Project Network Effects: Wind River Systems (WRS; WIND) (cont.)
Part IV. Other Networks that Leverage WIND's Growth
Fifth, the Network of 500 WindLink Partners. According to a recent press release, WindLlink has grown to 600 members; it is the largest network of embedded-software partners in the world. WindLink's members provide products and services that work hand-in-hand with Wind River to supplement WIND's end-to-end software solution in specialized niches. Wind River Direct describes all needed products and services in one place. Some partners are providing software or services that extend or expand functionality, including the sort of middleware that extends the RTOS into an operating system that manages specific niche functions, say, in a cable modem or an input-output processor in a server; other partners have more extensive partnerships with WIND. As an example of the latter, the Motorola Computer Group and WRS have a long-term partnership that encompasses field sales, marketing and business, product development and engineering, and technical support.
Ten examples of WindLink relationships from recent partner press releases are summarized: In December, (1) Celoxica announced new strategic alliances with Wind River and Xilinx. Celoxica's high-level language products for rapidly designing applications in reconfigurable hardware were integrated with Xilinx FPGAs and VxWorks to increase designer productivity and time to market. In November, (2) ANT developed a port of its Fresco browser to the StrongARM/VxWorks reference design platform; and, (3) Applied Microsystems supplied Wind River with CodeTest Trace, supplementing two earlier Trace tools that supported Tornado's use of VxWorks AE.
In October, (4) Ashley Laurent offered a full security suite that included their VPcom firewall and an IPSec compliant VPN for VxWorks; (5) Working closely with Wind River, RTview previewed its SurroundSupport solution for monitoring, diagnosing, self-healing, debugging, and repairing customer relationship management and network management solutions via the Internet; (6) In partnership with Wind River, Banderacom, an InfiniBand fabless semiconductor company, agreed to develop and supply InfiniBand transport software to support VxWorks and IxWorks on Banderacom's iBandit microprocessor architecture; (7) GoAhead integrated its service availability technology with Wind River to take high availability to the next level, service availability, using its cross-platform management software; (8) Snap2, an application and component supplier for digital STBs, in-flight entertainment systems, and Internet Appliances, has begun porting its STB, IFE, and games to Wind River's Tornado for Internet Appliances platform; (9) Extended Systems ported its certified Bluetooth software stack to VxWorks; and, (10) Metro Link joined WindLink to offer customers X Window System products.
Many WindLink partners join the leader in embedded software to insure their products ride along on Wind River's bandwagon. This requires pre-integrating their products with Tornado/VxWorks. Eric Johnson, the President and CEO of Banderacom tells us why, "A survey of our customers indicated that Wind River Systems already supplies the most popular embedded software and development tools for InfiniBand equipment in development or planned."
The many members of Wind River's value chain repeat this refrain of "most popular". WIND's popularity shouts, "direct network effects."
In addition, both their WindLink and WindVentures programs are a source of intelligence about what is novel and best in the embedded software field. Excepting ISI, six of Wind River's seven significant acquisitions over the last two years were former WindLink partners. Each new technology was already compatible with its own building blocks, and each partner had proven its added value to Wind River. How many more partners will be acquired as they prove their merit as valuable extensions of the WRS product line?
Sixth, The Network of over 50,000 Application Developers. A virtual network of embedded application engineers is a valuable resource. Embedded engineers are in short supply; the average code becomes longer as new requirements for communication and service are added; embedded microprocessors become more complex, varied, and ever cheaper; more connected smart devices are invented and marketed; innovation accelerates; Moore's law marches on; the law of the Telecosm sprints on. To summarize these trends using Jerry Fiddler's word, in a word, it is b>COMPLEXITY.
Accelerating complexity requires reliable integrated end-to-end solutions. It favors standardizing on a common platform. Given the shortage of engineers, the mobility of engineers will lead them to leave large companies in favor of new small innovators to gain higher pay and valuable stock options. Job mobility plus Moore's law must hasten the transition to an industry standard, because, first, old companies do not want to waste their time and spend their money training new personnel on an in-house proprietary OS that may soon be obsolescent, and second, new companies do not have the resources to waste on rolling their own. The Internet era requires a constant focus on core, the OEM's customized application; outsourcing embedded software increasingly becomes essential. If a company is to keep pace, Wind River's building blocks provide the necessary context to the OEMs who seek to manufacture smart devices quickly, efficiently, and cost-effectively.
