Apple Aims to Eventually Integrate Future Modems into Main Processor

[MacRumors]

Apple plans to eventually integrate the modem component in its devices into the main processor, making the electronics more power-efficient and less costly to produce, according to Bloomberg's Mark Gurman.

"I'd expect this in 2028 at the earliest," said the reporter in his Power On newsletter. Funnily enough, that's the same year that Apple could begin building its own custom modem into cellular MacBook models, as previously reported by Gurman. However, in his latest report, Gurman did not mention Macs in the same sentence as modems.

Apple has explored the possibility of developing MacBooks with cellular connectivity in the past. Indeed, the company reportedly considered launching a MacBook Air with 3G connectivity, but former CEO Steve Jobs said in 2008 that Apple decided against it, since it would take up too much room in the case. An integrated SoC would solve that problem.

Apple debuted its first custom modem, the C1, in the new iPhone 16e, which the company unveiled last week. According to Apple, the C1 is the most power efficient modem that's been included in an iPhone to date. The ‌iPhone 16e‌ has the longest battery life of an ‌iPhone‌ with a 6.1-inch form factor, and it lasts for up to 26 hours when watching video. Apple says the C1 is "just the start."

Gurman said that Apple is already testing the so-called C2 modem as well as its successor, the C3, which is expected to arrive in 2027, by which time Apple hopes it will "outdo Qualcomm's modem capabilities." MacRumors has previously reported that Apple is testing a C2 modem, which is said to be debuting in Apple's iPhone 18 Pro models next year, according to Gurman.

Apple has reportedly been working on its own modem since 2018, as it seeks to move away from Qualcomm's component currently used in iPhones. Apple's transition away from Qualcomm modems is a logical step, given their turbulent history. Despite recent tensions, the two companies recently extended their modem patent licensing agreement through March 2027, allowing Apple ample time to perfect its own solution.

Article Link: Apple Aims to Eventually Integrate Future Modems into Main Processor

 

[calzon65]

MM wave compatibility for the C series modems is certainly important, but I am more about power savings ... like Apple did with its M processors for laptops (which got a performance boost and power reduction).

So, I am curious to see power benchmarks on the iPhone 16e. I'm not just going to trust Apple's marketing department.

 

[Mac Fly (film)]

one ring-ring to rule them all ☎️

 

[zorinlynx]

Does it make sense to waste valuable SoC die space on a modem, when it's not something that really benefits performance-wise from being on the same die as the CPU?

 

[thefrost]

We are already thinking about what's ahead when we haven't even seen how well the C1 does. 🤣

 

[Frantisekj]

It just makes sense and could be expected. They have to fine tune designs not tu ruin whole chip.
Could be very interesting for Watches

 

[jayducharme]

If a board with a C1 chip currently fits in an iPhone, I don't see the problem with putting it in a MacBook right now.

 

[Hopefully Smarter]

Bring cellular connectivity to Apple Socks

 

[cjsuk]

Ah yes put the risky component from a yield perspective in the rest of the already binned eggs.

Always works that one.

 

[HaHaRich!]

Does it make sense to waste valuable SoC die space on a modem, when it's not something that really benefits performance-wise from being on the same die as the CPU?

Vertical integration. Staring down at 2nm dies and smaller over the next 5 years, if you can flash one SoC a billion times, and install one chip on a board, it means you don’t have to flash 2 billion chips, and don’t have to spend time assembling a second chip. Also, if you are adding a custom Bluetooth & WiFi chipset to the SoC as well, that’s saving even less interconnection and assembly you have to do (or buy from another company like Broadcom). And since Apple is paying for a big chunk of the new TSCM plant expansion in the US, those savings directly line up for ROI on the investment. They don’t have to do it forever if they can get a better deal down the road, but it makes sense as a road map for the investments they’ve made.

 

[timber]

Making everything connected beyond wi-fi in a very efficient way would certainly be a cool feature particularly nowadays when you can dispose of physical SIMs.

 

[cjsuk]

Vertical integration. Staring down at 2nm dies and smaller over the next 5 years, if you can flash one SoC a billion times, and install one chip on a board, it means you don’t have to flash 2 billion chips, and don’t have to spend time assembling a second chip. Also, if you are adding a custom Bluetooth & WiFi chipset to the SoC as well, that’s saving even less interconnection and assembly you have to do (or buy from another company like Broadcom). And since Apple is paying for a big chunk of the new TSCM plant expansion in the US, those savings directly line up for ROI on the investment. They don’t have to do it forever if they can get a better deal down the road, but it makes sense as a road map for the investments they’ve made.

It depends on more than that.

There's yield which is much much lower on complex RF parts. Do you really want to shovel all that aggregated risk into one package? Well probably not. The tile based CPU substrates that are starting to appear mostly exist to mitigate that risk and produce more generic subunits which can be assembled into a multi-chip module with fast interconnects in the substrate.

