holding their ~70-140 million net per quarter from sales. Quite nice, Quite nice. Although, perhaps I'm doing my math wrong, but, given that 56.7% gross margin, they should be making ~652 million before expensises... ~652 - ~477 = ~175 million. Taking away or adding Intel's 66 million payment doesn't change that.
But, I don't deal with this kind of stuff, so, my math is probably off somewhere :)
Yeah you were doing something wrong but not that. 1151 millions minus 43.1% is 652.6 and then non-GAAP OPEX is 425 , you are using GAAP OPEX but since you started with the non-GAAP margins i assume you wanted non-GAAP. Beyond that you got a few minor items and tax and you get the correct numbers.
Yes, what will happen to the company, that controls nearly 80% of its primary market, when it loses about 6% of its quarterly revenue? Nothing terribly exciting, my guess.
Besides, it is a licensing fee, so Intel's going to have to pay unless they decide to develop an entirely new iGPU from scratch.
I don't see it as just 6% quarterly revenue, I see it as much more as quarterly income. Since how much percentage of its should be considered as operating expenses ?
As long as I remember, Intel agreed to pay NVIDIA $1.5 bi of licensing fee. So that would end some day
As long as I remember, Intel agreed to pay NVIDIA $1.5 bi of licensing fee. So that would end some dayYou are correct, in that it is indeed a set amount - mea culpa.
Nonetheless, considering how they keep on raking in the billions and their market share on a steady rise, not to mention if their upcoming architecture is even half as good as they say it is, losing that quarterly 66 mil shouldn't be too dramatic in the grander scheme of things, IMHO.
I don't mean NVIDIA products not good. They are really good. But that does not mean the quarterly reports would be always green too (look at at HTC, Motorola or other companies else)
I mean if we subtract the income from Intel payment, how much would they get (or what would the EPS number look like) ?
nVidia is frighteningly ahead of AMD in GPU architecture. AMD hasn't released a new architecture since Maxwell 1.0 was introduced early LAST year, and odds are nVidia is going to release Pascal/Volta either in line with AMD's successor to GCN 1.2 (which hasn't even been announced yet, they're just riding GCN 1.0/1.1/1.2 parts into the sunset for the rest of the year)
Are they? Intel, yeah, they're frighteningly ahead of amd. AMD is still competitive, and, for gaming, their dx12 support has really closed many gaps. Their gcn was quite innovative. They don't necessarily have to come out with brand new archs each time. Intel has been riding sb since, what, 2011?
It's just revenue, Nvidia could very well be treating it as $66 million worth of R&D to pay 500 engineers for whatever they feel like dabbling in. So yes, in quarters where their net profit is closer to even, that $66 million might look like its a big difference in their net profit, but in reality, if they didn't have that guaranteed revenue stream, they might simply cut expenses accordingly as any responsible business would.
NVIDIA could only use 66 mi as operating expense if Intel pay them first. But if NVIDIA uses 66 mi of Q4 2014 to pay their employees in Q1 2015, so that 66 mi of Q4 2014 is the budget for Q1 2015, but not the revenue of Q1 2015. 66 mi of Q1 2015 is the budget for Q2 2015, but not the revenue of Q2 2015
That's not quite how it works. If you have enough cash on hand, you can incur the expense before you receive payment, you just set up a receivable (asset) which gets wiped out and becomes revenue when Intel actually pays up within the quarter. I am sure Nvidia knows Intel is good for it (given the tens of billions cash they are sitting on), so it is trivial for them to shell out for payroll before actually receiving the payments.
Ultimately it just comes down to budgeting and fiscal management, if the $66M went away, Nvidia would have to adjust but they would still be profitable.
That revenue is almost all profit for Nvidia, after accounting expenses, so the loss of it would cut net by almost half. So yes, it would have quite an impact.
Except they know exactly when that revenue will cease, so they can obviously make arrangements for it well in advance by cutting expenses similarly, so no, it shouldn't have any impact if Nvidia is managing their finances prudently as any business should/would.
The licensing fee is more to end Nvidia's lawsuit. I'm guessing it will be renegotiated as a free cross-license both ways. Nvidia's lawsuit about FSB and chipsets isn't going to happen again.
And, think of it has 25-50% of it's profit. It hurts a lot more. In some quarters, it has been more than 50% even.
That licensing agreement gives Intel those patents in perpetuum, I believe. So unless they want access to newer patents they don't have to sign a new deal.
