tools and price incentives.

HP’s financial performance shows how successful it has been at managing the transition. Between 1991 and 1996, total computing revenues grew by 16 percent per year, and open systems revenues by 54 percent. Meanwhile, revenues from proprietary computer systems dropped from 80 to 20 percent of total computing revenues. In the same period that two other major incumbents, IBM and DEC, lost a combined $18 billion in market value, HP’s market capitalization grew tenfold.

Five lessons can be drawn from the transitions made by these incumbents:

Migrate your customer base to open standards if the market opportunity justifies it—even at the risk of cannibalizing your core business

Embrace open standards rather than viewing them as a temporary discontinuity. Have the courage to migrate your customer base to open standards if the market opportunity justifies it—even at the risk of cannibalizing your core business.

Focus on a few value sources that you can control and that will enable you to secure strong market positions. As the experiences of IBM, HP, and DEC suggest, incumbents should begin the transition to open standards at the system level, while retaining intellectual property at the component level. Wait until a later stage of industry evolution before migrating to open standards at the component level to complete the transition—but make sure you do complete the transition, or the market will complete it for you.

Leverage incumbency. Extend existing supplier, customer, and partner relationships and assets. DEC used its service and support organization to forge a service agreement with Microsoft in the "open" NT arena; similarly, HP used existing contacts to forge new relationships with Intel, Microsoft, Informix, and others. Incumbents should also launch tools and programs to migrate customers systematically to open standards. Customers will stay loyal to a company if they have an idea of how its products will develop and if they are assured the transition will not be too painful.

Focus on dominating horizontal battlegrounds. As we have seen, the leaders in each slice of an industry capture disproportionate profits. Focusing on arenas that make good use of existing positions and intellectual property can place incumbents on the path to winning within a slice.

Reduce costs in line with those of aggressive attackers. With the rise of open systems, computing industry incumbents were faced with competitors that boasted lean cost models. IBM, DEC, and HP responded by reducing their SG&A costs by between 29 and 36 percent between 1992 and 1996.

Lessons for attackers: The USR/3Com experience

Although equipment manufacturers often follow the direction set by service providers—their customers—to win during a transition, they need to pay particular attention to product management. USR’s strategy was to dominate a slice. It succeeded in achieving price and performance leadership in the modem market both within a single product generation and across the entire technology life cycle. From one among several players in the 14.4 modem product generation, it became the share leader in the 28.8 generation. USR now owns over 40 percent of the market—almost double the share of its closest competitor.

As it has grown, USR has remained focused on its slice. Despite its ongoing battle over modem standards with Rockwell and its expansion into routers, hubs, and NICs via the recent 3Com merger, USR is still intent on winning in modem hardware.

Attackers can take several lessons from USR’s experience:

Use intellectual property assets to secure a strong initial position. USR developed many of the modem software interfaces that operate with PCs and networks. These interfaces gave it bargaining power, allowing it to negotiate favorable terms and lock in suppliers and customers around de facto standards.

Balance high-value with high-volume strategies by launching products that are differentiated by advantages in price and performance (transmission speed and standards). USR uses pricing to reflect leadership or dictate volume. Early in a new product generation, it prices to capture value and encourage market acceptance. Later in the cycle, as products become more commoditized, it maintains leadership by shifting to a cost leader and volume strategy in which it reduces prices but preserves margins by spreading costs across larger volumes. In other words, it has eschewed traditional strategies such as finding attractive niches or building a single-generation product and harvesting the business. Instead, it seeks to capture maximum value from each product generation from the beginning to the end of its life.

Leverage technology platforms to maximize reuse and minimize invested capital. As it develops each new generation of modem, USR draws on the physical and software design of the previous generation. Thus, while speed and capabilities more than double from one product generation to the next, much of the engineering builds on existing core design.

Ensure backward compatibility to lock in customers. With each product generation, USR has been able to dictate some portion of both the physical design and software components to its suppliers and customers. In addition, it builds "hooks" in the application interface layer of each generation to make its modems compatible with older systems, thereby creating an incentive for customers to be loyal when they upgrade to newer products.

Set the pace of innovation by balancing customer and supplier relationships. In developing each new modem generation, USR has employed two tactics that have helped determine the pace of innovation. First, it boosts its bargaining power by fragmenting its supplier base and finding second sources for key components. Second, it undertakes joint design efforts with "mentor customers" that can help it gain skills and improve its knowledge of product direction.

Should telecom executives add datacom to their already full plate of challenges? The answer is most assuredly "Yes." Few incumbents would have predicted the astounding growth of the datacom market over the past five years, or the positioning of attackers to take future share not only of data traffic but also of voice services. Companies that have prospered during the transition have been prepared to embrace open standards, compete in horizontal slices, and adapt as the market evolves.

Incumbents must act swiftly while there is still time to influence the outcome. Companies that lock in positions in the next two or three years could be winners for most of the coming decade. As they develop their strategies, both incumbents and attackers can learn from the computing industry. The payoff for getting it right in datacom may never have been higher, while wrong moves could cut players off from the next wave of growth.