Will IBM save AMD's bacon?

EVER SINCE the announcement that IBM and AMD would be working together to develop new chip-making technologies at IBM's East Fishkill NY, speculation has been rampant that Big Blue might buy the struggling chip company.

One immediate benefit of such a move, of course, would be the silencing of the CPU lunatic fringe that continues to claim that AMD is secretly owned by Intel and that the company's moniker actually stands for "Another Microprocessor Division." We'll be reporting on these geniuses, and their latest attempt to prove the earth is actually flat later in the week.

At any rate, if you're on the AMD side of the equation there are a lot of things to like about a potential IBM acquisition. For starters, IBM has money. Lots of money. In fact, not only does IBM absolutely dwarf AMD's financial status, its gross profit and net income both surpass even Intel's by a substantial margin. Toss in IBM's advanced foundries, its research centers, and its proven track record when it comes to designing and implementing new processor technologies, and AMD would stand to benefit tremendously from the association. IBM's long reputation as a business partner to thousands of firms could put Opteron into server rooms in volume much quicker and less painfully than might otherwise be anticipated and aggressive measures by IBM to port their database software to the x86-64 platform could speed 64-bit adoption as well. Clearly there's tremendous potential benefit for AMD in such an acquisition—but what about for IBM?

IBM would get AMD's Dresden facility, its memory fab, flash memory and market share. So there is some small inherent physical value for IBM when considering an AMD acquisition—but what about the value of AMD's IP? That's where the question turns murky, and where the direction IBM plans to take in the future becomes critically important.

Sorting Through the Variables
If IBM bought AMD and kept the company's core microprocessor division intact and its focus essentially unchanged, the acquisition would send shockwaves throughout the entire semiconductor industry, particularly in the direction of Intel. Santa Clara and Big Blue have been increasingly butting heads the last few years as Intel began to move more aggressively into the server market, and while IBM does carry Xeon-based systems it's doubtful that the company is looking favourably upon Intel's attempts to muscle the Itanium architecture directly into the high-end space occupied by the Power4 and Power5 architectures.

Acquiring AMD might give IBM the necessary means to go head-to-head with Intel on a much wider spectrum of products, but that's not a front to open lightly. IBM has steadily withdrawn from offering consumer-level products, including systems. Choosing to continue AMD's business in this market segment would mean reversing its current trend. Buying AMD would also mean it had another product set in the 64-bit market.

If it did buy Opteron and AMD64 technology, it could help rapid and widespread adoption of the 64-bit x86 standard in the midrange server market, isolating Itanium in the niche high-end and driving it out altogether with a mixture of legacy compatibility, high performance, and excellent price. IBM might choose to keep part of this strategy and block Itanium adoption at the mid-range server market and below, while deploying its own architectures to challenge Itanium at the high end. This would have the advantage of keeping 64-bit lines separate and allowing IBM to continue charging a substantial premium for "Big Iron 64-bit" while maintaining a consumer-level 64-bit technology that emphasized compatibility and desktop/workstation performance.

But the worst case for AMD is that IBM could wait until the company is on the auction block and selectively acquire the assets of the manufacturer at a bargain price. This would be the route to go if, for instance, IBM was only interested in AMD's manufacturing facilities rather than the employees and intellectual property of the CPU manufacturer.

Unlike Intel, IBM is in a very broad range of technological fields, with server CPU's being only one aspect of their business. Committing the kind of resources it would require to slug it out with Intel would be a major decision the company isn't going to take lightly, and Big Blue might not be interested in rocking the boat. On the other hand, Intel has already made it clear that they are willing to target Big Blue's own server line, and it'd be the height of stupidity to naively sit around believing Intel doesn't present a threat.

Six months from now, when the early sales data on Opteron and Athlon 64 starts to arrive, it may be easier to judge how much of a pearl AMD is to IBM. For now, it's in IBM's best interest to wait, watch the chip's progression, development, and market uptake—and examine its own options. With Itanium, Power4 / Power5, AMD 64, and Ultra SPARC, plus the remnants of Alpha knocking around, the 64-bit market is getting crowded — but the question of whose going to be left standing when the dust settles is still wide open. µ

See Also
AMD will die within five years, implies IBM exec