Intel team of 1,000 people working on iPhone modem, foundry SoC

[witeken]

VentureBeat: Intel has 1,000 people working on chips for the iPhone

Intel now has a thousand people or more working to outfit a 2016 iPhone with its lauded 7360 LTE modem chip, sources say. If all goes well, Intel may end up providing both the modem and the fabrication for a new Apple system on a chip.

One source said Intel needs a small army of people on the Apple account because of the importance of the project to Intel’s future in the mobile market, because of the complexity of the project, and because Apple is a demanding client with an extremely popular phone.

Our sources don’t believe Apple has officially signed Intel as a supplier of the modem chip, but a deal will happen if Intel continues to hit its project milestones.

Sources with knowledge of the situation say that Apple eventually would like to create a system-on-a-chip (SOC) that includes both the phone’s Ax processor and the LTE modem chip. A system-on-a-chip design could deliver significant returns in improved speed and better power management.

Apple would design the SOC, which would carry an Apple brand name, and would license the LTE modem intellectual property from Intel for the SOC, one source said.

This would be a sensible arrangement, in light of the fact that Intel hasn’t excelled at SOC design in the past, and because Apple does vertical integration very well. Apple’s skill in this was most recently seen in the integration of its M9 graphics chip with its powerful A9 chip for the iPhone 6s.

While the SOC would be created from top to bottom by Apple’s formidable chip designers, Intel would get the job of fabricating the SOC using its 14-nanometer process.

One of our sources close to Intel said that Apple is also attracted by the fact that Intel is already developing a 10-nanometer process alongside its 14-nanometer process. The 10-nanometer process, which could yield even smaller and faster processors, isn’t ready for large-scale production today, but might be ready for high-volume fabrication in as little as two years.

Another important piece of Intel’s LTE modem story fell into place at the end of September. Intel confirmed that it bought the CDMA modem assets of Via Telecom. Intel says it will use the technology to add CDMA network connectivity to its LTE modems.

Intel will need the CDMA technology in its modem in order to supply phone makers selling devices into the Chinese market. Carriers like China Telecom use a combination of LTE and CDMA technology in their networks.

So much for the people who said it would be the other way around: Apple would replace Intel for laptops etc.

The obvious next step is of course ditching the ARM architecture altogether, maybe at 7nm or 5nm, who knows.

 

[ShintaiDK]

It would pretty much close down all future lower nodes at TSMC and Samsung if Apple moves to Intel. And it would give Intel the extra volume needed to go down to 5nm.

 

[Arachnotronic]

The obvious next step is of course ditching the ARM architecture altogether, maybe at 7nm or 5nm, who knows.

Why would Apple do that?

 

[witeken]

Why would Apple do that?

Because Intel is an IDM and they optimize their Core architecture excessively for their own process and vice versa.

I'm not saying they will, it's just another possibility far in the utopian looking far future . They should first secure the iPhone modem win, then we can look at what comes next.

 

[krumme]

The obvious next step is of course ditching the ARM architecture altogether, maybe at 7nm or 5nm, who knows.

Yeaa who knows. In all the universe. This is absolutely the most straightforward conclusion to Intel now using their state of the art process to manufacture ARM cpu.

 

[Nothingness]

Why would Apple do that?

To please Intel, Intel shareholders and Intel fanboys... hmmm wait :biggrin:

 

[StrangerGuy]

To please Intel, Intel shareholders and Intel fanboys... hmmm wait :biggrin:

Some of those are downright delusional, especially if they have their favorite company as their forum avatar.

 

[Mopetar]

So much for the people who said it would be the other way around: Apple would replace Intel for laptops etc.

They're still sort of right in that Apple won't be using CPU cores designed by Intel, but I don't suspect that Intel cares as long as they're the company that's getting paid to fab them.

It's pretty clear from the A9 test results that Apple is leading in terms of SoC design in the mobile space and it would be interesting to see what they could do when paired with the best fab process.

