Featured Image Credit: Photo by mohamed_hassan
Apple Buys Intel’s Troubled 5G Smartphone Modem Business
Apple has announced that it is buying Intel’s smartphone modem business, giving Apple the in-house capability to produce cellular modem chips for existing 4G and new 5G networks, among other benefits. The $1 billion deal transfers technology, patents, and 2200 employees from Intel to Apple.
Apple doesn’t disclose how many of its 132,000 employees work in its chip division, but based on industry reports and announcements over the last decade—one report said 1000 in 2011—it could easily be double that now. The chip designs and intellectual property are undoubtedly necessary, but hiring this many qualified people in highly specialized professions in one go is at least as important.
Apple and Qualcomm
Apple had increasingly turned to Intel for 4G chips during Apple’s multi-year dispute with Qualcomm over chip prices and intellectual property, first out of competition and then—as Apple alleged in a lawsuit—when Qualcomm wouldn’t supply them. But then Intel’s relatively new CEO abruptly announced in April 2019 that Intel would exit its 5G smartphone modem business, saying there was no path to profitability. The division reportedly lost $1 billion last year.
The next day, Apple and Qualcomm said they had settled all outstanding matters, Apple would pay Qualcomm an undisclosed amount of money, and the two companies had executed a 6-year deal for Qualcomm to provide chips with an option to extend it by another 2 years (see “Apple and Qualcomm Settle Patent Royalty Suits and Agree to Licensing,” 17 April 2019).
That deal surprised industry analysts and regulators alike. But Apple and Intel were reportedly in discussions about the smartphone modem business since mid-2018, when Intel’s new CEO took over, and Apple must have had months of settlement talks already with Qualcomm to make a deal public so quickly.
Making matters more complicated, in May 2019, US District Court Judge Lucy Koh ruled in a lawsuit by the FTC that Qualcomm’s licensing practices were illegal. Her ruling said Qualcomm couldn’t threaten to withhold chips while negotiating a deal in the future. Qualcomm denied it had engaged in that practice, although Apple accused it of exactly that in its own lawsuit. Qualcomm asked for a stay on the decision in the FTC matter, which Koh denied. Qualcomm then filed an appeal on 9 July 2019.
If the appeals court grants its immediate request, Qualcomm’s current deals remain in place. If the court denies Qualcomm’s request, even if it agrees to consider the appeal later, Qualcomm will be forced to change its negotiation practices immediately—but the Apple deal could remain in place since the companies resolved their issues privately. And the outcome of the appeal could be years away and produce even more changes.
To add even more complexity, Apple will likely use Intel 4G chips in the iPhone models due out this fall, despite those chips performing poorly in comparison to Qualcomm’s in 2017, when iPhones used a mix of Intel and Qualcomm cellular tech. Apple freezes hardware designs many months before shipment, and it’s implausible that Apple had time to redesign the new models around Qualcomm chips given the date of the agreement.
If the Intel 5G chips truly aren’t competitive, expect Apple to wait until it has tweaked performance to match or beat Qualcomm to switch. Apple has often been a generation behind other smartphones in adopting cellular data technology upgrades, choosing to focus on battery life and reliable performance ahead of a marketing bullet point on possible speed.
Apple and Intel
Intel’s exit is the latest in a two-decade-long collection of failures to produce industry-leading modem chips, wireless technology standards, and mobile CPUs. (Remember XScale, Atom, and WiMAX?) Intel acquired its modem chip business from Infineon in 2011 for $1.4 billion.
This deal shifts Apple toward the position it loves to be in: no longer dependent on other companies to set its own roadmap and agenda. While Apple had long pursued this direction in software, in 2008, it bought P.A. Semi, then a tiny firm with 150 employees that designed ARM-based CPUs, for $278 million. That acquisition led to Apple’s A-series chips that now power iPhones and iPads, and which many believe will be the computational brains behind at least some of Apple’s future Macs.
Apple also started using its own graphics chips in the iPhone 8, 8 Plus, and X. In addition, Tim Cook and crew committed to $600 million to license power-management chip designs from Dialog Semiconductor and shift 300 employees over to Apple.
5G networks can use a wider variety of frequencies than 3G and 4G networks, requiring complicated new chip designs and networks. The potential is for much higher rates of speed and better support for massive numbers of devices. (If your iPhone currently claims it’s using 5G, don’t believe it—that’s marketing hype and nonsense. Carriers did the same thing with 4G.)
Despite the acquisition, Intel will continue to develop modem chips for computers, Internet of Things devices, and other non-smartphone purposes. Those markets could add up to billions per year in sales as 5G connectivity spreads outside of smartphones and tablets.
