Almost ten years ago now, Qualcomm and Apple began suing each other. Apple's position was Qualcomm was playing hardball with patents that ought have a FRAND license. Qualcomm's position was Apple was being mean, and then also to sue various downstream part contractors for not paying royalties they claimed were owed to them but that Apple was contesting. In general Apple won more of the litigation than they lost (but not all), and the litigation eventually ended in a large settlement. But Apple said at the time of the settlement, that it was untenable to have to source all of their LTE stuff from Qualcomm and they planned to secure their supply chain in the coming years. This is that statement now coming to fruition.
So in toto, my impression is that this isn't fundamentally about seeking any particular cost savings or technical difference, so much as it Apple trying to disentangle themselves from external suppliers who they view as hostile or undependable.
I feel like this is getting a little nuanced. “They bought a company to get a head start, thus they didn’t actually create their own modem.”
It’s been roughly four years since that purchase. Surely they’ve done some work since then that would justify labeling it as “Apple created their own modem.”
I feel like this would be comparable to me saying I cooked dinner and someone replying “you bought the food from Kroger and the cookbook from Barnes & Noble, so it doesn’t count” while I stare at a pile of dirty dishes like “sure feels like it counts.”
The PASemi team took two years from acquisition to first Apple silicon. I'm guessing the Intel stuff was a lot messier. It's not like Intel was a streamlined company that excelled at embedded SoCs.
I agree with where you are going with that, but the person I responded too mentioned that they could buy a part part of qualcom but instead just made their own.
But instead they bought part of intel. It doesn't diminish what Apple is building, but I think given the context it is important.
I was more emphasizing that if Apple was willing to "go it alone" on their CPUs (which they did by an acquisition, to be sure) then they probably have no issue at all "going alone" on the radio/modem, and I'm actually a bit surprised they didn't do that first - I guess modems are more complicated/difficult than CPUs.
First, Apple is known to squeeze every penny from their suppliers to improve their bottom line.
Second, likewise, Apple sued every single wireless patent holders (eg, Nokia, Ericsson, Qualcomm, etc) to negotiate better licensing terms -- but, to be fair, it's not uncommon for other smartphone OEMs to do so in court.
Third, Apple's legal argument was always that Qualcomm's per-device (ie, final product) royalty basis licensing offer was not FRAND. Apple believes that the royalty should be based on the cost of Qualcomm's chips instead. But, of course, there is no legal basis or industry practice to support Apple's unfounded claim and, consequently, LOST EVERY lawsuit based on this claim for past 10+ years.
Fourth, "... then also to sue various downstream part contractors ..." No, you don't sue everyone in supply-chain. The wireless patent holders collect their royalty only once and at the top of the supply-chain, ie, smartphone OEMs, to maximize their profit. and that's been the industry practice for much of the past 25+ years (possibly even longer).
Fifth, "... it was untenable to have to source all of their LTE stuff from Qualcomm ..." Umm.. Stop, stop, stop!
So in conclusion, it's all about saving Apple's margin, that's all. I don't blame any company for trying to minize their cost, but the recent event with Qualcomm, ie, FTC's attack on Qualcomm with Apple orchestrating behind the curtain, was highly unethical and abuse of their political power.
There's nothing seriously insulting in that post. Calling out biased Apple defenders for their misleading claims isn't an insult no matter how much some would want it.
> First, Apple is known to squeeze every penny from their suppliers to improve their bottom line.
This is just unequivocally false. I used to work for a semiconductor company that was a supplier for Apple. Apple would pay a premium compared to other companies.
The semiconductor game is kind of rigged, assuming you don't have your own fab, everyone knows your costs. They know how much a wafer costs and they know the size of your device, so they have a pretty good idea of what it costs you per device (they don't know your yield, but can guess a range).
Everyone uses that information to keep your profit margins fairly low. Apple would pay a premium compared to other manufacturers. You could probably look at the financial reports of various apple suppliers and reverse engineer that information.
Apple is extremely risk averse, them making their own modems is almost certainly about reducing supply chain risk.
Profit margins will be lower down on the priority list. They can just increase the price of the phone to improve those.
For the record, I am not an Apple fan, I disliked working for one of their suppliers and dislike their products. But they do pay their suppliers a premium.
> Third, Apple's legal argument was always that Qualcomm's per-device (ie, final product) royalty basis licensing offer was not FRAND. Apple believes that the royalty should be based on the cost of Qualcomm's chips instead. But, of course, there is no legal basis or industry practice to support Apple's unfounded claim and, consequently, LOST EVERY lawsuit based on this claim for past 10+ years.
Did you check FTC v Qualcomm antitrust case?
Can you stop your lies?
>> Did you check FTC v Qualcomm antitrust case? Can you stop your lies?
Sure, isn't that the case Qualcomm reversed and won? Sure, most legal analysts had expected Apple's sweetheart judge Lucy Koh to hand a victory to FTC as it was no coincidence how the case ended up on her desk in the first place. This was however unanimously reversed due to its flawed constrution of Qualcomm's monopoly in premium LTE market and Judge Koh's expansive interpretation (in favor of Apple/FTC).
But wrt Apple's claim on royalty basis? Not so much.
> Apple's legal argument was always that Qualcomm's per-device (ie, final product) royalty basis licensing offer was not FRAND. Apple believes that the royalty should be based on the cost of Qualcomm's chips instead. But, of course, there is no legal basis or industry practice to support Apple's unfounded claim and, consequently, LOST EVERY lawsuit based on this claim for past 10+ years.
