I have no way of knowing the questions you raise, however, the business of producing chips for others must be profitable enough for companies like Samsung to be interested. If things progress the way they have been, more people will buy mobile devices possibly dispensing with laptops. That has to hurt Intel.
Quote:
Originally Posted by solipsism
I think it cane used in either case so long as you refer to it being the revenue made or the profit made. But that?s neither here no their. Do revenue or profits from foundries out perform revenue or profits from Intel?s X86 foundries.
Melgross? comment about the price of the chip is interneting, but I wonder how much the foundry would get? The full $25, but how much to Samsung for RAM and how much to Img Tech for the GPU, etc? Seems like x86 chips would still be Intel?s biggest money maker.
That doesn?t mean I think Intel should ignore this area of business. I?m quite happy to read this rumor and hope it comes to fruition.
PS: Could Intel use the 3D Tri-Gate on ARM PoP/SoCs?
I think Apple uses Intel chips on the Mac line so that users can boot into Windows. So that would seem to make the need for custom chips less likely.
On their iPhone and iPad, Apple uses ARM because of their low power consumption compared to Intel. So that would seem to make the need for Intel chips less likely.
Not seeing a spot where it makes sense for Apple to use custom Intel chips.
I could definitely see them using custom chips for the Air.
I think Custom Chips are referring to Mac, not iPhone or other Mobile Devices that are currently using ARM. Since there are zero benefits for Apple moving to Intel. Intel's LP node are always one node behind, i.e even with 22nm their Low Power Node are still at 32nm. There are certain IPs required for manufacturing SoC outside TSMC, Samsung and GF, and since Intel are making CPU design themselves it is very unlikely those IP owner will allow Intel to see their design.
The only custom part of Desktop CPU i could think of is the GPU part. Apple would properly like to use PowerVR 6 across the whole range of Apple devices, that includes iPhone all the way to the Mac. Since Apple are largely responsible for Drivers, doing so would mean their resources are much better utilise.
And if Intel ever wanted to get rid of the x86 instruction bloat ( MMX, SSE1 etc ) , Apple would properly be the only customer in the world who could pull this off.
I agree that ARM will continue to gain massive traction but I?m not so sure about it being the single largest profit focus from Intel?s PoV. ARM Holding made $665.5 million in 2010 while Intel made 43.6 billion. I don?t know the revenue or profit from the foundries that make the chips but I can?t imagine that they would amount to single largest profit focus for sometime to come.
Don't forget though, that companies like TSMC only do fab work, and they're doing pretty well.
ARM is RISC and Apple's investment into it is not scaling back, but expanding. I like how you fix priced the Apple price for their SoC CPU combo. Sorry, but volume increases dramatically reduce the price to the buyer.
LLVM Project's Clang subproject just added assembly support for Cortex-A{8,9,15}, Cortex-R{4,5,7}, Cortex-M{0,3,4}, ARM11MPCore, ARM1176 in the general Clang tree.
They also just added SSE42_64 CodeGen support into LLVM.
LLVM 3.0 is truly shaping up to being the big break from GCC.
I didn't fix any price, but Apple is now using 125 million chips a year, and $25 is the price at that level of purchasing. It won't change much going to 200 million. And as the chips get more complex, they get more expensive.
I don't see how anything you mentioned here has anything to do with this.
It is clear you know the difference, but I think it is important to clarify neither Intel or ARM "made" the numbers you report. To me, "made" implies profit. The numbers you are using are gross revenue. That is what Intel and ARM brought in.
The profit for the year for Intel, however, was 4.4 Billion. About a tenth of gross revenue. Arms Holding's profit for the year was about 274 million. Closer to half its gross revenue. So comparatively speaking, ARM keeps a much higher percentage of the gross revenue as actual profit (or spends much less to make its profit).
ARM's number will likely only go up as companies like Microsoft start switching to ARM. Intel as a back plan should seriously start building its foundry business. It would be a win win for Intel and companies like Apple who are looking to escape Samsung.
Whoa! Where are you getting that Intel number from? You have to be making it up.
