The End of an Era - Part 2

So I just finished my entry on Sun Microsystems when I realized that there was another very sad passing that hasn’t gotten the kind of press that Sun - Oracle has: SGI. Good old Silicon Graphics! Yes, I spent time at SGI as well... They too had some religion albeit not at all the same passion for it as Sun did. SGI was willing to change and grow their solution to fit customer demand. Unfortunately their leadership (post-McCracken) was interim in every way and didn’t seem to care too much about the impact their short term thinking would have on long-term business. Just for clarification I’m not speaking about the SGI leadership of 2000. I’m specifically speaking about the year of “Rocket Rick.” This would-be savior came from HP’s printer division where he was hailed as a visionary leader who knew how to manage commoditized technologies and could take SGI’s graphics leadership to the next stage.

What he did instead was give away the farm.

SGI had already committed to a Windows path and was already working on an NT-based workstation. One small mistake there... the wonderful, industry-leading, 35M transistor graphics engine designed specifically for the Visual Workstation was hardwired to the desktop’s motherboard. This meant that for 6 months the customer had the top-of-the-line performance. After that they were stuck. There was no easy way to upgrade the components. Ooops. Trip. Not fatal but certainly embarassing. It was hard for many engineers to foresee the commoditization of graphics. But that was happening in real time.

No, that wasn’t fatal. What was significantly contributory however was SGI’s stewardship of the graphics API’s known as OpenGL. This was the cornerstone of SGI’s IP leadership. Yes, it was open to everyone but it was so advanced that SGI had a hand in virtually every big graphics and big data solution on the planet. (Big data was critical too.) Rick Belluzzo, eager to please his future employer Microsoft engaged in Project Farenheit. What this was supposed to be was a graphics interoperability project between OpenGL and Direct3D. What it ended up becoming was a way for Microsoft to successfully stall OpenGL development for a year or two while Microsoft enrichened Direct3D to make up some of the gap in technology.

By itself this wasn’t enough to kill SGI but couple that with the decision to 1. spin off the technology that would ultimately make business intelligence visualization pioneer ePhiphan.y and 2. adopt the Itanium processor as the successor to MIPS and you have the recipe for disaster.

For me the funniest and most tragic moment in my career at SGI occurred on the day Rick Belluzzo was introduced to the employees. Firstly he accidentally referred to “us employees at HP...” (I’ll give him that mistake) and his comment that if we didn’t do our jobs and execute we’d all be walking around with Sun Microsystems badges by year end. That was something! Most of us were asking what was so wrong with that?! In retrospect Sun gets sold for $7B and SGI gets sold for $25M.