Kuo: Apple's Work on 5G Modem Chip 'Failed,' Qualcomm to Remain Supplier for 2023 iPhones

[Macative]

I feel like an Apple in house radio could yield to crazy battery life. Isn't radio one of the big energy consumers, screen aside?

Especially when service is poor, which it usually is. The radio can boost power consumption 10x beyond normal.

 

[fuchsdh]

Engineering in the RF world is a whole different ball of wax than making processor chips and computers. A lot of people don't understand that. Apple is basically starting from ground zero with a little input from buying Intel's modem venture and bringing that in house. You can have all the money in the world, and it simply doesn't happen overnight.

It also can't be done remotely. Ultimately, RF is a "hands on" affair. Things need to be tested in the physical realm over and over and over again. The slightest thing you didn't think would effect performance will. It's really tricky. The whole remote work thing with the 'Rona had an impact on this development time line.

Finally, Apple has to get it right the first time and do it better than everyone else for half the price, or it will be considered another total failure. How long until people are talking "chipgate" or "modemgate" if it's the performance is the slightest bit off?

I dunno if they have to do it better than everyone right out the gate, but it absolutely does have to be as good as what they're replacing. Cellular networking is as basic a foundation of a modern cellphone as the camera; regressions in those areas are definitely going to lead to a massive amount of negative attention. They can't afford a Maps-like situation (or the appearance of one) these days.

 

[Diminish82]

Best news I've heard in awhile. If you think Apple could make a comparable modem without infringing on 9000 Qualcomm patents, you'd be incorrect.

 

[ttyRazor]

It wouldn't surprise me. A modem with key patents and tech locked is a hard endeavor.

Apple

Just a note: Apple's 2019 deal with Qualcomm bought them 6 years of global licensing rights, with an option to extend that for another 2 years.

With the extra 2 years, that would bring them to 2027.

I'm curious how Apple would implement its initial 5G support. Would it be in a discrete chip? Or would it be integrated into the SoC? A discrete chip would give them a lot of flexibility, although it would likely mean higher power usage, and it would mean more space used inside. If it were integrated into the SoC, that would require much more lead time.

i wouldn’t expect it as part of the SoC until there’s a few generations of a discrete chip.

 

[jav6454]

Apple

i wouldn’t expect it as part of the SoC until there’s a few generations of a discrete chip.

???

 

[JustSomebody12]

I'm not the only one who's completely disinterested in 5G right?

5G is more required by telcos to avoid saturation than otherwise.

Anyway, which are the companies around the world making 5G chips?

 

[ikjadoon]

For people that still don’t understand why this would be a failure, Apple had a long legal fight with Qualcomm which ended with Apple losing.

Apple went back to Qualcomm hat in hand, paid them $4.5B for the privilege of using their chips again. For comparison, Apple paid only $1B to acquire Intel’s 5G unit. Think about that for a moment. Apple has likely spend billions more on R&D for the past many years.

For a project of this importance (Apple pays $2B royalties to QCOMM per year) any delay is a failure. And that's not even counting Tim Cook's loss of face for crawling back to QCOMM. This is not a project where you spend billions and years then later walk into Tim's office to say "sorry, we're not sure when it'll be ready."

To add, Apple’s payments to Qualcomm are far more than just 5G modems. Apple also licenses a lot of SoC / power / media IP from Qualcomm. Qualcomm does a group patent, so you get access to all of Qualcomm’s IP, which Apple ended up using, too, beyond just the 5G modem.

In total, Apple paid Qualcomm $16.1 billion for its modem chips between 2010 and 2016. On top of that, Apple paid $7.23 billion in licensing fees.

Source for numbers.

The Apple / Qualcomm relationship is notably deeper than just the modem chipset supplier.

The companies also have reached a six-year license agreement, effective as of April 1, 2019, including a two-year option to extend, and a multiyear chipset supply agreement.

If I recall, Qualcomm makes 33% of its profits from their software IP (QCL division) and not from their hardware IP (QCT division).

Qualcomm is heavily propped up by licenses and patents, many that have nothing to do with their hardware & modems.

 

[dwaltwhit]

It’s always been hard for me to wrap my mind around continuing to work with companies while suing them.

 

[CarAnalogy]

Seems like buying Intel’s modem business was more of a setback than a gain.

 

[Bogstandard]

Failed: (to poach qualified Quaalcom engineers). Lol.

 

[Wildkraut]

It’s always been hard for me to wrap my mind around continuing to work with companies while suing them.

Yeah, but greed comes first, all just a partnership of convenience.

 

[NT1440]

I’m gonna wager this is more about having to engineer around ludicrously broad patents that Qualcomm has than anything.

 

[JPack]

I'm curious how Apple would implement its initial 5G support. Would it be in a discrete chip? Or would it be integrated into the SoC? A discrete chip would give them a lot of flexibility, although it would likely mean higher power usage, and it would mean more space used inside. If it were integrated into the SoC, that would require much more lead time.

Discrete or chiplet. We've seen Snapdragon 865 + external X55 use less power than an integrated LTE solution.

