By webmink on Apr 23, 2005
I see that IBM will treat dual core CPUs as one chip but I agree with Tim about his one. As I have been saying in my keynotes for about a year, pricing that's based on the ephemera of implementation is senseless in a massively-connected age when the network is the computer. Who cares how you achieve your results? What matters is that you can deliver those results.
Pricing in a massively-connected era can't be based on licenses to use particular bits of software, on how much memory you've installed, on the kind of CPUs you've selected or indeed on any arbitrary ephemera, because increasingly there will be multiple equivalent ways to achieve the same results - from self-built Unix installations all the way to rented time on utility computing services. It makes no more sense to charge per-CPU than it does to charge per-memory chip. it's like the electricity company charging for electricity based not on what you use but rather on how many filaments you have in your light bulbs (and woe betide if you install fluorescent!)
We're in a transitional phase, and IBM's move is a step forward (come on, Oracle, everyone is waiting), but it's time end users started pressuring suppliers to move their business models into the massively-connected 21st century, before open source and innovative silicon wipe them out.