Project Network Effects: Wind River Systems (WRS; WIND) (cont)
Part II. Wind River's Business Model and GG Criteria /the Concept of Network Effects
Synopsis of WIND's Value Proposition.
Wind River Systems rides the crest of the third-wave of hidden, but pervasive, embedded computing, whose products range from pacemakers and MRIs to antilock brakes and traffic signals, from routers and switches to set-top boxes and cable modems, from Web tablet to smart phones, from stellar observatories to the International Space Station. Dedicated to offering a complete software solution for application development in embedded systems, WRS provides an integrated development environment, a set of tools, and other building blocks, forming a platform that permits its OEM customers to focus on their core value-added embedded application. The OEM selects an embedded semiconductor with a specific application in mind, an application that must perform specific tasks and ancillary functions predictably, reliably, and without failure in a real-time environment. The OEMs benefit from selecting WIND's software building blocks because they increase speed to market, reduce their costs, and reduce the risk of incompatibilities among hardware, RTOS, tools, middleware, and their specific application. WIND's commercial off the shelf (COTS) pre-integration of its set of building blocks enables OEMs to manage the increased complexity generated by rapid advances in semiconductors and by proliferating communication requirements. Because WRS solves the growing complexity problem, it enables the OEM to create its own first mover advantage.
WIND's end-to-end solution includes not only a complete range of services, from designing to training, from porting to debugging, and from servicing to managing, but also a set of integrated building blocks composed of five layers of enabling software: (a) pre-integration of semiconductors and software; (b) real-time operating systems (Vxworks AE, pSOS, IxWorks); (c) a generic integrated Tornado III development environment; (d) specialized development environments that are pre-integrated with niche middleware, like Tornado for Managed Switches, Tornado for Internet Appliances, Tornado for Intelligent I/O, Tornado for OSEKWorks, and soon-to-be-announced Tornado for SOHO Gateways (DSL, cable modems, and the like); and (e) middleware, including 40+ communication protocols as well as stacks, drivers, management systems, security, and quality of service.
Only 10% of microprocessors are built for PCs, whereas 90% go into embedded systems. The PC market is estimated at only $460 billion in 2005, whereas the size of the embedded software and services industry and its connected smart devices is estimated at $3 trillion in 2005. Still, it remains difficult to estimate realistically the portion of that market that might flow to Wind River. Nonetheless, Wind River dominates the COTS market as the de facto standard, because: (a) WIND's has two-thirds of the COTS market, which is six to ten times higher than its nearest competitor; (b) WIND's two leading RTOSs are in over 150 million products; (c) WIND has the broadest and deepest set of products and the best service offering, with 650 field service engineers; (d) WIND, with 650+ development engineers, is the leader in innovation and its pioneer, with about 20 "firsts" to its credit, many of which became industry standards; (e) WIND is the only software vendor to support more than 100 microprocessors in 20 semiconductor families; (f) WIND has 50,000+ active developers; (g) WIND has four Centers of Excellence and 30+ value-added resellers; (h) WIND has 500 to 600 WindLink Partners whose specialized tools are pre-integrated with WIND, (i) an EE Times Survey revealed that 56% of engineers perceived Wind River as the leader in COTS usage, compared to 4% or less for QNX, Linux, Microsoft, Microtec, or Microware, and (j) there is room to grow in its RTOS market since in-house proprietary systems still comprise about 58 % of the total embedded market, compared to WRS's 36%, and its competitors' 6%.
Wind River's strategy for growing its market is to: (a) Focus on Customer
Satisfaction and Technical Excellence, (b) Drive Market Share by Enabling Major Processors, and (c) Address the Complexity and Time to Market Challenge. To do so, it plans to build, support, and globally service vertical, end-to-end solutions by proliferating its set of tools and integrated platforms that are specifically tailored to diverse microprocessor and application architectures (thereby achieving higher ASPs) that also are used end-to-end¾across Internet infrastructure, its servers and storage, and its connected smart devices.
An Up-Dated Reprise of How the New WIND Meets Gorilla Criteria
Discontinuous Innovation. The discontinuous innovation is WIND's COTS end-to-end solution for embedded software that replaces the legacy RTOSs that were designed for 8-bit or 16-bit microprocessors in a pre-Internet age. Fiddler and Wilner abstracted the software processes that permitted the WRS RTOS and its set of development tools to be applied to any embedded microprocessor and created an integrated set of building blocks that make the legacy in-house RTOS increasingly anachronistic in our connected age. This Internet era demands that embedded systems be reliable, available, secure, and serviceable, and, yet, in spite of increased complexity, it requires ever-faster time to market.