Apple doesn't need to fab a modem using the most expensive 3nm tech. As shown by Qualcomm, the power savings are negligible because there isn't a high speed link between the modem and SoC.

Being able to mix and match SoC with a modem is more important, especially since Apple isn't always paring the newest SoC with the current iPhone.

 

[haunebu]

I’m gonna wager this is more about having to engineer around ludicrously broad patents that Qualcomm has than anything.

Citation needed.

 

[MrKennedy]

Usually companies tries to produce as most as possible "in home" in order to be more competitive by lowering both manufacturing and selling prices. However it seems that's true only in case of the manufacturing (look Apple silicon), as selling prices are the same as before, and even more (look air m2). So personally i prefer that apple buy components instead of doing them at home. The only difference by producing at home is more money to cook and shareholders. I prefer my money is splitted among more parts

Unfortunately most businesses (especially Apple) is actively going to do this since it'll will net them less revenue 🤷‍♂️

 

[Freeangel1]

Probably couldn’t figure out how to make their 5g frequencies dangerous enough for all the conspiracy theorists out there /s

its science. pure science. the only conspiracy theory is you denying its a real problem.

 

[Psychicbob]

Was it too thin?

 

[genovelle]

It looks like actually designed and testing hardware and its supporting software remotely doesn’t seem to work very well.

 

[Pirate!]

its science. pure science. the only conspiracy theory is you denying its a real problem.

Appreciate the video from 2018 of a draft report. Please see Review if Published Literature between 2008 and 2018 of Relevance to Radiofrequency Radiation and Cancer: https://fda.gov/media/135043/download
Page 14, bottom of first paragraph:
“This, results of whole-body RFR exposure to rats and mice cannot be directly related to the results of the local RFR exposure a human receives when using a cell phone.”

Even the video comes to that conclusion and the guy is basically saying, “well, what if it did?”

 

[NT1440]

Citation needed.

You want me to cite my hunch?

 

[newdeal]

Going to add it to the SOC instead of having a discrete chip

 

[bmustaf]

Stalled/delayed vs failed are two different things...

Right, which is why I think he's finally saying "failed" versus "delayed" (which we've heard in the past).

I think the failure is not technical. It's anyone's guess, but my guess is that even with souping up Intel's modem business (and the IP there) Apple could not work around Qualcomm IP. Or, there's some decision making that has to happen about choosing fab capacity, and until the design is integrated on the SoC (which I doubt would happen with a 5G modem for a lot of reasons, thermals not being the least of it), it's a heavy distraction for capacity and production when the demand for M/S/A/H/W series ICs remains enough to basically saturate all capacity. Or some mix of both.

But I'm not sure this is just a regular "stall" or "delay". I think this is AirPower: we couldn't make it work.

 

[bmustaf]

Going to add it to the SOC instead of having a discrete chip

I'm not sure that is the best use of SoC space right now, especially with the thermals of 5G modems. Maybe a few more major revs down the road for both the PHY layer on the modem and the SoC, but part of what makes the A and M series so damn appealing is their great efficiency per Watt and relatively low thermal restrictions on packaging/placement.

Throwing a 5G modem, while very possible and quite "easy" for AAPL who owns the entire design vertically, into the package means less W for compute/memory/other application specific cores and means more heat to have to dissipate in the same physical area.

It's one reason why SoCs with integrated LTE/5G can't be competitive: how you gonna spend that precious space, power in, and heat out that you are largely bounded by physics around? Choose wisely! And, wisely so, I think AAPL has chosen to max out memory bandwidth, ASICs that accelerate image processing, etc, etc because an off-die solution is not just "good enough" but actually currently, probably preferable in terms of what the market wants.

No one says "I want to buy this iPhone b/c it has an integrated modem on the SoC", but they do say "I want to buy this iPhone because it has incredibly low power real time image processing that makes my portrait mode and cinematic mode video amazing" (maybe not in so many words).

I think SoC-integrated modems make great sense in the low and mid segments of the market, but Apple doesn't really play there and when they do, they just market 1-2 design cycle old products that are now high-margin rather than design specifically for it (even if they package it with a different name in marketing), so I just don't see the reasoning for SoC integration for at least a few more revs when the dynamics of die space, thermals, and power use evolve significantly.

 

[Morgenland]

It's just a "my lastest survey" by a known blowhard. Nothing else.
He always gets money or attention for each of his ideas, no matter how coherent his content is. Otherwise, he wouldn't be doing this job.
It's amazing how much people still listen to him.

Of course, he feeds on interesting topics. But the way he realigns facts with unproven assertions has nothing to do with informative enlightenment.

 

[SactoGuy18]

But in this case, fortunately! One thing I hated about my iPhone X was that when the reception quality drops, I get WAY too many dropouts on calls and just as bad, no Internet connectivity. 😒 At least with the current Qualcomm Snapdragon radio modem chip on my iPhone 12, in low signal situations I rarely get dropped calls and even Internet connectivity still works (mostly).