Supplemental Project Hunt Report: The New Wind River
Part I. Hidden in the Mist
The new Wind River Systems (WRS; WIND) is a gorilla-candidate that is hidden in the mist. The mist clouding the minds of investors results from three factors: (a) unlike the familiar PC, embedded chips are hidden and their real-time operating systems (RTOS) are unfamiliar; (b) the new WRS is transforming: redefining its mission, integrating a large acquisition, reorganizing its business model, and advancing the architecture of its operating system by adding memory protection and high availability; and, (c) the announcement of design wins, introduction of customer products, and subsequent run-time royalties are time-lagged, obscuring the visibility of the onset of WIND's Tornado.
The Mist of Hidden Ubiquitous Computing.
Living in the post-PC era, it is important to recognize that embedded technology is the third incarnation of computer power: each era is marked by a different relationship between computers and human users. In the Mainframe era, computer power was scarce and centralized. There were many who wished to use the system, delivering their requests and sacrificial punched cards to the high priesthood of technicians, the few who alone could interact with the sacred machine. This was the One-to-many era. The introduction of the PC personalized computing, creating the one-to-one era. During the post-PC transition, the PC continues to shrink into laptops, PDAs, e-books, teleputers and the like. As the third computer revolution, embedded computing is becoming ubiquitous pervasive computing. One person is surrounded by many, usually invisible, embedded computers, a one-to-many era, in which integrated circuits are everywhere but appear to be nowhere. Not only that, chips relate to chips, not necessarily requiring humans as their end-users.
Intel's Director of Research, Dr. David Tennenhouse reported that in the year 2000 the industry will produce 8 billion microcomputers, with more than 95% of them embedded in everyday objects. According to Tennenhouse (2000), "Ubiquitous connectivity and adaptive software will enable systems to be self-organizing and self-configuring, creating a pervasive and largely invisible computing paradigm." As computing power becomes abundant and pervasive, it disappears into the infrastructure, becoming part of everyday life, becoming ever more hidden and invisible.
The Mist Of Transforming from Old to New.
The old WIND sold an RTOS and tool set; the new WRS provides software solutions for the Internet and its appliances. WIND's model of embedded software solutions is encapsulated in this slogan, "How Connected Smart Things Think." The new WRS offers a software solution that includes: (a) a redesigned operating system, enabling new levels of reliability, availability, service, and security (RASS); (b) an integrated development environment of multiple tools focused, like the winds of a Tornado, on creating a product-developer's application; (c) a suite of middleware, including diverse drivers and communication protocols for connectivity; (d) hardware-software integration; and, (e) complete service, training, and support. This architecture embraces a world of smart devices.
Redefining its mission was an outcome of an inner process of flux-an accelerated period of change that was not only rapid but also unusually creative, like an evolutionary period of punctuated equilibrium. There were outward signs of upheaval, like the change in CEOs, but the winds of change inside WIND were stronger still: WRS was not only changing its game but also the rules of the game. The change in the game is captured in the redefinition of mission: the new WRS provides embedded software solutions for the Internet and connected smart devices.
However, the changes in the rules of the game were even more important. At their analyst's conference in June, WIND provided some information that is preserved in the slides on their web site that permitted me to infer how Jerry Fiddler and his WIND team saw the problem, the solutions, and the opportunities. I do not know if their insights came suddenly or gradually during the period of flux. I imagine that they arose following the origin of the transition, the collaboration with Intel, which is described as the first event below, and came before the two subsequent decisional events, which are described as how WRS chose to grow its value chain and how WIND perceived the need to develop advanced technology for their RTOS.
Key Decisions Follow From Market Drivers. Five macroeconomic factors are converging to drive WRS forward: (a) Moore's Law ordains that hardware will continue to become more powerful, (b) Metcalfe's Law creates an even more rapid increase in usable bandwidth, (c) Tennenhouse's Law of Pervasiveness ensures that computing and communication will be present everywhere at anytime, (d) Kurzweil's Law of Accelerating Returns promises explosive innovation in a myriad of new products, and (e) Holland's Law of Genetic Adaptation indicates that killer applications certainly will evolve among the connected smart devices.
Although such drivers are a blessing, creating tremendous opportunity, they also require solving immense problems. At the June analysts' day, Jerry Fiddler characterized the problem in one word as: COMPLEXITY. This was the first crucial insight. It is complexity precisely because computing hardware becomes better and better, new chips and consumer products emerge sooner and sooner, Internet growth occurs faster and faster, consumers demand more, the sooner the better, and connectivity becomes anytime and everywhere. Yet, despite an explosion in product designers, fewer and fewer software engineers are available to program increasingly complex embedded software. Dealing with mounting complexity is software's "last mile" problem.
In general, software is the bug in the works. Although it is the brain, software remains the slowest developing element in the advance of technology. It has not followed Moore's law; instead, coding software remains time intensive, with the number of coded lines per project continually increasing. Unlike microprocessors, software is not becoming smaller, cheaper, and faster. Thus, the crux of the problem is: how do software developers deal with an expanding variety of new microprocessors that are created ever more rapidly given an ever-shorter cycle of production and introduction?
On June 13, 2000, as the keynote address to the Embedded Processor Forum, WRS Chairman Jerry Fiddler urged closer partnerships between hardware and software producers. Fiddler argued that two market factors created the necessity of closer cooperation. First, the general-purpose computing platform of the desktop and the embedded platform's focus on devices built for a specific purpose were converging in the world of smart devices. Defined by some combination of hardware and software, smart devices are built for a specific purpose but they also are programmable. Second, the already frantic rate of development of new hardware is increasing. "Last year, there was a new chip derivative every six weeks," Fiddler said, "Now, it is every four weeks." At the analysts' day, a slide indicated that in 1995, system integration, that is, the integration of software with hardware, took 10% of the time in the development cycle; whereas, in 2000, it takes 30% of the time.
