RISC Strategies for Coping With IA64
There are three basic approaches RISC vendors are taking to deal with the onslaught of IA64 processors and Intel’s standard high volume, multiple system vendor business model. The first approach can best be described as “if you can’t beat them join them”. HP and SGI have adopted this path of least resistance and the once proud and competitive PA-RISC and MIPS architectures are now sustained at a subsistence levels to retain their customer base during the slow transition to IA64. Besides leaving both companies vulnerable to any slips and missteps by Intel along the way, the disruptions inherent to moving an installed base across architectures tends to cause even loyal customers to seriously consider moving to alternative vendors. These factors may be partially responsible for the decline in fortunes for SGI and more recently HP.
The second approach might be termed “what me worry?”. This is the strategy Sun Microsystems appears to be taking. Historically speaking, Sun has an excellent track record of outselling RISC rivals offering technically superior products. This success is the product of better marketing, strong software base, consistent product direction, and occasionally, the sheer ineptitude of its competitors. Sun management seems to view Intel and IA64 as just another competitor with a faster 64-bit chip. But IA64 is different in two main respects. Intel uses a chip merchant business model, and the fact Intel can market to corporate and information technology decision makers at least as well as Sun. An important indicator of the success or failure of Sun’s steady-as-she-goes strategy will be the number of applications previously exclusive to SPARC that get ported to IA64.
The third strategy RISC vendors can take to deal with IA64 can best be described in military terms as a withdrawal under fire to better defended positions. Both IBM and Compaq seem to moving their respective RISC architectures upscale into the realm of extremely high-end processor implementations such as POWER4 and EV7/8. The sales volume for such expensive processors is potentially so small as to not be economically viable for a chip merchant like Intel to pursue. But IBM and Compaq can afford to build these processors because they sell high margin, large-scale computer systems based on these devices, and thus can make the business case at the system, rather than component level. That is something Intel can’t do unless it goes into competition with its customers. The success of this strategy depends on the size of the “air pocket” that Intel leaves. That is, the size of the market for high end computers too powerful for competitors to duplicate using Intel’s standard high volume IA64 components, yet inexpensive enough to attract customers. It might seem unlikely that Intel will leave enough room to support multiple competing RISC families, but for companies like IBM and Compaq, high end servers have a drag along effect on sales of high margin ancillary products and services, such as storage and consulting. In this way, ongoing development costs for high-end RISC processors can likely be sustained by a surprisingly small share of the 64-bit MPU market by unit volume.
Discuss (15 comments)