Do you have a link for this? Everything I have read including the non-redacted portions of the settlement indicate the cross-licensing agreement ends and will need to be renegotiated. Indeed, the last settlement came just as the previous one expired. No where in the settlement language does it state Nvidia assigns their IP to Intel in perpetuum, in fact it is clearly stated it is a negotiated license fee that lasts for 6 years.
Exactly, it'll be interesting to see what happens here though. I've thought for some time Intel and even Apple might be prime candidates to take Nvidia up on their GPU IP licensing offer. Intel especially given they already pay for Nvidia IP, and it is really old IP at that. Apple I've felt is waiting for the litigation to settle with Samsung/Qualcomm, but it is pretty interesting to note Apple was never named in those original law suits.
No guarantee Intel re-signs with Nvidia though, I guess their options are to keep close to status quo, bump up their licensing to Nvidia's top-tier IP (Maxwell), or go really risky with someone else like Imagination or even AMD. Imagine Intel IGP with Maxwell-class mobile IP though! Wow that would really close the gap on AMD in the IGP department in a hurry.
The only reason why Nvidia signed it was that it was part of the deal to get Nvidia to drop it's chipset lawsuit. The licensing part will likely become a free cross-license. Like what AMD and Intel/AMD and Nvidia have.
For information about why the licensing was done and WHAT the 1.5 billion dollars was for: http://www.anandtech.com/show/4122/intel-settles-w... The 1.5 billion dollars is for the settlement of the suite, it was tied into the cross licensing. Likely Nvidia hopes to try to make Intel pay for it's patents once it expires, but, the odds of that happening are slim. Intel will crush Nvidia in court on that front.
Another way to look at it is that Intel paid Nvidia 1.5 billion dollars to stop any x86 plays.
Well, in what is becoming a recurring theme, I disagree with you. :)
The cross-licensing portion was Nvidia IP for Intel's chipset/socket. When Intel blocked Nvidia from making a chipset beginning with X58/Nehalem, that agreement was broken. Intel was continuing to benefit from Intel IP and Nvidia sought recourse. The $1.5Bn was simply that, Intel is paying for Nvidia IP now that they no longer benefit from Intel IP.
Now, what Intel IP is Nvidia benefitting from in this cross-licensing agreement that would change the scale and balance to outweigh the IP that Intel uses in every single one of the millions of CPUs with IGPs that they sell each and every year? Answer: none.
So yeah, Intel will need to renegotiate or find another IP vendor, both are certainly on the table.
Intel was continuing to benefit from Nvidia IP and Nvidia sought recourse. The $1.5Bn was simply that, Intel is paying for Nvidia IP now that Nvidia no longer benefits from Intel IP.
"To this end, NVIDIA specifically mentions that they are getting access to Intel’s “microprocessor” patents, excluding the x86 (and XScale) technologies we previously mentioned."
If Intel just needed to license Nvidia's GPU tech, they wouldn't have cross licensed outside of C2D/AGTL+ bus license/chipset license. Notice how before Intel was licencing Nvidia's GPU IP For their C2D/AGTL+ bus license?
Do you think Intel said "here, have access to microprocessor patents AND have money as part of the licensing" when this was also a settlement of a court case where Nvidia was suing for money? And, unlike AMD, NVidia had the money for a long court case. Which, while drawn our could cast more than 1.5 billion when all was said and done?
If when the agreement expires there is a cross license, expect no money changing hands. If Nvidia has no need of any of Intel's CPU patents (possible) than expect Intel to have to pay something. I'm guessing under 100 million/year.
Again, you don't seem to understand, there is cross-licensing in place, but the LICENSING FEE is the balance of those cross-licenses. Basically, Intel gets more out of the deal than Nvidia, so they make up the difference in licensing fees. That's why its a 5-year annuity rather than a lump sum. Simply put, there is nothing that Nvidia gets from the arrangement from Intel that is equivalent to what Intel is licensing back to Nvidia, so the difference is made up in cash. I mean I don't see how this is hard to understand, if you start with chipset licensing+microprocessor = graphics IP, but you take chipset licensing out of the equation (which was tangible product from Nvidia), what falls out of that equation, and why wouldn't you think this equation holds true once the existing licensing arrangement expires?
The ability to pay for court fees is irrelevant since those would be covered when Intel lost along with the possibility of treble damages, not to mention it is unlikely the fees would have amounted to that much anyways.
