NVIDIA GF100 & Friends speculation

[Arun]

Your reasoning is valid if and only if their gross margins were 8%. But those are their operating margins. Gross profit is from selling products; it simply equals revenue minus product production cost. So if that semiaccurate GF104 BoM analysis was correct, NVIDIA would nearly certainly be selling GF104 at negative gross margins.

The operating profit equals gross profit minus operating expenses (i.e. employee salaries etc.) - which is of course always much lower. It is possible (and quite frequent) to have negative operating margins even though every single one of your products is selling at very high gross margins - simply because you aren't selling enough of them.

One case where it makes sense to say one product has negative operating margins is if you can dissociate its incremental operating cost from everything else fairly accurately. GF106's architecture is quite similar to GF104's and it taped-out afterwards, so it's probably possible to loosely determine the incremental R&D required to develop the chip. So even if the chip sells at a profit (i.e. the more NVIDIA sells of it, the more money they make) it's possible that the product as a whole isn't worth the incremental development cost because they can't recoup the investment over its lifetime.

Of course that is VERY difficult to estimate in reality - how much would GF104/GF108 have filled that gap if GF106 hadn't existed? And from a strategic perspective, that gap in the line-up would have meant higher revenue for AMD - which might mean they'll have more money to invest in long-term R&D, which might make their future architectures more competitive and reduce your future gross margins. So break-even might already be considered a success for a derivative chip like that.

If you confuse gross margins and operating margins, your reasoning will be extremely flawed. And that's before we even consider fun stuff like Gross Margins vs Gross Profit (e.g. improved framebuffer compression would save on DRAM costs, which means for a target BoM you can ask slightly more for your chip and maybe have a higher gross profit - but if it didn't increase in percentage terms as much as your costs, you might have lower gross margins... despite still making more money overall!)

 

[tannat]

Your reasoning is valid if and only if their gross margins were 8%. But those are their operating margins. Gross profit is from selling products; it simply equals revenue minus product production cost. So if that semiaccurate GF104 BoM analysis was correct, NVIDIA would nearly certainly be selling GF104 at negative gross margins.

The operating profit equals gross profit minus operating expenses (i.e. employee salaries etc.) - which is of course always much lower. It is possible (and quite frequent) to have negative operating margins even though every single one of your products is selling at very high gross margins - simply because you aren't selling enough of them.

One case where it makes sense to say one product has negative operating margins is if you can dissociate its incremental operating cost from everything else fairly accurately. GF106's architecture is quite similar to GF104's and it taped-out afterwards, so it's probably possible to loosely determine the incremental R&D required to develop the chip. So even if the chip sells at a profit (i.e. the more NVIDIA sells of it, the more money they make) it's possible that the product as a whole isn't worth the incremental development cost because they can't recoup the investment over its lifetime.

Of course that is VERY difficult to estimate in reality - how much would GF104/GF108 have filled that gap if GF106 hadn't existed? And from a strategic perspective, that gap in the line-up would have meant higher revenue for AMD - which might mean they'll have more money to invest in long-term R&D, which might make their future architectures more competitive and reduce your future gross margins. So break-even might already be considered a success for a derivative chip like that.

If you confuse gross margins and operating margins, your reasoning will be extremely flawed. And that's before we even consider fun stuff like Gross Margins vs Gross Profit (e.g. improved framebuffer compression would save on DRAM costs, which means for a target BoM you can ask slightly more for your chip and maybe have a higher gross profit - but if it didn't increase in percentage terms as much as your costs, you might have lower gross margins... despite still making more money overall!)

I see, thanks.

Actually I was only thinking in operating profit distributed over product lines. It looks like some of the products have been sold at a "operating loss". Ok, for individual product lines this may not be possible to deconvolute/estimate but the operating profit is the important number.

To what degree cards are sold at negative gross margin I had no intention to get out of these numbers. It's of course still possible. It just depends on what is highest priority. Short sight profit for each product line or preserved market position. It's not like any of the alternatives are "unhonorable" for nvidia, so it's no big deal except if it would be a long term situation.

