Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MacRumors

macrumors bot
Original poster
Apr 12, 2001
67,965
38,678


As we wrote about earlier today, Bloomberg's Mark Gurman expects Apple's first 5G modem to debut in a new iPhone SE, an ultra-thin iPhone 17, and low-end iPads in 2025. Towards the end of the report, he also outlined Apple's modem plans for devices launching in 2026 and 2027, and we have summarized that information below.

5G-Modem-Feature-Blue.jpg

According to Gurman, Apple's second-generation 5G modem will debut in the iPhone 18 lineup in 2026 and in iPad Pro models by 2027. He said this modem will better compete with Qualcomm's modems in current iPhones by adding support for the ultra-fast 5G standard known as mmWave. Overall, he said the second modem will achieve theoretical download speeds of up to 6 Gbps, compared to up to 4 Gbps for Apple's first modem.

In 2027, he expects Apple to release its third-generation modem. The company is apparently aiming for this modem to top Qualcomm's modems in terms of performance and AI features, although it remains to be seen if it achieves this feat.

Further out, Apple is said to be discussing merging the modem into the iPhone's A-series chip.

All in all, this sounds like a natural progression for Apple following its decision to transition away from Qualcomm modems in iPhones. Apple and Qualcomm have had a rocky relationship over the years, but the two companies extended their modem supply agreement for iPhones through March 2027, so Apple still has plenty of time on its side.

Apple has been rumored to be working on its own 5G modem since 2018, and it acquired Intel's smartphone modem patents in 2019 to bolster this initiative. Starting with the iPhone SE 4 in March, we should finally see the modem in action.

Article Link: Apple's Second 5G Modem for iPhone 18 Pro and iPad Pro Expected to Support mmWave
 
  • Like
Reactions: SFjohn
Exciting stuff to come.
5G download speed even today is good enough for most people, it’s the infrastructure that’s the bottleneck
 
  • Like
Reactions: mganu and DaveN
Imagine buying a flagship phone in 2026 with underwhelming cellular performance, as is implied for the second generation of Apple’s modem. Yikes.

Keep testing your profit-maximizing modems on the low-end stuff, Apple, until you’ve truly topped Qualcomm’s modems.
What makes you think performance will be underwhelming? I know the Intel modems Apple previously used weren't as good as the Qualcomm modems available at the time, but now that Apple is developing the modems in house there's no reason they can't match or exceed Qualcomm's performance given the investment Apple has made and access to cutting edge TSMC manufacturing processes.
 
This is starting to remind me of the initial Apple silicon rollout. They started with basic M1 series chips on their cheapest consumer Macs, then used more powerful chips for the pro Macs. And basic M1 still blew Intel out of the water.

I know there have been struggles with Intel’s modem division. But usually when Apple makes bets like this, they don’t miss. Switching from Samsung to Apple-designed processors for iPhones, then switching from Intel to Apple-designed chips for Mac. And I don’t think they would make a decision like this haphazardly with such an important part of their business.
 
why cant they put 5g modems in Macs ?
My limited understanding is that Qualcomm licensing is pretty expensive for computers so Apple skips it. If Apple can successfully shift to their own modems that licensing issue would likely go away so they'd be able to add them at that point.

Would they? It is too early to say. Apple might find the Personal Hotspot support to be good enough, but they also may add it. The biggest challenge might be how complex do they want their SKUs to be come (as they are already complex). Cellular isn't a popular feature on computers so they might not like the idea of having another SKU just for it. Time will tell.
 
  • Like
Reactions: _Mitchan1999
"Further out, Apple is said to be discussing merging the modem into the iPhone's A-series chip."

People without knowledge on physics and digital vs analog electronic design.
Look at the red chips, A18 Pro vs Qualcomm modem size:
View attachment 2459514
"Merging" does not mean everything on one chip!

The most likely version is analog/RF electronics on one chip tightly coupled to the SoC (eg single package, most likely vertically stacked).
Whether to place the modem electronics on the analog chip or the A chip is less important - A chip gives you higher GHz and lower power, analog chip is cheaper area. May even change from one year to the next going forward.
 
  • Like
Reactions: rp2011
Imagine buying a flagship phone in 2026 with underwhelming cellular performance, as is implied for the second generation of Apple’s modem. Yikes.

