Intel cleans up its (Custom) foundry act

[Dayman1225]

Intel, ARM deepen foundry ties pages 2 - 5

ARM announced at the event it is extending its partnership with Intel’s foundry business to build IP tailored for the x86 giant’s 22nm FinFET node. It’s a fascinating example of co-opetition from the processor rivals.

ARM will deliver IP to enable a Cortex-A55 geared for midrange smartphones to run up to 2.35 GHz or down to 0.45V in its so-called 22FFL process. ARM is already helping make in Intel’s flagship 10nm process a test chip due out before the end of the year using a next-gen Cortex-A SoC running at 3.5 GHz or 0.5V and delivering 0.25mW/MHz.

An Intel executive showed road maps first released at a Beijing event in September, giving more details than ever on its foundry and IP plans. Observers said the level of candor was new for Intel but key to establishing a foundry business it has been struggling—largely with its own culture--to stand up for years.

Intel has three follow-ons planned for its 22FFL node which it claims will sport 100x lower leakage, 30% more performance and a 20% area shrink compared to 28nm nodes.

Intel is gearing up to compete with the wealth of foundation IP at foundries such as TSMC with basic cells of its own as well as some from partners.

Personal thoughts :
ICF is on the right track - but they have one crucial thing they have to get right to gain trust and gain partners and that is ensuring their nodes are competitive - and delivered on time (or even better, earlier than scheduled), otherwise they will continue as they are now - in partner limbo. Anyone else care to comment?

 

[moinmoin]

That's pretty great as a step in the right direction. Though time will tell if that's too little too late considering other foundries look to have caught up and the leading ARM designs nowadays are custom.

 

[Bouowmx]

3.5 GHz big core (likely Cortex-A75): that sounds exciting to see in a next-generation smartphone.

 

[Dayman1225]

3.5 GHz big core (likely Cortex-A75): that sounds exciting to see in a next-generation smartphone.

If they can get someone to use their fabs - yes, yes it does.

 

[FIVR]

They can't even get the latest iPhone modem on their own process and they've owned Infineon for like a decade. The idea they have "cleaned up their act" when they can't get something as basic as a 50mm^2 baseband produced on their process by themselves is comical.


More fairy tales from BK & Friends to pump that share price even higher. Next stop the moon... then the center of the earth!

 

[Dayman1225]

They can't even get the latest iPhone modem on their own process and they've owned Infineon for like a decade. The idea they have "cleaned up their act" when they can't get something as basic as a 50mm^2 baseband produced on their process by themselves is comical.


More fairy tales from BK & Friends to pump that share price even higher. Next stop the moon... then the center of the earth!

Just so you know the Title is from the Article not me. About modems - I guess we'll see with their 5G modems that are meant to be coming (when IDK). Either way this is a step in the right direction for ICF, the ARM collaboration has helped them "get on their feet" IMO.

 

[FIVR]

Just so you know the Title is from the Article not me. About modems - I guess we'll see with their 5G modems that are meant to be coming (when IDK). Either way this is a step in the right direction for ICF, the ARM collaboration has helped them "get on their feet" IMO.

I'm as surprised as anybody else that they have failed, again, to manufacture a decent modem using an advanced process. It's actually amazing incompetence that I never expected. It has made me reconsider purchasing an iPhone X because I have AT&T service and you can't get a decent Qualcomm-Powered iPhone unless you have Sprint or Verizon.


I will probably have to switch to verizon. How is it even possible that Intel can't integrate a company like infineon into their manufacturing process for a decade then simultaneously claim they have a great process on offer for custom (22nm? Where is 14nm?) and will have 10nm by 1H 2018? Do you actually believe these claims they make?

 

[Dayman1225]

I will probably have to switch to verizon. How is it even possible that Intel can't integrate a company like infineon into their manufacturing process for a decade then simultaneously claim they have a great process on offer for custom (22nm? Where is 14nm?) and will have 10nm by 1H 2018? Do you actually believe these claims they make?