Not that anyone will notice as Apple will bundle all aggregated costs into your device. Perhaps that's why the 16E is so expensive? Crap yield? Might as well bang the prices sky high because they'll still buy it! (not saying that is the issue but some vendors have done that in the past)

 

[DEMinSoCAL]

MM wave compatibility for the C series modems is certainly important, but I am more about power savings ... like Apple did with its M processors for laptops (which got a performance boost and power reduction).

So, I am curious to see power benchmarks on the iPhone 16e. I'm not just going to trust Apple's marketing department.

Would seem to be at some point there is a certain amount of power needed to transmit a signal to a tower and if you're too frugal on power, call quality may start to suffer. In other words, unlike CPU's that can still operate at very low power levels, transmitters require a certain amount of power.

 

[DEMinSoCAL]

If a board with a C1 chip currently fits in an iPhone, I don't see the problem with putting it in a MacBook right now.

WWAN in a laptop is mostly a "business" feature, which MacBooks are not particularly considered. And even at that, very few Windows laptops have that feature. Surface Pro 10th had it, but Surface Pro 11th gen doesn't.

The non-business crowd is happy using the hot-spot plan they already pay for on their cell phone to get their laptop on the internet when wifi is absent.

 

[HaHaRich!]

It depends on more than that.

There's yield which is much much lower on complex RF parts. Do you really want to shovel all that aggregated risk into one package? Well probably not. The tile based CPU substrates that are starting to appear mostly exist to mitigate that risk and produce more generic subunits which can be assembled into a multi-chip module with fast interconnects in the substrate.

Not that anyone will notice as Apple will bundle all aggregated costs into your device. Perhaps that's why the 16E is so expensive? Crap yield? Might as well bang the prices sky high because they'll still buy it! (not saying that is the issue but some vendors have done that in the past)

Depending on the cycle, you can leverage the yield decrease with the die shrinks. Going from 3nm to 2nm leaves you with room to iron out production decreases, and still improve margins per chip if the yield decreases by 10-15%, plus those yields historically improve iteratively over the life of the die, and lower costs on the following chip family. Also, Apple isn’t paying per wafer, TSCM charges by the chip. It’s their incentive to improve yield (or raise prices and risk losing future orders to competitors that have started to appear in the market segment). Not saying there’s no down side, but it’s risk you can mitigate.

 

[asdfjkl;]

Integrating modem on SoC makes sense from a power savings perspective. For a laptop, however, the cost of a separate data plan for your laptop has never made sense. Waaaaaaaay too expensive.

 

[turbineseaplane]

Sounds like a future of "if the device is one, it's always connected" whether you like it or not
😕

 

[HaHaRich!]

Would seem to be at some point there is a certain amount of power needed to transmit a signal to a tower and if you're too frugal on power, call quality may start to suffer. In other words, unlike CPU's that can still operate at very low power levels, transmitters require a certain amount of power.

An advantage for anyone customizing their modem is they can split the signal processing and the actual transmission anyway they like, not have to live in the thermal and power envelope of an OEM. The thing I think MacRumors isn’t stringing together well is the talk of Apple making custom WiFi and Bluetooth chipsets too. If you integrate a custom signal processor into your cpu that can handle both cell and WiFi, that’s two chips you don’t need to place and power separately on the board, even if the transmission hardware doesn’t change much.

 

[DanR16]

Imagine the number of SKUs for Macs if this is integrated

 

[HaHaRich!]

Integrating modem on SoC makes sense from a power savings perspective. For a laptop, however, the cost of a separate data plan for your laptop has never made sense. Waaaaaaaay too expensive.

On a per-device basis, absolutely. But with the market moving heavily on 5G Home Internet, having a plan data pool for all the devices using the hotspot at home or cell antenna on the road is a market shift that’s just one domino away. As soon as one provider offers it, they all will. If one is hinting to Apple it’s on the roadmap, that’s worth looking at if the cost to add the hardware is a few dollars a machine.

 

[timber]

WWAN in a laptop is mostly a "business" feature, which MacBooks are not particularly considered. And even at that, very few Windows laptops have that feature. Surface Pro 10th had it, but Surface Pro 11th gen doesn't.

The non-business crowd is happy using the hot-spot plan they already pay for on their cell phone to get their laptop on the internet when wifi is absent.

It will be like "retina display" or the aluminium enclosure, just features, not earth shattering but after MB and iPads become completely always on devices nobody will want any other way.

 

[switz]

A hackers dream. All those iPads and Macs powered on all the time. Every home will need a Faraday cage for their electronics for security. Apple will sell you one for how many thousands of dollars?

 

[obviouslogic]

Does it make sense to waste valuable SoC die space on a modem, when it's not something that really benefits performance-wise from being on the same die as the CPU?

It takes much less space to include it on the SoC than it does to create a separate chip and include it on the board somewhere else and it does actually reduce power consumption - both are the main reasons SoCs appeared in mobile devices.

 

[1037]

Like the secret ones Intel embedded in the Pentium around 2008? That could air gap jump to other PC's.

 

[jz0309]

Questionable. What does “integrated” in this context mean?
The current A and M chips/die are already quite large, modem, we’ll have to see.
typically modems are a couple nodes behind processors.
Integrating at the die level would result in a huge die that brings other problems.