 

[trinibwoy]

Actually I was only thinking in operating profit distributed over product lines. It looks like some of the products have been sold at a "operating loss". Ok, for individual product lines this may not be possible to deconvolute/estimate but the operating profit is the important number.

I hope by now that you realize the analysis you're trying to do here is pretty much in vain. It is nigh impossible to pro-rate a firm's operating expenses across its current product line either by volume or revenue. The numbers wouldn't have any valid meaning.

For example, if nVidia hires ten new engineers today to work on a 2012 product their salaries hit operating expenses but obviously they aren't costs related to the development or manufacturing of their current line of products. Not to mention all the other incidental costs of running a business - rent, utilities, legal fees etc that have no direct relationship to product development or sales. Just stick to gross margin, it's the only number that is relevant when talking about individual product profitability.

 

[tannat]

I hope by now that you realize the analysis you're trying to do here is pretty much in vain. It is nigh impossible to pro-rate a firm's operating expenses across its current product line either by volume or revenue. The numbers wouldn't have any valid meaning.

For example, if nVidia hires ten new engineers today to work on a 2012 product their salaries hit operating expenses but obviously they aren't costs related to the development or manufacturing of their current line of products. Not to mention all the other incidental costs of running a business - rent, utilities, legal fees etc that have no direct relationship to product development or sales. Just stick to gross margin, it's the only number that is relevant when talking about individual product profitability.

I don't see what you trying to say. Operational profit is related to the "profitability" of the products - granted a complex relationship when splitting the group of products up. That long term costs like development etc. impacts the profit a company makes goes without saying. These costs are real. It's tagging the cost to what the specific products carries that is impossible, and fortunately not critical in a long term or a specific product perspective.

In this example we don't need to do this however, It's enough to see that there are not really room for all the products to be profitable on their own. It's no big deal though.

 

[Silent_Buddha]

I don't see what you trying to say. Operational profit is related to the "profitability" of the products - granted a complex relationship when splitting the group of products up. That long term costs like development etc. impacts the profit a company makes goes without saying. These costs are real. It's tagging the cost to what the specific products carries that is impossible, and fortunately not critical in a long term or a specific product perspective.

In this example we don't need to do this however, It's enough to see that there are not really room for all the products to be profitable on their own. It's no big deal though.

Let's put it another way using another example.

AMD's has gross profit margins of 46%. So the products they sell certainly quite profitable in and of itself. However, once you bring in operating budget (everything from accounting, to building maintanaince, to utilities, etc.) and R&D costs, they ended up with only a 128M USD operating income on 1.62B USD revenue. Far far lower than the gross profit margin their products bring in. So minus operating costs and R&D, their products had a gross profit of ~745 million USD (very simplistic and naive view of it) yet only ended up with operating income of 128 million USD.

If you look at their graphics division. Year over year their revenue increased 33%, but their operating income declined from 2 million to 1 million. So they made more money but their operating costs increased far more leading to a Year over Year decline in profitability despite a Year over Year increase in revenue.

Blindly looking at the numbers someone might say something is wrong. But that's impossible to calculate since we don't know why operating costs are higher. Are they putting more into R&D for the next product lineup? Is cost of manufacturing up? Have taxes gone up? Have utilities gone up? Wages? Etc...

So with AMD making only 1 million USD on their graphics division, and the company having an overall gross profit margin of 46%, can we make any reasonable assessment of any graphics line selling at a loss? After all only 1 million USD operating profit. The correct answer is that we don't know and we have no way of knowing.

The same applies to Nvidia. They may or may not be selling a product at a loss. We have no way to know and no way to infer that information from the information we have available.

Regards,
SB

 

[3dcgi]

Low end GPUs and especially chipsets have appreciably higher profit margins.

I'm not sure how you came up with this assumption.

 

[Xenus]

Yeah I was pretty sure it was the opposite high end gpus have large margins while low end they are really tight.

 

[tannat]

I'm not sure how you came up with this assumption.

Sorry, low end GPUs give more profit by volume, you're absolutely right. Brainfart.

 

[tannat]

Let's put it another way using another example.

