Mr. Skinny, Cray has been profitable in only 2 of the last 10 years.
http://quicktake.morningstar.com/StockNet/cashflow10.aspx?Country=USA&Symbol=CRAY

Here is how HPC will be saved and becomes main stream: we will build a cloud named HPC-cloud (a Private Cloud) that meets what the users seek. I mean any user, the type of user who uses Google today, will be able access a directly or indirectly an HPC-cloud. That significantly much larger number of HPC users (in comparison to the HPC elite today)

- will always have all resources s/he needs
- will pay only for what it uses
- will use the HPC applications as a service

The users will have no idea where the HPC-cloud is located and what's inside the HPC cloud. They are shielded from all complexity. Power Optimization software will turn off the power on all unused resources when demand is low. Turning the power on means that there are enough paying users to cover the cost of the additional power from the billings sent to users.

An HPC-cloud, known as a Private Cloud, will not be a cost center. It will be a Profit Center. This means it can be sold to the commercial enterprises. The sale is much simple. Companies that need HPC (and today with all risk analysis and security concerns most companies do need HPC), will loose money every single day that did not buy an HPC cloud for their employees and and their customers.

http://my-inner-voice.blogspot.com/2009/02/hpc-cloud-will-democratize-and.html

http://my-inner-voice.blogspot.com/2009/03/coud-computing-revolution.html

Miha Ahronovitz

The total spending on Cray's Cascade and IBM's P7 will total about $300million this year on just two machines. (Total unclassified spending about $4 billion plus). This will take up up to 4-6 petaflops depending on how good and how fast code is developed. Look at the backlog for time at Jaguar at Oak Ridge which is your petaflop supercomputer for rent. No significant blocks of time until next year. That is at 1.1 petaflops. AS more research continues by drug companies, aircraft designers, Formula 1 racers time on capable machines are at a premium. Government research labs are getting money as the cost of a new HPC machine is 1/2-1/3 of one fancy new F-22. US Government contracts do not carry high margins but they do pay the full R&D cost plus associated overheads so the 6-10% margin is true profit and you get your R&D for free.

Relationships are far less fickle than many think. The difficulty of recoding software to get full performance for a change in architecture could take 100,000's of man hours. That is why NASA has never switched from Intel despite decidedly inferior performance since 2003. The risks of converting man rated software aren't worth it. NNSA had all of that they wanted with Thunderbird. That is why all of their recent purchases have been IBM or AMD. AS to software development IBM did quite well on Roadrunner and Cray is doing quite well on Cascade. NITRD was created by Congress to make sure that the US maintains the technology lead in HPC hardware and software. Congress remembers Japan's Earth Simulator.

Many flaws in this article. First - ITANIC killed SGI. SGI bet wrong and stuck with their guns, never offering Opteron systems which is completely insane especially during a time when Opteron killed Xeon in HPC. They never recovered.

Second - Cray is actually doing quite well. Read their quarterly financials and cliff notes of the analyst calls, you will learn that even with the economy in the shitter, they are managing to swim. Cray has enough secret sauce to distance themselves from the cavemen that turn screws in back alley shops for low margin.

Third - what financial crisis? HPC will see a boon in the coming year with vast sums of $timulu$ being directed towards new research systems.

Good point Mircea - if talking about commercial market, yeah. The HPC customers are usually too smart to take that "software and solutions" from those big vendors, and often have their own open-source or custom software doing the job better. That's why vendors can't stand them - it's just too hard to make money or have fun in the strip clubs for the sales force either - the academics are straight laced, no smoking or drinking types usually.

Very good article Nebojsa. But your analysis has one little flaw: HP and IBM sell often HW at little margins (or even at a loss) to sell SW and services at a much more margin (can I say insane?). It happens all around the world, IT publications reports about big software solutions, in reality many of them are just money thrown to please the staff and for image. Most public AND big private aquisitions are not based on a real business-case and often not in real interest of the aquisitor.
But, sorry, how can I say something bad about IT that make our life better, am I against evolution?
I spoke from my little experience...