Given a shortage in embedded engineers, the more engineer's trained to use WIND's building blocks the better. Over 50,000 engineers is a huge head start. In addition, Wind River has a University program in which it provides its products for education and training in its embedded software, seeking to enlarge the pool of engineers who can program embedded system software. Included in it is the Graduate Resume and Data Bank (GRAD), which allows Wind River and its WindLink partners to find qualified candidates with hands-on experience with Tornado/VXWorks.
Last August, in a move to expose a wider range of early-stage embedded systems developers to the benefits of integrated design, WRS announced the availability of Tornado Prototyper for free download from their web site, URL: wrs.com
Believing that the embedded industry is on the verge of explosive growth, but also that there is a relative shortage of embedded developers, Wind River is insuring that students or developers can learn their to use its tools or can even prototype a new application using its tools, by using its VxSim, the VxWorks target simulator.
On January 22, 2001, Wind River announced a series of two-day workshops offering advanced training in real-time programming for embedded systems as another strategy to increase the size of its network of application programmers.
However, their most significant strategic commitment to enlarging its network of developers was Wind River's decision in December to deliver its market-leading tools independently of its RTOS. Through its acquisitions of EST and ISI, with its subsidiaries Diab-SDS and TakeFive Software, Wind River added several prominent tool technologies that already were used outside of its RTOS customer base. The WRS IDEs will incorporate key features from these acquired technologies to serve as complete end-to-end solutions for its traditional RTOS customers. In addition, Wind River not only continues to support these newly acquired tool-customers but also has recently extended its product offering of tools to service both in-house proprietary (IHP) and rival COTS operating systems.
Developers often select tools prior to and independently of the RTOS. Wind River has opted to support all embedded operating systems with its best-of-breed tool set, broadening its market to encompass not only its market-leading COTS RTOS, but also other commercial RTOSs, and all engineers using various IHP operating systems. According to a September 2000 EETimes Survey of the embedded systems market, about 58% of the engineers polled were using IHP OSs, 36% were using WIND's RTOSs, and no other commercial RTOS vendor received more than a 6% share of the market for current RTOS usage.
By continuing the evolutionary integration of its new stand-alone tools and of other RTOS's APIs into its Tornado IDE/VxWorks AE platform, Wind River enlarges its markets and extends its offer of one-stop shopping by embracing its competitors as new partners. It extends the set of tools available for use with their rival's OSs in sure and certain hope of eventually converting them to Wind River's total solution. This strategy attests to Wind River's confidence that their growing popularity, based on the quality of their tools and other building blocks, will naturally embrace all embedded software engineers. Once these engineers discover the value residing in their tools for pre-integrating hardware, simulation, visualization, and device management, their pre-integrated standard stacks, drivers, and protocols, and their integrated middleware that extends an OS vertically into many diverse niches, then they must come to see the wisdom in the mantra: focus on core; outsource context; embrace Wind River. Wind River must believe, as I do, that they are going to become another natural technology monopoly because of the acceleration of their indirect and direct network effects.
The Twelve Networks of Lily Ponds
Once OEMs develop and sell their products, which include VxWorks or pSOS, they or their VARs pay Wind River run-time royalties based on average selling price and production quantities. Because Wind River is a ubiquitous embedded system enabler, it may have several customers who are selling competing products. At their Analyst's Day last June, Wind River introduced a model, using the xDSL market as an example, to illustrate how run-time royalties might be estimated. The model multiplies Growth X Market Share = Units, and then multiplies Units X ASP = Revenues. For DSL, according to credible market research, the projected compounded growth rate was 230%, and Wind River was in 47 of 52, or 90%, known manufacturer's DSL products. The average royalty was about a dollar. Using these figures, analysts can estimate future royalties, within the limits of the assumptions they choose. If analysts were to use this method of estimating future royalty, Wind River would becomes analogous to pharmaceutical company that has estimated future revenues from its pipeline of products, because WRS has a real option on future royalties in its pipeline of design wins.