No one knows what the complete answer is, but Fiddler believes pre-integration of software with hardware and the co-creation of integrated software/hardware are steps toward a solution. Fiddler announced that WRS would deal with the fast pace and changing markets by developing Centers of Excellence (CoE). This move changed the rules of the game.
My reading of the new WIND's strategic vision is, "We must closely partner with semiconductor developers, go deeper into the infrastructure of the Internet, and go out into the emerging world of smart devices. Our mission is to help connected smart devices think."
Although still in process, three events were turning points in this ongoing transition from the old to the new Wind River. First, the transition began when WIND seized the opportunity to collaborate with Intel on the development of a newly designed input-output system for servers and related appliances (I2O)-now called InfiniBand. WRS made a commitment to offer not only an operating system but also a complete software solution. This collaboration involved a commitment of WIND's resources to tailor its generic VxWorks into a specialized version, IxWorks, and specialized extensions of tools, Tornado for Intelligent I/O, to facilitate input-output product development. The enticing payoff was that Intel served as a value-added reseller, becoming the first of 30, by including a copy of Tornado/IxWorks with its chip and paying WIND for the developer's license and any run-time royalties that accrued.
Second, WIND grew its value chain. It acquired the best small companies for their value-added technology and talented engineers. It, acquired communication protocols from Epilogue and RouterWare, three types of GUI's from Zinc and NCI, hardware-software integrator EST, AudeSi's design tools for JAVA appliances, XML capability from IceSoft, and more. Acquiring ISI, their largest competitor, transformed the structural dynamics of competition in their industry by securing a two/thirds COTS market share and enough engineers to support both their extensive product development and the needs of new customers. Moreover, it opened Centers of Excellence and began to use other high tech firms as value-added resellers. After pre-integrating microprocessor/software, the value-added reseller bundles and sells the microprocess/Tornado/VxWork embedded system.
Third, on September 11, 2000, WIND introduced its High Availability VxWorks AE (advanced edition) and Tornado Tools 3. Formerly called Cirrus, this upgraded, redesigned RTOS and integrated development environment required WIND to invest 200 engineering years in its production, but it advanced their technology beyond the reach of any competitor. Field-tested by more than 125 customers, this major platform upgrade is designed to meet RASS requirements, offering high availability (HA) that enables 99.999% up time. It meets a six sigma standard, limited to only 30 seconds of downtime per year. The HA approach implements a fault detection and recovery process that keeps the system operating after a failure. It is targeted at manufacturers building service-critical Internet infrastructure and appliances, mission-critical defense and aerospace systems, and life-critical medical products, giving them advanced control of both performance and protection. Its second fundamental innovation is "protection domains," which is a new model for protecting memory that is superior to the process model. The domains protect by dynamically isolating memories in the kernel and in its different applications, preventing corruption of the system by any imported code. This new approach to memory protection uses less memory and executes faster, but also is less intrusive, more flexible and more easily understood. Wind River has seven patents pending on its protection domain architecture.
According to Jerry Fiddler, co-founder and Chairman, "This new development platform enables a new class of [HA] embedded developers to take advantage of Wind River's value proposition . . . professional services, advanced networking protocols and middleware, advanced development and visualization tools, support for the leading microprocessor families, and the value-add products and services of over 500 WindLink ™ Partners."
This advanced edition of WIND's Tornado/VsWorks can go deeper and can go outward. The completion of this research-and-development-intensive transition to the new Wind River will be marked by two events in 2001: the integration of pRISM+/pSOS+ into Tornado/VxWorks in the edition current called "Cumulus,"and the return of operating margins to the mid20s.
The Mist of Time-Lagged Returns.
Finally, like the Wind River of Wyoming for which it is named, WIND has been difficult for investors to grasp, like water flowing through our fingers and wind flowing through our hair. The third way WIND was hidden in a mist involved its time-lagged returns. Because it would give competitors crucial information, many product developers do not want WIND to announce their design wins. So, WIND often waits for products to be launched before announcing only some of its major wins. Remember that a slow start for any diffusion is a characteristic of the technology adoption life cycle. To matter much, run-time royalties necessarily wait for the inflection in growth to begin. Thus, if product growth accelerates at all, WIND's run-time royalties lag both the time involved in the product design and implementation cycle and the additional time involved from introduction of the product to its breakout. Still, WIND seemed to promise a river of increasing return from run-time royalties as its thousands of design wins grew.
As an investor in WIND since the spring of '97, it always seemed that WRS would breakout at any moment, but waiting seemed like an endless summer. What happened that slowed its revenues in late FY99 into FY00, breaking its typical pattern of 40% growth? The founder of the "Wind, Up, Up, UP!" thread, the astute analyst Allen Benn, in posts #7668 and # 7674, offered an explanation based upon his analysis of data while serving as a consultant to WRS during 1999. After performing a time-lagged statistical analysis of especially gathered data, Benn's answer was that the decline in revenues had been a delayed effect of the Asian crisis that took two years to manifest itself. The market reacted radically because it did not know why growth in revenues slowed. That Benn enables us to know what the market did not is a signal of a market inefficiency that still may not be completely priced in. Accepting that the time-lagged effect was explanatory, he can also conclude that the ship is righted: design wins are escalating to be followed by a time-lagged resurgence in royalties.
(Continued in Part II. Differentiating Developer's Core Applications From WIND's Contextual Layers of Enabling Technology.) |
