Nvidia Kepler Yields Lower Than Expected –CEO. Fermi 2.0?

[sontin]

One can generally expect yields to be lower early in a new process or node. However, at any point in time, if yields are lower than expected, it says something about your ability to make realistic projections. Fermi also appeared to suffer from overly optimistic expectations. Is there a trend here?

nVidia is in production right now. That's the reason why they guided their margin down. It's affect their financial business not their ramping process. They have build inventory over q4 with Kepler chips. They are shipping them for revenue in this quarter.

BTW: Fermi has better yields than they expected:

David White
Yes. Kevin, I think in the prior comments, and in the CFO commentary, we'd explained it really is associated primarily with GTX 470 and GTX 480. We ramped that product. We were favorably surprised with nicer yields than what we anticipated.

http://seekingalpha.com/article/205057-nvidia-q1-2011-earnings-call-transcript?part=qanda

 

[maddie]

Actually nVidia blamed nVidia for Fermi.

Eventually they did.

 

[blackened23]

Sontin, I think its pretty well known that Fermi yields were *horrible* at the time of initial release. Obviously over time that became less of an issue - as the process matured, yields improved. However, there will major yield problems at the start, Charlie reported that they had a defective rate of 60%+ per wafer (obviously I can't verify this). NV later switched to per working chip pricing to offset the costs of poor yields.

I'm sure yields are fantastic now, because the process has matured over time...it definitely was not good at the start.

 

[Keysplayr]

do you know about basic economi right ?

there are no competition, so its stupid if AMD release it like HD 5XXXX series. to be honest i like this new AMD CEO, he can prove to their shareholder that AMD can be aggressive too. and just with this move AMD share is increase more than 50%. so although its not good for consumer, its definitely good for AMD. and maybe if they gather enough profit, they can competitive again in CPU business.

if you want to blame, blame nvdia because they late again.

Coming from a stockholder, that sentence makes perfect sense. Coming from a consumer
That sentence is taboo. Which are you? If you're a stockholder, you make sense. If not, you'll get looked at cross-eyed.

 

[sontin]

Sontin, I think its pretty well known that Fermi yields were *horrible* at the time of initial release.

And it was pretty well known for a while that the earth was flat.

They had no yield problem with Fermi. They had a design issue to fix before they could started the ramping process.

 

[SirPauly]

Eventually they did.

I don't think I have ever read officially nVidia blaming TSMC for anything -- just Internet lore. You do realize nVidia has an outstanding relationship with TSMC; as does Jen-Hsun Huang with Morris Chang.

An Example:

http://www.youtube.com/watch?v=u-x7PdnvCyI

 

[blackened23]

And it was pretty well known for a while that the earth was flat.

They had no yield problem with Fermi. They had a design issue to fix before they could started the ramping process.

I'm sorry, but no, there were way too many sources reporting that Fermi yields were terrible. Mr CEO himself even mentioned it in an old financial release in 2010.

I'm not sure what this has to do with the topic at hand, but unless 200 news sources *and* Mr. CEO Jensen lied to us fermi yields were bad at release. This is pretty obvious considering that NV negotiated for a "per functional chip" price with TSMC. Normally you pay for the entire wafer regardless.

 

[maddie]

nVidia is in production right now. That's the reason why they guided their margin down. It's affect their financial business not their ramping process. They have build inventory over q4 with Kepler chips. They are shipping them for revenue in this quarter.

BTW: Fermi has better yields than they expected:

http://seekingalpha.com/article/205057-nvidia-q1-2011-earnings-call-transcript?part=qanda

Is this some sort of humour?

Do you honestly believe that statement, or are you provoking an argument?

Read the initial link. Nvidia themselves claim that yields are lower that expected. If, of all people, they know what's happening, certainly it will also affect the ability to build inventory and ability to ship in projected volumes.

 

[sontin]

I'm sorry, but no, there were way too many sources reporting that Fermi yields were terrible. Mr CEO himself even mentioned it in an old financial release in 2010.

Show me this that Fermi had yield problems.
Because David White and the same Mr. CEO Himself said other things...

Is this some sort of humour?

Do you honestly believe that statement, or are you provoking an argument?

Read the initial link. Nvidia themselves claim that yields are lower that expected. If, of all people, they know what's happening, certainly it will also affect the ability to build inventory and ability to ship in projected volumes.

They are ramping. The designs are finished. They will have lower volume to ship. But their missed estimation will not affect their launch timeframe. We have absoluty no clue what their guideline for yield was back in september. Maybe 60%? Or 30%?

