So, ATI is doomed..?
- Thread starter IbaneZ
- Start date
Ratchet is just whining because no one at ATi will return his e-mails lately. 
overclocked_enthusiasm
Regular
When people make the comparison between R520 and NV30 I think they miss the point. During the NV30 fiasco, NVDA gross margins only fell by 30 basis points sequentially when they had that huge warning and the stock cratered to $9. Their topline revenue number however REALLY dropped. ATI was seen as finally gaining share...and rightly so.
With the most recent quarter reported by ATI, gross margins fell a staggering 520 basis points in one quarter. In addition to the huge loss in gross margins, ATI lost revenue as well. So ATI has a much bigger problem on their hands. They MUST get a product out that competes with Nvidia right now and it MUST be of high enough margins to get back to the company's forcasted range of 34-38%. Right now ATI is at 28.9% and Nvidia is at 37.8%. That is almost a 9% difference in gross margins. Multiple that by $550 million in quarterly revenues and you have a MAJOR problem. The real issue for ATI is gross margins first and revenues second.
Here is the rub. Since ATI is late to market, NVDA can charge whatever it wants for the 7800 series. This gives Nvidia the pricing power they need to have fantastic gross margins initially. Nvidia can build this surplus in gross margins while their market has no competition AND they are given the further benefit of time in the manufacturing process which ALWAYS improves yields and futrther enhances gross margins. ATI has neither of those 2 gross margin building options. ATI will have no initial pricing power and no time invested in the manufacturing process to improve yields/gross margins.
NVDA conversly can afford to lower their ASP IF ATI plays hardball with price and still maintain solid margins. ATI would take a hit to the margins they desperately need right now OR sacrafice market share to more cost effective parts from Nvidia. At some point, ATI will have to turn a profit again and they cannot continue to sacrifice gross margins on the alter of market share because the CEO will be lynched. Sacraficing margins for market share only makes sense if your product can gain traction and uproot the competition. There is no sign of the 6800/7800/SLI market being vulnerable to a knockout punch from ATI.
In sum, yes ATI is screwed for the near future. ATI's presumed advantage on 90 nm is fading fast as NVDA said in the CC on Thursday that "we have 2 chips in production currently based on 90 nm". WTF happened to the blueprint for R3xx which used a more mature process and lower clocks? WTF happened to starting a new process with a midrange card? R420 and R520 have both been plagued by availability issues and yield issues because they used the most bleeding edge process and higher clocks. NV30 was guilty of that as well. You know who learned from that? Nvidia did with NV40 and G70. G71 looks to follow that path as well. ATI forgot the roots of their success and got too big for their britches. Someone has a hardon for bleeding edge processes there and it is going to kill them if they aren't careful.
CEO=engineer is a dangerous mix sometimes...
With the most recent quarter reported by ATI, gross margins fell a staggering 520 basis points in one quarter. In addition to the huge loss in gross margins, ATI lost revenue as well. So ATI has a much bigger problem on their hands. They MUST get a product out that competes with Nvidia right now and it MUST be of high enough margins to get back to the company's forcasted range of 34-38%. Right now ATI is at 28.9% and Nvidia is at 37.8%. That is almost a 9% difference in gross margins. Multiple that by $550 million in quarterly revenues and you have a MAJOR problem. The real issue for ATI is gross margins first and revenues second.
Here is the rub. Since ATI is late to market, NVDA can charge whatever it wants for the 7800 series. This gives Nvidia the pricing power they need to have fantastic gross margins initially. Nvidia can build this surplus in gross margins while their market has no competition AND they are given the further benefit of time in the manufacturing process which ALWAYS improves yields and futrther enhances gross margins. ATI has neither of those 2 gross margin building options. ATI will have no initial pricing power and no time invested in the manufacturing process to improve yields/gross margins.
NVDA conversly can afford to lower their ASP IF ATI plays hardball with price and still maintain solid margins. ATI would take a hit to the margins they desperately need right now OR sacrafice market share to more cost effective parts from Nvidia. At some point, ATI will have to turn a profit again and they cannot continue to sacrifice gross margins on the alter of market share because the CEO will be lynched. Sacraficing margins for market share only makes sense if your product can gain traction and uproot the competition. There is no sign of the 6800/7800/SLI market being vulnerable to a knockout punch from ATI.
In sum, yes ATI is screwed for the near future. ATI's presumed advantage on 90 nm is fading fast as NVDA said in the CC on Thursday that "we have 2 chips in production currently based on 90 nm". WTF happened to the blueprint for R3xx which used a more mature process and lower clocks? WTF happened to starting a new process with a midrange card? R420 and R520 have both been plagued by availability issues and yield issues because they used the most bleeding edge process and higher clocks. NV30 was guilty of that as well. You know who learned from that? Nvidia did with NV40 and G70. G71 looks to follow that path as well. ATI forgot the roots of their success and got too big for their britches. Someone has a hardon for bleeding edge processes there and it is going to kill them if they aren't careful.
