I still remember something someone from IBM told me during my prior analyst gig. Don't remember the specific but something to do with Linux and Power I think. Basically, he said that if it's not a billion dollar opportunity it's not worth all the costs and distractions up and down the line to pursue.
You're right that the million dollar opportunity is a lot more interesting for a startup. Of course, it's also going to be a lot harder for that startup to displace VMware (which the potential customer probably is using). I'm not sure going after those crumbs is a terrible idea but there are already other option like Xen too.
