If you caught my previous posting Who Is Really Acquiring Sun?, you will have concluded either that I’m missing some detail of the situation or that IBM is going to walk away with the high-end Unix market, while both Oracle/Sun and HP fight a profitless rearguard action. The IBM Power chip has treated its competitors the way that Jet Li treats his assailants in a martial arts movie, when they believe, quite incorrectly, that they have him cornered. There are bruises and broken bones, and only Jet Li walks away from the incident unscathed.
I-and-I
As far as I can tell, the high-end Unix market contest is over. SPARC will never recover and Itanium can only be classified as the walking wounded. It is never going to compete with Power. Consequently, another equally interesting battle is about to commence, which can be thought of as the contest of I-and-I: IBM and Intel. The outcome of this one is much less certain and there isn’t necessarily going to be a winner because, as in the game of chess, the rules of engagement admit the possibility of a draw.
Let’s paint the picture so that you can see what’s happening out there…
The Battle in Device land
Intel is the 500 pound gorilla in the chip market, but it is not in a completely comfortable position. You can divide Intel’s market between client devices and server devices. Once upon a time client computers (PCs) and server computers (servers with Intel chips) were almost identical. If you wanted a server, then you just bought a PC with a cheap monitor and that was your server. It was excellent value too. Cheap, cheap, cheap.
There was sufficient similarity between PCs and servers for Wintel (Intel and Microsoft) to capture a big percentage of the server market, simply by the force of numbers. From an engineering perspective, it was never a good idea for client and server to be the same kind of unit, but the numbers swamped that engineering nicety – and the commodity Intel server was born (running either Windows or Linux, according to taste.) Engineering sanity was simply swamped by the tsunami of cheapness.
That age has passed. With the monotonous persistence of Moore’s Law, the PC gradually became an engineering joke. It may have had a hugely powerful processor, but the processor rarely did anything – unless you ran some kind of truly heavy graphical workload. The PC chip was a jumbo jet engine strapped to the back of a bicycle.
The PC workloads that could be improved by processor power were, almost exclusively, the graphical workloads. So now, what really matters on client devices is graphical processing – the graphics card and the GPU. This reality has pushed NVIDIA into competition with Intel on the desktop. To add to that, AMD’s purchase of ATI (providing it with a little skin in this game too) has kept it relevant despite, it’s decline in the face of an Intel onslaught. There is more to be said about this, but I’ll save it for another article.
Mobile devices, many of them powered by ARM chips, are rising up, and just to complicate matters, IBM dominates the computer gaming market, providing the chips for Nintendo, Sony and Microsoft – although not identical chips. Intel still rules “the client” but, even though it finally managed to draw Apple into its orbit, it cannot claim dominance. Against the gain of Apple (which previously used a version of IBM’s power chip) it lost the games console market and it has a very weak position in the mobile phone market.
The point in mentioning the competition in this key market is to note the fact that Intel is in no position to employ the joyous tactic of “dumping” in the server market – i.e. using guaranteed profits in the client market in order to dump product in the server market at below cost, to destroy the competition.
Click to continue reading “The Coming Contest Between IBM and Intel”
























