John West at InsideHPC.com pointed to an article at Barron’s about SGI. Before I get into this, I want to note that I had wondered whether or not we would continue to see (massive) oscillations in the market, as it effectively dissipated valuation, or if it would start tending towards an asymptotic lower limit … testing the bottom as it were. It seems that the forces that are driving the economy are continuing to drive valuation out of the market.
This makes it hard for any company to do well. News across the board are significant job cuts ahead of a confirmed recession (e.g. traditional definition of two quarters of negative GDP growth can’t be determined until well after the fact). Sadly this will definitely result in a deeper recession, as spending comes to a grinding halt, people are cut from payrolls, and banks are so averse to risk as to not grant new credit. This is the world in which we find ourselves. It is in this world that companies have to make some incredibly difficult choices going forward to survive and thrive.
Now with this context, the article at Barron’s makes similar (if not identical) points to the ones we have made before. SGI has serious challenges ahead, a set of competitors that are much larger, and can sustain a more bruising battle for far longer. Several things I was not aware of … or maybe I missed in my past analysis of them … their looming payments to Morgan Stanley.
I’ve said it before, and I’ll emphasize it now. They have some very hard choices to make, and they need to make them fast. If they are simply Linux Networx 2.0, then they are in trouble.
They do have some things that they should have been capitalizing on all this time. Their NUMA architecture is one of the best things around for large shared memory systems. By not making it available in x86 format, they have effectively ceded the market to Shai Fulthem’s ScaleMP. His company makes a DSM layer (ok, it is more than that, and you need to go to his site to see it) that is very cool. They still have challenges facing them (ScaleMP) but I sense they are going to survive, and if anything, likely some bright tier 1 vendor is going to purchase them. No I don’t know anything about this, and I am not trying to start a rumor, and no one has asked us for due diligence on them. This is a guess.
But ScaleMP does in software layers (most of) what the SGI NUMA technology did at a lower level. And it does it with commodity gear.
So what does SGI have left? I don’t have a good answer to this.
I recommend following the link to InsideHPC, and read John’s comments and the article.