[jm4]

I'm not so sure it makes sense for IBM to buy AMD either. If they are already sharing technology what incentive does IBM have to acquire AMD? Why buy the cow if you can have the milk for free?

I guess they could get into the desktop chip market, but I'm thinking if they were interested in that they would have put more effort into making Apple happy a few years ago. Would an acquisition buy them anything that would improve their position in the server market?

It seems like a struggling chip maker that's years away from having a top product again is more a liability than an asset. I don't think anyone is going to touch them with a ten foot pole. Besides, the companies out there with the money and the know-how to take over a chip manufacturer will most likely be able to get what they want- technology deals, engineers looking to jump off a sinking ship, etc.- for a far smaller price than that of an acquisition.

[Tamerlin]

IBM isn't "getting the milk for free" from AMD. IBM and AMD are sharing a lot of R&D costs, along with most of the major semiconductor manufacturer's out there (like Samsung, TSMC, UMC, Chartered Semi, etc). IBM and AMD are also trading technologies; IBM provided AMD with low-K semiconductor technology, and AMD provided IBM with automation technology for IBM's fabs.

And don't forget that IBM WAS in the desktop chip market. They even had an x86 license that allowed IBM to design x86 processors, as well as manufacture them.

They tried to get back into that market with the G5, and both AMD and Intel pretty much steamrolled it -- because IBM launched the G5 while AMD and Intel were hot on each others' heels (AMD lead performance at the time, Intel lead price).

IBM didn't make Apple happy because IBM couldn't afford to. Back then, IBM's semiconductor division was losing (each quarter) approximately AMD's annual operating budget. IBM wasn't willing to invest in a fresh design for the G5, so they went with a cut-down implementation of POWER4, which doesn't work so well without the massive caches and buses that POWER systems get.

IBM is good at makeing cost-no-object systems like POWER (which only make money because the systems and services for IBM's servers are so lucrative), and custom jobs like Gekko, Cell, and Xenon. MS and Sony have 2nd sources for Xenon and Cell (Chartered or UMC, I forget which, and Toshiba respectively), because IBM's not so good at making high-volume chips at reasonable cost.

Would it be worth for IBM? It's hard to say, particularly since IBM is also one of Intel's biggest OEM's, and IBM no longer sells PC's.