 

[Arachnotronic]

Because Intel is an IDM and they optimize their Core architecture excessively for their own process and vice versa.

I'm not saying they will, it's just another possibility far in the utopian looking far future . They should first secure the iPhone modem win, then we can look at what comes next.

I think that a modem deal is possible down the line but ditching ARM is basically not going to happen. Apple designs easily the most robust CPU cores in the land of mobile SoCs and I would only expect this lead to grow not shrink.

 

[Nothingness]

So much for the people who said it would be the other way around: Apple would replace Intel for laptops etc.

So you take this rumor and consider it true? Don't you remember this summer when the rumors were telling iPhone 6s would have Intel modems? What about waiting before bragging?

The obvious next step is of course ditching the ARM architecture altogether, maybe at 7nm or 5nm, who knows.

Or it could be a beachhead for Apple to have its ARM-based chips manufactured by Intel. This is more likely than Apple getting rid of ARM.

 

[witeken]

So you take this rumor and consider it true? Don't you remember this summer when the rumors were telling iPhone 6s would have Intel modems? What about waiting before bragging?

Not all rumors are created equal. This one also from VentureBeat: http://seekingalpha.com/news/2360806-report-intel-partly-replacing-qualcomm-as-iphone-modem-supplier.

 

[stingerman]

It's pretty clear from the A9 test results that Apple is leading in terms of SoC design in the mobile space and it would be interesting to see what they could do when paired with the best fab process.

Actually while Intel was the first out with a FinFet based process, they're process is older and not as advanced as what's currently being used by TSMC. Intel's process made sense as an evolution to leverage their Fab investment, in the same way that TSMC evolved their 20NM Fabs to 16NM. But, arguably FinFet+ is more advanced than Intel's transistor. And, it appears that their both on track to mass produce 10NM around the same time.

Not to mention TSV and similar advances that are becoming very important in advancing SOC designs.

 

[ShintaiDK]

Actually while Intel was the first out with a FinFet based process, they're process is older and not as advanced as what's currently being used by TSMC. Intel's process made sense as an evolution to leverage their Fab investment, in the same way that TSMC evolved their 20NM Fabs to 16NM. But, arguably FinFet+ is more advanced than Intel's transistor. And, it appears that their both on track to mass produce 10NM around the same time.

Got any documentation for this?

And even Finfets evolve.

 

[stingerman]

Not all rumors are created equal. This one also from VentureBeat: http://seekingalpha.com/news/2360806-report-intel-partly-replacing-qualcomm-as-iphone-modem-supplier.

It's a dubious report coming a day after Intel tanked that is now helping Intel shares. I would take it with a heavy helping of salt. I'm sure Apple meets with many vendors but actually getting selected as a key supplier will take a lot more. Not least of which is producing a viable alternative at the right price, volume and quality points.

 

[stingerman]

And even Finfets evolve.

No doubt.

Here's an old article though:

My sources also tell me that TSMC 16nm FF+ is today the most competitive FinFET offering, meaning power, performance, area, AND cost. This is based on information from the associated PDKs and not from PowerPoint slides or press releases.

http://www.semiwiki.com/forum/content/3546-tsmc-vs-intel-vs-samsung-finfets.html

 

[jpiniero]

I could see Apple using Intel's modem IP and integrating it into the A10. I could also see Intel fabbing it too. But 1,000 people for a modem? That seems way too high.

 

[ShintaiDK]

No doubt.

Here's an old article though:


http://www.semiwiki.com/forum/content/3546-tsmc-vs-intel-vs-samsung-finfets.html

Semiwiki isnt exactly a neutral site to put it gently. So I dont give anything for it and their bias. Semiwiki and TSMC is pretty much the same.

 

[podspi]

Semiwiki isnt exactly a neutral site to put it gently. So I dont give anything for it and their bias. Semiwiki and TSMC is pretty much the same.