Qualcomm demanded a percentage of the retail selling price of phones using it's modems.
Similarly, Google bought Motorola for it's patent portfolio and famously lost a lawsuit against Microsoft after demanding a 2.5% cut of the retail cost of Microsoft hardware (estimated to be about 4 Billion dollars a year) using it's standards essential patents.
> Motorola sent a letter to Microsoft asking it to pay as much as $4 billion per year to license patents relating to the 802.11 standard that underpins Wi-Fi and the H.264 video encoding standard.
Google lost the case and the appeal, and the courts imposed a flat 3 million dollar per year licensing fee instead of the ~4 Billion a year that Google demanded.
>> " ... Qualcomm demanded a percentage of the retail selling price of phones using it's modems. ..."
No, that's not how it works. Apple does NOT have licensing agreement with Qualcomm. Apple's contract manufacturer, or CM's -- such as Foxconn -- are the primary licensees of Qualcomm's wireless portfolio. Apple's royalty -- or indirect payment to Qualcomm via CM's -- likewise is based on Foxconn's pre-existing licensing agreement with Qualcomm (which even predates Apple's first iPhone) and on Foxconn's manufacturing/wholesale price, which is typically 1/3 or 1/4 of Apple's retail price (often under $400 -- not surprisingly Qualcomm's licensing fee maxes out at $400).
>> Similarly, ... ( using it's standards essential patents. ... So, this is hardly an unfounded claim.
The Motorola case does not support Apple's royalty-basis theory at all, First, Judge Robart does not invalidate end-user device as royalty-basis, or favors SSPPU promoted by Apple. Second, his decision was based on two critical considerations: merits/value contribution/significance of the patents asserted against Microsoft. And, in order to determine an appropriate price/range, if there existed any comparable RAND licensing practice adapted by other patent holders in the industry. In Motorola's case, their patents were relatively frivolous and insignificant in value/merit contributed, and their royalty rates were simply out of whack -- certainly 1000x out of common industry practice in the video encoding/802.11 wireless business.
Qualcomm is an industry pioneer with numerous seminal contribution under their belt. Their licensing practice also conforms with long-held de facto industry practices; though there are some aspects of their licensing business that are bit quite anticompetitive -- royalty-basis is not among them however.
No, that's exactly what Qualcomm demanded. The fact that they have changed their demands over time after various court losses does not change the past.
>>No, that's exactly what Qualcomm demanded. The fact that they have changed their demands over time after various court losses does not change the past.
Sure, that's a fig of your imagination. You simply don't know how Qualcomm's, or the industry's, licensing model works. I also suspect that you didn't even bother to read the court case you cited, the Motorola case.
Let me just say again for one last time: Apple lost or settled every lawsuit involving wireless patent royalty basis claim. If Apple ever prevailed in one case cracking the industry's per-device royalty-basis model, they wouldn't have bought Intel's failing wireless business out of desperation after losing the last legal battle with Qualcomm in 2019, which Apple settled just one day after the trial started.
Let's just admit it. Now that the cellular industry in the States has transitioned past 3G, Qualcomm no longer has patents that are required for compatibility, but are not bound by commitments to license under FRAND terms.
The Google/Motorola/Microsoft case (and associated FTC order) shows that Qualcomm will not be able to pull the same price gouging shenanigans with patents that are encumbered by requirements to license under FRAND terms.
The standards essential 4G/5G patents are all encumbered by requirements to license to all comers on FRAND terms.
Which means Qualcomm is severely overvalued now that anyone with the means will be able to develop their own modems without anticompetitive interference. How long until Samsung starts selling their own in-house modem to others?
>> The Google/Motorola/Microsoft case (and associated FTC order) shows that Qualcomm will not be able to pull the same price gouging shenanigans with patents that are encumbered by requirements to license under FRAND terms.
sigh. Again you are throwing around court cases that don't support Apple or your claim. The Motorola case isn't nearly "associated" in anyway to the FTC case. Further, the FTC case was reversed in an unanimous decision -- due to its flawed construction of Qualcomm's monopoly in premium LTE segment and Apple's sweetheart judge Lucy Koh's overly expansive intrepretation of the antitrust law. The FTC's case is all, but dead and there would be no further legal challenge -- no SCOTUS justice is going to put up with a deceptive kabuki threater concocted by Apple and their Democratic operatives at FTC.
>> The standards essential 4G/5G patents are all encumbered by requirements to license to all comers on FRAND terms.
Qualcomm's licensing terms are FRAND -- just because Tim Apple doesn't like it doesn't mean that it's not FRAND. It's not about Tim Apple's feeling or Apple's margin, but rather about reasonable, acceptable industry licensing practices.
Sure, just b/c you haven't actually read anything you cited, simply don't understand how licensing/indusry works, doesn't mean that my arguements aren't factual. Thank you for playing.
Interesting that no one talks about the elephant in the room - why are technologies for supposed public standards patentable in the first place?
IMO, all standards that are mandated to be used by a government or a regulatory agency should be patent and royalty free and accessible by everyone, alone to have a healthy competition space instead of (effectively) a Qualcomm monopoly on the vendor side.
> why are technologies for supposed public standards patentable in the first place
Because they were developed in private by the stakeholders and then submitted as part of the standardization process. Money did get spent. At the core of the situation is multiple private companies all contribute technology towards a standard that a larger group of companies can implement. Without the option of license fees there's little incentive for companies to invest in developing these technologies.
