Jon Peddie Q2 graphics market share report

[kemosabe]

http://www.jonpeddie.com/about/press/MarketWatch_Q205.shtml

Mercury numbers are also out, though I have yet to see a detailed report.

 

[Skrying]

Uhh, very intersting to say the least. These numbers seem totally opposite what I would have thought, especially for the discrete graphics sector. I would have thought Nvidia would have gained market share with the 6x00 cards, not lost.

 

[kemosabe]

ATI's market share penetration was largely limited to its Xpress 200 integrated chipsets and low-end discrete (X300) parts in the PCI-E OEM segment, which is why its profit margins tumbled. Meanwhile, NVDA has a smaller slice of the overall pie but is selling a lot more higher-margin performance chips.

 

[Xmas]

JPR estimates that approximately 14.9 mobile graphics devices shipped in Q2'05

On the discrete side, ATI’s segment share remained essentially flat at 55.4% while Nvidia’s segment share fell from 46.8% to 46.1% during the period.

Interesting numbers

 

[PatrickL]

If you remember past years, we saw almost a 12 months gap between the feelings we had reading theses boards and their impacts in JPR numbers.

 

[WaltC]

If you remember past years, we saw almost a 12 months gap between the feelings we had reading theses boards and their impacts in JPR numbers.

Feelings don't matter a whole lot and perceptions can be wrong--you know--such as the obviously somewhat wide-spread "feeling" that for the past quarter "ATi hasn't been shipping anything and nV has been shipping everything." Heh... Seems not to have been the case at all, apparently.

 

[kemosabe]

And here are some Mercury numbers, compliments of Xbit.

Very much as expected.

 

[Mintmaster]

Wow, I had no idea that ATI had grown its integrated chipset market share so much. Good for them.

 

[Sxotty]

So anything above a 6200 from Nvidia is a performance part? Weird, I would have put a 6600 in mainstream.

 

[Diplo]

Q2 graphics? Do people still use that engine? I had no idea...

 

[Bjorn]

Feelings don't matter a whole lot and perceptions can be wrong--you know--such as the obviously somewhat wide-spread "feeling" that for the past quarter "ATi hasn't been shipping anything and nV has been shipping everything." Heh... Seems not to have been the case at all, apparently.

As has already been mentioned, increases in the market share are not necesserily the most important thing. It's also a question of what you sell, and then there's a little thing called margins also.

And with regards to feelings and perceptions, well, most people on these forums are probably thinking mainstream/highend in this case where i doubt that Ati has gained any market share.

 

[Bob]

Feelings don't matter a whole lot and perceptions can be wrong--you know--such as the obviously somewhat wide-spread "feeling" that for the past quarter "ATi hasn't been shipping anything and nV has been shipping everything." Heh... Seems not to have been the case at all, apparently.

I would normally agree with you, but in this case, ATI themselves seem to at least partially disagree:

Inventory levels of $456 million at the end of the third quarter increased from $255 million at fiscal 2004 year-end. Approximately two-thirds of inventory in the PC segment consisted of PCI Express goods versus one-third AGP. Inventory in terms of days sales outstanding was above our internal target of 60 days. We expect inventory levels in dollar terms to decline in the fourth quarter but to remain above our 60-day target.

Although this doesn't mean that ATI didn't sell anything last quarter (far from it), they don't seem to have been outsold either. Half a billion dollars of inventory isn't negligeable. It's a whole quarter worth of revenue.

 

[Richthofen]

Although this doesn't mean that ATI didn't sell anything last quarter (far from it), they don't seem to have been outsold either. Half a billion dollars of inventory isn't negligeable. It's a whole quarter worth of revenue.

at least their revenue and profit numbers from the last quater were pretty bad. They lost nearly 80 mio in revenue. I think that says enough about how much accuracy those market share numbers have.
I still think they have maintained or gained a little in the overall market share but in all desktop segments beginning from the 6600 i think they lost market share otherwise they wouldn't have lost 80 mio revenue in a 3 month timeframe.

 

[Geeforcer]

Feelings don't matter a whole lot and perceptions can be wrong--you know--such as the obviously somewhat wide-spread "feeling" that for the past quarter "ATi hasn't been shipping anything and nV has been shipping everything." Heh... Seems not to have been the case at all, apparently.

Well, this is a breakdown on the entire graphics chip market by units shipped apparently. It doesn't cover the "3D chip market" separately, or focus on that section of the market which would buy a 7800 GTX or X850 XE, etc., for the purpose of running a 3D game. Intel, for instance, is completely invisible in the 3D marketplace. It weighs a $10 gpu the same as a $100 gpu, and so obviously is not a reflection on the percentage of financial gain, either. Where the 7800 GTX would have impacted Nvidia's financial bottom line enormously would be if ~10% or so of its 18% total gain consisted of 7800 GTX sales. That would represent quite a bit of money, which is what the companies are interested in primarily, of course.