Open Proprietary Architecture. WIND's architecture is the integration and continuous evolution of its five layers of enabling software. Not only is this architecture open to continual innovation through R&D and acquisition, but also it is open at one end to being ported to over 100 processors, and at the other end to over 600 WindLink partners who add their specialized development tools for specific market niches. Yet, the architecture remains proprietary, protected by copyrights, patents, preserved by the coded knowledge in its software that comes from investing hundreds of engineering years to produce it, as it remains continually upgradeable and innovative through the intangible but valuable expertise and know-how of its engineering team.
High Barriers to Entry. Any new competitor facing the complexity of duplicating WRS's broad and deep products in the face of the increasing rate of new microprocessor introductions would find the complexity barrier exceptionally difficult to surmount, as they must work around WIND's copyrights and patents to develop an equivalent end-to-end system, when its new RTOS, VxWorks AE, alone took over 200 years engineering years to code. The specific BTEs are the competitive advantages of each contributing component of an end-to-end software solution, including porting to over 100 processors, pre-integration of hardware-software, a RASS, memory-protected RTOS, the most complete set of tools for developing embedded applications, vertically integrated development environments for niche markets, a broad-range of middleware, and the ability to service, up-grade, and manage the embedded software through the WEB. Its unduplicated value chain of CoEs and VARs prospers by standardizing on a single software solution because such a de factostandard increases both direct and indirect network effects, posing a high barrier to entry by itself.
High Switching Costs. Switching costs are the inverse of WIND's competitive advantages. Modularity is a basic principle in technology advance, and only WRS offers such powerfully abstracted and broad set of embedded software building blocks. Switching costs are equal to the value lost when forced to choose the next best alternative, usually the offering of the closest competitor. When WIND and its largest rival ISI merged, they raised dramatically the level of switching costs by changing the competitive structure of the industry by creating a single dominant leader with two-thirds COTS market share. There is no longer a competitor that comes close to rivaling WRS's broad and deep competitive advantages. This merger significantly widened their already strong competitive advantage gap by broadening their product line, increasing their design wins, and reaching a critical mass in engineers that lets them rapidly create new and enhanced products while providing outstanding service to embedded product developers. Having far surpassed their competitors in the range and depth of CAs, the barriers to switching for existing customer are immense. To rival the competitive advantages gained by using WRS, old or new competitors require vast improvement before becoming a feasible alternative for complex embedded solutions, improvements that include both R&D to duplicate WIND's end-to-end product portfolio and to establish relationships that would equal the power of WIND's value chain. These R&D and marketing costs must be added to costs associated with the lost time to market that a switch in vendors would entail, compared to the switching-company's competitors who stay with the ever-growing value of the WRS solution.
Strong Value Chain Formation. Tremendous value is generated by WIND's collaborative relationship with microprocessor firms in its four CoEs and in the reference designs of its over 30 VARs. When WRS moved "upstream" in its value chain, its partners became customers who sold bundled WIND products to OEM product developers who sold to mass-market customers. This reduced marketing costs and strengthened WIND's position in its value chain. WIND's relationship with its CoEs and VARs is win-win or complementary; each requires the other to offer an embedded system, consisting or hardware and integrated software. These relationships will strengthen over time because of the need to optimize performance across a wider array of microprocessors of greater specificity and complexity within a shorter design cycle, the need for Internet connectivity and the Internet's rapid growth, with its new requirement of the ability to upgrade the software in its embedded elements. WIND's strategic initiatives of acquiring Rapid Logic for its management software and and EST for its hardware-assisted pre-integration with microprocessors, plus its development of Tornado IDEs designed for specific niche markets by scaling up and adding middleware to its OS increases the relative value that WIND adds to the embedded system. This additional value increases both its revenues and its value-chain-influence over embedded designs.
Tornado Market Extant or Foreseeable. WRS participates in a very broad and rapidly growing market, the entire domain of embedded system, but it is not yet in hypergrowth. It is believed that WIND will become the recipient of cascading diffusion from, several anticipated market tornados within the Internet markets, including lily ponds encompassing I/O servers, Internet infrastructure, and smart devices. Thus, an exploitable gap exists for knowledgeable investors who are less risk adverse because WIND's potential for sudden nonlinear growth is not widely anticipated. On the other hand, WIND is not a Gorilla now because it is not yet in hypergrowth; rather, it is still integrating the ISI and other strategic acquisitions. It is investing heavily in R&D both for new products for rationalizing existing products, which requires high up-front costs. At present, WIND's strategy is to maximize revenues and market share at the expense of operating and profit margins. Conservative investors may choose to wait for the Tornado's beginning, the anticipated growth in royalties, and the return of operating margins to the traditional 20% + level before investing.
