SGI Files for Chapter 11

Silicon Graphics, one of the original makers of high-performance graphical workstations, has filed for bankruptcy protection. Sad news for this one-time market leader.

While I’ve been following the decline of SGI for some time, so this was not a surprise, it’s a useful reminder of how few tech companies survive changing market eras. The PC companies nuked DEC and the minicomputer companies, and then Sun blew up SGI (sort of), and now Google is giving Microsoft fits. It’s very rare, however, that one of those incumbents successfully fights back, making Schwartz’s job at Sun look tough, and making the Microsoft’s struggles with Google all the more painful.

If you’re interested in a little time in the wayback machine, check this 1994 Wired profile of SGI to see what it was like to be the then “hottest computer company in Silicon Valley”. I was an analyst at the time, and I remember reading the piece in Wired and writing a scathing (and more than a little jejeune) anti-SGI thing in Hotwired. Like usual, I was was a decade early, but we futurists always have troubling with timing such things 😉

[Update] Dave Taylor has a good users-eye view of SGI’s slide to Chapter 11-ishness, including the tidbit that Google has been buying up all the lovely SGI buildings around the Valley. Don’t let anyone tell you that the tech industry doesn’t eat its young. It does. Lustily.

[via Om]


  1. From the article:
    “Apple,” Jim Clark will sigh, as if he were talking about a horse on its way to the glue factory. “They’re not doing anything… Apple blew it.”

  2. fartikus says:

    SGI has a decent business in selling some exotic hardware to the government. its not clear how much of SGI can survive on this, but i would be surprised if the entire firm vanishes.

  3. There is an upside – used SGI machines should drop in price. I really do love my brace of O2 desktops. A couple more would look keen in my office. SGI – whatever their faults – made some darn nice and nice looking hardware.

  4. you mean the tech young eating the old in this case -) Not sure how to read this market – an operationally efficient vendor (Dell) and a product innovator (SGI) both with problems…I guess you have to be both – HP? screw the discipline of market leaders that you can only be one or the other…

  5. In my industry (scientific computational modeling), the Linux workstation and a move to cheap clusters was the beginning of the end for SGI (that and the crazy Windows idea). Good high end Linux graphic drivers from NVidia pretty much sealed the deal.
    A few years ago, all of us used SGIs (I spent my grad school days with two Octane’s under my desk), but today the SGI is a relic of the past. Too slow to use for most problems, and the graphics perfmance is no longer a factor. I will always have fond memories though.

  6. SGI demise can be best summarized by “lack of adult supervision”. Great technology, really smart people, always trying to reinvent themselves into the next big thing, without paying attention to the last thing that they were actually successful at. There was always to attempt to outrun the curse of being successful.

  7. Sorry for a late comment. What happened to SGI was unfortunately inevitable. Any company not willing to auto-cannibalize its markets for the sake of growing newer and bigger ones will run into exactly the same problem.
    In the 70s and 80s, the heavy metal monsters dominated. Along came the supermicros in the early 90s and started eating the same food (revenue) as the Cray mainframes. Cray suffered, and made some attempt at adapting, though SGI was busy eating its lunch (dinner, and breakfast).
    Some of us at SGI at the time thought the Cray purchase made no sense. Cray was cool, but we were killing them competitively. Our machines could do 80% the performance for under 20% the price.
    This rule is important in HPC, and you see it again, and again, and again …
    This purchase and the massive arrogance and ego in Mountain View prevented us from focusing outside of our narrow world view. Some claim that UltraSPARC started beating us. That is laughable and untrue. We gave up Beast and Alien in favor of the good ship Itanic. We ignored that Challenge S was the most popular web server at the beginning of the internet boom. We focused upon our own internal issues, avoiding looking at the competition which was sharpening their talons.
    That and we hired some real winners as management. Is there a sarcasm tag here?
    Around 1999/2000 some of us had noticed the PCs were on par with our boxes for integer performance. Pretty close for FP. A few of us pushed hard to enter the cluster market. We did with the 1400, and eventually the 1100. And we exited it. In 2001. About 1 year before it started going like gangbusters.
    Today, HPC is a 9B$ market. It is growing 20+% CAGR. The cluster market is about 1/2 of this, and is growing at a 3 year average rate in excess of 60% CAGR.
    Clusters came up underneath the supermicros, and for 20% the price could do 200% the work. So technology curves change. That happens.
    Here we are in 2006. Anyone believing that the cluster market is the be-all and end-all of HPC will likely be proven wrong in a few years.
    SGI could not deal with this. SGI could not adapt. They could not believe that a cluster could do what their supermicro could do. They were right. It did it better, faster, cheaper. Rather than adopt it, destroy their existing market in favor of the much larger new market, they chose the head in the sand approach.
    None of these lessons are lost on my as I build my startup. You can adapt to a quickly evolving market or you can be overrun by competitors. SGI chose the latter.
    The question is whether or not they will successfully emerge from Ch11, or if some of their jilted (former) owners will file suit to force a Ch7. I am not sure a Ch7 is in the offing now, but there aren’t many assets left of value there. SGI claims it will emerge from Ch11 as effectively a cluster vendor. It has to do a rapid product transition to a new chip. Moreover it has indicated a desire to go after a market where stability of the company is valued as much as the product.
    I am not convinced that they are long for this world. It may be early to grieve for them, but I have great memories of the wonderful teams I worked with while there. That SGI died a long time ago. What emerges, if it emerges may be named SGI, but it won’t be the place I once knew and called home.