Just_An_Engineer :
Keep in mind that AMD's overhead is much smaller than Intel's while their cost of fabrication is nearly the same (Intel currently lower due to use of 45nm, but they should be pretty much equal when Deneb comes out at 45nm) so they can afford to sell at a lower price and still have a good profit margin. That being said, I don't think that a $200 price tag for their flagship processor is going to make much profit. If AMD can start selling processors in the $300-$500 bracket again, with the flagship processor going for around $500 then it will be a tremendous increase in margin.
Actually not necessarily. Intel's cost of fabrication a chip is wholly lower than AMD's, partly due to the number of fabrication plants they have, and the superior technology they possess. According to industry insiders, Intel's 65nm yield is probably the highest in the industry, while they're already fabricating 45nm processors at very decent yield rate. With the second generation of 45nm fabrication technology coming online, the yield will only gets better.
On the other hand, AMD's 65nm is known to be inferior, and by using SOI at 45nm only exacerbates the problem. SOI is pretty much useless at this point for general purpose processors, and it only serves as a poor insulator against electron ionization. In addition to that, SOI also inhibits AMD processor from clocking higher than their Intel counterparts (just compare Intel 65nm vs. AMD 65nm).
Then, on the overhead side. Your theory is correct, but you also have no calculated the fact that Intel is selling their processors at a very nice premium (gross margin @ 55%), while AMD's is below satisfactory (36%). Not only that, Intel is also slowly eating away AMD's market share. So larger overhead? Yes, but Intel also makes a lot more money to cover that up, and then some, compared to AMD's small overhead, but doesn't make much money.
The same argument can be made for ATI and Nvidia. ATI has less overhead than Nvidia by virtue of being a smaller company but actually has a much lower cost of fabrication due to the use of the 55nm process and smaller GPU's in general. ATI is likely making more profit per card than Nvidia despite being priced lower. Remember, Nvidia has been forced to cut the prices on their newer cards but thus far ATI has not had to make any significant price cuts.
Again, you're forgetting to factor in the gross margin. When G80 launched, Nvidia was able to charge 3x the price for a product that costs about 150 bucks now. Same with G92. It is not until the launch of 48xx series where Nvidia realized they cannot compete until price drops.
Sure, smaller die and smaller process node does make a difference, but only when the performance is on par against Nvidia's larger die/larger process node. Even after 38xx series launch, ATi still cannot outperform 8800GT with a 3870, let alone 8800GTS, 9800GTX, and 9800GX2.
How will 48xx series do, we shall see. I really hope ATi can succeed this time, as I really like their new GPU line. I will support them whenever I can, but doesn't mean the market will.
I'm going to stick by my statement that IF Deneb is competitive with Yorkfield and IF ATI's sales remain strong, and IF the Puma laptop platform is a success AMD will post a profit before the end of the year. Now, that's three things that need to go right for AMD and there is certainly no guarantee that all of them (or any of them for that matter) will happen. This is strictly a hypothethical scenario.
To be honest, I fail to see the correlation between Deneb and ATI and Puma. Deneb is AMD's CPU division's brainchild, and so far it doesn't look good at all. Unless AMD can also offer a mobile version of Deneb (which I highly doubt) like Intel, Deneb will not affect ATI and Puma sales in anyway.
I agree that those three need to go right in order for AMD to survive. But we can pretty much conclude the Deneb will still be like Phenom vs. Yorkfield, which is being pushed towards the mid-end / mainstream market, where the margin isn't as rosy.
The 48xx series is going well so far, but I wonder what's the margin on the RV770s.
As for Puma, I'm not sure if it will go well, if at all. Puma, again, is largely based on performance / price. The OEM I work at will only launch 1 line of Puma, compared to....about 12 lines of Centrino 2s and Centrino. I'm sure others will be very similar.
It should also be noted that AMD will likely have some more staff reductions this year and will be jettisoning the former ATI divisions that were written down this past quarter, further reducing overhead.
They're also significantly reducing their capital spending. If they're reducing their CapEx, how can they support the development of Bulldozer? Fusion? Torrenza? According to industry insider, it appears that AMD is putting all the egg into the 45nm Deneb basket. If AMD cannot at least attract a decent portion of the market, its very likely that AMD will be forced to file Chapter 11.
Its not all about overhead and performance/price, but rather gross margin and capital expenditure, and a good management team. At the moment I don't know what Dirk will do to save AMD, but Hector really screwed up, big time.