Franz -
All good questions - let me see if I can come up with some reasonable answers. I had a discussion just a few days ago on this issue with one of CPQ's technical ladder guys, a 'Compaq Fellow', and he covered most of these points.
1) If we are going to keep Alpha it means that we have decided to become a chipmaker (who just sold its fab plant).
That's the good news - the design team fits with CPQ's model, the fab does not. Do you know that CPQ is already a substantial chipmaker? they design special chips for memory management, ethernet adapters, and integrated functions for the consumer products. Two key chip designs that CPQ owns are the ServerNet clustering technology chip, which is the basis for the CPQ / Intel / MSFT VI Architecture specification, and the SMART Array controller chip. CPQ has been in the chip design business for many years and it has always paid off well in both competitive advantage and lower cost.
2) We have only one chip to sell. To be a chipmaker you have to carry the staff (designers, etc) that a chipmaker has - all for one chip.
See discussion above, CPQ has maintained a large silicon design team for more than 10 years. Also CPQ DOES sell chips to others - a lot of them. I can't go into details but CPQ sells large numbers of chips to HP and IBM among others.
3) You need to continually come up with R&D improvements, advancements, etc.
That's how CPQ can achieve and maintain a competitive edge. CPQ Array controllers are the result of great R&D work and they hold a number of key patents in this area. Their patented logic and advanced PCI designs gave them about a 30% advantage in performance over such rivals as Mylex - this is how CPQ was able to achieve the number 2 overall sales position, with only IBM selling more, and ahead of such well-known names as EMC. CPQ's storage business alone will probably exceed 6 billion, and that would not be possible without all of their silicon design capability.
4) Your relationship with other chipmakers becomes a little more strained if you are in direct competition ("Let's favor DELL more than CPQ in prices, etc, because CPQ is a direct threat to us.").
Not true about prices - that's illegal and even Intel would not take that chance. It is true that Intel may be more reluctant to share design information with CPQ, but they have always had a difficult relationship for all of the reasons stated above, plus CPQ's readiness to use the best technology for a given product regardless of whether it comes from Intel. As I mentioned above, CPQ uses their own silicon for support chipsets, not Intel's, and this has always created friction. But it also gives CPQ a competitive edge.
5) With announcements like the above, you can see that Intel is closing the gap even though Merced is delayed.
I don't think there is any truth to the idea that Intel is closing the gap. These are all 32 bit processors, and CPQ will use them to make great 32 bit products. Look at today's Xeon announcements - CPQ has killer server products using Xeon and CPQ's own high speed I/O designs. The fact is that Intel NEEDS CPQ to help develop the 64 bit market, even more so now that Merced is delayed. Alpha products over the next 2 years will make speed advances faster than the Intel products according to process roadmaps for both products. Intel themselves will help to make this happen.
6) There is a whole army of other chipmakers out there that have a whole line of chips.
Let's see, there's Intel, and Alpha, and.... maybe PowerPC? I think that the MAC market does not quite count for CPQ, and the only other major product using PPC is IBM's RS6000 UNIX machines. MIPS has a date with Dr. Kevorkian sometime next year. AMD will also be producing Alpha, and their new Intel-architecture chips will use the Alpha socket and memory scheme rather than Intel's socket-1. Who else is in the game?
But more importantly, I think there is one opinion that counts more than any other as far as what chips are important for mass marketing, and that is MSFT, since they represent 90% of the volume market. As MSFT shifts its customer base to NT (virtually assured since they have stopped development on WinXX), there will be a huge ongoing market for Alpha products. Unlike Dell, CPQ can also develop a strong Unix play, moving into the multi-billion-dollar market currently owned by SUN, HP, IBM and a few others.
7) DELL is now offering 400Mz units cheap and so are others -with faster chips to follow soon.
So is CPQ. CPQ will always have competitive products in the Intel product line.
But according to MSFT, Alpha is 50% faster than Intel at the same clock speed, and there are 500MHZ and 600MHZ Alpha chips now, with 800MHZ and 1 GHZ chips soon to follow. That means for critical workstation and server applications, Alpha will have nearly a 2X speed advantage over the Intel products. Don't you think that products with that kind of advantage, from a company with the market size of CPQ, will interest corporate customers?
8) Anybody that buys a Alpha can only use it for NT/UNIX stations (a very limited market).
Hmmm, limited to about 90% of all of the operating systems in use today, on a unit basis. Would you like to see a port to MVS or the MAC just to get that other 10%?
I hope this clears up the Alpha issues. I am satisfied that CPQ is looking at this as a business decision, they are pretty hard-nosed about costs and they seem to have looked at this in depth.
|