Do you have any documentation for Intel's FF process being superior?

It seems to me if you are going to demand proof for other's assertions (going so far as to throw out their sources when they come up with them) you should offer your own.

Personally I agree that Intel's process tech is tops, but I think it is hard to separate out how much of that is objective process superiority and the process being tailored specifically for their processors.

 

[ShintaiDK]

TSMC is still on first gen FF with extremely high cost still with limited production. They have very limited experience. Intel is on second gen.

TSMC have shipped what, 6-7 million FF chips? Intel have shipped over a billion since production start in 2011.

Also some random undocumented source tells me. That's as vague as it gets.

Nobody can tell you about the extract cost between TSMC and Intel for example either. And those that may be able to get somewhat close would all be working at Intel. Since TSMC is the public product so to say.

And what devices have TSMC produced that rivals Intels in performance and performance/watt?

In terms of area alone I could just post this.

Or this:

Rough calculation between both dense nodes shows TSMC 16FF+ is ~40% bigger. Also explains why Samsung A9 is smaller than TSMC 16FF+. Just again proving the article from Semiwiki is completely fiction.

 

[Nothingness]

Semiwiki isnt exactly a neutral site to put it gently. So I dont give anything for it and their bias. Semiwiki and TSMC is pretty much the same.

TSMC is still on first gen FF with extremely high cost still with limited production. They have very limited experience. Intel is on second gen.

TSMC have shipped what, 6-7 million FF chips? Intel have shipped over a billion since production start in 2011.

Also some random undocumented source tells me. That's as vague as it gets.

Nobody can tell you about the extract cost between TSMC and Intel for example either. And those that may be able to get somewhat close would all be working at Intel. Since TSMC is the public product so to say.

And what devices have TSMC produced that rivals Intels in performance and performance/watt?

In terms of area alone I could just post this.

You can't spit on SemiWiki claiming they have a bias and at the same time use Intel marketing slide to prove your point :biggrin:

Or this:

Rough calculation between both dense nodes shows TSMC 16FF+ is ~40% bigger. Also explains why Samsung A9 is smaller than TSMC 16FF+. Just again proving the article from Semiwiki is completely fiction.

You should provide source for that data (which looks correct).

 

[Dresdenboy]

I could see Apple using Intel's modem IP and integrating it into the A10. I could also see Intel fabbing it too. But 1,000 people for a modem? That seems way too high.

It's no 56k modem anymore.

 

[NTMBK]

I'm sure Intel would love to get the Apple foundry job. Plenty of volume, and they should be a damn sight more reliable than GloFo as a partner.

 

[Arachnotronic]

I'm sure Intel would love to get the Apple foundry job. Plenty of volume, and they should be a damn sight more reliable than GloFo as a partner.

If Intel could get the Apple foundry job + become a second source to Qualcomm in modems, I'd say that's a pretty reasonable way to profit from the mobile market. It's not glamorous, but it's something.

 

[ehume]

Hm. Does Apple have enough money to buy Intel outright yet? But they're not doing so. So sourcing to Samsung, TSMC, Intel according to who makes the coolest chip at the least cost still keeps the competition alive, gets them the best chip.

Yet Apple bought that design firm, made it their own. Now they in-source chip designs. Must be that would not benefit from competition.

This is a fun rumor. Gives food for thought, something to munch on for the weekend.

 

[Khato]

They're still sort of right in that Apple won't be using CPU cores designed by Intel, but I don't suspect that Intel cares as long as they're the company that's getting paid to fab them.

Indeed, I'd imagine that winning Apple's foundry business would be adequate motivation to give them excellent pricing on the modem side too. The extra wafers help to offset fab costs and, more importantly, represent revenue that's not going to Samsung and TSMC. I've always been of the opinion that Intel's long term play is to maneuver themselves into being the only company that can continue making the necessary investments for future node shrinks. And depriving Samsung and TSMC of leading edge customers is the best way to do that.