Currently, WIND's sources of revenues are 25%-35% Services and 65-75% Products, with Licenses accounting for about 45%-65%, and Royalties for about 15-25%. Licenses derive from software licenses per seat (or designer) for tools and from licenses per project for embedded software components like RTOSs, TI/O, and TMS, with an additional run-time royalty per device. The advanced Tornados with their vertical middleware for specific niches produce higher royalties than the run-time royalty on the RTOS alone. In January 2001, Curt Shacker, VP of Business Development, presented a slide show that is archived on WIND's web site. Slide 24 presents the role of Wind River in Internet infrastructure, where Tornado for Managed Switches will be used increasingly. The slide indicates that WRS had 94% market reach in cable modems, 49 customers out of 52 vendors; 90% reach in XDSL, with 47 out of 52 vendors; 82% reach in LAN switches, 9 out of 11; 82% in Layer 3 switches, 9 out of 11; and 64% in routers, with 7 customers out of 11 vendors. Also, Wind River had over 290 design wins in this arena in Q3 and was working with 9 of the top 10 network processors. This reach is evidence of their increasing ubiquity, not only in Internet infrastructure, but also wherever their deep reach extends. Broad reach in Internet-related markets and ever-deeper penetration from higher value-added products are the reasons why Wind River's royalties are expected to inflect in many of these Lily Ponds.
Estimating the exact timing of inflections is perilous. Any such estimates would contain errors of varying magnitude because a lot of unexpected and unknowable events can occur between the present and the future. Estimates of CAGR and market share become less reliable across time. Royalty figures are not ordinarily disclosed. Estimates based on about half the cost of the bill of goods can serve as a proxy for an average selling price. Wind River would never offer its internal estimates of the timing of the inflection in growth of a lily pond, if it even makes them, because errors are inevitable and failure to meet an announced estimate would sharply lower their market price per share. Currently analysts and investors just wait for the inflection in run-time royalties to surprise us.
On the SI "Wind, Up, Up, Up" board, recently, its founder, the analyst Allen Benn, has put the model suggested at Analyst's Day to good use in the form of spreadsheets located at a WEB site created by peter grossman for Allen's Lily Ponds. Its URL is:
net-gain.com
Allen Benn uses "lily Pond" as shorthand for a set of defining features; the Lily Pond metaphor is chosen to remind us that compounding growth may continue doubling each day all summer but only become noticeable "suddenly" in the last few days of the summer. Although compound growth is explosive, when it begins from a small base, it must compound for a time to become more than a mere snap, crackle, and pop. The earliest doubles in exponential growth are no more noticeable than the first few grains of rice on the Emperor's checkerboard. The criteria that need to be fulfilled to qualify as a Lily Pond are: (a) signs of high and sustainable current and future growth; (b) dominant and growing market share; (c) success in market place already demonstrated by significant existing royalties; and (d) expected increase in ASP, not to tail off with higher volume. The intent is to identify Lily Ponds as signs of upcoming inflection points, where a few inflections points combined can produce an inflection point for Wind River.
Twelve Lily Ponds and a summary of total royalties are forecast and tracked by these spread sheets, whose assumptions are open to modification and revision as better information becomes available. This is a work in progress, not the final word. Benn should be credited for developing the numbers; he will revise and upgrade these spread sheets with the help of Wind Thread members. For example, see Benn's post # 9122, on January 24, where he suggested dropping the Digital Printers/Multifunction Lily Pond for failing to satisfy the minimal-growth-rate criterion. He favors paying closer attention to the relative size of the projected lily ponds. In particular, he pointed out that the Service Appliances STILL is projected to be the largest in five years. Adding, "To put this in perspective, if all lily ponds save Server Appliances grow only 30% annually, then based mostly on this one lily pond, WIND still would be expected to earn $4 per share in four years according to my model."
The 12 Lily Ponds were: (1) Wireless Hand-Held Internet Devices; (2) Server Appliances; (3) Home Gateways; (4) Consumer Internet Appliances, NES (not elsewhere specified); (5) Network Processors; (6) Digital Printers/Multifunction (to be dropped); (7) IP Phones; (8) Set-top Boxes; (9) Broadband Optical Switches; (10) Digital Cameras; (11) Navigation Systems; and (12) Total Royalty Forecast.
