Nvidia GeForce GTX 400 Yields Are at 20% - 30%, Further Delays Possible – Analysts.

[ajaidev]

"Even though Nvidia Corp. commercially released its GeForce GTX 400-series roughly six months after the first public demonstration and had to disable parts of the chips to ensure higher supply, financial analysts from Needham and Company claim that Nvidia only has about ten thousand of Fermi-series chips on the market at the moment and that the yields of Fermi chips is between 20% and 30%."

http://www.xbitlabs.com/news/video/d..._Analysts.html

WoW seems Charlie had some insight...

[Hannibal Lecter]

"Even though Nvidia Corp. commercially released its GeForce GTX 400-series roughly six months after the first public demonstration and had to disable parts of the chips to ensure higher supply, financial analysts from Needham and Company claim that Nvidia only has about ten thousand of Fermi-series chips on the market at the moment and that the yields of Fermi chips is between 20% and 30%."

http://www.xbitlabs.com/news/video/d..._Analysts.html

WoW seems Charlie had some insight...

and you think that financial analysts would know this information? They probably read Charlie in the first place

I bet you that financial analysts know a lot LESS about the GPU market than the average XS reader! I am not saying this is not true, but making a story about a financial analyst claim is laughable.

[saaya]

impossible! charlie cant have been right or even close to the truth! hes making everything up!
this is just xbitlabs backing him up for some reason, maybe he sold them drugs or kindapped antons dog (owner of xbitlabs)

iirc i did some math a while back and 20-30% was the point where yields were acceptable, below 20% the retail price per card would be in the 750$+ range...
so what do you know, nvidia might actually not be losing money on 470 and 480 cards... i still doubt they are making much if any with the cards they sell atm though...
im pretty sure that they are selling the current cards for pr and to support their AIBs...

[ajaidev]

Financial analysts are big cats they just dont read charlie's posts they have sources and a understanding about the sector they do work in. Its not by magic that they are in the position they are in it due to years of speculation and understanding among other things.

[SKYMTL]

No, Chalie threw around many numbers including conflicting numbers in a number of this articles.

This one says two things: "well under 20%" and "well under 2%".

http://www.semiaccurate.com/2010/03/...hacked-gtx480/

If you throw enough darts at a dart board, you will eventually hit the bulls eye. Charlie may have sources (there is no doubt about that) but he embelishes the info they give him far too much.

[Eastcoasthandle]

No, Chalie threw around many numbers including conflicting numbers in a number of this articles.

This one says two things: "well under 20%" and "well under 2%".

http://www.semiaccurate.com/2010/03/...hacked-gtx480/

If you throw enough darts at a dart board, you will eventually hit the bulls eye. Charlie may have sources (there is no doubt about that) but he embelishes the info they give him far too much.

Perhaps, just perhaps, it was the case then vs now where there is a slight improvement. Honestly, going from under 20% back then to 20%-30% now does sound reasonable. I really don't see the need where he has to be accurate for what may have been the case back then to "it must be the case now". Or else he's "throwing darts...".

On another note according to the article Nvidia was downgraded to "hold". That's not good at all. Looking at the ticker their stock went from over $17 to $16.62. Apparently having a negative impact on them even though they are at "hold" instead of "sell". But time will tell if that trend continues.

[Macadamia]

The majority of nVidia's VLSI people must have left the company to think of it.

Tegra 1 taking 1+ year to get into a product
Tegra 2 now taking ~1 year to get into a product after delays
GT21x being delayed for 4-6 months when it was supposed to be released before RV740 (4770)
GF100 process trainwreck
GF104 even taking that long to come out by virtue

I think management pissed off the wrong department unfortunately.

[Helmore]

I think management pissed off the wrong department unfortunately.

What do you mean by that?

[metalop1g]

and you think that financial analysts would know this information? They probably read Charlie in the first place

I bet you that financial analysts know a lot LESS about the GPU market than the average XS reader! I am not saying this is not true, but making a story about a financial analyst claim is laughable.

Typical syndrom often seen around here : People always think they know better than insiders. There's actually a name for this syndrom, I read about it a while ago..

[RejZoR]

I could wait for mid end GeForce for like another 6 months or i could just buy HD5850 and be happy. Well, i did and i'm happy. Supply of GeForce cards is really horrific.

[Helmore]

Typical syndrom often seen around here : People always think they know better than insiders. There's actually a name for this syndrom, I read about it a while ago..

Now you must know more about this syndrome than the actual insiders .

[stinkas]

If you throw enough darts at a dart board, you will eventually hit the bulls eye.

Stop copying the SemiAccurate logo for your metaphors!

Anyway hitting the bulls eye with 2 darts is pretty impressive...

I can understand why people who favour Nv don't like him much, and also some people who don't happen to favour Nv and simply don't like bias.

The sting comes of course when he is proved 'SemiAccurate'!

For me at least his articles are the most entertaining tech based articles I'v ever read, and it's Nv who supply the material by the lorry load.

What I like about Charlie is the fact that while obviously biased he does not pretend to be unbiased, kind of refreshing to see such honesty.

Let me ask a question to the General forum...
Who would you trust more to tell you the honest (known) truth about Nv products from the following two choices.

1) Charlie
2) Nvidia

[SKYMTL]

Perhaps, just perhaps, it was the case then vs now where there is a slight improvement.