Intel was simply doing the smart thing and paying fair value on what they were getting from Nvidia with their graphics IP, knowing full well they were about to make a major IGP push. An injunction would've crippled their desktop CPU business since they were already shipping IGP parts by the time this got settled, but this was most likely the reason Havendale was cancelled and Clarkdale delayed for so long.
I forgot to post this in my original reply, but, remember that this was done before before Tegra really fell on it's face and lost marketshare. Nvidia likely predicted HUGE marketshare wins/etc from Tegra. The access to the majority of Intel's CPU patents has certainly helped them. Sadly, they managed to piss off most vendors, and make chips that drew to much power for phones.
If the Nvidia/Intel Cross license was signed after Tegra 3 hit the market in this state, I would say that Nvidia deffo would drop Intel's patents and aim for a real GPU license.
Uh, you need to check your facts again. The Tegra team actually rose from the ashes of the defunct motherboard/chipset/IGP team. So yeah, Tegra has nothing to do with the expectations of this settlement, if anything Tegra only exists because Intel forced Nvidia out of the chipset business, but as a result of the IP licensing deal, they are still funding that team. So yeah, despite the fact Tegra hasn't been a stellar performer in terms of profit, it is certainly easier to justify the expense when Intel is helping to foot the bill, and certainly better to keep talented folks in house rather than lay them off.
Er, Tegra itself doesn't. Custom cores sure do. Intel has a lot of patents for CPU design. And, stopping Nvidia's lawsuit due to all the trouble they could have caused again in the EU/etc is well worth 1.5 billion dollars over 6 years...
Custom Tegra cores that Nvidia licenses ARM IP from, so again, non-applicable to what they are getting back from Intel. Also, Nvidia bought Transmeta so they also have their own CPU IP. Let's also not forget Nvidia already designs some of the world's most advanced application processors. But fact remains, Intel is clearly getting more out of the licensing arrangement than Nvidia, so they paid the balance in a licensing fee.
And of course it made sense for Intel to make the lawsuit go away, especially given they are so clearly in the wrong.
-Intel and Nvidia agreed to cross-license tech, bulk of which was Nvidia GPU IP for Intel chipset IP. No money changes hands, as value of IP is deemed relatively equal. -Intel rescinds socket license for IMC chips, cripples Nvidia's IGP/chipset business knowing full well they have their own IGP products on the horizon that fully make use of Nvidia IP. -Intel settles and pays the balance of the licensing agreement over 6 years knowing full well an injunction on Intel IGP capable parts would be crippling to their business, maybe even enough for AMD to get back in the game.
It is obvious that once this agreement expires, it will need to be renegotiated or Intel will have to find another IP vendor and integrate it into all of their IGPs.
It will be replaced by lawsuit victories from samsung/qcom and either settlements etc from IMG.L, Apple etc. Either way all of these will add up to more than Intel who was NOT willful in infringement the way mobile people are, and it still remains to be seen if Intel can get out of payments in the future. In particular samsung has been infringing willfully for 2yrs, not sure about the rest or what negotiations went on between qcom, apple etc, but samsung was named as having known (and blown off) about this for at least 2yrs (when NV started IP licensing efforts). None of these people have any gpu patents (nowhere near what I suspect Intel even has), and Intel lost 1.5B with FAR fewer units sold. Remember that even MS got $4 a pop just for android being on samsung units. 1B+ owed and MS just sued them recently for failing to pay what was owed on ~250mil units. Depending on vendor MS gets $4-15 I guess from infringing code in android per unit. Nobody really knew much until the non-redacted info came out on samsung. MS collects about 7B a year on this crap, so NV should do fairly well and they have three things to point to: 1. Intel and 1.5B+not willful (just due to Intel breaking chipset agreement). 2. MS fees over some code in android. 3. Qcom's agreements that take a % of the ENTIRE device for having their tech inside; meaning they don't just take a price for the modem, they take a fee of the entire device it is sold in.
