This is the problem OpenSolaris is facing today in the nutshell: it has no breadth. It has a very limited number of excellent technologies, such as ZFS. They are very interesting, but they confer no advantage outside of very narrow niches, such as the storage appliance. To get the breadth, you have to have thousands of simple humans like Pete Zaitcev. Your OS can't survive with just Jeff Bonwick and Val Henson (who has quit Sun). I'm aware of the productivity gap according to Brooks and Paul Graham, but the problem is, while Jeff is 10 or 100 times more productive than I am, he is not 1000 times more productive. Plus, we have Stephen Tweedie.
So, what to do? Sun alone cannot hire as many programmers as Microsoft or Oracle. So they have to grow a "community", and it has to be "vibrant" and "large", of which OpenSolaris' community is neither currently. It may be "high signal-to-noise", which is easy when you're small. Just ask NetBSD people.
Life is not as simple, of course. For one thing, I ignored the OpenOffice here for simplicity, but it's an important application running on Solaris. And then there's Java where Sun exhibits a passable stewardship.
BTW, since we're on it, isn't ZFS patented? OpenSolaris may be open source, but you can't just throw it into your storage appliance and ship it without paying Sun. With Linux and XFS (or ext4), you can. This is something Jeff neglected to mention on his blog as if it's not an issue at all. It would be nice if he reassured prospective appliance manufacturers in this regard.
UPDATE: Nick Istre reminds me about the patent grant in comments. How silly of me.