AMD's has gross profit margins of 46%. So the products they sell certainly quite profitable in and of itself. However, once you bring in operating budget (everything from accounting, to building maintanaince, to utilities, etc.) and R&D costs, they ended up with only a 128M USD operating income on 1.62B USD revenue. Far far lower than the gross profit margin their products bring in. So minus operating costs and R&D, their products had a gross profit of ~745 million USD (very simplistic and naive view of it) yet only ended up with operating income of 128 million USD.

If you look at their graphics division. Year over year their revenue increased 33%, but their operating income declined from 2 million to 1 million. So they made more money but their operating costs increased far more leading to a Year over Year decline in profitability despite a Year over Year increase in revenue.

Blindly looking at the numbers someone might say something is wrong. But that's impossible to calculate since we don't know why operating costs are higher. Are they putting more into R&D for the next product lineup? Is cost of manufacturing up? Have taxes gone up? Have utilities gone up? Wages? Etc...

So with AMD making only 1 million USD on their graphics division, and the company having an overall gross profit margin of 46%, can we make any reasonable assessment of any graphics line selling at a loss? After all only 1 million USD operating profit. The correct answer is that we don't know and we have no way of knowing.

The same applies to Nvidia. They may or may not be selling a product at a loss. We have no way to know and no way to infer that information from the information we have available.

Regards,
SB

Point taken.

In hindsight, I was not really trying to stir the pot with my first post. Just pointing out that the numbers used to show that the cards were profitable could as well cover the possibilities that cards were sold at a loss, or even that the whole Fermi line may not have been profitable yet.

I think this is still true? The numbers don't prove any of the outcomes to be false. Even if they are wrongly used to show one outcome in this thread.

 

[Arty]

Which doesn't matter to the customer as long as price / performance and features are good.

Thats not the point. Nvidia could very well slash prices and sell a 600mm2 GF100 (470) against Barts. That would hardly be any marvel and no one here would be amazed at such accomplishment. Reverse the role and then it would be an technical discussion we can have about here. (This isnt Anandtech forums)

 

[Silent_Buddha]

Thats not the point. Nvidia could very well slash prices and sell a 600mm2 GF100 (470) against Barts. That would hardly be any marvel and no one here would be amazed at such accomplishment. Reverse the role and then it would be an technical discussion we can have about here. (This isnt Anandtech forums)

Taking that example one step further. While cutthroat pricing as a result of heavy price wars may be good for the customer in the short term, it could end up being very VERY bad for the customer in the long term.

If a company is forced to sell a product at a level where profit margin falls too far behind operating costs, then you may have a situation where R&D has to be slashed to remain operating. But once you slash R&D too much you start a downward spiral where you can no longer compete, thus meaning you have to undercut your competitor even more thus leading to more cuts and on and on.

We've seen many companies go by the wayside due to this. 3dfx, S3, Matrox (only hanging on due to specialized niche hardware), Imagination, Rendition, 3D Labs, and many others. Some may continue on in the handheld market, but they're all basically gone or at best niche or budget players in the PC market

With only 2 players left, if one were to suddenly find itself in a position where it could not maintain the R&D costs required to remain competitive and basically had to drop out, that leaves the remaining player to set prices anywhere they wish.

Sure it can take years to get to that point, and consumers certainly don't look that far ahead, but it is a very real possibility.

Heck, look at the direction Creative Labs was headed before the addin soundcard market basically died. Their prices were starting to skyrocket with no competition. And with no competition their drivers got even worse. And with no competition the one thing that could have helped keep the soundcard market afloat, environmental sound modeling, was made propriety once the competition bowed out.

So everyone should always hope and pray that sometimes Nvidia is on top and sometimes ATI is on top.

Regards,
SB

 

[trinibwoy]

Operational profit is related to the "profitability" of the products

No, gross margin is related to profitability of the products currently being sold. Operating profit is related to the efficiency of the overall business and includes management of costs unrelated to the development or manufacturing of the current product line.

