Apple in Talks to Purchase Intel's German Modem Unit

[SurferPup]

I look forward to  producing it's own modem chips and the new features that result from this transition.

 

[needsomecoffee]

Profit via vertical integration. Makes sense for modems and processors. This enables the former, catalyst the latter. I think Nadella knows Tim Apple's plan. I think Satya has his own plan. Going to be interesting.

 

[Analog Kid]

I would imagine the money saved from not paying Qualcomm's profit margins alone would make it worth it for Apple. If they can make a chip that is on par or better, it makes absolute sense for Apple.

If they can make a chip on par or better. There's a lot to say against Qualcomm, but technically inept is not on that list. Both Infineon and Intel are world class companies with experience in these types of components and they failed to compete.

 

[dantroline]

Wow, I thought Apple was going to drop Intel.

Yes right after they finish acquiring it.

 

[Analog Kid]

Profit via vertical integration. Makes sense for modems and processors. This enables the former, catalyst the latter. I think Nadella knows Tim Apple's plan. I think Satya has his own plan. Going to be interesting.

What's missing in this discussion so far is the reason why it makes sense for modems. Qualcomm pricing is only a minor part of the calculus, I suspect...

 

[fokmik]

Apple should buy even the chips and with the help from TSMC...they can push x86 chips forward ,faster ,better

 

[Alan Wynn]

Man, if this happens I hope they pull it off. Their success with the A-series chips suggests they can, but it also sounds like Intel was struggling to pull it together. Does it make sense to go through the effort if the end game is a standards compliant chip with only yourself as a customer?

You could make much of the same argument about the A-series (they are mostly standard ARM chips). I would guess that one benefit would be building more a more tightly integrated SoC or some other chip that would let them improve performance while reducing parts and power costs.

 

[calstanford]

A mess of a strategic blunder by Apple that cost them billions.

Infineon (formerly Siemens) were the baseband chips in the first iPhones before they switched to Qualcomm. Infineon's modem unit then got sold over to intel who messed with it before selling some chips to Apple again and now struggling with 5G.

Apple should have invested or downright bought them ten years ago (they weren't expensive anyway) and would have a modem team bar none by now (as they did with their PA Semi / A Series chips).

 

[ROGmaster]

Apple will just ship the entire business to China. Everyone is dumping 5G businesses because governments will not hold China accountable for IP theft. Good move for Intel to sell it before China steals it anyway.

How can China steal something they already have?
Why would China need to steal Intel's inferior 5G tech? It doesn't make sense.

 

[Applebot1]

There is some tech/patents Apple wants....never face value Apple acquisitions.

 

[Wowereit]

Let me get this straight. Apple are trying to acquire the same baseband processor design team that failed to meet their own originally specified deadline ?

They are probably more interested in the IP, but if some top-level baseband experts come with it, why not? Keep the employees they need, integrate them into the Apple team, get rid of the rest.

 

[urbanman2004]

Good for you Apple, would've expected some sort of deal to be struck sooner

 

[Glideslope]

Apple will just ship the entire business to China. Everyone is dumping 5G businesses because governments will not hold China accountable for IP theft. Good move for Intel to sell it before China steals it anyway.

More likely to Taiwan back to TSMC. I agree with other posts there is no way Apple runs a Fab. Especially in the EU.

 

[ksec]

Man, if this happens I hope they pull it off. Their success with the A-series chips suggests they can, but it also sounds like Intel was struggling to pull it together. Does it make sense to go through the effort if the end game is a standards compliant chip with only yourself as a customer?

Not much room for differentiation or deeper integration into the software stack as far as I can tell, or am I missing something? Seems foolish if the only points of differentiation are power consumption and reception quality-- Qualcomm the existing players have a huge lead on experience in those areas. And unlike Samsung, Intel and Qualcomm, if Apple doesn't source the parts to 3rd parties they can't amortize the R&D expense across as many parts...

I get both of these points, but why stop here? Why not make their own memory? Roll their own steel? Is the plan, in effect, to pull a reverse-Samsung?

Samsung makes commodity parts, and uses their smartphones to showcase them. Is Apple planning to showcase smartphones and simply produce the parts to build them?

And I'm not sure I'd assume they can even reach parity with Qualcomm. At some point, I'd imagine that monopoly power will degrade Qualcomm's technical dominance, but I see no sign of it yet.

The usual logic is to specialize and focus on your core competence and places where you can differentiate. If it's something that someone else can do as well or better than you, let them. Apple has been pretty disciplined in following that approach, at least in hardware. Software could be debated, I suppose.

It may be because there are too few players in the modem space for this to be a truly commodity part but, if so, someone out there missed an opportunity and Apple is taking a huge risk. You're saying they'll make it up on product margin, but they'll be paying multiples of the R&D cost per unit that the commodity players do-- it seems a strange place to spend it. It seems like it would make more sense for Apple, Samsung, and a few others to go in on a joint venture.

Who else out there is capable of making 5G chips? Qualcomm, Huawei, Samsung? Intel dropped out, obviously. Anyone else in the game?

