Last great question asked by Jim Privat:
>(3) If the market grows at the aggressive end of your table (100% plus), then I would suspect that more competitors will enter the market. Does your table assume WIND’s market share over time increasing, decreasing, staying the same?
Answer: Increasing, from 1/5 at the beginning to 1/3 at the end.
Once you realize that commercial RTOS’s will dominate the landscape in future, and the RTOS paradigm is not threatened by obsolescence (whether implemented in hardware or software) in our lifetime, then the remaining issue is market share. How will market share distribute in future, and will there be important new entrants?
This is very important, so we need to look carefully at what will happen.
First, look at it from the perspective of an embedded systems application developer or a chip maker planning a hardware/software processor e.g. I2O. What affects your choice of a RTOS vendor? I think three things, in the following order:
1. Which (adequate) RTOS is the market leader, with the greatest number of engineers conversant in its intricacies and the greatest number of platforms supported.
2. Which supports the most 3rd party libraries and tools for simulation, debugging, compilers, probes, etc.
3. Which are affordable?
To the embedded applications developer, the first two will determine the design win irrespective of cost as long as it is affordable. Affordable means that the per unit cost of the RTOS is below some percentage cutoff, such as 1 or 2%. This is even more the case with the chip maker. Unless the chip maker is responding to a special order, it is imperative that any tightly integrated RTOS be the most popular, and the most supported RTOS around. After all, the chip maker is in the business of selling product, not making a technology statement.
What this means is that a new entrant cannot gain market share from established players merely by cutting prices. It also means that this market will consolidate to a few winners-take-all. This is why the I2O is so very important. Intel endorsed WIND to such an extent that it tightly integrated its I2O chip offering with IxWorks, and announced that I2O developers (which is about everybody) can use Tornado to develop their I2O applications, which supports more tools than any other development environment.
What would it take for a new entrant to enter this market and dominate market share? For starters, you must have an adequate RTOS ported to about 20 processors. You must have a complete toolset of simulators, compilers, and debuggers (which is delicate beyond belief for real-time systems). You must provide a development environment open to other toolsets that may appeal to your would be customers. You must be able to make available in-house or through 3rd parties all sorts of functionality, including all networking protocols, graphics libraries, file systems including flash memory, GUI interfaces libraries, fuzzy control logic libraries, etc. You must obtain some sort of certification that your RTOS can be trusted. Why should I spend hundreds or even thousands of man-hours testing your RTOS, only to learn it isn’t robust? Finally, you must not have corporate interests that conflict with other embedded systems developments. For example, you can’t be Motorola, Intel, IBM, DEC, HP, Sun or any of the Japanese chip makers. While it is conceivable that you could be Microsoft; not if you insist on using the in-house developed Microsoft at Work. Sun’s JavaOS may make a spash among Internet clients, but it cannot take a dominate position in the RTOS space, nor can Sun. Recall that Sun came out with Network Extensible Windowing System (NEWS) before Xwindows became a Unix defacto standard, but it was soundly rejected by a consortium of anti-ATT and anti-Sun vendors under the rapidly-formed umbrella of OFS.
More likely, when the size and importance of this market becomes obvious it will be too late for new entrants. If a large company such as Microsoft or Oracle decides to enter the market, they will have to acquire the startup capability, not make it.
Given the implications of the I2O action by Intel, and the power of Tornado to incorporate 3rd party toolsets, WIND is the clear market leader. Given the winner-take-all economics governing the decision-making by embedded systems developers or chip makers, WIND seems destined to dominate market share.
I believe MWAR and INTS may still have ample opportunity to excel by concentrating on market niches. By focusing on functionality specific to market niches they can deflect WIND’s overall dominance and maintain a healthy, vibrant business - at least for the foreseeable future. In particular, MWAR’s DAVID seems to be the defacto standard in digital TV, and working with Systemsoft, and partnering with Motorola they should strike gold in the wild and wireless hills of portable devices. I believe INTS dominates the point of sale and automatic teller machine market, and is targeting commodity home appliances.
I invite comment on respective market share. For example, does anyone have access to latest market share statistics produced by surveyors?
Allen, |
