Things are a bit quite, following the big ESC in San Jose during mid-September, and my recent travels. I have been trying to digest all the news from the show, the INTS quarterly report, and both INTS and MWAR announcements about CFO resignations. The following are quick responses to many questions asked lately, as well as on-going concerns that I am pondering.
1. I am waiting for a response from some queries I made before commenting further on Mark Murphy's review of the Embedded Systems sector. I did want to alert everyone to Mark's useful review, but hopefully after I have confirmation regarding one of his points. Incidentally, I don't know enough about Mark Murphy to recommend that his advice should be followed, but I find his newsletter interesting. I stumbled across mention of the California Technology Stock Letter in Hulbert's newsletter review years ago, and was impressed by its performance. I subscribe because the newsletter follows one of my companies, and comments generally intelligently about much of my universe of stocks. I only have one serious problem with the newsletter, which generally is the same one I have with all newsletters: lack of detail. Since Mark must perforce follow numerous stocks, his analysis must be limited, and so he relies on a custom variation of the trailing PE ratio, called growth flow, among other measures. Neat, but not good enough for serious investing. (See below.)
2. I keep rethinking the implications of the NC, intrigued by Oracle's serious challenge to Microsoft for supremacy of the PC paradigm. (The Netscape war is a red-herring, or at most a warm-up for the coming battle royal!) Once Oracle's looming dominance of the NC alternative to the PC is understood, and why, then the awesome implications for WIND become easy to see. (Hint: I do not think Oracle has any interest in manufacturing NCs, competing with the likes of Compaq, IBM, and commodity Japanese electronic firms.) Incidentally, I have not yet added projected NC run-time licenses to my WIND model, and probably will not until more information becomes available on size and timing.
3. While Microsoft absolutely abhors the notion of the NC, Intel is more ambivalent. In a way, Intel would like it to go away, since the company is busily beginning to assault the remaining ramparts of the current computer paradigms that it does not dominate: servers and high-end workstations. Why should it have to turn attention back to the PC, which it already owns, and defend turf? The reason is that if the NC only approximates some pundits claims, Intel must participate. Consequently, in Andy's paranoia, he has set a separate shop within Intel to create the world's best NC. But knowing that price will be the determining factor, Intel will have to rely on excellent design and cost-effective production, and lots and lots of volume to equal monopolistic PC processor profits. Even though Intel does not relish the distraction and possible threat to monopolistic pricing, Intel must recognize that the growth potential in numbers for the NC is at least an order of magnitude greater than the PC, with dollar potential being at least of the same order of magnitude. Neither servers nor workstations offer Intel that kind of growth potential.
4. Keep an eye out for I2O-based products. For example, I think the Compaq's newly announced ProLiant 2500 uses I2O. (I believe it is used in a new version of their Smart-2 Array Controller for arrays of disks. The ProLiant 2500 can have up to 6 such controllers. So far it appears to me that other "smart" adapters would require additional I2O chips.) Read Compaq's white paper on I2O to confirm the importance to software and hardware developers alike of isolating and abstracting each level of I/O. Intelligent I/O needs to be standardized, a la I2O, for advanced data handling to be automatically compatible across products. While this certainly is encouraging, and Compaq's ProLiant 2500 announcement should help speed add-in product makers to conform to I2O, there remains intriguing questions about when the compelling arguments in favor of I2O will first bring it to the desk-top. If the answer is sooner rather than later, then I2O run-time license revenues to WIND will increase even more rapidly than I currently project. Even bigger, I still sense that I2O may prove critical to a full commodity deployment of the NC; albeit not necessarily as a separate on-board processor but instead, for reasons of cost, integrated into a multi-functional processor chip. It seems to me that commodity deployment of the NC will place responsibility for I/O compatibility and capability firmly on the back of the server, which needs the easiest way possible to determine and update client I/O processing. That would dictate a very high level of abstraction, and as much as possible of the particulars, e.g. protocol handling, to be implemented in software. What better than I2O? If this proves true, then with WIND becoming dominate in the coming NC operating system (thru Oracle) and already owning the I2O RTOS space, future run-time license revenues from I2O and NC would be gigantic.
5. On Oct 25 MWAR reports its latest quarter, which needs to be closely examined. This will settle the markets lingering concern about the recent resignation of the CFO. Also, the recent announcement of a Java deal with Correl is intriguing. Everyone knows Correl is reprogramming its desk-top tools as Java applets. But why the connection to MWAR, unless Correl is just beginning a strategy of connecting to everybody and anybody that will be deploying a Java interpreter with user interaction? It seems to me that the NC is the starting point for Correl applets, not wireless hand-held devices or internet appliances.
6. Microsoft's newly announced Windows CE probably will extend the PC paradigm to some hand-held devices, but it is not a threat to the NC or WIND. I cannot see any reason why Windows CE should become the OS of choice for much beyond a few high-end, hand-held PCs. However, wouldn't you be worried if you were Geoworks? As for Microsofts stealth development of a true embedded RTOS, they have been at this for years, starting with Microsoft At Work, with nothing to show for it. I believe the culture clash is too great for Microsoft to succeed internally in the embedded RTOS world. If Microsoft abhores the NC, how do you think they would like a world without a Windows interface? Only when you can look beyond Windows, and its ever-more-complicated extensions on the drawing board, can you envision the next wave of hidden computers. If Bill Gates were Andy Grove, he would (1) write a check and buy WIND, (2) keep it separate and invest heavily for years, (3) bask in the delight of continuing to dominate the second wave while also preparing to dominate the third and largest wave of ubiquitous computers. If he did it immediately, Justice probably would not stop him, like it did with Intuit, because Embedded Systems is not yet consolidated enough to scare the trust busters.
7. On questions of investment strategy, I have concluded that Technical Analysis, Classical Value Analysis (making any judgments at all on the basis of existing, i.e. trailing, numbers such as revenues, earnings, PE, ROE, etc.), and Asset Allocation are all too costly in time and effort for too little benefit to be useful to the long-term investor. The serious long-term investor should abstain from all these much-heralded techniques, favoring a common sense approach of getting deeply involved in a few select companies, and understanding those businesses profoundly. This includes doing the numbers on projected sales and earnings, which is concerned with value, but looking to the future not the past. It is interesting that Asset Allocation is the most difficult of all the techniques to dismiss, because the underlying theory is compelling, essentially following step by step from reasonable and acceptable axioms governing rational behavior of man (investor). Rejection of Asset Allocation is justified because the very statistics that are used to define it are its ultimate undoing. Since diversification virtually guarantees that your portfolio will achieve results governed by statistical outcome, there is the unfortunate downside that, even for periods of ten or twenty years, it is much too likely that your return will be greatly disappointing. While most financial advisors assume that ten or twenty years is sufficient to bring about expected results, they are wrong. If you rely on statistics of the stock market, you should be prepared to suffer by those same statistics, and the chances of suffering are significant. Thus, the very process of minimizing risk through asset allocation serves to create an unbearable risk of under-performing. (Just to keep things confusing, in arguing against Asset Allocation, I am the first to concede that the rational investor almost always should diversify an investment between two equally attractive possibilities. Also, the argument I made a couple months ago justifying Jerry Fiddlers selling a chunk of WIND shares still stands.)
I expect to be able to flesh out a few of these items in future posts.
Allen |