Last year, in addition to the $43.6 billion is revenue Intel made, they made $11.5 billion in profit. That's 26.28% net profit margin.
I think it cane used in either case so long as you refer to it being the revenue made or the profit made. But that?s neither here no their. Do revenue or profits from foundries out perform revenue or profits from Intel?s X86 foundries.
Melgross? comment about the price of the chip is interneting, but I wonder how much the foundry would get? The full $25, but how much to Samsung for RAM and how much to Img Tech for the GPU, etc? Seems like x86 chips would still be Intel?s biggest money maker.
That doesn?t mean I think Intel should ignore this area of business. I?m quite happy to read this rumor and hope it comes to fruition.
PS: Could Intel use the 3D Tri-Gate on ARM PoP/SoCs?
Nobody knows exactly what the split is, but Intel would get the large majority of that, or even all of it, as it isn't clear as to how Apple pays Imagination for its IP.
Well, I imagine that adding 10% to your business, and cementing the relationship with one of your largest, and fastest growing, customers for x86 chips wouldn't be a bad thing.
ARM is RISC and Apple's investment into it is not scaling back, but expanding. I like how you fix priced the Apple price for their SoC CPU combo. Sorry, but volume increases dramatically reduce the price to the buyer.
LLVM Project's Clang subproject just added assembly support for Cortex-A{8,9,15}, Cortex-R{4,5,7}, Cortex-M{0,3,4}, ARM11MPCore, ARM1176 in the general Clang tree.
They also just added SSE42_64 CodeGen support into LLVM.
LLVM 3.0 is truly shaping up to being the big break from GCC.
Great! Now When will Xcode 4 be as fast as Xcode 3?
Intel has been steadily ramping up capacity, but with The one-two punch of softening PC sales and the take off of mobile devices based on ARM (and MS publicly pledging to support ARM for Windows) they are clearly concerned.
While CedarTrail news has been something of a disappointment I dunno why folks don't believe that Apple stays well informed with Intel MID processors and probably has a couple experimental designs around.
It's also not like Intel couldn't resurrect it's ARM CPU business. It still holds an ARM license even after the sale of XScale PXA and they bought Infineon AND they still make ARM processors today. They just call them IO Processors...but these are 1.2Ghz XScale processors from the ARM V5TE family. They might make some Marvell XScale CPUs still but I think these got moved to someone else.
I think they might still be making CE 2110 (Olo River) but these are pretty long in tooth. But they are coupled with PowerVR GPUs. The Intel CE line uses PowerVR quite a bit so it's not like Apple couldn't use Intel for one stop shopping.
If Apple dumps a $500M order on Intel for A5 and A6 production it's only a little bit less than what Intel sold XScale to Marvell for. With the lawsuits with Samsung perhaps Intel or TSMC is a better choice than Samsung anyway.
An executive with chipmaker Intel said on Thursday that his company "wouldn't blink" if given the opportunity to build a custom chip based on the Intel architecture for a major client like Apple.
.
i thought intel was already making chips for apple .
I can understand Intel not being too interested in doing foundry work for others if the production was small. Small being on the order of hundreds of thousands to in the low millions. Then the profit would be small. as they mentioned.
But, Apple will need a good 200 million chips in 2012. Being that these are all the same chips, except for the speed sort that Apple uses, running the iPhone slightly slower than the iPad, and possibly the Touch, Intel would benefit from the enormous production run. Intel doesn't make more than a fraction of that number of any of its own chips. And as efficiency increases in chip production the further down the road you get with a chip, they would have increasing profits as the year went on.
The estimate is that Apple is paying about $25 per A%, as opposed to a cost of $15 for the Tegra 2. 200 million of those chips, if Apple paid intel what they pay Samsung, would be $5 billion dollars. That's a significant portion of their total sales, which last year was $43.6 billion.
$25 bucks each?? are these the pre-payed chips Apple bought last year ??
and 200 million chips ? you say -o- gran poo ah !!
wow so for 2012 that means apple is gonna be selling a lot of ipads and iphones .
Comments
I think it cane used in either case so long as you refer to it being the revenue made or the profit made. But that?s neither here no their. Do revenue or profits from foundries out perform revenue or profits from Intel?s X86 foundries.