CEO=engineer is a dangerous mix sometimes...
I personally think most of their current woes are down to the fact that they're designing 2 console GPUs that need to be done before Nvidia's 1 console GPU.
If you look at Nvidia's numbers without the Xbox they wouldnt look very good. They're going to have to contend with this for well over a year while ATi is making money with the Xbox 360 and the Ps3 isnt out yet.
I think ATi in the short term has bitten off more than they can chew but its not nearly as gloomy as some are trying to paint.
If ATi comes storming back, Nvidia will not get a free pass like the last time they got whipped. I still cant believe the handjob Nvidia got from most of the online press for what proved to be a totally subpar set of cards.
If you look at the requirements for Battlefield 2, it becomes clear that ATi was far, far ahead of Nvidia. The 8500 is a useable card in that game but the Geforce 4 is not.
If you look at Nvidia's numbers without the Xbox they wouldnt look very good. They're going to have to contend with this for well over a year while ATi is making money with the Xbox 360 and the Ps3 isnt out yet.
I think ATi in the short term has bitten off more than they can chew but its not nearly as gloomy as some are trying to paint.
If ATi comes storming back, Nvidia will not get a free pass like the last time they got whipped. I still cant believe the handjob Nvidia got from most of the online press for what proved to be a totally subpar set of cards.
If you look at the requirements for Battlefield 2, it becomes clear that ATi was far, far ahead of Nvidia. The 8500 is a useable card in that game but the Geforce 4 is not.
overclocked_enthusiasm
Regular
karlotta said:yeah right.
Let me clarify. It seems to me that Dave Orton's infatuation with technology and bleeding edge processes has gotten the better of him. A CEO needs to be more of an executive and less of an engineer. I cannot otherwise explain big gambles such as with PCIe (excluding AGP in early R420), 130 nm low-k and now 90 nm with R520. Again, what happened to the conservative approach that turned ATI around with R3xx?
Last edited by a moderator:
Given what was said to me by Orton when I interviewed him back at the launch of X800 I think it is clear that the engineering indicators for how ATI would implement an SM3.0 architecture said that 90nm would be the only viable option for them.
overclocked_enthusiasm
Regular
Dave Baumann said:
I certainly don't doubt the predisposition at ATI to use 90 nm for their vision of SM 3.0. What is unclear to me is how Nvidia is able to implement SM 3.0 on both 130 nm and 110 nm while ATI feels compelled to use 90 nm? Are the engineers at Nvidia more economicle with the space dedicated to SM 3.0 than ATI or is ATI overly aggresive in their SM 3.0 "plus" approach that requires the larger die size?
For whatever reason, the conservative approach that worked so well at ATI has been replaced by what I would call overly ambitious engineering.
Last edited by a moderator:
Comparing one against the other is never going to lead to anything conclusive. All that can be said is that they have different design teams with different architectures and different goals at the times when these roadmaps were set in place.
Jawed
Legend
I like to think (foolish hope!) that R520 will do (PS) dynamic branching properly (i.e. batch sizes will be fairly small making dynamic branching actually worth using).
But sadly, NV40/G70's entirely hopeless (PS) dynamic branching means that no developers will bother, anyway.
Jawed
But sadly, NV40/G70's entirely hopeless (PS) dynamic branching means that no developers will bother, anyway.
Jawed
overclocked_enthusiasm
Regular
Mordenkainen said:
Based on Nvidia's recently reported gross margins of 37.8%, where they specifically mentioned 7800 GTX as one of the prime drivers of gross margins, I would say that Nvidia is getting good yields. Granted the ASP on that SKU is very high right now with no competition. The yields on NV40 seemed to be quite good as well because Nvidia was able to expand gross margins for the past 14 months while ATI had stalled, slipped and then tanked last quarter to 28.9%.
Nope, they're doomed. It's official, might as well just consider them out of the graphics game altogether.
Bouncing Zabaglione Bros.
Legend
Folks, as I said at the beginning, it's been only six/seven weeks since G70 launched. No one is going to win anything in that time. Even if ATI launch next month or the month after, the time difference is negligable.
If I said that ATI and Nvidia were launching their next-gen chips within 12 weeks of each other, no one would bat an eyelid, and no one would predict a winner based on who shipped before the other.
It's just silly to suggest ATI is doomed, or even hurting until we know whether R520 and it's derivatives are going to be a good product or when it's going to arrive. One company was always going to ship a few weeks before the other - that was never going to decide the best product.
If I said that ATI and Nvidia were launching their next-gen chips within 12 weeks of each other, no one would bat an eyelid, and no one would predict a winner based on who shipped before the other.
It's just silly to suggest ATI is doomed, or even hurting until we know whether R520 and it's derivatives are going to be a good product or when it's going to arrive. One company was always going to ship a few weeks before the other - that was never going to decide the best product.