So NV could argue a reasonable fee will include a % of the entire device too, as your phone really doesn't function without the screen's graphics just like it doesn't function without a modem and a major portion of time spent on a phone is gaming. Don't ask me why people do this on a puny screen, but app store sales being mostly games don't lie. I guess people could also be using hdmi out to tv with a gamepad or something.
http://english.caixin.com/2015-02-12/100783598.htm... China doesn't like Qcom's royalty crap and just shot it down (over % of the entire unit). Not sure how this will play out in USA in the future, but maybe NV comes away with % of unit in USA and plays according to china rules over their. Who knows, but everyone will get stung here to some degree most likely. But even in the china deal, it's still able to charge on a % of 65% of the device (instead of the whole unit). Still seems odd to me you can even do this, I mean I sell you a chip and that is it here so why do I get to charge you on the whole unit which is comprise of tons of crap not even made by me? Whatever, it bodes well for NV's case. Qcom's patent royalties alone are 7.6B a year in that article (probably higher now, I haven't looked recently, note most of that is modem). Arberl says they'll still charge on the whole unit price in that article, so who knows how this works out in china for them. The article is a bit confusing to say the least with qcom's own comments being [seemingly] opposite of the verdict...LOL. The point here I guess, is it shouldn't hurt NV much as other income comes in, from suits, subscriptions, grid, console, auto etc.
First off, Qualcomm, Samsung, Imagination Tech, ARM and many other companies Nvidai will likely end up suing ALL have GPU patents. Well, they won't sue ARM, but, everyone who uses Mali graphics probably. Qualcomm and Samsung both have GPU patents. Qualcomm likely quite a lot.
Depends on how much you value the patents. And, how much Nvidia asked for. And, if Nvidia negotiated in good faith. And if the patents are valid and apply to these cases. If NVidia was confident in their cases Apple and Imagination Tech would be sued currently.
IF NVidai licensed to them we would also know as they would say got 1-2 big licenses/got Apple and Imagination. Haven't seen anything that would make these valid. Hence, I'm a bit dubious. No one at all in the industry accepted the license? Not even a small chinese manufacturer or something? If it was industry standard it would let them get a large GPU advantage for a low cost, although, price is king in their SoCs, so, that could be a problem. Most SoCs that make their way to phones aimed more at the west have huge SoCs.
Anyhow, I don't see anything that confirms Nvidia's patents holding for a license. Or holding up at all. If they do, good for Nvidia. It gets more years to transition into hopefully something beyond GPUs. Tegra doesn't seem to be working out to great.
Nvidia does what it usually does, make money and develop new markets. Gpu markets are still in their infancy and the processing requirements of the future will be massive.
In all of this, remember that after Apple bought 11% of Imagination, Intel bought 15%. So they have that IP that they're using. It certainly isn't Nvidia or bust for them.
Its true Intel does have options, because they are using Imagination IP in their SoFIA chips, but they have to weigh the cost of implementing that IP into all of their IGP desktop CPUs and extracting enough performance to be competitive. But I am sure that is part of the reason Nvidia is going after Qualcomm/Samsung in their IP litigation, because that is also a de facto claim that Qualcomm, ARM and Imagination are infringing on Nvidia graphics IP.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
43 Comments
Back to Article
testbug00 - Thursday, May 7, 2015 - link
holding their ~70-140 million net per quarter from sales. Quite nice, Quite nice. Although, perhaps I'm doing my math wrong, but, given that 56.7% gross margin, they should be making ~652 million before expensises... ~652 - ~477 = ~175 million. Taking away or adding Intel's 66 million payment doesn't change that.But, I don't deal with this kind of stuff, so, my math is probably off somewhere :)
Brett Howse - Thursday, May 7, 2015 - link
Your math is off. Gross margin is the ratio of revenue to the cost of goods sold, and higher means your operating income per unit sold is higher.testbug00 - Friday, May 8, 2015 - link
Thanks. Figured I was doing something silly.jjj - Friday, May 8, 2015 - link
Yeah you were doing something wrong but not that.1151 millions minus 43.1% is 652.6 and then non-GAAP OPEX is 425 , you are using GAAP OPEX but since you started with the non-GAAP margins i assume you wanted non-GAAP. Beyond that you got a few minor items and tax and you get the correct numbers.
MrSpadge - Friday, May 8, 2015 - link
"I’ve heard they are known for gaming even"Haha.. that's the way to go to spice up boring reports ;)
meacupla - Friday, May 8, 2015 - link
I would have been surprised if they had loses with such a good product roll out for 2015.Dark Man - Friday, May 8, 2015 - link
66 ... 66 and 66 ...I wonder when the payment from Intel is done, what would happen next
D. Lister - Friday, May 8, 2015 - link
Yes, what will happen to the company, that controls nearly 80% of its primary market, when it loses about 6% of its quarterly revenue? Nothing terribly exciting, my guess.Besides, it is a licensing fee, so Intel's going to have to pay unless they decide to develop an entirely new iGPU from scratch.