Part of the standardization process is making the patent licenses available at a "fair and reasonable" rate (FRAND). When the patent holders have large patent pools of their own they cross-license in lieu or in addition to paying actual license fees.
Universities aren't always better than private companies with respect to patents and licensing. Sometimes they're much worse because there's no real reciprocity of licensing.
Governments could set up national labs to develop communication technologies but they're not going to be competitive with private companies pay wise. If you're a radio engineer you can be stuck on a government schedule salary or get a nice six figure "Tech Company" salary and fat stock options. Why work at the government lab when your same skills will get to twice as much at Apple or Qualcomm?
I'm not arguing for the status quo but the alternatives aren't necessarily better for the public nor the engineers doing the work.
> Also are there other benefits like integrating onto the main chip or something?
Yes, but the iPhone chips may be tiled by the time this happens, so it might not be "on chip." This is probably for the best, as the RF analog stuff a modem chip has to deal with requires certain compromises.
> I'm curious what kind of cost savings this is going to provide Apple with per-device
My understanding is that Qualcomm charges a percentage of the total device cost, rather than a fixed price, so as phones get more expensive, that cost increases.
Not sure why you are getting downvoted, because its true. Qualcomm is one of the main things holding Android devices back, and their driver policy is shitty.
It's the same reason why your home router runs an ancient version of Linux that can never be updated. The hardware requires a proprietary firmware blob that only links with a specific kernel. You can never update the system independently, you're at the mercy of Qualcomm releasing updated blobs.
Basically Qualcomm only releases proprietary Linux drivers for their chips, provided as a precompiled binary blob, and they only offer ~2 years of support before dropping them all together.
Because the Linux kernel does not have a stable binary interface for driver modules, getting an old driver to work with a new version of the kernel can become a hacky nightmare.
I have to imagine this will bring considerable battery life improvements since the radio will be fab on the same latest generation TSMC node that Apple gets access too (as opposed to Qualcomm fab'ing on 3+ year old nodes).
Qualcomm fabs on Samsung and TSMC. Unlike Apple who put all their eggs in one basket, TSMC, Qualcomm always used multiple fabs -- usually rotate between releases -- to produce their chips. Qualcomm also was one of SMIC's largest customers before they were blacklisted.
The thermals on that would be difficult to deal with I think. Snapdragon had thermal problems for a long time. Have they solved it? What did they end up doing?
Although logic circuits are available at every process node, some specialized usages like radio frequency transceivers aren't available on every process node from TSMC.
The most recent node change from TSMC intended for use in cellular modems was the jump from 16nm to 6nm in 2021.
> Compared with 16FFC RF, its predecessor, N6RF supports 3.2X logic density, 55% power reduction, and 0.38X SRAM scaling in terms of digital KPI. N6RF can deliver ideal RF performance and power efficiency for 5G and Wi-Fi 6 & 6E applications that demand both performance and power to enable flexibility for system optimization.
Could it also be, with the advent of Satellite SOS, that Apple is starting to explore non-conventional mobile RF protocols, and that Qualcomm's product offerings are only really geared towards generic, standards-based mobile networks.
Satellite SOS is based on the same bands used by 5g. Also it is now part of 5G NR Release 17. Qualcomm modems since Sanapdragon X60 had support for these bands. Even before Qualcomm officially announced Snapdragon satellite with X70 modem, Huawei was offering satellite communication with Snapdragon X65 in China, which is the same modem found in iPhone 14 series.
Does that make it difficult to make a cellular modem or just expensive? Those are two different things.
I'm assuming that they were asking why it's so difficult that so few vendors do it and why it's taking Apple so long.
For example, Intel had trouble competing with Qualcomm's modems - they were always a generation or two behind. Apple bought Intel's modem business around 3.5 years ago and it looks like it'll take 5 years for them to come out with their first Apple-modem phone - and Apple wasn't starting from scratch. Apple was starting with a billion dollar purchase of Intel's modem business including all that IP, 2,200 Intel modem employees, 17,000 wireless patents, etc.
It's gotten to the point that Qualcomm is basically the only game in town for high-end smartphones. Even Samsung who has their own modems went with Qualcomm worldwide for the Galaxy S23 - which they hadn't done for previous models.
If it were just a matter of paying money, Apple and Intel would have created a competitive chip quickly.
Apple and Qualcomm have been fighting over patents for a decade. I presume Apple is designing their own chip that does not require any licensed tech from Qualcomm, which is what makes it harder.
But one reason was that if you take the specification and implement it from scratch you end up with a significantly worse performing solution than simply buying it from Qualcomm.
It could also explain why Apple is putting this first in their iPhone SE4 as their users will already expect not to own the fastest phone.
It's worse than that. Often the specification isn't complete. You need to pass the conformance and interoperability tests. The chipset needs to support older standards as well and pass those tests. I suspect they'll find a number of "quirks" in various equipment, that will need to be worked around and then retested. It's a long process.
Regardless of the business/legal reasons, they're also just incredibly complex from a technical perspective. The power requirements, the RF side is black magic, the protocol side isn't easy, etc.
It's interesting to see discussions here that are adamant that this is only about improving profit, or only about quality (battery), or only about getting back at qualcomm.
Apple has repeatedly shown that you don't have to make decisions on one to compromise the other. You can launch your own chip (M1), and improve quality (battery, perf), and improve profits, and get back at intel.