 

[tannat]

Taking that example one step further. While cutthroat pricing as a result of heavy price wars may be good for the customer in the short term, it could end up being very VERY bad for the customer in the long term.

If a company is forced to sell a product at a level where profit margin falls too far behind operating costs, then you may have a situation where R&D has to be slashed to remain operating. But once you slash R&D too much you start a downward spiral where you can no longer compete, thus meaning you have to undercut your competitor even more thus leading to more cuts and on and on.

We've seen many companies go by the wayside due to this. 3dfx, S3, Matrox (only hanging on due to specialized niche hardware), Imagination, Rendition, 3D Labs, and many others. Some may continue on in the handheld market, but they're all basically gone or at best niche or budget players in the PC market

With only 2 players left, if one were to suddenly find itself in a position where it could not maintain the R&D costs required to remain competitive and basically had to drop out, that leaves the remaining player to set prices anywhere they wish.

Sure it can take years to get to that point, and consumers certainly don't look that far ahead, but it is a very real possibility.

Heck, look at the direction Creative Labs was headed before the addin soundcard market basically died. Their prices were starting to skyrocket with no competition. And with no competition their drivers got even worse. And with no competition the one thing that could have helped keep the soundcard market afloat, environmental sound modeling, was made propriety once the competition bowed out.

So everyone should always hope and pray that sometimes Nvidia is on top and sometimes ATI is on top.

Regards,
SB

Good points all - but:

Deficiencies in delivery will always open up opportunities to new players. However, creative fell bad since they were made obsolete by parallel development of technology.

Discrete graphic cards may be made obsolete at some point. Until then competition is of course critical.

 

[tannat]

No, gross margin is related to profitability of the products currently being sold. Operating profit is related to the efficiency of the overall business and includes management of costs unrelated to the development or manufacturing of the current product line.

This is contradictory to what I said, only evading the fact that operating profit still is related to the products being sold long term.

 

[NathansFortune]

Gross profit margins are directly related to the profitability of the products sold, i.e. a negative margin would indicate selling products below cost and a positive margin indicates products are selling for a profit.

Operating profit margins are related to the health of the business and how well management are doing. It is all about core running costs and R&D budgets.

I said it earlier in the thread but a fabless semi-conductor company should have a gross profit margin of 40-50% and operating profit margins of 5-15%. I seriously can't understand why, despite all of the evidence to the contrary, you seem to think Nvidia are selling their products below cost.

 

[PSU-failure]

With only 2 players left, if one were to suddenly find itself in a position where it could not maintain the R&D costs required to remain competitive and basically had to drop out, that leaves the remaining player to set prices anywhere they wish.

I don't know which world you live in, but as long as a company needs to sell, it needs to improve its products regularly for each price point...

 

[tannat]

Gross profit margins are directly related to the profitability of the products sold, i.e. a negative margin would indicate selling products below cost and a positive margin indicates products are selling for a profit.

Operating profit margins are related to the health of the business and how well management are doing. It is all about core running costs and R&D budgets.

I said it earlier in the thread but a fabless semi-conductor company should have a gross profit margin of 40-50% and operating profit margins of 5-15%. I seriously can't understand why, despite all of the evidence to the contrary, you seem to think Nvidia are selling their products below cost.

1) I've never claimed that Nvidia is selling products below costs.
2) I have however not seen any proof that they are selling all products above cost. You claim we have a lot of evidence for this. Please clarify, now I am confused.

 

[Man from Atlantis]

according to nordichardware gtx 560 will be launched on january 25

 

[KimB]

1) I've never claimed that Nvidia is selling products below costs.
2) I have however not seen any proof that they are selling all products above cost. You claim we have a lot of evidence for this. Please clarify, now I am confused.

While there are sometimes reasons to sell products below cost, I don't think nVidia is in that sort of situation.

 

[trinibwoy]

This is contradictory to what I said, only evading the fact that operating profit still is related to the products being sold long term.

No, you were actually talking about one specific product - the GTX 460. You're only changing your tune now after folks pointed out the flaws in your analysis. Of course operating profit is dependent on the ongoing success of the overall product line but that's not what you started out saying.