AMD: The not-so-great pretender

As any rebel leader knows, survival means always being lucky. While the established state can absorb a few hits, the pretender to the throne only has to get it wrong once and it's "game over".

With size so important in the processor wars, AMD has had to rely on guerrilla qualities — smarter, nimbler, more in touch with market — to outplay its giant enemy. AMD did well with Opteron, producing key technical innovations on the industry-standard platform while Intel wandered off into the silicon la-la land of Itanium.

But things have changed, and not for the better. Luck and smarts seem to have deserted AMD, followed at the end of last week by its chief technology officer, Phil Hester. Tellingly, the company isn't planning to replace him, leaving technical direction to the individual divisions — something that only makes sense if its planning to fundamentally reorganise.

It may have little choice. With sales slumping, products months or years late and the $5bn (£2.5bn) debut from the ATI purchase far outweighing any advantage it brought, the chances of the company continuing as is are increasingly slim. It could divest itself of its Dresden fab plant, choosing the far less capital-intensive path of buying in capacity as needed; it could turn one or more of its five divisions into a joint venture with any of a number of other chip companies with pretensions to some of Intel's market share; or it could hold on until it really can't pay its bills, and try to build a new company out of Chapter 11.

None of the above addresses the fundamental problem: loss of faith. Intel dug itself out of its own hole by rediscovering managerial discipline and delivering on promises; there are few indications that AMD has got that message. However, there's still room to operate against Intel, which may be more vulnerable than it appears. With Atom and Nehalem, Intel is making big architectural changes on two major fronts: things may not work out as well as the roadmaps predict.

Until AMD can resolve its own problems, though, it will be in no position to make any use of future Intel mis-steps. Its creditors, its customers and the rest of the market will be looking for radical change; otherwise, the next piece of bad news may be final.