Cook is an operations guy, I'm sure he's thought this through, but it still seems an odd move.

Because you can differentiate your products with the processor design. I'm not sure how you do that with the modem-- it has to talk to the tower, so it has to be standards compliant. If it's standards compliant, then it's not functionally different from any other standards compliant modem.

Maybe I'm thinking about the wrong device. Maybe I shouldn't be thinking about the phone, but the watch. There might be more room to innovate there?

If they can make a chip on par or better. There's a lot to say against Qualcomm, but technically inept is not on that list. Both Infineon and Intel are world class companies with experience in these types of components and they failed to compete.

It is nice to see finally 1 person asking the right question.

Let's put some number into prospective.

Samsung made roughly 250M Smartphone a year, Huawei made roughly 200M Smartphones, Apple roughly 200M. These three are the top 3 smartphone vendors. However Apple also ships 40M Tablet a year, the largest Tablet Vendor.

This hopefully answer your amortize question. Samsung cant sell their modem or SoC with modem to others as per the agreement with Qualcomm. However it should be noted that both Samsung and Huawei are in the telecom equipment industry, making testing slightly easier for their Modem Design Team.

There are plenty of 5G or even 4G designs. And Intel didn't drop out ( yet ), There is also Mediatek, Spreadtrum, and a few others. The problem isn't making a modem that is difficult, it is like making ARM SoC is dead easy if you buy the blueprint. Making a well tested, performant Modem is extremely difficult. And if anyone has actually read the 233 page of the case between Qualcomm and Apple, you might be surprised Qualcomm isn't the monopoly as you might imagined.
( It is actually easy to read, and don't believe everything what ever that FOSS guy has to say )

Intel and Infineon, ( which is basically the same since Intel acquired them ), had difficulties for number of reasons. 1. Intel's culture has so far meant they have not had any successful acquisition that makes synergy. 2. Intel has had little to no experience in Fabbing Modem, which is quite different to CPU and GPU. And as you can tell by their progress on Custom Foundry and 10nm, things haven't been working out well. ( Modem is done on 14nm Intel Custom Foundry ). 3. Intel decide to change whatever they had with Infineon Modem ( either ARM or MIPS ) with their own IP, mini x86 Core and FPGA. That saves cost, but you are basically throwing away lots of things and rebuilding them. And one reason why Apple only wants to buy *part* of the Modem IP, and not everything. They have no need for x86 core. 4. Being Intel doesn't have anything to test with. Modem is really about Trial and Error. And it is the testing that is the most expensive. Qualcomm listed as spending additional $200M every year on standalone modem chip just for design and testing. That is reusing their existing IP. You can imagine Intel will need a lot more than that. Qualcomm has many partners to filed test all their modem, Intel? none? Or Relying on Apple.

The question of whether Apple should or should not fab their modem is hard, but it does make financial sense for Apple. If Apple leave $10 per iPhone on R&D for Modem, that is $2B. With the scale of Apple it certainly make sense for them to not only make their Modem, but also their own WiFI and Bluetooth Chip. Apple has been slowly working on WiFi and Bluetooth, but whether Apple can make a Desktop / Tablet Class WiFi Chip and Qualcomm Modem remains to be seen. I am personally skeptics.

 

[citysnaps]

What's missing in this discussion so far is the reason why it makes sense for modems. Qualcomm pricing is only a minor part of the calculus, I suspect...

Apple can add their own and proprietary secret sauce (beyond standards conformance) not available to competitors if it were a publicly available part. That can include extra features through greater integration and better performance.

And then, of course, Apple will also be more in control of their roadmap not relying on a third party for critical technology. And at a lower cost.

 

[cajun67]

Wow, I thought Apple was going to drop Intel.

Apple has a long history of using Intel chips. It didn't start with using their CPU's.

 

[bb9]

The sooner Apple develop (design) more of the own hardware the better.

 

[prasand]

The thing about Qualcomm modems is that they are designed generally, since it's an item used by many companies. With Apple designing their own chips, they can remove things they deem unnecessary, and add security / privacy enhancements while remaining standards complaint. However, I'm just speculating, I'm not a chip nor communication standards guy. But I highly doubt Apple would produce something so general, which should result in optimizations.

 

[alphaod]

So they're buying the former Infineon modem unit basically.

 

[lec0rsaire]

Yeah if it was me I’d leave it to the Qualcomm experts.

This is the sort of thinking that encourages monopolies to remain unchallenged. While Apple and Qualcomm made peace for now, you can never put the future of your product in one company’s hands. While not the same situation just look at Huawei. They’re getting cut off from all of the stuff they need to sell phones. ARM SoCs, their Android license, Corning Gorilla Glass. The more self-reliant you are the better.
[doublepost=1560363343][/doublepost]

The thing about Qualcomm modems is that they are designed generally, since it's an item used by many companies. With Apple designing their own chips, they can remove things they deem unnecessary, and add security / privacy enhancements while remaining standards complaint. However, I'm just speculating, I'm not a chip nor communication standards guy. But I highly doubt Apple would produce something so general, which should result in optimizations.

