It's not financially "the strongest" of companies and investors
probably worry, as I do, that slowing US economy could hurt sales in
the coming year or so.
That could be the reason for the fall. Ones the ball starts rolling
it's hard to stop it.
I hope the stock finds support soon, for the sake of my portfolio.

I'm no expert though

Who wants a nonprofitable company with over 5 billion in debt, and are
so desperate for cash that it inflates its stocks buy selling 10% of
its company to one buyer?

AMD's new Phenom processor can't compete in terms of performance, and
there's no indication that they can compete in terms of power
consumption at this point:
http://www.tomshardware.com/2007/10/29/intel_penryn_4ghz_with_air_coo...

The only thing left for AMD is to compete in terms of price, which
isn't good news for the stock.  They'll most likely continue selling
processors at a loss.

This is only B2 Stepping cpu's, and 65 NM, they havnt started on
showing 45 NM cpu's

Means that they will be fighting in the budget market which means low
profit margin.
Quad core chips I think also take around twice the area on each wafer.
So each wafer gives only half the number of chips as before.

So half the number of chips per wafer but they don't charge double the
armount per chip exactly. They are having issues getting product to
market as well. IBM had to invalidate their opteron scores cause they
couldn't get products to market. http://www.theregister.co.uk/2007/11/06/ibm_opteron_x3455/

If AMD can't provide enough barcelona cores to feed the higher margin
server market even for oems like IBM there won't be enough phenom
cores to feed the comsumer market.

Penryn is a LOT smaller than Phenom. Phenom comes in at 283mm2 whereas
Penryn is 2 x 107mm2 (214mm2) If penryn was bigger at 45nm than AMD's
65nm offering I'd be shorting Intel and buying AMD.

Don't forget also, because AMD is a single die opposed to 2 glued
together...

A) A defect destroys more die space (283mm2 opposed to 141mm2)
B) There are less die per wafer

This is on the assumption AMD could have done the glued approach,
Let's not go into that, I'm simply trying to demonstrate how die size
and yield affect each other.

To save arguement I'll put this into perspective using Intel as an
example and the yield graph on page 3-7 of this document.
http://smithsonianchips.si.edu/ice/cd/CEICM/SECTION3.pdf

I will assume Intel can get 600 45nm Dual core die on 1 300mm wafer @
a die size of 107mm2

At a defect density of 0.2cm2 @ 107mm2 die this would mean a yield of
80%
80% of 600 = 480.
Intel glues their quad cores so that means they could also have 240
quad cores.

If Intel's quads were native (not glued, and for the purpose of
demonstration we'll assume it would be exactly double the size) the
die size would be 214mm2. Since the die size has increased the die per
wafer has decreased. It would be less than half, but for sake of being
conservative and actually over proving my point we'll say they get 300
die per wafer.

At a defect density of 0.2cm2 @ 214mm2 die this would mean a yield of
approx 70%
70% of 300 = 210.
That means Intel would get 210 quad cores as opposed to 240. That is
30 less!

Now plug AMD into all of this. At a whopping 283mm2 they would be
getting around 226 die per wafer if we assumed the same perfect
scaling.

At a defect density of 0.2cm2 @ 283mm2 die this would mean a yield of
approx 60%
60% of 226 = 136.
Thats 104 less than Intel's current dual core gluing approach.

If we pretend AMD could produce 2x142mm2 die instead.
At a defect density of 0.2cm2 @ 142mm2 die this would mean a yield of
approx 75%
75% of 452 = 339
Or we could call that 170 quad cores. Approx 25% more.

That's only the tip of the iceberg though. If we plug in the 0.5cm2
defect density from the AMD slide a while back things just get plain
ugly. AMDs yields are not as good as Intels either. That is just a
given. Or to put it this way, there is no way Intels yields are as bad
as AMDs.

I'd estimate Intel have a yield of around 80% opposed to AMD at around
50%, and thats being VERY kind to AMD.

Lets put some money figures into all this since this is a finance
forum.

I'll use the price of the cheapest quad core from each manufacturer.
$251 for a quad core Phenom.
http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_609,00....
$266 for a Core 2 quad.
http://www.intel.com/intel/finance/pricelist/processor_price_list.pdf

AMD: 136 quad cores per wafer @ $251 per processor = $34136
Intel: 240 quad cores per wafer @ $266 per processor = $63840

This is of course assuming that both Intel and AMD have a defect
density of 0.2cm2. So even in a situation where AMD is manufacturing
to the same standard as Intel they are nearly $30k per wafer down.

If we put more realistic estimates in (AMD @ 50% yield)
AMD: 136 quad cores per wafer @ $251 per processor = $28363

Now people are going to say, you're nuts. 50% yield is way too low.
Before you say this look at the yield model. A 50% yield @ 283mm2
equates to approx 75% @ 107mm2.
Defect density is the constant here. Yield is an effect of defects. A
larger die has a larger chance of being hit by a defect and the amount
of damage is larger.

If AMDs defect density would yield 75% @ 107mm2 I'd have to put Intels
yields at the least 85% just by default. It is quite obvious Intels
production is stronger that AMDs. You'd have to be insane to think
otherwise.

Hopefully someone actually puts in the time to think about my post
rather than 1 starring me and saying I'm wrong. If I'm wrong then you
should probably disregard all mathmatical probability. Start folding
pocket aces and playing 27 off. Or you could provide a counter
argument.

I didn't put in manufacturing costs etc. I wouldn't like to guess what
they are. Then employees to pay, and don't forget hector, he'll
probably get a nice comp package when he leaves, just to bury AMD some
more.

Whats funny is even if AMD had 100% yield Intel would still take more
money per wafer than them.

AMD didn't put P4 to the grave. AMD let Intel sell P4 for stupid high
prices and AMD did the same with AMD64. The fact as AMDs product was
by far the best when Intel had P4. AMDs product is not the best today.
Taking into account Intel launched their product in July 2006 at
volume and AMD launched theirs In Sept 2007 with about the same
availability as oxygen on the moon that puts them even further
behind.