Dark Man - Friday, May 8, 2015 - link
I don't see it as just 6% quarterly revenue, I see it as much more as quarterly income. Since how much percentage of its should be considered as operating expenses ?As long as I remember, Intel agreed to pay NVIDIA $1.5 bi of licensing fee. So that would end some day
D. Lister - Friday, May 8, 2015 - link
As long as I remember, Intel agreed to pay NVIDIA $1.5 bi of licensing fee. So that would end some dayYou are correct, in that it is indeed a set amount - mea culpa.Nonetheless, considering how they keep on raking in the billions and their market share on a steady rise, not to mention if their upcoming architecture is even half as good as they say it is, losing that quarterly 66 mil shouldn't be too dramatic in the grander scheme of things, IMHO.
Dark Man - Friday, May 8, 2015 - link
I don't mean NVIDIA products not good. They are really good. But that does not mean the quarterly reports would be always green too (look at at HTC, Motorola or other companies else)I mean if we subtract the income from Intel payment, how much would they get (or what would the EPS number look like) ?
Samus - Saturday, May 9, 2015 - link
nVidia is frighteningly ahead of AMD in GPU architecture. AMD hasn't released a new architecture since Maxwell 1.0 was introduced early LAST year, and odds are nVidia is going to release Pascal/Volta either in line with AMD's successor to GCN 1.2 (which hasn't even been announced yet, they're just riding GCN 1.0/1.1/1.2 parts into the sunset for the rest of the year)tuxRoller - Sunday, May 10, 2015 - link
Are they? Intel, yeah, they're frighteningly ahead of amd.AMD is still competitive, and, for gaming, their dx12 support has really closed many gaps.
Their gcn was quite innovative. They don't necessarily have to come out with brand new archs each time. Intel has been riding sb since, what, 2011?
Samus - Monday, May 11, 2015 - link
There is so much wrong with your statement, all the way down to Intel riding sandy bridge, that this is already too much of a dignified response.medi03 - Monday, May 11, 2015 - link
That's quite a clueless comment on for a tech site. Gap between Intel and AMD exists and is huge.Gap between nVidia and AMD is marginal.
chizow - Friday, May 8, 2015 - link
It's just revenue, Nvidia could very well be treating it as $66 million worth of R&D to pay 500 engineers for whatever they feel like dabbling in. So yes, in quarters where their net profit is closer to even, that $66 million might look like its a big difference in their net profit, but in reality, if they didn't have that guaranteed revenue stream, they might simply cut expenses accordingly as any responsible business would.Dark Man - Friday, May 8, 2015 - link
NVIDIA could only use 66 mi as operating expense if Intel pay them first. But if NVIDIA uses 66 mi of Q4 2014 to pay their employees in Q1 2015, so that 66 mi of Q4 2014 is the budget for Q1 2015, but not the revenue of Q1 2015. 66 mi of Q1 2015 is the budget for Q2 2015, but not the revenue of Q2 2015chizow - Friday, May 8, 2015 - link
That's not quite how it works. If you have enough cash on hand, you can incur the expense before you receive payment, you just set up a receivable (asset) which gets wiped out and becomes revenue when Intel actually pays up within the quarter. I am sure Nvidia knows Intel is good for it (given the tens of billions cash they are sitting on), so it is trivial for them to shell out for payroll before actually receiving the payments.Ultimately it just comes down to budgeting and fiscal management, if the $66M went away, Nvidia would have to adjust but they would still be profitable.
melgross - Monday, May 11, 2015 - link
That revenue is almost all profit for Nvidia, after accounting expenses, so the loss of it would cut net by almost half. So yes, it would have quite an impact.chizow - Thursday, May 14, 2015 - link
Except they know exactly when that revenue will cease, so they can obviously make arrangements for it well in advance by cutting expenses similarly, so no, it shouldn't have any impact if Nvidia is managing their finances prudently as any business should/would.testbug00 - Friday, May 8, 2015 - link
The licensing fee is more to end Nvidia's lawsuit. I'm guessing it will be renegotiated as a free cross-license both ways. Nvidia's lawsuit about FSB and chipsets isn't going to happen again.And, think of it has 25-50% of it's profit. It hurts a lot more. In some quarters, it has been more than 50% even.