I think they just want the same or similar performance without having to rely on Qualcomm ever again. Or at least get good enough to have leverage when making the next deal.

 

[apolloa]

This is the sort of thinking that encourages monopolies to remain unchallenged. While Apple and Qualcomm made peace for now, you can never put the future of your product in one company’s hands. While not the same situation just look at Huawei. They’re getting cut off from all of the stuff they need to sell phones. ARM SoCs, their Android license, Corning Gorilla Glass. The more self-reliant you are the better.
[doublepost=1560363343][/doublepost]

I think they just want the same or similar performance without having to rely on Qualcomm ever again. Or at least get good enough to have leverage when making the next deal.

Huawei are a state funded spying company, the West just hides the ones it supports to do the same thing, but it has nothing to do with Qualcomm who make the best modems, you don’t want to use the best modems? You’d rather have inferior ones that won’t work as well as proven already?
That’s a strange opinion to take blanketed under the monopoly banner...

 

[lec0rsaire]

Huawei are a state funded spying company, the West just hides the ones it supports to do the same thing, but it has nothing to do with Qualcomm who make the best modems, you don’t want to use the best modems? You’d rather have inferior ones that won’t work as well as proven already?
That’s a strange opinion to take blanketed under the monopoly banner...

Apple managed to design the best SoCs in under a decade. They definitely have enough resources to pull off modems as well. I’m not saying Qualcomm isn’t the best. What am I saying is that if no one steps up they’ll continue to be the only game in town and that isn’t good.

Qualcomm charges crazy prices for their modems exactly because there’s no other choice. They charge not based on what the modem costs but on the retail price of the phone. So the modem for a 512 GB 11 Max costs much more than the modem for a 64 GB 11R.

 

[Degerz]

They are probably more interested in the IP, but if some top-level baseband experts come with it, why not? Keep the employees they need, integrate them into the Apple team, get rid of the rest.

Why not ? Well for one this is the same team that failed Apple before and two, Apple doesn't participate all that much at 3GPP so this means that they'd have to start setting global standards so that they aren't under the tyranny of other corporations interests ...

Apple has shown exactly ZERO foresight in how the wireless network industry operates so far ...

 

[nebo1ss]

Apple will just ship the entire business to China. Everyone is dumping 5G businesses because governments will not hold China accountable for IP theft. Good move for Intel to sell it before China steals it anyway.

Huawei, Nokia and Ericsson happen to be the leading suppliers of 5G equipment. The US is desperately trying to kill off Huawei by badgering foreign governments not to do deals with them. However, the options are limited.

 

[Analog Kid]

Samsung made roughly 250M Smartphone a year. This hopefully answer your amortize question. Samsung cant sell their modem or SoC with modem to others as per the agreement with Qualcomm.

Ok, that's what I was wondering-- does Samsung sell their part to anyone else, and the answer is apparently no (though this might change for 5g?). So this is different from Samsung's model for displays, memory, etc, in that the modem is only made by them for them, and they've somehow justified the development.

According to the Qualcomm filings, Samsung only uses their own modems in about about half of their products, so they've managed to justify the development on a 100M unit volume.

And Intel didn't drop out ( yet )

They've dropped out of 5G smartphone modems.

The question of whether Apple should or should not fab their modem is hard, but it does make financial sense for Apple. If Apple leave $10 per iPhone on R&D for Modem, that is $2B. With the scale of Apple it certainly make sense for them to not only make their Modem, but also their own WiFI and Bluetooth Chip. Apple has been slowly working on WiFi and Bluetooth, but whether Apple can make a Desktop / Tablet Class WiFi Chip and Qualcomm Modem remains to be seen. I am personally skeptics.

The problem isn't making a modem that is difficult, it is like making ARM SoC is dead easy if you buy the blueprint. Making a well tested, performant Modem is extremely difficult.

This is key, I think. Samsung still uses Qualcomm parts for their Galaxy S series, presumably because Qualcomm is still the design to beat on pure performance (and power). I don't know the full details of Samsung's lineup, you might, but I'd guess they ship their own modem in the devices they are trying to save cost on and shipping Qualcomm in their flagships.

For now, Apple only makes flagships. They don't make much effort to compete with Samsung at the lower end of the product line.

If that is their strategy going forward, it's hard to see how Apple continues selling a $1200 phone with second rate connectivity so if they make their own modems and fall short, it's a bust.

It's also possible that Samsung producing their own parts is what allows them to remain competitive in the lower cost devices and this might be Apple's first step toward moving into those markets. On paper, Samsung's low cost phones don't seem terribly profitable for their telecom division (which is part of why Apple dominates on smartphone profit share), but it may be because while there are profits to be had in that market they show up in the semiconductor division.

3. Intel decide to change whatever they had with Infineon Modem ( either ARM or MIPS ) with their own IP, mini x86 Core and FPGA. That saves cost, but you are basically throwing away lots of things and rebuilding them. And one reason why Apple only wants to buy *part* of the Modem IP, and not everything. They have no need for x86 core.

I hadn't heard this-- what a predictable disaster...