I've been working with virtual machine technologies a lot lately at a start-up called Leostream. Leostream's product is a management layer that controls a network of VMs. If you're only experience with VMs is running Windows on your Linux box, then consider the advantages VMs have for server consolidation and application training. Server consolidation is all about saving co-lo costs by reducing the number of physical machines needed to run your applications. Buy some IBM Big Iron and run several VMs on it with whatever operating systems are required. You've just saved a boat-load of cash. Trainers that need workstations with various flavors of Windows can have one classroom with machines that have several VMs on them. Switching between the VMs and the maintenance of those VMs is a snap. In short, VMs are really cool. But this is just the tip of the iceberg.
Microsoft has yet to figure out an effective way to embrace and extend open source software. Imagine if Microsoft shipped a FreeBSD VM with their operating system. Aside from licensing the VM technology, it would cost nothing. They don't have to contribute any code to BSD, but now Windows customers have access to all the same software the Open Source community has developed. Take THAT Gnu. In this way, Microsoft out-Apples Apple. Zero-developement costs to integrate BSD into Windows and all the backward compatibility Windows customers demand. Heck, customers can run their old Windows 3.1 apps in a VM if they want to.
There are many other awesome implications of VM technology, like the death of Java. Why bother with the goofy JVM when you can program for a real architecture (like Intel) with all the well-developed tools already in existance and ship that VM?
Perhaps the death of the PC has been greatly exaggerated after all.
UPDATE: Wow! Microsoft is going to acquire Connectix, a VM product for Windows. It looks like someone in Redmond reads my blog!