chizow - Friday, May 8, 2015 - link
Asked below, but again, what IP is Nvidia getting back in that free cross-licensing deal?Yojimbo - Friday, May 8, 2015 - link
That licensing agreement gives Intel those patents in perpetuum, I believe. So unless they want access to newer patents they don't have to sign a new deal.chizow - Wednesday, May 13, 2015 - link
Do you have a link for this? Everything I have read including the non-redacted portions of the settlement indicate the cross-licensing agreement ends and will need to be renegotiated. Indeed, the last settlement came just as the previous one expired. No where in the settlement language does it state Nvidia assigns their IP to Intel in perpetuum, in fact it is clearly stated it is a negotiated license fee that lasts for 6 years.tviceman - Friday, May 8, 2015 - link
Intel won't get to keep using Nvidia's patents for free when their licensing deal expires.chizow - Friday, May 8, 2015 - link
Exactly, it'll be interesting to see what happens here though. I've thought for some time Intel and even Apple might be prime candidates to take Nvidia up on their GPU IP licensing offer. Intel especially given they already pay for Nvidia IP, and it is really old IP at that. Apple I've felt is waiting for the litigation to settle with Samsung/Qualcomm, but it is pretty interesting to note Apple was never named in those original law suits.No guarantee Intel re-signs with Nvidia though, I guess their options are to keep close to status quo, bump up their licensing to Nvidia's top-tier IP (Maxwell), or go really risky with someone else like Imagination or even AMD. Imagine Intel IGP with Maxwell-class mobile IP though! Wow that would really close the gap on AMD in the IGP department in a hurry.
testbug00 - Friday, May 8, 2015 - link
The only reason why Nvidia signed it was that it was part of the deal to get Nvidia to drop it's chipset lawsuit. The licensing part will likely become a free cross-license. Like what AMD and Intel/AMD and Nvidia have.For information about why the licensing was done and WHAT the 1.5 billion dollars was for: http://www.anandtech.com/show/4122/intel-settles-w...
The 1.5 billion dollars is for the settlement of the suite, it was tied into the cross licensing. Likely Nvidia hopes to try to make Intel pay for it's patents once it expires, but, the odds of that happening are slim. Intel will crush Nvidia in court on that front.
Another way to look at it is that Intel paid Nvidia 1.5 billion dollars to stop any x86 plays.
testbug00 - Friday, May 8, 2015 - link
Er, what I say "the only reason why Nvidia signed..." I mean 'The only reason why Intel signed..." oops! Sorry! :Schizow - Friday, May 8, 2015 - link
Well, in what is becoming a recurring theme, I disagree with you. :)The cross-licensing portion was Nvidia IP for Intel's chipset/socket. When Intel blocked Nvidia from making a chipset beginning with X58/Nehalem, that agreement was broken. Intel was continuing to benefit from Intel IP and Nvidia sought recourse. The $1.5Bn was simply that, Intel is paying for Nvidia IP now that they no longer benefit from Intel IP.
Now, what Intel IP is Nvidia benefitting from in this cross-licensing agreement that would change the scale and balance to outweigh the IP that Intel uses in every single one of the millions of CPUs with IGPs that they sell each and every year? Answer: none.
So yeah, Intel will need to renegotiate or find another IP vendor, both are certainly on the table.
chizow - Friday, May 8, 2015 - link
First paragraph should read:Intel was continuing to benefit from Nvidia IP and Nvidia sought recourse. The $1.5Bn was simply that, Intel is paying for Nvidia IP now that Nvidia no longer benefits from Intel IP.
testbug00 - Friday, May 8, 2015 - link
"To this end, NVIDIA specifically mentions that they are getting access to Intel’s “microprocessor” patents, excluding the x86 (and XScale) technologies we previously mentioned."If Intel just needed to license Nvidia's GPU tech, they wouldn't have cross licensed outside of C2D/AGTL+ bus license/chipset license. Notice how before Intel was licencing Nvidia's GPU IP For their C2D/AGTL+ bus license?
Do you think Intel said "here, have access to microprocessor patents AND have money as part of the licensing" when this was also a settlement of a court case where Nvidia was suing for money? And, unlike AMD, NVidia had the money for a long court case. Which, while drawn our could cast more than 1.5 billion when all was said and done?