It's the same here. Watch the iFixit teardown videos to see just how big the standalone qualcomm modem chip is. I knew it was big, but it was even bigger than I thought.
Apple moving to an integrated chip can be a win in both quality (battery improvement and space on the SoC to put other stuff), cost, and getting back at qualcomm (who is getting more and more like Intel of mobile world, comfortable, monopolistic, litigative). They've proven that a vertically integrated SoC is better in every way.
So maybe stop arguing whether it's "this or that" it's this and that.
yea but apple has also chosen to use lightning with shit xfer speeds just to lock in its customers. m1 might have been nice but not every descision they made was good (look how they nerfed qualcomm iphones back in the day because the intel models couldnt compete with it)
First I'm hearing about nerfed iphones, but if true this sounds like a reasonable decision from a business perspective.
They wouldn't want customers worrying about whether they got the better qualcomm iPhone X or the slower intel iPhone X.
Apple could have chosen to make everything with the faster qualcomm chips, but they prob wanted to derisk their supply chain by diversifying onto two chip manufacturers (and perhaps play them against each other for negotiating power)
Either ways, once you arrive a the result that you need to chip manufacturers making chips for the same phones, you need to make sure your customers have practically the same experience regardless of the underlying hardware they end up with
The reason they went with Lightning was because the other option was Micro-USB at the time. Lightning is so so so much more robust than that awful USB connector and can be inserted any way round. Bring on 2015 and USB-C appeared which Apple jumped on pretty quickly with their MacBooks.
The iPhone is an outlier because due to the insane number of devices out there, if you change it now everyone's Lightning accessories are dead in the water. But it doesn't really matter now. Only thing Lightning gets used for most users is charging and it works fine for that and people have cables and chargers already floating around. This isn't a big deal really. They ship a USB-C to Lightning cable with the last 2 iPhones I bought and all my chargers are USB-C.
was there a point you are making or just a natural language response?
your use case isn't the only use case, expanding the possible use cases in obvious ways expands what users would do, such as upgrading 10 year old technology. News at 11.
The point is that sure you can punch yourself in the balls, complain it hurts then call for the regulators to provide a mechanism for you to stop punching yourself in the balls.
Or you could just stop punching yourself in the balls i.e. use WiFi and FGS just pay $3 for some storage rather than coming up with a wobbly straw man.
My experience is fairly vast. I use iOS for music/video/photography (iPad Pro), and support 11 other iOS devices (excluding watches, homepods) across my family who range from biochemical engineers to students to children to retirees to software engineers. That use case NEVER comes up these days in my experience unless you're hurting yourself.
this isn't a problem I have, it is a problem I know other people have. solved by faster transfer speeds than what Wifi or any airdrop combo offers. I am completely aware of how to avoid the problem, I am also completely aware of how to make it less of a problem with faster transfer speeds.
why are you worried about accessories? Apple absolutely is not worried about that, one of their favorite things to do is pull the rug under accessory holders.
> Apple absolutely is not worried about that, one of their favorite things to do is pull the rug under accessory holders.
This simply isn’t borne out by fact. The entire iPod, iPhone and iPad (non-Pro) range, going back over 20 years, have had exactly two connectors.
I bet there isn’t someone at Samsung or Motorola who can even tell you how many different connectors their products have used over that period of time, let alone what they were.
You know they won't. Take a look at what they did on the iPad lineup for what is coming on the iPhones:
Basic iPads only support USB2.0 transfer rates (480 Mbps). The iPad Pros get full Thunderbolt speeds. Expect the same for iPhones, I'd say.
The EU mandate is about forcing a single connector type. It doesn't say anything about quality of service of the connector, because ultimately it depends on the device.
I definitely tend to agree. So many decisions, not just in product design or strategy, DO happen to take one option above all others. Increasing the overlap and moving all indicators at once is how you get ahead these days. Companies, products and strategies that focus on one thing above all else will probably tend to start not really cutting it when it comes to success.
Also an overlooked point is diversification: Apple's chip aspirations have allowed functionality that isn't trivially/inexpensively possible with standard parts. This also makes copying any unique ideas much more difficult, since competitors can't just approach the chipmaker for the same part.
Apple could very well bring innovations to cellular modems that make their products behave more seamlessly or are simpler to use. Additionally they may introduce features that aren't currently possible with Qualcomm's designs.
Maybe although it’s going to be harder to do anything on that front because a modem has to follow standards and so non-standard extensions on the radio side are tricky. And since that’s the case, landing really new features can be tricky.
I wonder if they’re looking to do a city-level mesh network that only iPhones get to participate in where your phone can simultaneously transmit and receive on LTE (although I don’t know what kind of regulatory things come into play for that idea).
I think the main advantage is power, security (QC’s software stack was a giant mess last I looked), stability (see previous point) and time to market for new standards into as many products (improving battery life, cost and size of wearables using their chip). I’m sure how long it took them to get LTE from QC in a fit and finish that was acceptable wasn’t a lost frustration (since that’s repeated constantly with 3g and 4g previously too).
Also, at Apple’s scale, I’m sure being able to single source a chip that is vertically integrated across billions of phones means the costs of integrating it into other smaller product lines is even more amortized so more things will get the option of a modem.