There is no substitute for going to the Web site to examine the Lily Ponds. However, an overview of the potential of the Lily Ponds can be garnered from the Royalty Totals and estimated Annual Growth Rates. In FY 20000, base year royalties are estimated and $13.7 million; in FY 2001, royalties are estimated at $24.1, CAGR = 75%; for 2002, royalties @ $69.6 million, CAGR = 189%; for 2003, royalties @ $228.6 million, CAGR = 228%; for 2004, royalties @ $684.7 million, CAGR = 200%; for 2005, royalties @ 1.67 billion; CAGR = 144%.
All estimated royalties are in addition to royalties from Wind River's traditional platform business. On January 13, in post # 9075, Benn wrote that he estimated that in FY2001, lily pond royalties would account for about 25% of royalties, and royalties for 21% of total revenues; in 2002, lily pond royalties may account for about 50% of total royalties and 25% of total revenues; and in 2003, lily pond royalty should account for about 67% of royalties, and royalties should grow to more than 33% of revenues.
At some point, Benn will combine these numbers with WIND's other sources of income to estimate Wind River's total future revenues. Again, the further out in time you go, the larger the potential error in any estimate. Nonetheless, these numbers speak for themselves, declaring the beauty and fecundity of the Lily Ponds.
Each Lily Pond is a specific market segment whose growth is a function of network effects within several different virtual networks. Leaving aside the networks of Wind River's customers and ported processors and VARs and application software engineers, and the like, focus on the two most relevant networks: (a) Competing Application-OEMs and (b) their End-Customers. The reach figures within market segments for Internet infrastructure that were given above (94% for cable modems; 90% for xDSL; 82% for LAN; 82% for L3 switches, and 64% for routers) provide a selected sample of direct network effects within market segments among competing application developers. The higher the reach, the greater is the popularity of the WIND software solution. Popularity says, "Influence has met susceptibility." When WIND's reach exceeds 50%, it is the leader; whereas, when it appears in the eighth and ninth deciles, WRS is dominant. Endemic popularity foments bandwagon-contagion. The majority's implicit message to all competitors becomes, "If you want to compete with us, you must choose Wind River." Why is this so?
Allen Benn provided some insight in his post #9036 that, I believe, answers this question. The answer is: Time To Market (TTM). Benn sites Sony's decision (discussed below) to use WIND's solution rather than their in-house proprietary Aperios RTOS in set-top boxes. He notes that it took Sony one and a quarter years to develop the set-top box; about the same time it took Intel to develop their new Web tablet. According to Benn,
In both cases the development teams were top notch, experienced and conversant with all needed technologies and tool sets. This suggests that development teams with lesser skills and more constrained budgets probably would take at least two years to complete similar developments under otherwise optimal conditions. Complicate matters by not providing adequate tools, access to drivers, protocols, and middleware, etc. and the time will extend indefinitely. In short, "time to market" is no joke, and clearly is, and will always be, a major barrier to WIND's competitors.
Given a mass market, the second network of OEM's customers is the biggest plum of all, and the most important to understand. This is so because large networks promise the most explosive growth in sales. The market for applications is discussed in the next section.
The Ratio of Many Networks of Applications to Few Killer Apps
So what exactly are the OEMs making as their applications? Wind River's "Cool Customer Design" Awards provide one window into present (and future?) applications. These are not necessarily the best sellers, just ten, recent cool products: (1) This month, Nixvue's Electronic Photo Album won. Without a PC, Nixvue's album stores up to 10,000 high-resolution images, grouping the photos into easy-to-browse albums for viewing, deleting, or storing long-term. (2) In December, the winner was the Monterey Bay Aquarium's Tiburon Deep Sea remote-operated-vehicle. ROV Tiburon is an integrated, unmanned, submersible research platform that features fiber-optic video cameras and sensors that sample and log scientific data in depths as great as 4,000 meters. (3) In November, Bang & Olufsen won for its BeoVision Avant TV, a dazzling home theatre with an outstanding picture quality powered by a digital video processing unit, plus compatible receivers, tape, and disc players, five-speakers, a projector, automatic light controls, and a motorized stand; all operated by the same remote. (4) Also, in November, VoiceTel won for its Voice Activated Universal Remote that recognizes vocal commands in 30 languages to activate TVs, VCRs, set-top boxes, audio, and entertainment systems. In the near future, it will integrate Bluetooth for wireless connectivity, using WIND's BlueThunder networking module, and extend its coverage to telephone and home automation systems. (5) In October, Lambus Enterprises won for its WANGO handheld wireless gaming platform and