What amazes me no end is that I read some sites who somewhat brazenly (and stupidly, if you ask me) said they were "sticking with their story" about how AMD "said" it was going to spin off its FABs even after the sites received contradictory statements directly from AMD itself.

Who are you going to believe--a newspaper reporter writing about topics and subjects he has no professional experience with, or printed statements sent directly from AMD itself? Go figure. If AMD said he was misquoted, then, by gosh, I'll believe AMD knows its own business better than anyone else until AMD tells me different...;)

I hope this vertical disintegration doesn't translate to higher markups on the products, because then they would REALLY have no chance against Intel and nVidia.

AMD may still be contemplating it. It really depends on several factors:

- Are they not using full capacity of their fabs? Do they expect them to be underutilized in the future? This is especially true of fabs running older processes that are no longer appropriate for CPUs and are expensive to move over to new nodes but could still be employed to produce other chips (for other customers). This gets to be a larger issue as time goes on: once AMD is fully on 45nm, do they have the cash to convert their 65nm lines?

- Is there a partner willing to invest in a spun-off fab operation? You're unlikely to be able to take a spun-off fab company public, at least initially, so if all you end up with is a wholly-owned subsidiary you haven't really accomplished anything except an accounting trick: some chunk of the debt now sits on AMDFabCo's books instead of those of the mother company. But with a partner investment you can retire some of the debt and perhaps bring in more business. IBM is the obvious candidate here, but there are others (TI, Motorola) ; with the current dollar weakness it's also possible they could attract some interest from foreign companies (Samsung, Toshiba, etc).

- How does possible contention for production capacity get resolved? If the fab company drums up a lot of outside business and then AMD (and/or its partner) sees increased demand, who gets the short end of the stick?

The trouble is, this is a lousy time for anybody to be feeling bullish about new companies anywhere in the semiconductor sector (with the possible exception of solar cells). Which of course means it's probably a good time for an outsider to invest, because they'll be buying at a discount, but that means AMD likely won't be able to get what the fabs are really worth, if they're able to get any interest at all. Selling at the bottom is never a good strategy. Unfortunately, sometimes you paint yourself into a corner.

It seems to me that the actual silicon is so tied to the fab process that one cannot really design circuits and then just hand them out to a foundry for manufacturing. Seems to me there would be cost savings in being able to tweak your process to solve a circuit design problem and vis-versa. If you don't own the whole thing, then you don't have the flexibility that can give you that advantage and I suppose also, that cycle time for all stages of a design, from conception to silicon, would increase.

No FAB = no competition with Intel.

AMD would become another VIA.

Unless they get orders from other companies the Fab company wold have to rely on what AMD paid them for their chips. Since GPUs are currently not made with SOI, they couldn't even move orders from TSMC for Radeon. Perhaps they could make the chipsets, but even then they would need additional customers to make it feasible.

Of course, there has been word that Power 7 will be Opteron compatible so perhaps AMD may manuf those or perhaps Cell. AMDs biggest problem now is that Intel called them on their price cuts and now it's difficult to get the desktop parts back up to 2006 price levels. Fortunately, Phenom X3 and X4 are made on 300mm wafers which doubles the effective output of 200mm wafers which is what all Opterons and most desktop chips were made from.

And because Brisbane and Turion are so small, they can make more per wafer than Phenom or Opteron and Griffin is where the market is going (mobile). AMD really should shrink Griffin now to get some 3GHz mobile chips. That would give them a huge leg up on Centrino.

So we're not seeing AMD minus fabs, we're seeing AMD1 and AMD2. AMD1 goes fabless and designs and sells chips, AMD2 sells fab space aggressively, like they should have been doing all along.

If this is handled right, it could end up being a hugely good thing for both halves. If it's handled poorly, AMD will sink to Nvidia's level (fabless but hopefully profitable) and the fab-AMD will just plain sink. If things get really bunged up, both halves sink, nothing new there.

It's not possible for a leading mainstream microprocessor company like Intel, IBM, or AMD to go fabless.
If the fabless model allowed a company to remain competitive, IBM would have moved in that direction years ago.
In the words of the last competent AMD CEO;
Real Men have fabs.

Only real men have fabs.

After having $1.2 billion loss I think is time to sell them.
I wonder what AMD will sell next to remain afloat

I don't see why people are freaking out. The company I work for is actually 3 separate companies. Sales, design and manufacturing is a separate entity (on paper).

Dirk Meyer starts off with a blunder.... NIZE...