The Dawn and Dusk of Sun Microsystems [video]

In other words, they died by their own sword.

Some things I would have said differently;

Sun's biggest potential competitor early on was Digital Equipment Corp's VAXStation. It was only DECs unwillingness to cannibalize the sale of their larger minicomputers that prevented them from sucking a lot of oxygen out of the market.

Sun invested heavily in the i86 architecture, creating the "east coast division" on route 128 outside of Boston. However, the ideas that group had clashed with the "old school" UNIX folks in Mountain View and as a result SunOS actually feature forked between the version ECD shipped and the version Sun shipped.

John Hennessy, who knew Andy from Stanford, convinced Andy and Bill Joy that "CISC" was dead and RISC was the future. Nobody was going to build a new processor at that point and so Bill & Andy decided Sun would design one and make it open (unlike the 68K series or the x86 series). The first workstation, codenamed Campus 1 was the result.

But by the time SPARC was "real", too much of the company was seeing risk in being too open. And so even though SPARC was an open standard, Sun tended to compete viciously with anyone that implemented it. It made for a some weird dynamics.

The UNIX wars were heating up, and Sun, a company that had a high cash burn rate because of its HW business, had a quarter that we "burped" (which was SunOS 4.0 was late (it was the big shared library release and a LOT of code had been rewritten[1])) Anyway we had a quarter were the software delay had delayed the release of the next workstation which had interrupted revenue and as a result we have a quarterly loss (for like the first time ever) and that scared the crap out of the senior management. Berkeley was yelling at us because they felt not enough of the changes Sun were making were making it back into BSD, and AT&T started threatening everyone with lawsuits because, well UNIX System 3 kinda sucked and they didn't want it to. The result of THAT tornado was that Sun struck a deal (codenamed "lulu") with AT&T where AT&T would give Sun a billion dollars (which at the time was a lot of money), and in exchange, Sun would STOP improving BSD, and instead would work with AT&T to make was System 3 (which was going to be System 4 but became System 5) the best of both worlds, and AT&T and Sun would both cross license the IP so that they wouldn't sue each other but they could sue anyone else making a UNIX derivative.

In the opinion of a LOT of my peers at Sun, THAT was the event that killed Sun. They had fully switched to the dark side of "not open" systems and proprietary everything.

Sun made some excellent stuff after that, massively symmetric multiprocessing systems, "Spring" a really creative re-imagined OS, Etc. But none of it, even Java which was leveraged so hard to be a "Microsoft killer" that Sun lost sight of what it was to be a leader.

When John Hime left Sun (he had been Carol Bartz's boss as I recall) he told me that leaders don't imitate or copy their competitors, they identify what it was they were missing that gave the competitor some traction and improved their understanding of their own path. As I saw Sun become more and more their competitors by following them on their path, rather than blazing their own, it really became clear what he meant.

I remember when Opteron was introduced (AMD's first processor in the AMD64 set) at Microprocessor Forum in 2003 I emailed Scott and suggested perhaps while Sun had become DEC (they were retreating to the machine room with their big iron, and Solaris Desktop was a smaller and smaller part of their revenue mix) that coming out with an open workstation based on the AMD64 architecture using BSD again might not be a bad idea. He thought that was pretty funny and said that train had left the station.

[1] I still have my "4.0 Team Member" sign that we put on the doors of engineers who were working on 4.0 to remind others not to bother us with non-important problems :-).