Melgross? comment about the price of the chip is interneting, but I wonder how much the foundry would get? The full $25, but how much to Samsung for RAM and how much to Img Tech for the GPU, etc? Seems like x86 chips would still be Intel?s biggest money maker.
That doesn?t mean I think Intel should ignore this area of business. I?m quite happy to read this rumor and hope it comes to fruition.
PS: Could Intel use the 3D Tri-Gate on ARM PoP/SoCs?
I think Apple uses Intel chips on the Mac line so that users can boot into Windows. So that would seem to make the need for custom chips less likely.
On their iPhone and iPad, Apple uses ARM because of their low power consumption compared to Intel. So that would seem to make the need for Intel chips less likely.
Not seeing a spot where it makes sense for Apple to use custom Intel chips.
I could definitely see them using custom chips for the Air.
The problem is: Nobody wants IA chips in the post-PC devices. Not even custom-made ones.
Now, an Intel-manufactured chip with ARM architecture. That would be very yummy...
sadely, AMD would be better than Intel in the Arm market i believe, better graphics, better scaling...
this is kinda stupid... what would Apple want lol
The only custom part of Desktop CPU i could think of is the GPU part. Apple would properly like to use PowerVR 6 across the whole range of Apple devices, that includes iPhone all the way to the Mac. Since Apple are largely responsible for Drivers, doing so would mean their resources are much better utilise.
And if Intel ever wanted to get rid of the x86 instruction bloat ( MMX, SSE1 etc ) , Apple would properly be the only customer in the world who could pull this off.
Apple and several others own a nice amount of ARM.
I think Apple sold off the last of its ARM shares years ago.
I agree that ARM will continue to gain massive traction but I?m not so sure about it being the single largest profit focus from Intel?s PoV. ARM Holding made $665.5 million in 2010 while Intel made 43.6 billion. I don?t know the revenue or profit from the foundries that make the chips but I can?t imagine that they would amount to single largest profit focus for sometime to come.
Don't forget though, that companies like TSMC only do fab work, and they're doing pretty well.
ARM is RISC and Apple's investment into it is not scaling back, but expanding. I like how you fix priced the Apple price for their SoC CPU combo. Sorry, but volume increases dramatically reduce the price to the buyer.
LLVM Project's Clang subproject just added assembly support for Cortex-A{8,9,15}, Cortex-R{4,5,7}, Cortex-M{0,3,4}, ARM11MPCore, ARM1176 in the general Clang tree.
They also just added SSE42_64 CodeGen support into LLVM.
LLVM 3.0 is truly shaping up to being the big break from GCC.
I didn't fix any price, but Apple is now using 125 million chips a year, and $25 is the price at that level of purchasing. It won't change much going to 200 million. And as the chips get more complex, they get more expensive.
I don't see how anything you mentioned here has anything to do with this.
Intel is the one behind. They are severely behind Apple's A# Series SoC.
Sorry, but Apple's not abandoning the overwhelming amount of recently granted IP for Intel.
One thing has nothing to do with the other. Fabs don't "get" the IP of the companies they're fabbing for. I don't know where you get that idea.
It is clear you know the difference, but I think it is important to clarify neither Intel or ARM "made" the numbers you report. To me, "made" implies profit. The numbers you are using are gross revenue. That is what Intel and ARM brought in.
The profit for the year for Intel, however, was 4.4 Billion. About a tenth of gross revenue. Arms Holding's profit for the year was about 274 million. Closer to half its gross revenue. So comparatively speaking, ARM keeps a much higher percentage of the gross revenue as actual profit (or spends much less to make its profit).
ARM's number will likely only go up as companies like Microsoft start switching to ARM. Intel as a back plan should seriously start building its foundry business. It would be a win win for Intel and companies like Apple who are looking to escape Samsung.
Whoa! Where are you getting that Intel number from? You have to be making it up.