Keep testing your profit-maximizing modems on the low-end stuff, Apple, until you’ve truly topped Qualcomm’s modems.
Normally, I'd agree with you. I also do think we have to have some skepticism with the new stuff as it is new...

That said, I think Apple proved themselves with the M-Series processors. I was fully expecting the M1 to be a dud or painful in some way longer term since it was the first chip Apple was using to swap off Intel. It came in roaring by beating Intel chips at the time at every metric while also sending battery life through the roof. It was so fast that it ran x86 code in an emulation layer faster than Intel chips ran x86 natively. Bonkers. Here we are 4 years later and those chips are still performing extremely well and their replacements have been evolutionary and not "this one finally fixes all the bugs of the M1".

So I'm inclined to see them as having something VERY special to deliver on this front.

I don't see this being a repeat of the Intel modem fiasco.
 
What AI features does a modem have?
"AI" is used whenever you need to choose from a large range of possibilities, with no practical algorithm to choose the best. This is already being done for video compression; and radio communication is similar - what's the optimal combination of FEC, modulation, diversity, etc right now for current radio conditions.

It's become cool (amongst the stupid) to say that this is just "heuristics". In a sense this is true, in that it's an empirical way to make a choice rather than a formally optimal way. But "heuristics" ignores the point that, like many other AI schemes, the quality of your result improves as you provide better training.

Of course this is not an LLM or even genAI - which is why it's not called such.


BTW this is a teachable moment. As I've said repeatedly, about 70% of the patents Apple has filed over the past two years have been modem related. Just SO DAMN MANY...
They are all over the place (and I only skim them, I don't read them closely the way I read CPU, GPU or ANE patent), but it's obvious that one of the themes in the patents is this use of "AI" to figure out optimal transmission.
Another big theme is minor tweaks here and there to save power. Each tweak may save .1% of the joules - but throw enough of them at the problem and eventually you're running at 30% longer battery life.

Point is, there are a lot of people here with a lot of opinions. And basically zero of them have ever even looked at a single Apple modem patent. There opinions have ZERO basis beyond "I like Apple" vs "I don't like Apple".
Be careful to whom you listen...

My guess is that the first two or so iterations of the modem will probably be behind QC performance in various edge cases - ways that will not matter to the average user, but will be massively hyped by the press and the idiot internet. BUT they will be 30% or whatever more energy efficient -- right out the gate they will be noticeably superior along the most important dimension to most users. Just like, eg, an M1 relative to the absolute highest end of the Intel range at the time...
And (again like M series) with that firm energy-siping foundation in place, Apple will then relentlessly add performance boost after performance boost each subsequent year.
 
It remains to be seen if the theoretical 2027 Apple modem will outperforms Qualcomms? Heck, it remains to seen if the underperforming 2025 one exists.
 
Bottom line is that if the first two gens of Apple modems are slower than Qualcomm’s, it would be a very hard marketing exercise for whatever device it is in.
Imagine Qualcomm running an ad with an 1st gen Apple modemed iPhone and a Qualcomm modemed Samsung side by side with Speedtest on both.

I would be very surprised if Apple released a product that could be so easily ridiculed, and it would wait until the superior third gen before releasing a product with it.

Or, this article is full of it. Intel modems back in the day, after all weren’t that inferior to Qualcomm’s. I can’t imagine after spending all that loot buying the intel modem business, Apple has been sitting on its hands.
 
  • Like
Reactions: SFjohn and Chuckeee
What makes you think performance will be underwhelming? I know the Intel modems Apple previously used weren't as good as the Qualcomm modems available at the time, but now that Apple is developing the modems in house there's no reason they can't match or exceed Qualcomm's performance given the investment Apple has made and access to cutting edge TSMC manufacturing processes.
Qualcomm’s Snapdragon X75 modem (found in the iPhone 16 Pro) is rated for a max speed of 10 Gbps, whereas the article has Apple’s 1st generation modem at 4 Gbps max speed and the second generation at 6 Gbps.

Now keep in mind those are maximum speeds under perfect conditions, it’s an open question if, under the same less-than-perfect signal strength conditions, a 2nd generation Apple modem would still be 40% slower than the X75.
 
  • Like
Reactions: Chuckeee
Maybe by then I'll actually find mmWave in areas I'm at. Though I suspect it's more for big cities and places like stadiums that have many people in small areas.
 
  • Like
Reactions: LlamaLarry
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.