If when the agreement expires there is a cross license, expect no money changing hands. If Nvidia has no need of any of Intel's CPU patents (possible) than expect Intel to have to pay something. I'm guessing under 100 million/year.
chizow - Friday, May 8, 2015 - link
Again, you don't seem to understand, there is cross-licensing in place, but the LICENSING FEE is the balance of those cross-licenses. Basically, Intel gets more out of the deal than Nvidia, so they make up the difference in licensing fees. That's why its a 5-year annuity rather than a lump sum. Simply put, there is nothing that Nvidia gets from the arrangement from Intel that is equivalent to what Intel is licensing back to Nvidia, so the difference is made up in cash. I mean I don't see how this is hard to understand, if you start with chipset licensing+microprocessor = graphics IP, but you take chipset licensing out of the equation (which was tangible product from Nvidia), what falls out of that equation, and why wouldn't you think this equation holds true once the existing licensing arrangement expires?The ability to pay for court fees is irrelevant since those would be covered when Intel lost along with the possibility of treble damages, not to mention it is unlikely the fees would have amounted to that much anyways.
Intel was simply doing the smart thing and paying fair value on what they were getting from Nvidia with their graphics IP, knowing full well they were about to make a major IGP push. An injunction would've crippled their desktop CPU business since they were already shipping IGP parts by the time this got settled, but this was most likely the reason Havendale was cancelled and Clarkdale delayed for so long.
testbug00 - Friday, May 8, 2015 - link
I forgot to post this in my original reply, but, remember that this was done before before Tegra really fell on it's face and lost marketshare. Nvidia likely predicted HUGE marketshare wins/etc from Tegra. The access to the majority of Intel's CPU patents has certainly helped them. Sadly, they managed to piss off most vendors, and make chips that drew to much power for phones.If the Nvidia/Intel Cross license was signed after Tegra 3 hit the market in this state, I would say that Nvidia deffo would drop Intel's patents and aim for a real GPU license.
chizow - Friday, May 8, 2015 - link
Uh, you need to check your facts again. The Tegra team actually rose from the ashes of the defunct motherboard/chipset/IGP team. So yeah, Tegra has nothing to do with the expectations of this settlement, if anything Tegra only exists because Intel forced Nvidia out of the chipset business, but as a result of the IP licensing deal, they are still funding that team. So yeah, despite the fact Tegra hasn't been a stellar performer in terms of profit, it is certainly easier to justify the expense when Intel is helping to foot the bill, and certainly better to keep talented folks in house rather than lay them off.testbug00 - Monday, May 11, 2015 - link
Er, Tegra itself doesn't. Custom cores sure do. Intel has a lot of patents for CPU design. And, stopping Nvidia's lawsuit due to all the trouble they could have caused again in the EU/etc is well worth 1.5 billion dollars over 6 years...chizow - Wednesday, May 13, 2015 - link
Custom Tegra cores that Nvidia licenses ARM IP from, so again, non-applicable to what they are getting back from Intel. Also, Nvidia bought Transmeta so they also have their own CPU IP. Let's also not forget Nvidia already designs some of the world's most advanced application processors. But fact remains, Intel is clearly getting more out of the licensing arrangement than Nvidia, so they paid the balance in a licensing fee.And of course it made sense for Intel to make the lawsuit go away, especially given they are so clearly in the wrong.
-Intel and Nvidia agreed to cross-license tech, bulk of which was Nvidia GPU IP for Intel chipset IP. No money changes hands, as value of IP is deemed relatively equal.
-Intel rescinds socket license for IMC chips, cripples Nvidia's IGP/chipset business knowing full well they have their own IGP products on the horizon that fully make use of Nvidia IP.
-Intel settles and pays the balance of the licensing agreement over 6 years knowing full well an injunction on Intel IGP capable parts would be crippling to their business, maybe even enough for AMD to get back in the game.
It is obvious that once this agreement expires, it will need to be renegotiated or Intel will have to find another IP vendor and integrate it into all of their IGPs.