It also wouldn’t surprise me if at some point Apple became their own ISP and could really start to add non-standard functionality and not be tied to individual ISPs. Their SOS satellite play is part of that I think but I’m not yet sure how terrestrial internet and satellite will interplay - maybe they need to vertically integrate the cellular modem with the satellite piece to get things to work at good enough power? Or maybe they’ll have base stations repeating satellite internet via traditional cellular/Wi-Fi?
If so, then yes that's a shockingly large piece of the phone.
Edit: @rdsubhas: Your link confirms that the modem really does take up a sizeable portion of the entire volume of the phone. Now Apple's desire to get more control of it makes a lot more sense. Thank you!
I tend to think that big companies as they grow lose their focus and fail to adapt and further specialise. Just look at Google, they are imploding under their weight.
It’s so common to see Google bashing on HN now that it’s become a cliche. Google has never been good at monetizing products beyond their search. I don’t think it has anything to do with growing too large for its own good. Google is this generation’s Xerox PARC.
> You can launch your own chip (M1), and improve quality (battery, perf), and improve profits, and get back at intel.
It's less about "getting back at intel" and more about having so much money that they need somewhere to invest it or increase the dividend and send that profit to the shareholders. The past decade, investors have constantly demanded Apple to turn over more/all of their cash to shareholders. A defense against activist investors is to show shareholders you are investing for the future. Bringing more production in-house is a great way to do it since it would eliminate future expenditures and increase profits in the future. Of course it all depends on execution. If Apple keeps minting money and their in-house hardware stack proves a success, I wouldn't be surprised if they start creating an in-house app stack.
If you have enough cash and in house engineering skills to pull this off it makes sense. You get full control over the baseband firmware, power management, supply chain, cost, integration, process, security surface, everything.
Hard to resist the extrapolation from Visual Voicemail to see Apple one day in the near future shipping software for carriers to deploy on their networks to support features E2E integrated with their modems.
I've a sneaking suspicion that Apple's vertical consolidation of the iPhone supply chain makes the company much more fragile. There are clearly lots of economies of scale by building custom silicon that's optimized for their use cases. However, if their sales drop, they become saddled with factories, product lines and lease commitments that further drag on costs and make it hard to react and change direction.
I don't think that's such a problem for apple, since there is such a market for 3 year old products from them. iPhone 8 was effectively a product being sold for 5 straight years from apple, as the 8 and then as the second generation SE. For their computers, they also sell them for a few years past when they were cutting edge too, since there's always a market for a macbook that's a few hundred bucks cheaper and most consumers don't know what the difference between M1 or M2 might be anyhow.
One way to ease that pain is for apple to make itself a supplier to others rather than the supplier and consumer of it's own parts. A top of the line iPhone costs ~$500 to 1000 bucks! why not try and go after the middle of the market by licencing the chipsets and other tech so that they can then supply mid teir companies that can compete in the android space with the lure of apple chipsets and hardware.
Ofc Apple can do it inhouse but I as an "economist" wonder who can specialise better? Apple is bigger and more powerful than Qualcomm but do they really have that comparative advantage to deliver better product. Qualcomm has years of know-how but looking from Apple's point of view it is better to be self-sufficient then to rely on the 3rd party. I guess time will show.
Apple has the volume. Random search shows they have roughly a quarter of the world smartphone sales (I also would expect Apple users to upgrade more frequently). Apple has just several SoC packages with the same cores across all its products. Samsung (2nd is sales volumes) has many more product lines and uses vastly different silicon.
> Amon did not confirm whether Apple would pay Qualcomm QTL licenses if it moves to its own modems, but said royalty was “independent from providing a chip.”
They'll probably get around by paying the full price. This was what Samsung did with Exynos, they didn't used Snapdragon SoC yet they paid the full price of the SoC (not just the FRAND patent fees for the modem).
109 comments
[ 3.3 ms ] story [ 209 ms ] threadAlso are there other benefits like integrating onto the main chip or something? Or other business benefits?
So in toto, my impression is that this isn't fundamentally about seeking any particular cost savings or technical difference, so much as it Apple trying to disentangle themselves from external suppliers who they view as hostile or undependable.
So they’ll roll their own. They built a CPU so a modem shouldn’t be incredibly difficult.
in 2019 they bought the modem business from Intel.
It’s been roughly four years since that purchase. Surely they’ve done some work since then that would justify labeling it as “Apple created their own modem.”
I feel like this would be comparable to me saying I cooked dinner and someone replying “you bought the food from Kroger and the cookbook from Barnes & Noble, so it doesn’t count” while I stare at a pile of dirty dishes like “sure feels like it counts.”
But instead they bought part of intel. It doesn't diminish what Apple is building, but I think given the context it is important.
Second, likewise, Apple sued every single wireless patent holders (eg, Nokia, Ericsson, Qualcomm, etc) to negotiate better licensing terms -- but, to be fair, it's not uncommon for other smartphone OEMs to do so in court.
Third, Apple's legal argument was always that Qualcomm's per-device (ie, final product) royalty basis licensing offer was not FRAND. Apple believes that the royalty should be based on the cost of Qualcomm's chips instead. But, of course, there is no legal basis or industry practice to support Apple's unfounded claim and, consequently, LOST EVERY lawsuit based on this claim for past 10+ years.
Fourth, "... then also to sue various downstream part contractors ..." No, you don't sue everyone in supply-chain. The wireless patent holders collect their royalty only once and at the top of the supply-chain, ie, smartphone OEMs, to maximize their profit. and that's been the industry practice for much of the past 25+ years (possibly even longer).
