Good summary of the story. It is really sad the Itanium was so hard to get going. It is a vastly better architecture than the many times warmed over x86 dog's breakfast mess of a processor. But masive economies of scale win out. SGI were a great company, and in many ways still are. Just reduced in size. They understand the ecosystem of HPC better than the remaining big vendors and provide support second to none.

Apple were once mooted as a suitor, because of SGI's ownership of the graphics, animation and video post processing markets. That was a long time ago. Apple got those markets anyway, as SGI ceased to be able to hold them. Apple have a company mentality that just doesn't gel with the service and support needed for big iron. I suspect that if they did try, they would mess it up and probably just ruin the remaining non-technical asset that SGI have - the world wide support organisation that does understand.

SGI has been presenting the Xeon based UltraViolet system to customers for some time.

I think the original plan may have been for a Tukwila based system, and later changed to first have a Tukwila system then a Nehalem system, then changed to bring out the Nehalem system first, and finally the Tukwila version was killed altogether.

With UltraViolet, SGI has proven one can design a large, NUMA, shared memory Intel Xeon system based on a design intended for Tukwila. Expect HP to follow suit at some point. HP's challenge is it has no operating system for a Xeon Superdome. Linux is great for HPC, but customers used to running big iron want UNIX on their big iron.

Itanium is dead, and has been for years. Some vendors (HP) are still in denial.

Abso bleedin lootly, but they won't. But it does make you think, if some one buys SGI, perhaps a speculative buy out for Sun isn't such a daft idea.