Yojimbo - Friday, May 8, 2015 - link
Actually, I've heard that they will...Yojimbo - Friday, May 8, 2015 - link
It ends in a year. FY 2017 Q1 is the last payment I think, or maybe it's FY 2016 Q4. (The quarter just completed is NVIDIA's FY 2016 Q1.)TheJian - Monday, May 11, 2015 - link
It will be replaced by lawsuit victories from samsung/qcom and either settlements etc from IMG.L, Apple etc. Either way all of these will add up to more than Intel who was NOT willful in infringement the way mobile people are, and it still remains to be seen if Intel can get out of payments in the future. In particular samsung has been infringing willfully for 2yrs, not sure about the rest or what negotiations went on between qcom, apple etc, but samsung was named as having known (and blown off) about this for at least 2yrs (when NV started IP licensing efforts). None of these people have any gpu patents (nowhere near what I suspect Intel even has), and Intel lost 1.5B with FAR fewer units sold. Remember that even MS got $4 a pop just for android being on samsung units. 1B+ owed and MS just sued them recently for failing to pay what was owed on ~250mil units. Depending on vendor MS gets $4-15 I guess from infringing code in android per unit. Nobody really knew much until the non-redacted info came out on samsung. MS collects about 7B a year on this crap, so NV should do fairly well and they have three things to point to: 1. Intel and 1.5B+not willful (just due to Intel breaking chipset agreement). 2. MS fees over some code in android. 3. Qcom's agreements that take a % of the ENTIRE device for having their tech inside; meaning they don't just take a price for the modem, they take a fee of the entire device it is sold in.So NV could argue a reasonable fee will include a % of the entire device too, as your phone really doesn't function without the screen's graphics just like it doesn't function without a modem and a major portion of time spent on a phone is gaming. Don't ask me why people do this on a puny screen, but app store sales being mostly games don't lie. I guess people could also be using hdmi out to tv with a gamepad or something.
http://english.caixin.com/2015-02-12/100783598.htm...
China doesn't like Qcom's royalty crap and just shot it down (over % of the entire unit). Not sure how this will play out in USA in the future, but maybe NV comes away with % of unit in USA and plays according to china rules over their. Who knows, but everyone will get stung here to some degree most likely. But even in the china deal, it's still able to charge on a % of 65% of the device (instead of the whole unit). Still seems odd to me you can even do this, I mean I sell you a chip and that is it here so why do I get to charge you on the whole unit which is comprise of tons of crap not even made by me? Whatever, it bodes well for NV's case. Qcom's patent royalties alone are 7.6B a year in that article (probably higher now, I haven't looked recently, note most of that is modem). Arberl says they'll still charge on the whole unit price in that article, so who knows how this works out in china for them. The article is a bit confusing to say the least with qcom's own comments being [seemingly] opposite of the verdict...LOL. The point here I guess, is it shouldn't hurt NV much as other income comes in, from suits, subscriptions, grid, console, auto etc.
testbug00 - Monday, May 11, 2015 - link
First off, Qualcomm, Samsung, Imagination Tech, ARM and many other companies Nvidai will likely end up suing ALL have GPU patents. Well, they won't sue ARM, but, everyone who uses Mali graphics probably. Qualcomm and Samsung both have GPU patents. Qualcomm likely quite a lot.Depends on how much you value the patents. And, how much Nvidia asked for. And, if Nvidia negotiated in good faith. And if the patents are valid and apply to these cases. If NVidia was confident in their cases Apple and Imagination Tech would be sued currently.
IF NVidai licensed to them we would also know as they would say got 1-2 big licenses/got Apple and Imagination. Haven't seen anything that would make these valid. Hence, I'm a bit dubious. No one at all in the industry accepted the license? Not even a small chinese manufacturer or something? If it was industry standard it would let them get a large GPU advantage for a low cost, although, price is king in their SoCs, so, that could be a problem. Most SoCs that make their way to phones aimed more at the west have huge SoCs.
Anyhow, I don't see anything that confirms Nvidia's patents holding for a license. Or holding up at all. If they do, good for Nvidia. It gets more years to transition into hopefully something beyond GPUs. Tegra doesn't seem to be working out to great.
beck2050 - Saturday, May 9, 2015 - link
Nvidia does what it usually does, make money and develop new markets. Gpu markets are still in their infancy and the processing requirements of the future will be massive.melgross - Monday, May 11, 2015 - link
In all of this, remember that after Apple bought 11% of Imagination, Intel bought 15%. So they have that IP that they're using. It certainly isn't Nvidia or bust for them.chizow - Thursday, May 14, 2015 - link
Its true Intel does have options, because they are using Imagination IP in their SoFIA chips, but they have to weigh the cost of implementing that IP into all of their IGP desktop CPUs and extracting enough performance to be competitive. But I am sure that is part of the reason Nvidia is going after Qualcomm/Samsung in their IP litigation, because that is also a de facto claim that Qualcomm, ARM and Imagination are infringing on Nvidia graphics IP.