It’s a bit ironic that rumors say Apple will give up Intel CPUs and instead move to A series on Mac. While on the iPhone they are expected to drop what they have with Qualcomm and instead go with something developed by Intel.
If you take the word “Intel” out of the two situations, it’s entirely consistent. In both cases, Apple is bringing chip development in house.
In the first case, it’s by developing their own CPUs instead of relying on Intel. In the second, it’s by buying Intel’s entire smartphone modem business rather than relying on Qualcomm.
I guess what makes it interesting to me is that Intel is in general considered highly capable at CPUs. OTOH their efforts in the mobile modem space have been widely considered mediocre (at best). The former is something they’ve been doing since almost 50 years now and they’ve dominated vast parts of the market with their products. The latter is something they bought up from a rival in 2011 and so far didn’t seem to have a whole lot of fortune with, to the point where they wanted to sell it off.
Maybe the Qualcomm deal and the fact that they set it up over many years represents a kind of insurance Apple has in place because they definitely are aware of what Intel has demonstrated they can do and where they have failed.
I think that’s it. Intel has proven itself surprisingly and unfortunately incompetent at mobile-focused chips repeatedly. Qualcomm has been alleged to participate in a lot of anti-competitive behavior (and a judge has found it so), so Apple didn’t want to be entirely in bed with it. But it also wisely couldn’t trust Intel would deliver the goods. When push came to shove and Apple had to rely on Intel smartphone chips, it looks like Intel wasn’t able to meet Qualcomm’s mark.
I agree that it’s ironic! Potentially leaving Intel for in-house ARM chips only to buy an Intel division to make other chips—it just shows Apple thinks there’s a lot of headroom left in the Intel product direction and worth spending a very minimal amount of money.
For all we know, however, Apple could have had smartphone chip designs in progress in house, was stymied by costs and hiring, and will shift the Intel employees it hired to working on an Apple-originated chip design! It’s quite plausible.
Expertise remains one of the things in shortest supply in chip design and fabrication. It’s just not a profession with a lot of extra people hanging around who could do the same job.
And they also need to get 5G phones on shelves asap, since Samsung, etc. are already selling them, and other companies have unveiled models to be released very shortly. From what I’ve read, there’s a giant wave of low priced 5G models getting ready to flood the market in 2019.
I think it’s especially ironic because Apple switched to Intel processor chips because IBM wanted out of chip business and refused to develop RISC chips any further. I doubt that Samsung, Huawei, would consider putting their 5G divisions up for sale, and I also have my doubts that they would want to sell their current state of the art, top of the line modems they use in their most expensive phones to Apple. And Apple has a history of bragging rights to having the best, fastest and smartest. And because they are being beaten to the market in 5G device sales, they are also loosing out on 5G app revenue, and services are increasingly important to Apple’s bottom line.
I’ll bet this is the case. Apple was not early to market with 4G or LTE because of sourcing issues, and I’ll bet it rankled the powers that be back then at least as much as the 5G situation does now. And importing talent during an acquisition tends to cost less than having to offer big increases to lure people away from jobs they are happy with.
For Apple, I think that having the best expertise that can quickly develop faster, smaller, more power efficient modems and chips is a top priority. Their profits will benefit too.
I think the notion that ARM chips will ever power a Mac is delusional. It’s a fantasy that’s been around for years now. Meanwhile, if you look where Mac designs are going, they use ever more powerful Intel chips. The new Mac Pro uses up to 28 cores of Intel Xeon W chips. ARMs are good for low power applications like mobile phones and tablets, but they are a long way from being capable of powering real computers, even laptops. Laptops, too, are powering up. Even the MacBook Air uses an eighth generation Intel Core i5 chip. The 15" MacBook Pro uses up to an 8-core Intel Core i9 processor. Sadly, some Mac pundits are wedded to the idea that Apple will somehow bring their ARM chip business up to be competitive with Intel high power chips. You know, like Microsoft did. Not.
Apple will bring a technology in-house when their suppliers are unreliable. But Intel has continued to improve their high power chips. There has been no slack in these chip designs. Apple has skipped a few generations of Intel chips, but that was not Intel’s fault. Apple just got lazy. Now they’re back on track.
For Mac Pros…you’re probably right, and also for at least the lower end MacBook Pros. However…the A series chips continue to improve and do have battery life improvements over Intel’s chips, not to mention the advantages of being in house and hence Apple can customize the chips the way Apple needs them. For lower end laptops as well as lower end desktops the A series…particularly if they put in 2 cpus instead of 1.