A Review of the Concept of Network Effects
WIND is "how connected smart things think." WRS enables the distributed intelligence of the Internet. This is not to claim that WRS is the brain of the Internet, but only that WRS provides necessary building blocks that enable connected smart devices to think and communicate. Taken together, a multitude of distributed intelligences give birth to a life-like, emergent complex adaptive system, which, after reaching a critical mass of intelligent nodes, becomes self-organizing and self-managing.
The Internet emerged over time from the aggregated interactions of its multiple components. Such a distributed intelligence grew from the bottom-up, connecting device to device, connecting network to network, until it transformed into an emergent complex system that became much more than the sum of its distributed parts. We call this emergent higher-order intelligent system, which is a network of networks, the Internet. It is an example of a complex adaptive system that acts as if it were life-like, as an organized, global whole.
The Internet itself is an ideal demonstration of the concept of network effects: the power of the Internet expanded explosively after its number of users achieved critical mass. In 1983, the Internet connected 562 computers; in 1997, the Internet connected over 16 million computers; the number of new computer connections was doubling in size every ten months. One of the interesting properties of an exponential doubling of growth is that it means that approximately half of the people connected to the Internet have been added in the last year. Using 1997 figures, if you have been online three years, you have more experience than 87% of Internet users; after six years, that figure reaches 97%.
An explosive inflection in network growth occurred with the development of URL addresses for the World Wide Web sites. In June of 1993, there were zero Web sites, but critical mass was quickly created since Web sites increased to 200,000 in June 1996, increasing skyward to 1.4 million in September 1997. In 15 months, Web sites increased 7 times and continued sharply upward.
Demand for bandwidth was growing even faster than the number of connected computers, doubling every 100 days according to UUNet. RHK estimates that demand for bandwidth will increase from 125 terabytes per month in 1998 to an estimated 16,200 terabytes per month in 2003, a CAGR of 165%. The explosion in value that the Internet creates powers that growth.
The "power" of the Internet is that it leverages its user's agency-a state of being in action, of exerting power¾an agency that permits creative innovation not only for individuals but also for connected users working together. Connectivity everywhere is becoming the new standard because no one wants to be without this valuable resource. So, not only does the Internet rapidly diffuse innovations, it accelerates the creative destruction of the old and familiar as it leverages the creation of the new and discontinuous.
Recall how economic value was created a few millennia ago when the rise of agriculture permitted the accumulation of capital, a proliferating specialization in the division of labor, and the emergence of entrepreneurs: humankind progressed from hunting and gathering for subsistence to creating mathematics, writing, and civilizations. With a global Internet, economic value expands explosively because of exponential growth in the specialization of labor and from an accelerating rate of economic interactions resulting from the collapse of space and time barriers. Consider contemporary financial markets as a prime example of how both specialization and the rate of timely interactions created new economic value.
The value of the Internet as an asset is determined by the expected benefits that it will generate. Yet, the use of the Internet will permit scientists, engineers, and entrepreneurs to leverage innovation beyond our expectations. The Internet is destined to generate benefits that we do not yet know that we want or need and that we cannot even imagine.
However, what we do know is that the value of any physical network increases as the number of connected elements increases. Metcalfe's law approximates the magnitude of that increase in value by specifying: as the number of elements or nodes increases arithmetically, the value of the network expands exponentially. Metcalfe proposed this model after observing that networks needed to reach a critical mass before they exploded in value. Using a model of a telephone network in which each customer talks to all others once a day, when you have 10 customers, the value of that network is approximated as n-squared, 10 X 10 = 100. If you add one new customer, the value would increase to 121; if you doubled the number of customers to 20, the value explodes to 400. When that value compounds by doubling every 100 days, you have the non-linear explosive growth of the Internet.
Metcalfe's telephone network model assumed one-to-one communication. With the Internet, not only one-to-one, but also one-to-many, many-to-one, and many-to-many communication(s) occur among many sets of people. Necessarily, moving from one-to-one communication to many-to-many communications increases the number of potential interactions. Thus, it must expand the exponent beyond the square. Not only that, Internet communication extends beyond human communication: many devices with embedded systems talk to one another or to many others. Therefore, the exponent of growth in network effects is not limited to a fixed exponent of 2, but may have larger exponents.
What we know for certain about network effects is the bigger the better: the more elements, the greater the value. And, what we also know is that interactivity drives the value higher. Furthermore, the small world effect spreads the outcomes of adaptive interactions faster and further than ever before. The human being is a social animal, and the Internet enlarged the diverse human and nonhuman networks of communicators as it shattered the restraints of space and time.