Here their statement that they are ramping since q4:

Inventories at the end of the quarter were $340.3 million, up 6.5 percent from $319.6 million in the prior quarter, and down 1.5 percent from $345.5 million in the prior year. Our DSI at quarter-end was 67 days, up 10 days from 57 days in the prior quarter. Our DSI for fiscal years 2012 and 2011 was 64
days and 59 days, respectively. Our DSI increased as we ramped production builds for new generation products and transitioned to 28nm.

 

[blackened23]

http://www.xbitlabs.com/news/graphi...t_20_30_Further_Delays_Possible_Analysts.html

“We are downgrading Nvidia to a “hold” and removing our 12-month price target (was $22) after a series of channel checks indicating that Fermi is not ramping well and there could be further product delays. Based on our findings, Nvidia has very limited supply of Fermi desktop/notebook parts and yields remain poor at around 20-30%,” analyst Rajvindra Gill wrote in a note to clients.

 

[AnandThenMan]

IF one listens to the conference call, there is actually praise for TSMC from nvidia.

I read the transcript (not sure if it is the full call however) but have not listed to it. Link?

There are problems from nVidia with 28nm and clearly offered that their yields are lower than expected and would like more 28nm wafers.

Lower yields are fine, that just means less Kepler chips. Right now there are zero that I can buy.

Coming from a stockholder, that sentence makes perfect sense. Coming from a consumer
That sentence is taboo. Which are you? If you're a stockholder, you make sense. If not, you'll get looked at cross-eyed.

I own both AMD and Nvidia stock, have held NVDA for a really long time, I think about 12 years. So I'm both, and I can say for certain that what some may think affects the stock price, things like poor yields, late to market etc. rarely have a short term affect on price. I'm not going to get into what the trends are and how I know when to buy/sell, but it has almost nothing to do with the types of issues we discuss here. In fact many times issues that you are sure will drive the stock down end up doing the opposite.

 

[BenSkywalker]

This horrible situation as it is known by nV at this point is expected to impact their margins all the way down to 49%, that's only ~10%(4% absolute) higher then AMD, and increase their operating expenses by an awe inspiring 4%......

As a business report- yield issues explain why operating expenses are going to increase and why margins are going to see a decline. nVidia could still produce millions of GPUs with yields below expectations, it would simply impact their margins.

BTW- Those expecting something like another six months, the parts have broad based design wins which means they have already been sampled at a high enough rate to ship parts to tier two/tier three OEMs(in other words, they are on a production spin).

How good the part will be is still up in the air. If nV was shooting for a 800mm part clocked at 3.4GHZ and their yields are 'only' 85%, it wouldn't by default make anything stated in the comments inaccurate. Obviously those are absurd numbers, but until we know *what* the specs are that nV is shooting for commenting on yield issues is largely meaningless- particularly when we have no clue what AMD is yielding at. For the record- nVidia was profitable with the "horrible" yielding Fermi at price points roughly matching what AMD is charging for their current line up of parts, and nV had a considerably larger die. Maybe AMD is seeing sub Fermi yields? Who knows, with how low their margins are expected to be, it's not like they would need to explain any hiccups in revenue/expense as a company like nVidia or Intel need to do.

 

[sontin]

http://www.xbitlabs.com/news/graphi...t_20_30_Further_Delays_Possible_Analysts.html

Analyst. That's not a source. And their predictation for the second half was wrong. Fermi increased their margin after they write off their 10.1 low-end.

 

[blackened23]

Analyst. That's not a source. And their predictation for the second half was wrong. Fermi increased their margin after they write off their 10.1 low-end.

Good grief man, what do you want me to show you? Just use google and you can make your own decision. There are TONS OF OUTLETS that reported low Fermi yields. There is one marketing guy at NV that denied it which i'm sure is a reliable source (NOT). I'd rather believe 200 news sources + CEO himself. I have a feeling that you wouldn't believe Jen-Tsun Huang if he showed up at your doorstep to tell you. (I kid)

 

[SirPauly]

I read the transcript (not sure if it is the full call however) but have not listed to it. Link?

Lower yields are fine, that just means less Kepler chips. Right now there are zero that I can buy.

I own both AMD and Nvidia stock, have held NVDA for a really long time, I think about 12 years. So I'm both, and I can say for certain that what some may think affects the stock price, things like poor yields, late to market etc. rarely have a short term affect on price. I'm not going to get into what the trends are and how I know when to buy/sell, but it has almost nothing to do with the types of issues we discuss here. In fact many times issues that you are sure will drive the stock down end up doing the opposite.

From the Q/A from the transcript:

And this is a world-class company. And Morris and the management team are all over it, and they're doing everything in their power, and if history is any indication and it tends to be, they're going to work their way into a fabulous node here any time.