Fifth, "... it was untenable to have to source all of their LTE stuff from Qualcomm ..." Umm.. Stop, stop, stop!
So in conclusion, it's all about saving Apple's margin, that's all. I don't blame any company for trying to minize their cost, but the recent event with Qualcomm, ie, FTC's attack on Qualcomm with Apple orchestrating behind the curtain, was highly unethical and abuse of their political power.
PS. insulting languages removed
Sounds great. Apple should also base their app store tax on the cost of hosting.
This is just unequivocally false. I used to work for a semiconductor company that was a supplier for Apple. Apple would pay a premium compared to other companies.
The semiconductor game is kind of rigged, assuming you don't have your own fab, everyone knows your costs. They know how much a wafer costs and they know the size of your device, so they have a pretty good idea of what it costs you per device (they don't know your yield, but can guess a range).
Everyone uses that information to keep your profit margins fairly low. Apple would pay a premium compared to other manufacturers. You could probably look at the financial reports of various apple suppliers and reverse engineer that information.
Apple is extremely risk averse, them making their own modems is almost certainly about reducing supply chain risk.
Profit margins will be lower down on the priority list. They can just increase the price of the phone to improve those.
For the record, I am not an Apple fan, I disliked working for one of their suppliers and dislike their products. But they do pay their suppliers a premium.
Did you check FTC v Qualcomm antitrust case? Can you stop your lies?
Sure, isn't that the case Qualcomm reversed and won? Sure, most legal analysts had expected Apple's sweetheart judge Lucy Koh to hand a victory to FTC as it was no coincidence how the case ended up on her desk in the first place. This was however unanimously reversed due to its flawed constrution of Qualcomm's monopoly in premium LTE market and Judge Koh's expansive interpretation (in favor of Apple/FTC).
But wrt Apple's claim on royalty basis? Not so much.
When are you going to stop spreading misinformation?
Qualcomm demanded a percentage of the retail selling price of phones using it's modems.
Similarly, Google bought Motorola for it's patent portfolio and famously lost a lawsuit against Microsoft after demanding a 2.5% cut of the retail cost of Microsoft hardware (estimated to be about 4 Billion dollars a year) using it's standards essential patents.
> Motorola sent a letter to Microsoft asking it to pay as much as $4 billion per year to license patents relating to the 802.11 standard that underpins Wi-Fi and the H.264 video encoding standard.
https://www.computerworld.com/article/2955773/microsoft-best...
Google lost the case and the appeal, and the courts imposed a flat 3 million dollar per year licensing fee instead of the ~4 Billion a year that Google demanded.
https://www.vox.com/2015/7/30/11615248/google-loses-bid-to-o...
So, this is hardly an unfounded claim.
No, that's not how it works. Apple does NOT have licensing agreement with Qualcomm. Apple's contract manufacturer, or CM's -- such as Foxconn -- are the primary licensees of Qualcomm's wireless portfolio. Apple's royalty -- or indirect payment to Qualcomm via CM's -- likewise is based on Foxconn's pre-existing licensing agreement with Qualcomm (which even predates Apple's first iPhone) and on Foxconn's manufacturing/wholesale price, which is typically 1/3 or 1/4 of Apple's retail price (often under $400 -- not surprisingly Qualcomm's licensing fee maxes out at $400).
>> Similarly, ... ( using it's standards essential patents. ... So, this is hardly an unfounded claim.
The Motorola case does not support Apple's royalty-basis theory at all, First, Judge Robart does not invalidate end-user device as royalty-basis, or favors SSPPU promoted by Apple. Second, his decision was based on two critical considerations: merits/value contribution/significance of the patents asserted against Microsoft. And, in order to determine an appropriate price/range, if there existed any comparable RAND licensing practice adapted by other patent holders in the industry. In Motorola's case, their patents were relatively frivolous and insignificant in value/merit contributed, and their royalty rates were simply out of whack -- certainly 1000x out of common industry practice in the video encoding/802.11 wireless business.
Qualcomm is an industry pioneer with numerous seminal contribution under their belt. Their licensing practice also conforms with long-held de facto industry practices; though there are some aspects of their licensing business that are bit quite anticompetitive -- royalty-basis is not among them however.
No, that's exactly what Qualcomm demanded. The fact that they have changed their demands over time after various court losses does not change the past.
Sure, that's a fig of your imagination. You simply don't know how Qualcomm's, or the industry's, licensing model works. I also suspect that you didn't even bother to read the court case you cited, the Motorola case.
Let me just say again for one last time: Apple lost or settled every lawsuit involving wireless patent royalty basis claim. If Apple ever prevailed in one case cracking the industry's per-device royalty-basis model, they wouldn't have bought Intel's failing wireless business out of desperation after losing the last legal battle with Qualcomm in 2019, which Apple settled just one day after the trial started.
The Google/Motorola/Microsoft case (and associated FTC order) shows that Qualcomm will not be able to pull the same price gouging shenanigans with patents that are encumbered by requirements to license under FRAND terms.
The standards essential 4G/5G patents are all encumbered by requirements to license to all comers on FRAND terms.
Which means Qualcomm is severely overvalued now that anyone with the means will be able to develop their own modems without anticompetitive interference. How long until Samsung starts selling their own in-house modem to others?
sigh. Again you are throwing around court cases that don't support Apple or your claim. The Motorola case isn't nearly "associated" in anyway to the FTC case. Further, the FTC case was reversed in an unanimous decision -- due to its flawed construction of Qualcomm's monopoly in premium LTE segment and Apple's sweetheart judge Lucy Koh's overly expansive intrepretation of the antitrust law. The FTC's case is all, but dead and there would be no further legal challenge -- no SCOTUS justice is going to put up with a deceptive kabuki threater concocted by Apple and their Democratic operatives at FTC.
