It's not going to hardware either: their hurricane chipset is designed for Windows on Xeon, not Linux. Instead, it's Sun, not IBM, that's about to release a wave of AMD based products using a chipset and board design specifically tailored to boost Linux (and Solaris x86) throughput.
So, where did the money go? Consider this from an interview Robert McMillan did with IBM's Jim Stallings just after the latter became responsible for IBM's Linux initiative in April of 2003:
I'd like to ask you a few things about IBM's Linux Technology Center. You have about 250 people working there. Can you tell me exactly what they're doing with their time? Are they helping customers or developing the kernel, or doing other work?
Stallings: They're making contributions. Their full time job is making contributions to the kernel. That's it. They don't have another job sweeping the floor or working on WebSphere or anything like that. Another way to look at it is, IBM is making a real physical contribution to the open source community, both in terms of what we're doing with our products, but also what we're doing with the kernel. And these people are the ones who work on the kernel full time.
That's 250 people, many of them presumably experienced kernel level Unix developers, working on the kernel for over three years now. Sounds impressive, but what have they been doing? Has anyone seen results remotely commensurate with that level of effort?
Then too, 250 people, full time, seems like a big number, but actually only accounts for about 30 million a year -- leaving almost $900 million unaccounted for.
So where did it go? Is it under a mattress somewhere or did you get it? I know I didn't.