From Sam Byford's "Apple reportedly switching to its own iPhone modem design in 2023" posted Wednesday on The Verge:
Apple is planning to partner with Taiwan Semiconductor Manufacturing Co. for the production of its own 5G modems for future iPhones, according to a new report in Nikkei. Apple is said to be planning to use TSMC’s 4nm process node, which hasn’t yet been deployed for any commercial product; the modem is apparently being designed and tested at 5nm before moving to mass production in 2023 at 4nm.
Apple’s switch to modems of its own design is widely expected to happen in 2023, and TSMC is the natural manufacturing partner. Qualcomm, which is the dominant player in the industry and produces modem components for the entire iPhone 13 lineup, recently said that it expects to account for just 20 percent of iPhone modem orders in two years’ time.
Apple bought Intel’s 5G modem division in 2019, foreshadowing the eventual switch. Earlier that year Qualcomm and Apple agreed to end a costly modem technology patent dispute, with Qualcomm receiving more than $4 billion as part of the settlement.
My take: Old news. See, for example, MacRumors' Apple-Designed 5G Modem Said to Debut in All 2023 iPhone Models dated 3/11/21.
Friend-of-the-blog Bartley Yee, fearing that the first batch of Apple-designed modems might be slower than Qualcomm's, writes:
Could it happen that fast? Might they try the new chip in a lower level design like the iPhone SE 2022 where possibly lesser performance would not be a huge problem?
Apple has the luxury to wait till they can do better. That’s the Apple way. Plus they’re TSMC’s best customer. Plus Apple can integrate with the rest of the chip / sw stack to optimize performance further.
2023 starts a 2-year plus migration all all mobile devices to Apple 5G. Might we see 5G Macs if and when 5G allows cord cutting. And then there are rumored new products.
I wonder if this modem, if it’s not bespoke, could eventually be sold to 3rd-parties, and take some of the gloss off Qualcomm’s sales? If it’s not bespoke, and on such a small process, it could be lower power than Qualcomm’s modem, and the modem has to be one of the three largest sources of power drain in a phone, besides the chip and the display. Then again, 3rd-party phone mfrs might not have IP leverage with Qualcomm, so they’d be buying Apple’s modems and still paying licensing fees to Qualcomm. Of course, their Qualcomm IP license could be from Apple, but we know Qualcomm will still sue and try to double-dip.
I agree Apple could and would intro Apple 5G parts in normal high end iPhone IF performance (speed, power consumption, OEM costs) is as good or better than Qualcomm at the time. Where I suspect Apple is going with sub-5nm modems is better performance with less power used than current Qualcomm X60 parts. Where Apple will excel is not needing an actual separate modem chip but an integrated module within the CPU with dedicated memory, high bandwidth bus and processing, perhaps physically buffered a bit for heat management. Higher value iPhones would have all bands enabled, less expensive starter iPhone SE with mmWave excluded or reserved for a more expensive variant (SE Plus) with bigger battery.
Primarily, an Apple 5G modem gives Apple more options, more performance, more battery life, and less cost, assuming they actually can build a solid competitive chip design.