Abstract
In the first quarter of 2005 Intel released its new 4-way Xeon MP processor, an event followed by HP's decision to not adopt 8-way Xeon CPU. IBM will be rolling out a new X3 chipset for the Xeon CPU in the second quarter, designed to support 32-way and higher servers. The release of Intel's new offering is provoking different responses from major server brands -- responses that connote the differing strategies and product positioning of the various players.
The progressive withering in market growth for mid-range and high-end servers is irreversible. Yet with cut-throat price competition in the entry-level space, the leading players are unwilling to let go of the healthy profits in the mid to high-end range. However, considering the immensely higher resources needed for such systems, the number of players willing and able to take such a route can be counted on half a hand.
Judging from the trajectory of mid-range and high-end servers described above, the development of x86 systems will be hindered in the upper segments. Lower customer faith and a dearth of software will hem in market scale, leaving enough space for only a few players. Especially with the muscle exerted by IBM, other makers will have a tough time gaining entry.
In terms of R&D strategy, the circuitous path that clustering technology has opened to the upper market segments is still the best way to proceed. Linux developers are currently pushing cluster technology and simplifying the design needs of system makers. As these advances help wear down entry barriers, second tier players or other makers with scant R&D resources will have no choice but to adopt clustering. Mid-range and high-end servers, however, will still remain the preserve the heavyweight incumbents.