Just adding to my post, there are a couple of things I cannot fathom. Sure, SPARC revenue is declining and has been declining for years, fair enough, Oracle lost that war, end of story.
But Solaris underpins ZFS storage, which is selling well enough. Oracle will certainly not let others just use ZFS if they kill Solaris, but they can't just EOL ZFS storage and not expect any backlash from its userbase, which may demand Oracle to open up Solaris and ZFS (the closed portion, not the CDDL one released with Indiana), and Oracle is always loathe to do that.
Next up, there's Exadata. Losing control over hardware means the next generation is very likely going to s--- compared to the current one (in terms of bang per buck, not overall performance, which may be better). I don't think Oracle can allow that to happen since it would make lots of room for system integrators that might once again have level playing field and may offer better than Exadata performance at lower price level.
Third, there's tape. This is the one I can't understand at all. T10000E was almost finished and close to be released.. While it would be underwhelming in ~3 years compared to LTO9, it would still sell well enough to ease transition to LTO. Same goes for new tape libraries, for which there is some demand (and which Oracle desperately needs to supports its existing userbase), but there were just too many delays to let them come to market. The biggest problem is the installed base. There's a lot of very powerful users that started off with Storagetek tape decades ago and that Oracle will simply be unable to just shake them off.