Historical reasons. Solaris was the definitive platform circa the late nineties. I used Linux back then, but it was definitely not ready for prime time yet. Many enterprises were built around Solaris. Today, Linux is a better alternative today for most use cases, but only marginally better. Until a year ago, it was not worth the cost of a switch.
Why a year ago? Oracle bought Sun. Sun was very good at creating open, useful platforms, but was very bad at making money off of them. Sun was bought by Oracle. Oracle isn't very good at building software, but is extremely good at selling it and maximizing profit from it. Oracle needs to make back the 7+ billion dollars it made on Sun somehow. How? We have part of the answer in the Google suit: by milking Sun's customer base.
Java has 10 million developers. To make 10 billion (there's a fair bit of risk in buying Sun -- if Oracle wants a 3x return, that's 10 billion each for Java, hardware, and Solaris), it needs to make $1000 per developer. How? I don't know how, but I'll run one possible scenario.
Step 1: Cut off competing JVM implementations with patent and copyright suits. This is already starting to happen.
Step 2: Release new versions of Java in two versions. The first is a free Java Express, which is okay for student use, but has no JVM, and omits many of the more enterprise-centric libraries. The second is a full-fledged version, and costs $1000 a pop. Developers would, of course, free to use the old, bit rotting version as well.
Most Java developers cannot switch platforms -- most are working on enterprise systems with hundreds of thousands or millions of dollars of sunk costs. Java immediately becomes a legacy platform, but even if only 50% of developers buy into it (even if only during the transition period), Oracle makes a cool $5 billion, almost paying for the rest of Sun.
Where does this leave Solaris? I don't know, but I suspect similar things are afoot. Oracle will want to make more money, and Oracle's customers will end up paying more. Oracle is already beginning to move in this direction with the death of OpenSolaris.
Aside from that, Oracle is incapable of attracting good
developers. Former Sun employees are leaving as quickly as they can find new jobs. Will Solaris continue to have a technological advantage in the long term without Sun's brainpower? I seriously doubt it.
I know this is a lot of FUD, but with good reason. I am afraid, uncertain, and doubtful about the future of Java and Solaris. While I'm certain that Oracle will continue to develop Solaris so long as it continues to make them money, I don't know what they'll charge for it, or how well they'll develop it. I really wouldn't want to base my business around it.
(A secondary, technological answer is that Solaris advantages are much better suited to the climate of the nineties than of the teens. In the nineties, we tend to rely more on big computers with many CPUs, massive amount of memory, and many hard drives. In the teens, the trend is much more towards big arrays of cheap, unreliable, redundant boxes. Solaris' advantages are much more pronounced for the former than the latter.)
That's less than 0.1% of what Gates donates. That's about 0.01% of Google's market cap. That's about 4.6 work-hours of Google revenues.
Google is counting on people not having a feeling for what "million" and "billion" mean, aside from both being big numbers.
And part of what was advertised was charities using Google Apps.
I like Google, but this feels pretty cynical.