>> The standards essential 4G/5G patents are all encumbered by requirements to license to all comers on FRAND terms.
Qualcomm's licensing terms are FRAND -- just because Tim Apple doesn't like it doesn't mean that it's not FRAND. It's not about Tim Apple's feeling or Apple's margin, but rather about reasonable, acceptable industry licensing practices.
IMO, all standards that are mandated to be used by a government or a regulatory agency should be patent and royalty free and accessible by everyone, alone to have a healthy competition space instead of (effectively) a Qualcomm monopoly on the vendor side.
Because they were developed in private by the stakeholders and then submitted as part of the standardization process. Money did get spent. At the core of the situation is multiple private companies all contribute technology towards a standard that a larger group of companies can implement. Without the option of license fees there's little incentive for companies to invest in developing these technologies.
Part of the standardization process is making the patent licenses available at a "fair and reasonable" rate (FRAND). When the patent holders have large patent pools of their own they cross-license in lieu or in addition to paying actual license fees.
Governments could set up national labs to develop communication technologies but they're not going to be competitive with private companies pay wise. If you're a radio engineer you can be stuck on a government schedule salary or get a nice six figure "Tech Company" salary and fat stock options. Why work at the government lab when your same skills will get to twice as much at Apple or Qualcomm?
I'm not arguing for the status quo but the alternatives aren't necessarily better for the public nor the engineers doing the work.
Yes, but the iPhone chips may be tiled by the time this happens, so it might not be "on chip." This is probably for the best, as the RF analog stuff a modem chip has to deal with requires certain compromises.
> Or other business benefits?
Just getting away from Qualcomm is a big one, lol
Apple isnt the only company they started a feud with. Intel accused them of being anticompetitive in a remarkable rant back in 2018: https://newsroom.intel.com/news/another-attempt-stifle-compe...
My understanding is that Qualcomm charges a percentage of the total device cost, rather than a fixed price, so as phones get more expensive, that cost increases.
This was resolved back in 2019, with Apple paying $5B to Qualcomm
https://www.apple.com/newsroom/2019/04/qualcomm-and-apple-ag...
Basically Qualcomm only releases proprietary Linux drivers for their chips, provided as a precompiled binary blob, and they only offer ~2 years of support before dropping them all together.
Because the Linux kernel does not have a stable binary interface for driver modules, getting an old driver to work with a new version of the kernel can become a hacky nightmare.
I have to imagine this will bring considerable battery life improvements since the radio will be fab on the same latest generation TSMC node that Apple gets access too (as opposed to Qualcomm fab'ing on 3+ year old nodes).
Pretty sure I've read on Macrumors that they are developing their modem on TSMC 6nm.
Sometimes it's not worth the hit to yield to put it on die. On package might be a better tradeoff.
The most recent node change from TSMC intended for use in cellular modems was the jump from 16nm to 6nm in 2021.
> Compared with 16FFC RF, its predecessor, N6RF supports 3.2X logic density, 55% power reduction, and 0.38X SRAM scaling in terms of digital KPI. N6RF can deliver ideal RF performance and power efficiency for 5G and Wi-Fi 6 & 6E applications that demand both performance and power to enable flexibility for system optimization.
https://rf.tsmc.com/english/dedicatedFoundry/technology/RF.h...
https://www.anandtech.com/show/16030/tsmc-updates-on-node-av...
Intel did it with some firmware / microcode, and it took too much energy.
I'm assuming that they were asking why it's so difficult that so few vendors do it and why it's taking Apple so long.
For example, Intel had trouble competing with Qualcomm's modems - they were always a generation or two behind. Apple bought Intel's modem business around 3.5 years ago and it looks like it'll take 5 years for them to come out with their first Apple-modem phone - and Apple wasn't starting from scratch. Apple was starting with a billion dollar purchase of Intel's modem business including all that IP, 2,200 Intel modem employees, 17,000 wireless patents, etc.
It's gotten to the point that Qualcomm is basically the only game in town for high-end smartphones. Even Samsung who has their own modems went with Qualcomm worldwide for the Galaxy S23 - which they hadn't done for previous models.
If it were just a matter of paying money, Apple and Intel would have created a competitive chip quickly.
But one reason was that if you take the specification and implement it from scratch you end up with a significantly worse performing solution than simply buying it from Qualcomm.
It could also explain why Apple is putting this first in their iPhone SE4 as their users will already expect not to own the fastest phone.
Apple has repeatedly shown that you don't have to make decisions on one to compromise the other. You can launch your own chip (M1), and improve quality (battery, perf), and improve profits, and get back at intel.
It's the same here. Watch the iFixit teardown videos to see just how big the standalone qualcomm modem chip is. I knew it was big, but it was even bigger than I thought.
Apple moving to an integrated chip can be a win in both quality (battery improvement and space on the SoC to put other stuff), cost, and getting back at qualcomm (who is getting more and more like Intel of mobile world, comfortable, monopolistic, litigative). They've proven that a vertically integrated SoC is better in every way.
So maybe stop arguing whether it's "this or that" it's this and that.