Well who else am I meant to believe, you? I'm pessimistic of what is claimed, but i'd rather Intel succeed than fail. Also notice it says RISK in H1 2018 for ICF partners, I assume they get HVM in H2 2018 like BK claimed in their earnings call recently. Sure their 10nm is very late from what they originally claimed back in 2013 (HVM in 2015) - but i'd rather see this process late than never.

 

[FIVR]

Well who else am I meant to believe, you? I'm pessimistic of what is claimed, but i'd rather Intel succeed than fail. Also notice it says RISK in H1 2018 for ICF partners, I assume they get HVM in H2 2018 like BK claimed in their earnings call recently. Sure their 10nm is very late from what they originally claimed back in 2013 (HVM in 2015) - but i'd rather see this process late than never.

A 14nm modem should be <50mm^2. Apple would have payed extra (as they did with Qualcomm) to get their modem on the most advanced node intel could possibly provide especially considering they got tons of flack for their poor performance of their last modem in the iPhone 7. That node, which would've gone into production in 2017Q1 or 2017Q2, turned out to be TSMC 28nm.

That tells us that Intel can't even produce a 50mm^2 custom product on 14nm. They were supposed to have the latest Altera FPGAs on 14nm more than two years ago. They've produced nothing.

There is much evidence that intel has no ability to do custom foundry work even when they own the companies they are trying to produce chips for. Intel can't even do custom work for intel.


There is zero chance any of the stuff you posted is anything more than flashy shareholder PR

 

[Dayman1225]

They were supposed to have the latest Altera FPGAs on 14nm more than two years ago. They've produced nothing.

Intel Stratix 10 FPGA Now Shipping with over 10 TFLOPs

There is zero chance any of the stuff you posted is anything more than flashy shareholder PR

Guess we'll wait and see, regarding the modem part - yep I agree, Intel is miles behind.

 

[scannall]

Dog and pony show. Intels's first priority is themselves, As it should be btw. Sorry, your order has been pushed back two quarters so we can soggy launch Cofflake. Try to stay in business in the meantime...

 

[FIVR]

Intel Stratix 10 FPGA Now Shipping with over 10 TFLOPs




Guess we'll wait and see, regarding the modem part - yep I agree, Intel is miles behind.

I stand corrected about the Altera FPGAs. It is good that they have gotten something out on 14nm even if they were supposed to have it years ago.


However, I am skeptical for a reason... I have been fooled before by intel. It's like the old saying goes.... "If you fool me once, shame on you. If you fool me twice..... you can't get fooled again"

I can't get fooled again.

 

[Arachnotronic]

A 14nm modem should be <50mm^2. Apple would have payed extra (as they did with Qualcomm) to get their modem on the most advanced node intel could possibly provide especially considering they got tons of flack for their poor performance of their last modem in the iPhone 7. That node, which would've gone into production in 2017Q1 or 2017Q2, turned out to be TSMC 28nm.

That tells us that Intel can't even produce a 50mm^2 custom product on 14nm. They were supposed to have the latest Altera FPGAs on 14nm more than two years ago. They've produced nothing.

The next Intel modem is on 14nm, so don't worry too much.

 

[FIVR]

The next Intel modem is on 14nm, so don't worry too much.

Is it? That would be encouraging. I haven't seen a teardown yet but there was a long post on reddit where a guy claimed it was still on 28nm.


They did test the iPhone 8 and the intel modem preformed better than the QC modem from the 7, but worse than the QC modem from the other 8. So I guess it's possible it's on 14nm but still gets outperformed by the QC part.

 

[Arachnotronic]

Is it? That would be encouraging. I haven't seen a teardown yet but there was a long post on reddit where a guy claimed it was still on 28nm.


They did test the iPhone 8 and the intel modem preformed better than the QC modem from the 7, but worse than the QC modem from the other 8. So I guess it's possible it's on 14nm but still gets outperformed by the QC part.

iPhone 8/X modem is 28nm TSMC, next year’s modem is 14nm.

Node gap should narrow since the comparable Qualcomm modem should be on 10LPE.

 

[FIVR]

iPhone 8/X modem is 28nm TSMC, next year’s modem is 14nm.