The article I linked was written 23 DAYS ago. Not months ago. Going from "well under 20%" and "well under 2%" to 20-30% is a massive jump in such a short time considering the well-documented issues TSMC is having with this chip.

Still, this thread isn't about Charlie. It should be about the problems NVIDIA may now have in the investment community considering the overly bullish outlook some had. A target of $22? Seriously?

[stinkas]

What do you mean by that?

Maybe that Management p'd off the engineering Dep.

[Chickenfeed]

Now you must know more about this syndrome than the actual insiders .

I see what you did there...

But yes this was to expected. If they had said higher than 40% I'd would be extremely skeptical.

[Eastcoasthandle]

The article I linked was written 23 DAYS ago. Not months ago. Going from "well under 20%" and "well under 2%" to 20-30% is a massive jump in such a short time considering the well-documented issues TSMC is having with this chip.

Still, this thread isn't about Charlie. It should be about the problems NVIDIA may now have in the investment community considering the overly bullish outlook some had. A target of $22? Seriously?

Oh so you think that news posted 23 days ago had to be for that day and thar day only. Got it. Perhaps it was regarding information prior to the date it was published from him. But this thread really isn't about him, I find it interesting that Nv was placed on a "hold" status. Only time will tell how that will impact them.

[xVeinx]

Analysts are subject to the same sort of misconceptions as anyone else. While they do generally have good contacts and are well paid to know what is going on within a particular industry, their perceptions of what a company needs to be doing can vary. Tech companies have their stocks fluctuate wildly based on whether or not their latest creation is perceived to be a hit with consumers, or whether it is advancing far enough ahead of the competition. Analysts and investors are skittish right now, looking to see large profits and high volume among other things. Nvidia has had good quarters recently, but the stock market is based on the future, not the past. If an analyst thinks rough times are ahead, they'll downgrade a companies stock pretty quick.

[kemo]

Xbitlabs have been biased towards ATI ever since there HD4890 , that is why i wasn't surprised when they showed there 480GTX review almost 10 days later than everyone else

[Splave]

^ couldnt get one because of the 20-30% yields?

[MrMojoZ]

Xbitlabs have been biased towards ATI ever since there HD4890 , that is why i wasn't surprised when they showed there 480GTX review almost 10 days later than everyone else

So Xbitlabs made this up because they are ATI biased, there was no hold placed on Nvidia stock. I'm glad you cleared that up.

[bigKr33]

Man I hope nvidia improves their cards. I'm not an nvidia fanboy, but I really do enjoy their products. Then again both gpu manufacturers goes through a rough series in video cards.

[Piotrsama]

No, Chalie threw around many numbers including conflicting numbers in a number of this articles.

This one says two things: "well under 20%" and "well under 2%".

http://www.semiaccurate.com/2010/03/...hacked-gtx480/

The article I linked was written 23 DAYS ago. Not months ago. Going from "well under 20%" and "well under 2%" to 20-30% is a massive jump in such a short time considering the well-documented issues TSMC is having with this chip.

Are you misreading his article on purpose?
He's telling the whole GF100 story.
From the 2% yields for the hot lots (linking his sept 2009 article), to under 20% for production wafers (march 2010 numbers).

On another note according to the article Nvidia was downgraded to "hold". That's not good at all. Looking at the ticker their stock went from over $17 to $16.62. Apparently having a negative impact on them even though they are at "hold" instead of "sell". But time will tell if that trend continues.

Possible solution: Rename NVDA to something else.... should work... for some time.

The majority of nVidia's VLSI people must have left the company to think of it.

Tegra 1 taking 1+ year to get into a product
Tegra 2 now taking ~1 year to get into a product after delays
GT21x being delayed for 4-6 months when it was supposed to be released before RV740 (4770)
GF100 process trainwreck
GF104 even taking that long to come out by virtue

I think management pissed off the wrong department unfortunately.

I would call it failure to execute.
Like Intel during the final Netburst times.
They'll probably recover, just as Intel did.

Xbitlabs have been biased towards ATI ever since there HD4890 , that is why i wasn't surprised when they showed there 480GTX review almost 10 days later than everyone else

Or maybe they were cut out from the initial review...
Wouldn't be the first time Nvidia does it, they even cutted out Anandtech when necessary.
http://www.anandtech.com/show/2731/2
This part was fun:

Perhaps NVIDIA was afraid we'd point out that it was nothing more than a 9800 GTX+ that ran a little cooler. Or perhaps we haven't been positive enough about CUDA and PhysX and NVIDIA was trying to punish us.

[annihilat0r]

piotrsama: 20-30% is still "well above" the "well below 20%" figure, don't you think?

Actually 20-30% don't sound too bad to the ear considering the gigantic chip that runs superh ot, and the rumors that we had been hearing. I am inclined to think this means well for GF104.

[Manicdan]

Possible solution: Rename NVDA to something else.... should work... for some time.

like Xfinity, worked for Comcast right!?

[kemo]

So Xbitlabs made this up because they are ATI biased, there was no hold placed on Nvidia stock. I'm glad you cleared that up.

They didnt made it up , they are just more willing to pass Information

Or maybe they were cut out from the initial review...
Wouldn't be the first time Nvidia does it, they even cutted out Anandtech when necessary.
http://www.anandtech.com/show/2731/2
This part was fun:

The HD4890 article was a reason for cutting for sure , and as far as i know they didnt cut anand on Fermi , they would Cut [H] before anand