Sun did not invent the engineering workstation, but they certainly perfected it. But where are workstations today? Gone, for the most part. Sun's workstation business is about the same size as SGI's, which is to say small. Sun is now a server company, but that won't last long either under the onslaught of Linux. Cheap Intel and AMD hardware running Linux is going to kill Sun unless the company does something so stop it, which they aren't. -- http://www.pbs.org/cringely/pulpit/2003/pulpit_20030213_0007...
A lot of this comes down to how companies view themselves and what they are really good at. You'll notice, for example, that I didn't include Sun in my list of vital companies. That's not so much because Sun can be defined in terms of the others but that Sun is simply doomed. Their software isn't better, their hardware isn't better, and they can't see themselves as anything but a maker of hardware or software, so my simple recommendation is that they take the rest of their cash and try entering a hot new field like -- say -- space flight. Or making really fine cakes. The world will always need fine baked goods. Or just give it back to the shareholders. Really. -- http://www.pbs.org/cringely/pulpit/2006/pulpit_20060504_0008...
Remember my doom-and-gloom prediction last week for Sun Microsystems? That's based almost entirely on the company's inability to see itself moving from being vertically integrated (doing its own proprietary hardware and software) to competing on a level (that is horizontal) playing field. While that might make them just another PC vendor, don't worry about that happening because Sun would rather die first. And will. -- http://www.pbs.org/cringely/pulpit/2006/pulpit_20060511_0008...
If he's right then it will be interesting to see the details unfold. If he's wrong then it will be interesting to see Sun re-invent itself.
The second case is where you're in a corner and need to scale up. I wouldn't be surprised if Sun offers servers with 2048 cores. Another case is where you've got a huge volume of data on a SAN and the throughput of Solaris justifies the cost of the licenses. A final case involves the energy saving gained from RISC. However, I've previously argued this latter case ignores the opportunity cost to a fast growing organisation and is therefore irrelevant to a start-up ( http://news.ycombinator.com/item?id=123587 ).
The other cases aren't important to new companies either. So, I agree with you that they're running out of options. I agree with Cringely that vertical integration isn't a big draw.