It is appropriate to gather as many facts, or even rumors, as a means to increase understanding of a sector of business and the competing companies. When asked why Warren Buffett felt he knew enough about the newspaper business to invest in the Washington Post, he reportedly indicated that investigative reporting is the primary skill of the successful investor, suggesting that he held a deep appreciation and understanding of the newspaper business.
However, gathering the facts is only the first step of investigative reporting. Analyzing them correctly is an even more important second step. David, you do yourself and us a service when you bring relevant observations to everyone's attention on this thread. You would do yourself more service if you thought deeper about these facts, arriving at more appropriate conclusions, rather than always trying to imply something apocryphal about WIND.
Keep in mind, we are investors, not advocates. This is not an intellectual debate. If I out-debate you, that doesn't make me the winner if in fact you were right all along. The scoring mechanism is stock price, not wit.
>>You are probably right - there will not be a single direct hit on the Wind River franchise. It might be that WIND faces a fight with the possibility of bleeding to death from a thousand small cuts. You must agree that this "embedded" business is certainly getting more crowded.<<
Embedded systems is gaining recognition, and no doubt is attracting new players at every level: operating system, development tools, compilers, debuggers, microprocessors, procedural and non-procedural languages, and on and on. A couple of years ago, Java was completely unknown, now it is all the rave. Windows CE is targeting the sector. Geoworks can point to smart telephone/PDAs being shipped in production quantity by Nokia, and recently announced a developers support program to enable ISD's to develop applications that will run on Nokia's smart phone. Without any doubt, Microsoft envisions an emerging 3rd party support program for Windows CE, wanting to make that OS the industry standard for hand-held devices. Inexpensive development environments exist for Windows CE and also for Java, at prices much less than Tornado. AT&T introduces Inferno and the Limbo language for embedded systems development and deployment.
These are the facts, but what do they mean? As you no doubt realize by now, exactly the opposite from what you concluded. Let's see why.
1. All the new players and embedded systems development products do signify that the sector is on the threshold of very rapid advancement. Roll-your-own operating systems and self-made solutions will be replaced using commercial off-the-shelf (COTS) technology. Even our government has instituted this as formal policy in the military, NASA and the FAA, the benefits of which already have been mind-boggling. The Mars Pathfinder is heading for Mars using COTS to trim costs to a tiny fraction of prior costs of similar missions - not coincidentally with VxWorks controlling both the spaceship and the rover vehicle to be used for terrestrial exploration.
2. At the same time, the number of embedded systems that can afford to deploy complex logic needing COTS is exploding due to the availability of inexpensive embedded processors capable of handling all the complications. The number of such high-level microprocessors being deployed in products is doubling at least every 18 months today, and is just beginning to accelerate.
3. These two facts mean that many organizations are pressed to the wall to get their new products developed and out-to-market - ASAP. They know their competition is hard at work, and they are beginning to realize that they use commercial RTOS and powerful development tools to get the job done. That means they must do the same or better. Incidentally, the cost of development tools for these companies - at this stage of decision-making - does not even enter the picture. If they think they need Tornado to develop a new series of printers with a total life-time revenue projection of $1 billion, the cost of the setup can be ignored - totally.
4. In the not-so-distant-past, companies trying this approach simply flirted with it in R&D shops, so the RTOS and tools were selected by the project manager, guided and persuaded by a few lead technicians. This meant that WIND, INTS, MWAR and others rarely even competed against each other for design wins. Their competition was the in-house group of technicians devoted to traditional, invented-here ways to develop products.
5. Today it is different. As companies experience the difficulties of shedding their old ways, not experimenting anymore, but changing, they also realize the importance of forming a replacement strategy that will obviate the necessity of making similar changes any time soon. In company life, this means "The Selection Committee". The Selection Committee evaluates all suggested approaches, and picks the best one, thereby committing the company to years of working with the chosen vendor, good or bad. Now, for the first time, all the RTOS vendors, as well as the JavaOS, Inferno, the Win32 crowd and others such as Geos go after these design wins - knowing that they are likely to lead to a strategic, lasting relationship with the company.
6. Unfortunately for the thousand wounds theory, the result is the exact opposite from what you think. Committees like industrial strength solutions, if only because they make up lengthy lists of mandated functionality followed by even longer lists of preferred functionality. For example, hard real-time may not be required for the immediate project at hand; but it is virtually inconceivable that The Selection Committee would omit hard real-time from its list of required functionality. Notice that this single, often unnecessary requirement just torpedoed pure solutions from both Windows CE and JavaOS. Peer group recommendations pare down surviving vendors (often there is only one). (Actually you could argue that recommendations from acquaintances in other organizations preceded The Selection Committee, and provided not only the beginning list or requirements but the name of the vendor to beat.)
7. Less established vendors with incomplete and/or non-standard tools use the "veto feature" approach to try to win these battles. They befriend a participant, or a technical advisor to a participant, with a convincing argument about the necessary of a particular, proprietary feature. They hope to get the feature on the mandatory list, and get "their" people to demand the feature in order to veto all competitors. They also try to get features they don't support off the mandatory list, by arguing their inappropriateness. Occasionally this strategy succeeds, but usually it only engenders ill-will. The market leader always is the vendor likely to be selected by The Selection Committee, and will always be accepted by management. Incidentally, the market leader must always be somewhat evasive about adding missing features that cannot be scratched from the mandatory list. After all, it is not usually necessary that all features be demonstrable.
8. The evidence has sprung up over the last year is that WIND is clearly established as the market leader, based on Tornado, a feature-rich product without peer. I am not talking about whether you think Tornado is bug free, or appropriate for exceptional situations, or even appropriate for any project. The simple fact is that nothing covers the waterfront of development features like Tornado and its support cast of a hundred, or so, third party developers.
9. Organizations will now circle their wagons around the market leader, WIND, to protect it from the thousand knives you referred to previously. The reason is that once they commit to the market leader, it is in their interest to preserve the stature of the leader to guarantee they get the support, vendor-trained engineers off the street, continued upgrades, access to compatible third-party products, ports to new processors, and job safety that derives from being in the main stream. So, the thousand knives you view as such a threat are the very things that give solid foundation to the market leader.
A few years ago, just after Oracle recovered from its hiccup after dominating the relational database business, I was surveyed by the Janus Fund about what I thought of Oracle's future. I had bet right on Oracle, knowing the power of market dominance in a software sector, but in that interview I expressed fear that Microsoft, with desktop databases, and mainly Borland with Paradox, dBase and Interbase (which some argued was the equal of Oracle) might do to Oracle what the PC did to dedicated word-processors. You can see the result. No desktop database, however capable, can get through The Selection Committee for corporate databases. Once ensconced, no database can root out a corporate-wide implementation of Oracle 7. It is even true that Oracle 7 hasn't even eliminated the old standby COBOL and ISAM. Consequently, Oracle is the number 2 software company in the world, and Borland is barely hanging on.
Now, put it all together and you see why it has been so exciting, and profitable, to invest and enjoy watching the embedded systems sector spring into prominence under WIND's leadership.
Allen |
