Ronald, you make very good points, but there is so much more to say.
WIND has just demonstrated another great quarter - beating estimates, etc. Doesn't matter. Given the chilly reception that WIND now enjoys from Wallstreet, maybe it will take an uncharacteristic blowout quarter to get any positive attention - however, such a blowout quarter didn't help CHKPF any (revs/profit up 100%+,etc).
Meeting this quarter's estimates, or next's, is indeed currently irrelevant to the market, because of the Microsoft threat. You are correct that not even a blowout quarter would have helped. (The market is concerned about the future not the past.)
It will be a long while before WIND recovers IMO. MSFT succeeded in freezing this segment until they intro their new OS in 18 months. It will be awhile after that when the dust clears.
This is where you are way too pessimistic. Any number of things can change the market's view virtually overnight. But first, let's understand a little more about what went wrong with the stock.
WIND tracked the Technology Index (e.g. MSH) almost perfectly over the last two years, although with a high Beta exaggerating the index's moves. That is, WIND tracked the index until January of this year when it diverged uncharacteristically and never got back in line. January was when H&Q indicated the I2O production ramp up would be delayed six to nine months, and had nothing to do with the CE announcement. As the stock began to recover from that blow a month or so later, it got hit with the Microsoft CE threat causing the disheartening results we all suffered.
As we now know, I2O is ramping production with OSM availability across the Novell, SCO Unix and Windows NT 4.0 operating systems (provided by the OS vendors, not external work-around patches). For example, Novell announced full availability across current versions of Netware just last week on May 20th. That negative factor should now be a positive, although H&Q reiterated the delay scenario just prior to the CC and has yet to issue a correction. This means uninformed parts of the market are still operating under false premises about I2O.
As developers get more experience trying to develop embedded systems with NT/CE and analysts pick up on the facts, the Microsoft black cloud will dissipate surprisingly fast. Actually, the black clouds dissipate surprisingly fast in the market even without unearthing facts. Witness how the market reacted to the Asian Flu, which clearly was never cured.
Recently Richard Smith posted an interesting press release about Microsoft serving up NT piecemeal for embedded applications a la VenturCom. VenturCom offers products and services for adding needed things to Microsoft's general purpose operating systems, and taking away things that aren't needed. Apparently VenturCom is the leader in this weird, emerging field. For the moment, Microsoft is learning from them, even investing in them, along with Intel and other VCs. Later, one presumes Microsoft will buy them out or design them out, but then again maybe not.
I want everyone with any interest in embedded systems software to link to VenturCom's homepage, vci.com , and read every word. This probably is the best source for understanding what it takes to develop embedded NT/CE systems.
Here's some things you are likely to learn from VenturCom's homepage:
1. VenturCom makes a strong case against hybrid systems in which NT or CE is combined with another RTOS to add real-time features, a la Radisys' InTime or INTS pSOS deal with Microsoft. I agree with them. I have always believed a hybrid built around Windows NT/CE would be the ultimate kludge.
2. VenturCom offers an extension to NT/CE called RTX for adding real time features to what becomes a real-time subsystem (RTSS). Here are a couple of interesting points about RTX in question/answer mode:
Q. Does RTX 4.1 manage the so called "blue screen" situations?
A. Yes. VenturCom's Real-time HAL intercepts all NT STOP conditions (which lead to the "Blue Screen") and shutdown situations, while allowing the RTX real-time sub-system (RTSS) and its real-time processes to continue running. RTSS is also notified of the situation, and it can then take a number of different actions. Typically RTSS processes will receive the notification, sequence the equipment to a safe "idle" state, and then allow the system to be rebooted. Depending on the POST speed of BIOS, the system can be operational again in under 15 seconds.
Please God, don't let NT/CE/RTX control my car!
Q. What kind of debuggers does VenturCom recommend for use with RTX 4.1?
A. VenturCom typically uses Microsoft's WinDbg and NuMega's SoftICE, which are powerful source code, dynamic break point debuggers. Again, most initial debugging and testing is done as a Win32 process, using a whole range of development tools, since the same API and functionality is available to Win32 and RTSS processes. There is a need for final testing on the RTSS environment, where the above mentioned debuggers apply.
So, developers are attracted to the Win32 environment in order to use familiar tools and libraries costing a few hundred dollars, but then are suckered into having to switch environments and tools when the going gets tough. Ladies and gentlemen, this is not what one expects when developing with a Win32 API.
"But what does all this cost", says the developer who buys his supplies at CompUSA? Well that depends. An unsupported evaluation version of RTX 4.1 (NT or CE) costs a mere $4,900. Get it with two developer licenses and support (it is highly recommended that you subscribe to support, and I think you can see why) and the cost increases to a noticeable $33,000. Want the full ball of wax with up to 800 RTX run-time licenses and even including the Component Integrator? Well that's going to cost you a cool $103,000. And by the way, don't forget you still have to buy licenses for Windows NT/CE from Microsoft, which you may have noticed continue to increase in price as everything else in technology gets cheaper (probably to have enough money to give to Microsoft).
I suggest you take the $33,000 option and buy the Component Integrator and run-time licenses later when you need them. (When the project fails you will save yourself $70,000, which is not bad for a single decision.). But let's suppose the project doesn't fail and you seriously want to deploy the embedded application. To do this you have to shrink it to a mere 8 to 17 megabytes using the Component Integrator, which actually dis-integrates operating system code. That will cost you $14,500 with 90 days support, or $28,000 with support for one year. This may seem high, but it is actually cheap when you consider all the useless code it can toss out of NT or CE, even though there still will be megabytes in the customized "dis-integrated" Windows products.
Well, that's all you need to develop in Microsoft wonderland - except for DCX. Did I forget to mention DCX, which will be available for RTX soon? DCX is a virtual I/O thing that allows the customized, shrunken real-time system to get to I/O in a real-time way. DCX looks vaguely like a poor man's I2O without an IOP. In any case, I think we call all agree that it's necessary or it wouldn't exist. Anyway, I'm sure it won't cost anything like the $103,000 option.
Another of Ronald's observations:
MSFT penetration into WIND's core market will fail, but it will be atleast a year after MSFT intros before the market figures that out. In my estimation, it will take about 2 1/2 years for WIND to recover as long as they continue to execute as flawlessly as they have up to the present.
I don't know how long it will take for developers and analysts to see through the Microsoft charade, but it will not take 2 « years. I find it hard to believe it will take more than 2 « hours, but admittedly timing is my weakness.
Microsoft will be using much of VenturCom's tools and techniques to ready its operating systems for embedded applications, but that won't make things better or much cheaper. The problem is fundamental and can't be cured with all the money and talent in the world. The problem with Microsoft's operating systems is that they are general purpose BY DESIGN and therefore inappropriate for being the foundation of the next wave of embedded computing. Notice Microsoft's response to Windows CE inadequacies: They will ADD real-time features to the already bloated system, then engage the likes of VenturCom to help them dis-integrate instantiations. While this response may appear stupid, actually it is entirely consistent with Microsoft's business model, and therefore is the only response possible.
Allen |
