I didn’t have access to all the information (and still don’t), just what seeps out in reports. Here is a good report, which suggests that this might not be as pretty a face (and it wasn’t pretty) as SGI suggested in its customer letter. Critical aspects not mentioned in public until now appear to be
I had heard it was shopping itself around. But what real value did it have? And what about its liabilities?
Ok, this was actually much worse than I had thought.
Another article points out
(since I don’t subscribe, and the WSJ hasn’t quite figured out this interweb thingy, I don’t have the rest of their write … but then I don’t need it)
That last quoted sentence is critical though. Not too many companies were going for new SGI gear. It was more expensive than other gear, and there was little added benefit to the extra expense. More to the point TI09 just bet heavily on SGI, and they lost. It remains to be seen whether any successor of SGI will be able to fulfill the TI09 contractual obligations without requiring additional significant expense and cost.
This suggests, rather strongly, that something is (badly) broken with the HPC acquisition system in the TI-XX series, and HPC modernization program in general. Others have made this point, I am merely echoing what I have heard.
The goal of the TI-XX program for HPC should not be, to drive the winners out of business … as it has done, apparently, with both LNXI and SGI. HPC is critical to the business of the US government, and many agency missions. It would be a shame if patronage of these agencies was perceived as a death sentence by businesses and investors. Talk about stifling innovation in an area …
If a vendor becomes effectively an arm of the government, due to the government being effectively their only customer, then all you need are for political winds to blow a different way, or for government budgets to be delayed by politics … and whammo …. the company is in trouble. Since HPC benefits business and government alike, the government should use the same conduct and methods for HPC procurement that business uses.
SGI died as it had effectively one customer. It couldn’t sustain itself on that customer. It couldn’t grow new business (remember the not trying to out-Dell Dell comment I keep making … there is a real good reason I make it), and couldn’t justify its higher cost for gear it didn’t make.
At the end of the day, what people care about in businesses is better, cheaper, faster. It would be hard to argue that SGI was any of these (though I have seen people try).
So, now TI09 is left with 6 of 7 spots basically unfillable due to the vendor going away. Sure it can insist that the successor organization take this on, though the bankruptcy court could treat such obligations as a liability. So they may be SOL.
Looking over the magnitude of the debt, and the true sale price of $15M (not $25M, as $10M is being held in escrow for tax purposes), I suspect (strongly) that this sale will be challenged. This is pennies on the dollar here folks. $15M/$500M is roughly 3 cents on the dollar. There is no downside to the creditors pushing back, and hard.
This isn’t pretty, and unless the creditors don’t wish to write this loss down (most will not), it is likely start to get all lawyer-like and messy. While I’d like to be a fly on the wall during the creditors conference, I don’t think anyone is going to go away happy. But $0.03 on the dollar?
Note also that GM is actively talking about a 363-based process for it. But they have no buyer, they simply plan to split the company in two (good company/bad company) and leave the bad assets in the bad company.
[update] I should point to John West’s very intelligent story on the 363 process at InsideHPC. Here he looks at how this could play out. I don’t think its going to be clean. At $0.03 on the dollar, I think the long knives have just come out …