The Internet revolutionizes how humans live, work, and play. The change is an order of magnitude beyond the usual order of magnitude levels of change in revolutionary products because of the significance of it unparalleled escalation in connected interaction
However, determining the exact exponent of expanding value remains an empirical matter that probably has different solutions in different circumstances. Some arithmetical additions may have different degrees of magnifying power; exponential effects are not created equal: some arithmetic additions double network effects, some triple them, and sometimes effects quadruple in value. (If you add Shuji Nakamura to the network of scientists at CREE, does he add more exponential value than a new engineering graduate?) Therefore, "exponential" growth is properly used only in the vernacular sense of "compounding explosive growth."
Thus, the idea of compounding non-linear growth is at the heart of what is meant by network effects. Network effects exist when the value of a good increases because the number of people using that good increases. Michael Mauboussin argued that successful networks generate huge shareholder value once they reach critical mass and that a locked-in network effect produced a sustainable competitive advantage. He noted that the strongest network effects are driven by (a) interactivity, where there is a lot of contact among the members or nodes, although (b) compatibility, among transactions, communities, or devices also can produce strong effects.
Using a common biological analogy of the spread of flue for the spread of innovation, Mauboussin indicated that the flue spreads as function of the degree of interaction and the degree of contagiousness (susceptibility or adoption threshold). The breakout of an epidemic-runaway explosive growth¾requires reaching a critical mass of infected individuals, creating an inflection point where interaction explosively combines with susceptibility.
The concept of critical mass, the point in cumulative adoption where network effects suddenly accelerate at an increasing rate, can be described as (a) an inflection point, or elbow, in an S-curve, (b) a transition from early adopters to the early majority, or crossing the chasm, or (c) a tipping point, where incremental market share comes at incrementally lower costs. Based on many studies of innovation diffusion, the empirical results suggest that the inflection point in an S-curve of cumulative adoption occurs when a cumulative level of 10 to 20% adoption is reached. Also, quoting the Gorilla Game, Mauboussin offers this rule of thumb describing hypergrowth as a means of defining the threshold for critical mass: "when year-to-year growth exceeds 100%, and when quarter-to-quarter growth is also rapidly accelerating." Popularity is cumulative and inflects upward upon reaching a critical mass.
The rate of diffusion of interactive innovations is more rapid than the rate of non-interactive innovations, creating a steeper S-curve for the former. Nonetheless, a non-interactive innovation is judged valuable whenever a critical mass of people has adopted it because it is now perceived to be the standard solution, expected to become the "next big thing." For instance, if people anticipate that the standard is becoming "VCR," not "Beta Max," then they buy a VCR, whose value increases still further as more buyers jump on the bandwagon and more videotape of movies and the like are created for this growing audience. The sudden, or even an expected, inflection in popularity resolves disputes over which standard is the standard. Compatibility between the standard and its applications, when combined with inflecting popularity, solves the chicken-and-egg problem.
However, when a strong network effect derives from interaction, not only do early adopters influence later adopters but also later adopters influence earlier adopters because they continue to interact. The potential interactions are from each-to-every, not just from older-to-newer users. Thus, given interactivity, the benefit deriving from each adopter increases the value for all adopters. For instance, consider Napster as an example of how both early and later adopters can make more musical selections available to all. Such demand-driven cumulative adoption among youth who wannabe like whoever's cool is truly mega-explosive.
Because of this significant difference in their rate of growth, interactive network effects are called direct, and non-interactive network effects are called indirect. Runaways, whether in epidemics or innovation diffusion, require interactions in which Influence meets Susceptibility.
Influence can meet susceptibility in many venues. But, sometimes it is useful to distinguish between physical networks and virtual networksas venues, because all innovation diffusion requires social communication but all social communication does not require physical networks.
Within the technology adoption life cycle, a network effect describes a dynamic of suddenly noticeable and rapid inflection in cumulative adoption when the diffusion reaches a critical mass and begins to generate explosive growth. Thus, a network effect is the dynamic of hypergrowth. Efforts to describe the inflection point of this accelerating growth dynamic have included phrases like "when a total solution is created," "when a compelling price point is reached," "when the value chain organizes to support a single standard," "when the buzz declares X is the next big thing," "when the last barrier to product adoption is overcome." These can be summarized as, "when the accumulation of competitive advantages tips decidedly in favor of a solution that either is in the process of becoming or is expected to become the popular or standard choice." Influence, as competitive advantage plus marketing, has met Susceptibility at the adoption threshold of the majority, generating explosive growth, that nonlinear growth-surge called "hypergrowth." |