Intel’s high end chips are pretty decent…but their smaller process, yield and output aren’t doing as well as they should be…at least from my cursory readings of the various articles. Mostly I’m not really interested in what kind of chip is inside the box…I’m interested in what it does for me.
I definitely think we’ll have two tiers of devices as Neil notes. The ARM chips are very much catching up or exceeding Intel performance, not including battery usage. You may be relying on outdated comparisons of current A-series products. ARM chips have been multi-core for a while and continuing upwards. See this comparison.
What I think we’ll see if a definite bifurcation. Mac Pro and probably iMac will remain on Intel, have a premium for performance, and have no battery usage issues to deal. MacBook Air and later MacBook Pro, and certainly Mac mini, will move to ARM. The battery live will really justify it alongside higher performance. And Apple wants to control its destiny. Having ARM chips for Macs is a natural extension. They control features, roadmaps, and integration.
Let me know when an ARM chip can run Windows or the macOS, plus mainline software, at competitive speeds. iOS is written for ARM. The macOS is not. Though Apple is trying to make it easier to port iOS apps to the Mac. And Adobe has a lite version of Lightroom for the iPhone. But that’s a compromise since so many people use their phones as cameras. Adobe wants to sell cloud space for people to store their images, just as Apple does.
ARM-driven Windows is rapidly approaching comparable speeds for comparable hardware. With native-compiled apps, some ARM laptops now compete with low-end laptops. That’s expected to change in the near future with a new generation of chips and designs.
Apple obviously has a working version of ARM-based macOS in the lab, and it’s efforts at partial code convergence and framework convergence in recent years, which will see a real start of change with Catalina, is an indication of that. I’m sure you realize that iOS and macOS share the same Unix origins, a lot of underlying software is the same on the Unix side (as you can tell when security updates come out that have to be installed across four platforms, soon to be five), etc.
I did post benchmarks above from chips Apple is already making. There’s little reason to expect Apple will happily continue to stick to Intel chips as they underperform A-series ARM chips that Apple designs.
Apple just effectively hired 2,200 new people by acquiring Intel’s poorly performing 4G/5G cellular modem chip business. I expect that was in part to tap more expertise as well as create their own modems in house to eventually compete.
ARM performance hasn’t stalled. It’s multi-core, and the benchmarks I linked show how rapidly its improved. Apple has carried out the kind of transition three times before (Motorola 68030 transition; Mac OS 9 to Mac OS X; PowerPC to Intel) that usually ruins other companies. You could even count porting iOS from its macOS origins to ARM as another one of those transitions…
Running in the lab already I wager…remember Intel was running for years(?) before the first Intel Mac. Look at thenA series benchmarks…plenty of power for non Pro machines. iOS came from macOS…don’t forget that.
Will we see Prommachines with A series innards? Mebbe, mebbe not. Consumer machines…one would be a fool to bet against that.
Apple has a proven record in doing things like chip swaps…give them a little credit. I would. It be surprised if the first A series Mac was running in the lab 5 years ago.
We just learned in the other thread that Windows on ARM in fact only runs a small subset of Win apps. Case in point, not even MS Office for ARM Windows is seeing updates. It doesn’t matter if the benchmarked performance is comparable if you can’t run any decent apps on it.
Underperforming in benchmarks is one thing, underperforming on actual hardware running real world apps another. Apple’s A series has been heavily optimized for fan-less, heatsink-less tablets/phones where battery life is paramount. Who’s to say those CPUs will just scale to dual/quad-core ~3 GHz designs and retain their low power consumption? That’s one heck of an extrapolation for an architecture that has never been optimized for the parameter space a 13" MBP lives in.
Don’t get me wrong, I’m sure ARMs can be tweaked for a different envelope. I’m also confident Apple has tons of great people working on this that are nowhere naive. I also perfectly agree with your argument that Apple has a lot of highly relevant experience in successfully moving (semi-)transparently to new ISAs. I do, however, caution against assuming that because Apple has a great A12X in an iPad they can just easily migrate that design to a 13" MBP and we’ll get the same great performance plus twice the battery life on macOS with existing apps, all just like that.
So far! And Apple would be allow native ARM compilation; that’s clearly part of the path they’re starting with Catalina.
Look at the latest iPad performance, though. Certainly, the A series chips could currently power long-battery life, good-performance Mac laptops.
There’s a transition ahead, and whenever it starts, I’m sure it’s going to take years, much like the PowerPC/Intel one. The idea Apple would announce ARM laptops and three months later they’d ship with all-native apps is of course a silly idea.
Remember how Apple seeded developers with Intel systems? I could see the same kind of arrangement happening with ARM-based Macs, too.