 

[WaltC]

As has already been mentioned, increases in the market share are not necesserily the most important thing. It's also a question of what you sell, and then there's a little thing called margins also.

And with regards to feelings and perceptions, well, most people on these forums are probably thinking mainstream/highend in this case where i doubt that Ati has gained any market share.

This is really sort of amusing when I think back to the years between nV25 and nV40, in which nV proponents were assuring me that the only thing that did matter at all was the high-volume, low-performance markets the GF5200 was said to dominate--it being a "DX9" part and all... Wonder where that sentiment has gone today...?

 

[WaltC]

Well, this is a breakdown on the entire graphics chip market by units shipped apparently. It doesn't cover the "3D chip market" separately, or focus on that section of the market which would buy a 7800 GTX or X850 XE, etc., for the purpose of running a 3D game. Intel, for instance, is completely invisible in the 3D marketplace. It weighs a $10 gpu the same as a $100 gpu, and so obviously is not a reflection on the percentage of financial gain, either. Where the 7800 GTX would have impacted Nvidia's financial bottom line enormously would be if ~10% or so of its 18% total gain consisted of 7800 GTX sales. That would represent quite a bit of money, which is what the companies are interested in primarily, of course.

A few years ago it certainly paid to distinguish the 2d chip markets from 3d, because a few years ago you could readily separate and identify the 2d chips from the 3d chips. But today they are all 2d/3d chips, at least to some degree, AFAIK. Certainly, this trend will only increase. But at least it is heartening to note that nV supporters no longer believe the high end is neglible--that's something, anyway, so I believe we are making progress...

 

[WaltC]

I would normally agree with you, but in this case, ATI themselves seem to at least partially disagree:


Although this doesn't mean that ATI didn't sell anything last quarter (far from it), they don't seem to have been outsold either. Half a billion dollars of inventory isn't negligeable. It's a whole quarter worth of revenue.

My own thought from the outset was that ATi was jumping on the PCIe bandwagon too rapidly, and the fact that two thirds of the inventory cited was excess PCIe stock indicates that I was right in saying that. But if you look at the "one-third" AGP figures of ~$130M in AGP inventory in comparison with the 2004 figure of $255M in inventory at the close of fiscal 2004--which by definition would be almost all AGP at the time--then ATi's total excess AGP inventory last quarter was ~half what it was at the end of 2004.

Since we know that ATi did not lose share but actually gained, then it's easy to see that they just manufactured a lot more this year than last in anticipation of a PCIe market that didn't grow as fast as they'd estimated. ATi's mistake was overbuilding for a PCIe rush that will be slower in coming than they thought--why they thought PCIe would mushroom so quickly is anybody's guess--but my guess would be that they simply overestimated Intel's influence in the 3d-market segments they serve.

Personally, I've been consistent in saying from the start that PCIe is a technology far better suited to the IGP markets than to the discrete 3d markets at present, for quite a few obvious reasons. I do not know why ATi was slow on the uptake there--as nV seemed to figure it out from the beginning as well by the use of bridge tech for PCIe support initially.

 

[WaltC]

at least their revenue and profit numbers from the last quater were pretty bad. They lost nearly 80 mio in revenue. I think that says enough about how much accuracy those market share numbers have.
I still think they have maintained or gained a little in the overall market share but in all desktop segments beginning from the 6600 i think they lost market share otherwise they wouldn't have lost 80 mio revenue in a 3 month timeframe.

What is "80mio" and where did you hear that ATi lost it?...

 

[Geeforcer]

A few years ago it certainly paid to distinguish the 2d chip markets from 3d, because a few years ago you could readily separate and identify the 2d chips from the 3d chips. But today they are all 2d/3d chips, at least to some degree, AFAIK. Certainly, this trend will only increase. But at least it is heartening to note that nV supporters no longer believe the high end is neglible--that's something, anyway, so I believe we are making progress...

Really? So unlike 2002, Intel *IS* visible in 3D marketplace? And when did they make 2d-only chips? When this post was written, Intel's presence in the marketplace was primarily I740 derivative, which has always been "2d/3d chips, at least to some degree". Seems to me that the only thing that really changed since then was your faith in relevance of Jon Peddie's numbers.

 

[Richthofen]

What is "80mio" and where did you hear that ATi lost it?...

well they missed their already 2 times cut down expectations for the last quater.
And their revenue was 78 Mio below the quater before and their profit stalled from around 60 Mio to a small loss. Further on they cut down their expectations for the remaining year.
Around 80 mio lower revenues in a 3 month timeframe is really big.
In contrast to that Nvidia was able to increase both - revenue and profits. They are now above ATIs numbers.

So while i think it is possible that ATI maintained overall marketshare its undeniable that they lost customers by a decent amount in price segments beginning at the GF6600GT level up to the ultra high end.