terminal that receives streaming video through a Web-based VPN and permits gaming competition across town or around the world. Because they used TIA, they can make use of BlueThunder and easily refine, re-build or up-date the product by adding or deleting gaming applets for all sorts of sporting, gambling, or fantasy games. (6) In October also, eRemote won for its second-screen interactive television solution [a Web tablet for email, PIM, etc. that provides a second screen that is combined with a remote controller] that manages personalized preferences and simultaneous Internet access, interaction with advertisers, and control of peripherals. [Gemstar, are you listening?] (7) In September, Applied Biosystem won for its remarkable ABI PRISM 3700 DNA analyzer, which I described previously in my supplemental Project Hunt Report. (8) In September, Wind River made a strategic investment in Pingtel and awarded a Cool Customer Design" award to its xpressa VOIP phone that was introduced last March. The xpressa phone leverages the infinite potential of IP networks through a Java-enriched open environment for the delivery of personalized voice services and extensible applications not possible with a "dumb" phone. (9) In August, Telestream won for its ClipExpress digital media delivery system, a low-cost version of its innovative and award winning predecessor, ClipMail Pro, which is also WIND-based. Time-efficient and cost-effective, Telestream's products transform professional quality video, audio, and timecode into a digital format for transport over and IP networks. Telestream's customer list reads like a Who's Who of the film and television industry, with additional applications in business communications, education and training, and telemedicine. And, in July, (10) goReader won for its electronic smart device for textbooks [a combined Web tablet and e-book]. The goReader downloads all textbooks, study guides, calculators, calendars, and dictionaries into an e-book with a full-color, full-size touchscreen that displays crisp text and images and that also allows note-taking and highlighting directly on the text, note-sharing, creating personalized study guides, up-dating of information, forming of communities, and instituting e-commerce. [Are you listening, Gemstar?]
If you thought you could build some interesting structures using Lego blocks, just imagine the range of innovation, the connectivity, and the personal empowerment that will be built into a future of smart devices using Wind River's building blocks. "We want to be the standard software behind all of those devices in the Internet," says David Fraser, VP of WIND's Networks Division, "… the standard software that unifies all these devices so that people can manage them in the same way." So, if you are an investor, enjoy imagining WIND as the universal enabler than links and manages all these devices.
All innovations diffuse, but only few diffuse widely, and only a rare innovation diffuses pervasively. I believe it was economist Brian Arthur who first said, "Of networks, there shall be few." If you accept this proposition, it implies that there will be only a few significant standard solutions and only a few devastating killer apps because so many phenomena emanating from complex systems follow power curves: as the power of an exponential change increases, its set of members decreases logarithmically. So, the stronger the network effect, the more rare it becomes.
The complex system of the Internet is unrivaled in its power and uniqueness. Imagine it as the nexus of all value chains of all businesses and consumers worldwide. Eventually, only a few, gigantic physical networks will evolve and be combined into the Internet. All will be connected using a few types of wired and wireless portals, which will monopolize most of the traffic. Network effects and standardization are soul mates. Dominant businesses (and good long-term investments) are necessarily rare; the larger their growth in value, the further out the power curve stands the company; so, rarity is guaranteed. Although increasing returns reward the winner, simultaneously they also punish all competitors. Fame and fortune follow the law of network effects: popularity continues to explode value. However, the exponents become ever-higher powers for a few fortunates, and higher still for the truly rare famous fabulous fortunate.
Innovation is an evolutionary process that is characterized by punctuated equilibrium. Change occurs, not slowly and incrementally, but seemingly suddenly and in big, discontinuous leaps. Down and Mui (2000, p. 4) define a killer app as "a new good or service that establishes an entirely new category, and by being first, dominates it, returning several hundred percent on its initial investment." A killer app is by definition both popular with consumers and a rare outcome. Killer apps are the rare survivors on the power curve of all applications. Of killer apps, there will be few. There will be many applications that use Wind River's building blocks, but only a few will become killer apps. (Leaving aside the question of whether WIND's COTS software solution for embedded system is itself a killer app.) However, as a universal enabler and manager of smart devices, Wind River's wide net may yet scoop up few outstanding applications. |
