>Unit pricing for i960RP in 10,000 100,000 1,000,000?
Don’t know, but WIND’s license fee is fixed per unit, unrelated to volume. At least that is the impression I got at the H&Q show when management answered follow-on questions about the fee. Buyers of i960RP chips do not pay WIND anything, since they buy directly from Intel. Intel pays WIND a fixed fee per unit shipped. In other words, buyers of i960 RP will not negotiate fees for IxWorks with WIND; although they may well deal directly with WIND to obtain a license for the IxWorks development toolset and for any engineering support or training needed.
>How much is WIND licensing IxWorks for?
Don’t know, but I think your guess of $1 is low. I would guess $3 would be too high. Lately I’ve been trying to hone my estimates of the size of the IxWorks market. I2O is an Intel invention that solves a problem with the PentiumPro processor, and will enable the Wintel machine to take over the server world, including all mission critical Unix and IBM mid-range and mainframe computers. Since I2O is not OS dependent, it may or may not be picked up by Apple and Unix as a defensive move, but it really doesn’t matter. Starting in the Q4 of 1996, I2O cards will be available targeting existing PentiumPro servers, but will be applicable to Pentium servers as well. In Q1 1997, virtually all PentiumPro servers will sport I2O on the motherboard. Later in 1997, when the PentiumPro begins to be used on the client side, I2O will go with it on the motherboard. At this point the i960RP volume will start to mushroom. By 1998 a significant percent of Wintel motherboards (most but not all made by Intel) will house the i960RP, resulting in huge license revenues. Meanwhile, peripheral devices will begin to be produced using I2O, and conceivably many data-restricted devices such as set-top boxes and cable modems. In time there will be other I2O processors besides the i960RP, including integrated chips with more than just I2O. IxWorks should be the OS of choice for those chips.
>% of current WIND revenue is licenses? My guess if 50%
I believe from what has been implied during past conference calls, that license revenue is about 25 to 30 percent of total product sales, making license fees less than 20% of total sales. If so, I would still conservatively project license fees to comprise more than 40% of total sales by the turn of the Century. In other words, even while total revenues, including sales of toolsets and engineering services continue to grow nicely, license fees should grow much faster, adding significantly to the bottom line.
While the run-time license fee aspect of WIND’s business model is exciting and we need to know about it, other parts of the business also offer opportunities for attractive growth. The most obvious of these is the toolset business. Less obvious are the following:
1. Contract Engineering. Chip makers pay WIND to port their RTOS and tools to their chips - in order to make the chip attractive to the world of VxWorks developers. As WIND gains market share, this business becomes increasingly profitable (as remarked in the last analyst conference call). Also, due to the plethora of new processors and flavors or processors, the amount of porting is expanding rapidly.
2. Engineering Services. As companies are forced by market pressures (time to market, cost of development, increased functionality) to convert to commercial tools and RTOS, they need help getting up to speed. Just as Oracle Consultants has turned into a large service business helping organizations implement client-server systems based on the Oracle RDBMS, WIND could become sizable just selling their expertise as a service to embedded systems developers.
3. Training. Every organization choosing to get on the band wagon of commercial embedded systems development needs to get their development teams trained. This mounts up and, with volume, can be very profitable.
4. Manuals and Documentation. Even something as non-high-tech as selling extra manuals and documentation can become sizable and profitable.
All these growing business opportunities is why run-time license fees are less than 50% of revenues, and probably will be always.
>% market share of 32-bit RTOS shipments that WIND currently has? My guess 15%
First separate commercial from in-house shipments, i.e., Company X avoids royalites and insists on developing and using its own RTOS that "doesn’t have needless frills". Commercial RTOS’s probably account for less than one-half of all RTOS’s shipped today. By the turn of the Century commercial RTOS’s should account for 70 to 80% of all RTOS’s shipped.
WIND’s market share of commercial RTOS’s is probably around 25%, but it depends on what market share we are talking about. WIND’s share of units shipped is probably less than its share of license fees, since it is less dominate for low-end than for high-end products. For example, I asked an executive at Dialogic about their use of VxWorks. He replied that they still used VRTX for their lower line, but that they used VxWorks exclusively for the high-end of their products.
The most important issue here is how WIND is doing within each market segment. My guess is they dominate the military segment, the high-end commercial telecomm/datacom segment, and they own the I2O segment. They probably are the leader in the office products segment (nice to be used extensively by HP). I believe MWAR dominates the digital TV segment (which in time will pay off big, but, like automobile internals, is slow to develop fully), but not without serious competition from both WIND and INTS. I believe INTS is making headway in home commodity appliances and point-of-sale systems. MWAR is making an additional play for the hand-held wireless market segment, but there will be mighty competition not only from WIND (see post #141 re QCOM and ARM press releases) and INTS (see ARM press releases), but from Apple, Microsoft, Geoworks and General Magic among others.
These are my guesses. I would be interested in any source that can quantify any of these numbers.
Allen, |