They wouldn't want customers worrying about whether they got the better qualcomm iPhone X or the slower intel iPhone X.
Apple could have chosen to make everything with the faster qualcomm chips, but they prob wanted to derisk their supply chain by diversifying onto two chip manufacturers (and perhaps play them against each other for negotiating power)
Either ways, once you arrive a the result that you need to chip manufacturers making chips for the same phones, you need to make sure your customers have practically the same experience regardless of the underlying hardware they end up with
Consistency is key
The iPhone is an outlier because due to the insane number of devices out there, if you change it now everyone's Lightning accessories are dead in the water. But it doesn't really matter now. Only thing Lightning gets used for most users is charging and it works fine for that and people have cables and chargers already floating around. This isn't a big deal really. They ship a USB-C to Lightning cable with the last 2 iPhones I bought and all my chargers are USB-C.
The USB 2.0 speeds are a big deal.
If your iCloud Drive is full and your phone is full, your wired backup that should take 20 minutes is going to take 14 hours.
I hope the EU forces their hand on putting USB-C in at USB 3 or Thunderbolt speeds.
Last time I plugged anything in other than for charging was an iPod. Even my DSLR talks to the phone via WiFi.
your use case isn't the only use case, expanding the possible use cases in obvious ways expands what users would do, such as upgrading 10 year old technology. News at 11.
Or you could just stop punching yourself in the balls i.e. use WiFi and FGS just pay $3 for some storage rather than coming up with a wobbly straw man.
My experience is fairly vast. I use iOS for music/video/photography (iPad Pro), and support 11 other iOS devices (excluding watches, homepods) across my family who range from biochemical engineers to students to children to retirees to software engineers. That use case NEVER comes up these days in my experience unless you're hurting yourself.
why are you worried about accessories? Apple absolutely is not worried about that, one of their favorite things to do is pull the rug under accessory holders.
This simply isn’t borne out by fact. The entire iPod, iPhone and iPad (non-Pro) range, going back over 20 years, have had exactly two connectors.
I bet there isn’t someone at Samsung or Motorola who can even tell you how many different connectors their products have used over that period of time, let alone what they were.
Basic iPads only support USB2.0 transfer rates (480 Mbps). The iPad Pros get full Thunderbolt speeds. Expect the same for iPhones, I'd say.
The EU mandate is about forcing a single connector type. It doesn't say anything about quality of service of the connector, because ultimately it depends on the device.
I like comparing it to a Raspberry Pi 4 that has the same thing.
The EU is only forcing Apple to ship with USB-C so that every single mobile device uses the same connector for charging.
Apple could very well bring innovations to cellular modems that make their products behave more seamlessly or are simpler to use. Additionally they may introduce features that aren't currently possible with Qualcomm's designs.
I wonder if they’re looking to do a city-level mesh network that only iPhones get to participate in where your phone can simultaneously transmit and receive on LTE (although I don’t know what kind of regulatory things come into play for that idea).
I think the main advantage is power, security (QC’s software stack was a giant mess last I looked), stability (see previous point) and time to market for new standards into as many products (improving battery life, cost and size of wearables using their chip). I’m sure how long it took them to get LTE from QC in a fit and finish that was acceptable wasn’t a lost frustration (since that’s repeated constantly with 3g and 4g previously too).
Also, at Apple’s scale, I’m sure being able to single source a chip that is vertically integrated across billions of phones means the costs of integrating it into other smaller product lines is even more amortized so more things will get the option of a modem.
It also wouldn’t surprise me if at some point Apple became their own ISP and could really start to add non-standard functionality and not be tied to individual ISPs. Their SOS satellite play is part of that I think but I’m not yet sure how terrestrial internet and satellite will interplay - maybe they need to vertically integrate the cellular modem with the satellite piece to get things to work at good enough power? Or maybe they’ll have base stations repeating satellite internet via traditional cellular/Wi-Fi?
Is this it? https://fdn.gsmarena.com/imgroot/news/22/09/iphone-14-satell...
(From https://m.gsmarena.com/ifixit_iphones_sos_via_satellite_feat...)
If so, then yes that's a shockingly large piece of the phone.
Edit: @rdsubhas: Your link confirms that the modem really does take up a sizeable portion of the entire volume of the phone. Now Apple's desire to get more control of it makes a lot more sense. Thank you!
It's less about "getting back at intel" and more about having so much money that they need somewhere to invest it or increase the dividend and send that profit to the shareholders. The past decade, investors have constantly demanded Apple to turn over more/all of their cash to shareholders. A defense against activist investors is to show shareholders you are investing for the future. Bringing more production in-house is a great way to do it since it would eliminate future expenditures and increase profits in the future. Of course it all depends on execution. If Apple keeps minting money and their in-house hardware stack proves a success, I wouldn't be surprised if they start creating an in-house app stack.
One way to ease that pain is for apple to make itself a supplier to others rather than the supplier and consumer of it's own parts. A top of the line iPhone costs ~$500 to 1000 bucks! why not try and go after the middle of the market by licencing the chipsets and other tech so that they can then supply mid teir companies that can compete in the android space with the lure of apple chipsets and hardware.
I’ve always heard everyone goes through qualcomm because they own a good deal of the IP and patents for the radio/wireless comm stack.
Would love to hear any insights on this.
They'll probably get around by paying the full price. This was what Samsung did with Exynos, they didn't used Snapdragon SoC yet they paid the full price of the SoC (not just the FRAND patent fees for the modem).