Last year, in addition to the $43.6 billion is revenue Intel made, they made $11.5 billion in profit. That's 26.28% net profit margin.
if you want some proof of that, here:
http://money.cnn.com/quote/quote.html?symb=INTC
Then you have the other made up number. That wasn't ARM's profit either. They made $132.7 million in profit, for a 21.14% net profit margin.
Here:
http://money.cnn.com/quote/quote.html?symb=ARMH
Both of these numbers are the reverse of your argument.
How are you coming up with these numbers?
And, for the record, for the purpose of comparison, Apple made $14 billion in profit last year, for a 21.54% net profit margin.
http://money.cnn.com/quote/quote.html?symb=AAPL
I think it cane used in either case so long as you refer to it being the revenue made or the profit made. But that?s neither here no their. Do revenue or profits from foundries out perform revenue or profits from Intel?s X86 foundries.
Melgross? comment about the price of the chip is interneting, but I wonder how much the foundry would get? The full $25, but how much to Samsung for RAM and how much to Img Tech for the GPU, etc? Seems like x86 chips would still be Intel?s biggest money maker.
That doesn?t mean I think Intel should ignore this area of business. I?m quite happy to read this rumor and hope it comes to fruition.
PS: Could Intel use the 3D Tri-Gate on ARM PoP/SoCs?
Nobody knows exactly what the split is, but Intel would get the large majority of that, or even all of it, as it isn't clear as to how Apple pays Imagination for its IP.
Well, I imagine that adding 10% to your business, and cementing the relationship with one of your largest, and fastest growing, customers for x86 chips wouldn't be a bad thing.
ARM is RISC and Apple's investment into it is not scaling back, but expanding. I like how you fix priced the Apple price for their SoC CPU combo. Sorry, but volume increases dramatically reduce the price to the buyer.
LLVM Project's Clang subproject just added assembly support for Cortex-A{8,9,15}, Cortex-R{4,5,7}, Cortex-M{0,3,4}, ARM11MPCore, ARM1176 in the general Clang tree.
They also just added SSE42_64 CodeGen support into LLVM.
LLVM 3.0 is truly shaping up to being the big break from GCC.
Great! Now When will Xcode 4 be as fast as Xcode 3?
It's also not like Intel couldn't resurrect it's ARM CPU business. It still holds an ARM license even after the sale of XScale PXA and they bought Infineon AND they still make ARM processors today. They just call them IO Processors...but these are 1.2Ghz XScale processors from the ARM V5TE family. They might make some Marvell XScale CPUs still but I think these got moved to someone else.
I think they might still be making CE 2110 (Olo River) but these are pretty long in tooth. But they are coupled with PowerVR GPUs. The Intel CE line uses PowerVR quite a bit so it's not like Apple couldn't use Intel for one stop shopping.
If Apple dumps a $500M order on Intel for A5 and A6 production it's only a little bit less than what Intel sold XScale to Marvell for. With the lawsuits with Samsung perhaps Intel or TSMC is a better choice than Samsung anyway.
An executive with chipmaker Intel said on Thursday that his company "wouldn't blink" if given the opportunity to build a custom chip based on the Intel architecture for a major client like Apple.
.
i thought intel was already making chips for apple .
9
I can understand Intel not being too interested in doing foundry work for others if the production was small. Small being on the order of hundreds of thousands to in the low millions. Then the profit would be small. as they mentioned.
But, Apple will need a good 200 million chips in 2012. Being that these are all the same chips, except for the speed sort that Apple uses, running the iPhone slightly slower than the iPad, and possibly the Touch, Intel would benefit from the enormous production run. Intel doesn't make more than a fraction of that number of any of its own chips. And as efficiency increases in chip production the further down the road you get with a chip, they would have increasing profits as the year went on.
The estimate is that Apple is paying about $25 per A%, as opposed to a cost of $15 for the Tegra 2. 200 million of those chips, if Apple paid intel what they pay Samsung, would be $5 billion dollars. That's a significant portion of their total sales, which last year was $43.6 billion.
$25 bucks each?? are these the pre-payed chips Apple bought last year ??
and 200 million chips ? you say -o- gran poo ah !!
wow so for 2012 that means apple is gonna be selling a lot of ipads and iphones .
bruce p
9