Jeff Atwood of Coding Horror has come under attack for saying that most people won't derive any benefit from having more than two cores in their desktop. In particular, he claims that they're of no benefit to developers.
Now, this seems to be an absurd claim. Ever compiled a C++ project? It takes a long time, and it is very easily parallelisable. Atwood is a VB.NET programmer (they still exist?) primarily, so this isn't something he's as likely to see; apparently the current managed .NET compilers do not parallelise well.
Personally, I don't have much use for seven million cores on my desktop. I'm writing this on a Macbook Pro, running off battery. Since the machine isn't doing much, the CPU will have wound itself down to about 800MHz. And that's just fine. It feels fast. If I was a C++ programmer, four cores or more might be nice, but I mostly do Erlang and Lisp these days.
I have a 4-core machine at work. It's nice to watch an Erlang system using all the CPUs, but that's about all the good it does me. It can be useful for getting an idea of how something will run on a multicore server, but actually running that something on a similar server would generally be better.
Of course, this isn't really the point. For some time, your average desktop machine has been faster than anyone except gamers, graphic designers and so on really need. If I was buying a desktop I'd be reluctant to spend much extra to get four cores, but I probably wouldn't need to; the price difference tends to be minuscule. Where multicore chips really come into their own is on the server; an application server will generally scale linearly with number of processor, although a database server won't.
So yes, four cores on desktop; not so useful for most people, great for C++ programmers and graphics people, but so cheap that it hardly matters.