 

[garagisti]

@Keysplayer

The cheapest GTX480 on newegg...
http://www.newegg.com/Product/Product.aspx?Item=N82E16814162073

The cheapest 7950, which kicks Nvidia in the groin...
http://www.newegg.com/Product/Produ...1&name=Radeon HD 7950&Order=PRICE&Pagesize=20

psst... AMD/ ATI is selling chips performing better than Nvidia's flagship, at a lower price!

7970 are about ($70) more expensive, but they're easily overclocking 200 Mhz over stock and getting another 25-30% jump in performance.

So really, who's shafting who?

p.p.s: you should take a couple of minutes to do your homework!

EDIT: personally, i'd have liked a lower price, since i was to build a new pc... but then, nvidia with lower performance costs the same. Why should AMD take a hit to their bottomline?

 

[Grooveriding]

Disagree with the above. For the majority of posters here the financial situation of nvidia or AMD and related metrics amount to a whole pile of nothing. Unless there is something I'm not aware of, does nvidia's share price currently give me more FPS if I use an nvidia card, if so - please enlighten me. Yields and process issues mean a lot to PC enthusiasts and gamers who want to see, compare and purchase new hardware.

When yields are poor and things are not going well on a new process for nvidia, it means we wait to see the product on the shelves and possibly deal with a space heater. GTX 480/470 - Fermi is a shining example of that. So they are pretty relevant to most of the posters in this subforum, this is not a financial market and analysis subforum last time I checked, there are a just a few posters who are very enthusiastic about a company's market worth.

It's also plain obtuse to take what the CEO and upper management of a publicly traded company say at face value. Everything they say publicly is spun and crafted not to affect their value, so using comments from them to say there were not issues with 40nm for nvidia is laughable. They had massive issues with 40nm at the outset. No one is saying their current 28nm issues are as bad, time will tell on that one.

All we know is as yet they are still behind the eight ball and don't have any 28nm products out and their current high end offerings available to purchase are all obsolete against their competition's offerings.

 

[SirPauly]

This horrible situation as it is known by nV at this point is expected to impact their margins all the way down to 49%, that's only ~10%(4% absolute) higher then AMD, and increase their operating expenses by an awe inspiring 4%......

That's the amazing part is nVidia offers a gauge what to actually expect and 49.2 percent for margins is still impressive and as 28nm improves, nVidia may be back around 52 percent through the year.

 

[sontin]

Good grief man, what do you want me to show you?

We can start with the statement made by huang that they had yield problems with fermi after they ramped it into production.

 

[SirPauly]

Disagree with the above. For the majority of posters here the financial situation of nvidia or AMD and related metrics amount to a whole pile of nothing. Unless there is something I'm not aware of, does nvidia's share price currently give me more FPS if I use an nvidia card, if so - please enlighten me. Yields and process issues mean a lot to PC enthusiasts and gamers who want to see, compare and purchase new hardware.

When yields are poor and things are not going well on a new process for nvidia, it means we wait to see the product on the shelves and possibly deal with a space heater. GTX 480/470 - Fermi is a shining example of that. So they are pretty relevant to most of the posters in this subforum, this is not a financial market and analysis subforum last time I checked, there are a just a few posters who are very enthusiastic about a company's market worth.

It's also plain obtuse to take what the CEO and upper management of a publicly traded company say at face value. Everything they say publicly is spun and crafted not to affect their value, so using comments from them to say there were not issues with 40nm for nvidia is laughable. They had massive issues with 40nm at the outset. No one is saying their current 28nm issues are as bad, time will tell on that one.

All we know is as yet they are still behind the eight ball and don't have any 28nm products out and their current high end offerings available to purchase are all obsolete against their competition's offerings.

Actually, they're everything.

 

[BenSkywalker]

It's also plain obtuse to take what the CEO and upper management of a publicly traded company say at face value. Everything they say publicly is spun and crafted not to affect their value, so using comments from them to say there were not issues with 40nm for nvidia is laughable.

This was reported during an earnings call. If they lie, they go to jail. It is criminal.

 

[SirPauly]

This was reported during an earnings call. If they lie, they go to jail. It is criminal.

It's amazing -- obtuse -- ignore exact statements in a conference call.

 

[AnandThenMan]

And this is a world-class company. And Morris and the management team are all over it, and they're doing everything in their power, and if history is any indication and it tends to be, they're going to work their way into a fabulous node here any time.

I don't even know what this is supposed to mean.

This was reported during an earnings call. If they lie, they go to jail. It is criminal.

Give me a break, unless you blatantly lie through your teeth and it's proven that you lie, nothing happens. I have heard so many suspect statements on the record over the years it's not even funny. Every company does it.

 

[SirPauly]

Analyst. That's not a source. And their predictation for the second half was wrong. Fermi increased their margin after they write off their 10.1 low-end.

What can you do?

 

[SirPauly]

I don't even know what this is supposed to mean.

Giving Praise to TSMC.