Node gap should narrow since the comparable Qualcomm modem should be on 10LPE.

Well I guess it's sort of impressive they were able to outperform the old QC part form the 7 on 28nm... which IIRC was on 20nm planar.


I'm still going to switch to verizon to get the X with the QC modem tho

 

[Arachnotronic]

Well I guess it's sort of impressive they were able to outperform the old QC part form the 7 on 28nm... which IIRC was on 20nm planar.


I'm still going to switch to verizon to get the X with the QC modem tho

Verizon phones are better because they work on all the networks, which means better resale value down the line. I wouldn't buy any iPhone with an Intel modem until they work on all networks.

 

[IntelUser2000]

They shouldn't get too sidetracked from what they already have. ICF is a nice hobby project but there's still plenty to do for bettering the business they already have.

Apple is a long-term risk. They're looking into a fully vertically integrated company where they make nearly every chip by themselves. Once that happens, Qualcomm and Intel both lose the modem business.

Intel modems are behind Qualcomm modems not because of process reasons, but because their designs are behind Qualcomm's. The signal integrity issues Intel modem iPhones are having indicates that its more than just specs on paper that Intel devices need to improve on. They better do it quick, because once Apple business is over, they'll have to rely on ever diminishing Android vendors that'll increasingly do in house chips that integrate modems.

 

[NTMBK]

They shouldn't get too sidetracked from what they already have. ICF is a nice hobby project but there's still plenty to do for bettering the business they already have.

Apple is a long-term risk. They're looking into a fully vertically integrated company where they make nearly every chip by themselves. Once that happens, Qualcomm and Intel both lose the modem business.

Intel modems are behind Qualcomm modems not because of process reasons, but because their designs are behind Qualcomm's. The signal integrity issues Intel modem iPhones are having indicates that its more than just specs on paper that Intel devices need to improve on. They better do it quick, because once Apple business is over, they'll have to rely on ever diminishing Android vendors that'll increasingly do in house chips that integrate modems.

On the flip side, perhaps Intel could exploit that tendency for integration and sell their modem designs to Apple as a package deal. Fab your SoCs with us, and we'll integrate our modem onto the same die.

 

[Phynaz]

However, I am skeptical for a reason... I have been fooled before by intel. It's like the old saying goes.... "If you fool me once, shame on you. If you fool me twice..... you can't get fooled again"

I can't get fooled again.

Really? Interesting, could you tell us about the chip you were going to have Intel manufacture for you before they "fooled" you?

 

[Phynaz]

On the flip side, perhaps Intel could exploit that tendency for integration and sell their modem designs to Apple as a package deal. Fab your SoCs with us, and we'll integrate our modem onto the same die.

Maybe...

Apple Is Designing iPhones, iPads That Would Drop Qualcomm Components

 

[raghu78]

On the flip side, perhaps Intel could exploit that tendency for integration and sell their modem designs to Apple as a package deal. Fab your SoCs with us, and we'll integrate our modem onto the same die.

Intel has fallen behind TSMC given that ICF 10nm risk production is H1 2018 while TSMC N7 hit risk production in Q2 2017 and is slated for volume production in Q2 2018. TSMC N7+ with EUV is slated for HVM in 2019 and N5 in 2020. Intel has no chance of getting Apple back given the pace at which TSMC/Apple juggernaut is moving forward. Frankly given the fact that Intel cannot get their own products out on time on 10nm I think ICF has zero chance of adding a major high volume customer at 10nm.

 

[Lodix]

I didn't read the note that the symbols of the wafers mean "risk production" which makes the situation even worse...

 

[oak8292]

The Pixel 2 may include an Intel foundry part in the Visual Core processor. It is reported that Google and Intel collaborated on the chip. The question is whether the collaboration was with Movidius prior to purchase by Intel and used TSMC as a fab or is this a die from Intel fab services?

 

[NTMBK]

Counterpoint to this article: https://www.fool.com/investing/2017/10/31/intel-corp-doesnt-seem-that-serious-about-its-cont.aspx