The Sad Saga of Silicon Graphics: The Final Chapter
Posted by: Peter Burrows on April 01, 2009
There was a time when Silicon Graphics Corp. was the Apple Inc. of corporate computing. It received coverage out of all proportion to its size, certainly by BW. And for good reason: It involved larger-than-life characters such as Jim Clark, who went on to co-found Netscape. SGI was forever on the cutting edge of technology innovation, and pioneered use of powerful computing technology in the making of movies, game consoles and for early Web companies in the mid-1990s. And it was a lightning rod in the best sense, always a central player in the big debates roiling the computer industry (workstation vs minicomputer, Risc vs Cisc and UNIX vs Windows, come to mind).
Today, the company was sold for a piddly $25 million to server-maker Rackable Systems. Given evaporating sales of its proprietary machines in recent years, it was weighed down with more than $500 million in debt. This despite the fact that just two years ago it emerged from a pre-packaged bankruptcy similar to the ones being considered for GM and Chrysler right now. (Managed bankruptcy is designed to help companies quickly put their finances in order so they can continue to operate in shareholders’ best interests.)
If there’s a farther fall in tech history, I can’t think of it. Despite breathless headlines, the fact is that few really successful tech companies come to this. IBM survived its dark days in the early 1990s, to emerge as a services powerhouse. Apple had just a few weeks of cash left when Steve Jobs returned in 1997, but he did return. Other once-proud players — Lotus, DEC, Netscape — at least sold for a number that started with a “b” rather than just millions. Even Sun (which put out acquisition feelers for SGI many times in the past) looks likely to get close to a 100% premium if it is acquired by IBM as expected.
I haven’t bothered to spend much time with SGI in recent years. That’s something I’ve felt somewhat guilty about. Silicon Valley is supposed to champion real innovators, and SGI never stopped innovating. (Full disclosure: part of the reason I stopped visiting is that on a few occasions I almost lost my lunch watching demos of flight simulations and such on the huge, immersive, theater-like visualization centers that were themselves a breakthrough in “visual computing.”)
In the end, SGI’s mistakes of the 1990s were too much to overcome. There was the basic mismanagement, which my bureau chief Rob Hof chronicled so well in this 1997 cover story (it’s worth reading again; probably the only time a BW story has mentioned a CEO “moon[ing] SGI employees at one of the company’s annual lip-synch contests.”). And there was a chronic indecisiveness about what to do when the server market really began to commoditize around Wintel. Former HP executive Rick Belluzzo, a Wintel fan, tried his best for a couple of frustrating years. After he left, the company turned back to its proprietary ways—continuing to fight technology wars, but without the scale of a Sun to be able to support a viable business.
But this end is even more evidence that SGI is a company snakebit by bad timing. After all, SGI ran out of future just as the enterprise computing business is getting focused on breakthrough innovation again. For years, it was all about which Wintel provider could deliver the cheapest blade. Now, there’s a battle royal going on to rethink the data center, in ways that will surely require major changes to the underlying gear. Rather than the same old servers, routers and storage systems, all of these elements are being glommed together in new ways; that’s certainly the point of Cisco’s “Project California” initiative, which has shaken up the sector more than anything in years. As one of my editors put it today, “servers are sexy again.”
It was only a matter of time for the pendulum to swing back that way. While commodity hardware isn’t going away, all of the focus on low purchase price hasn’t solved problems with power consumption or with controlling operating costs related to running thousands of cookie-cutter devices. Nor will today’s computers help big Net companies keep up with sky-rocketing Net traffic, without further advances. It’s too bad that SGI won’t be there to be part of it.