So notebook SoCs isn't really making a difference on the overall need to move forward. Decoupling the modem from the die all the more so.
Not understanding your point here. A big selling point to a Qualcomm SoC would be having a cellular modem built into the notebook, something Apple does not offer, so far
That selling point worked so well that Qualcomm had to go out and spend $1B on a better non-modem core to be competitive. When getting your butt kicked on CPU/GPU/etc performance then may not want to allocate die space to a cellular modem that the vast majority of PC laptop users are not going to use.
That selling point really didn't work that well. There wasn't a huge , additive upswell in Windows laptop expansion for a modem and battery first and performance "just OK" laptop at the price points that Qualcomm systems were selling at.
To take something like 7-10% share away from AMD/Intel Qualcomm will need a SoC with no modem overhead in it. Could they have a secondary SoC that does have a modem in a bigger , more expensive multiple die package. Sure. But that can't be the main compete. It won't work. Qualcomm spun that 'we have a cheaper BOM' story before , but it was only cheaper if compared to AMD/Intel+ modem; not AMD/Intel as they are typically sold. That is a substantive contributing factor to why they got limited traction.
That cost gap isn't just on BOM. cellular modems are more expensive for end users to operate also. A modem that is actively hooked to a cellular service is an additional $xy.zz per month. Lifecycle costs of 3-5 years is 30x or 60x $xy.zz for total costs. ( 60x $10 = $600 which would very useful to put toward buying the next laptop at the end of the cycle.) Are the vast majority of end users looking for that additional cost? Nope. As a "we sell radios" company, Qualcomm would love it of most people bought two cellular modems from them at a time.
As long as Qualcomm has the viewpoint of primarily just trying to sell incrementally more radios, they are not going to get deep traction against AMD/Intel in the Windows space.
Qualcomm hasn't bee shy to talk up Oryon as "Apple M-series" like solution. Similarly, if Apple is throwing gobs of die space at more CPU/GPU/NPU cores is Qualcomm going to be able to allocate die space to cellular modem that cuts into the allocation for those other cores and still compete? Probably not. Since they explicitly 'called out' Apple , they are probably going to get measured against them. There will be lots of reviews stating they that 'failed' if the GPU or CPU comes of dramatically short of the bar that Apple set.
. Decoupling the modem would not be a net positive, except possibly in terms of improving yields, but on that front, Qualcomm could just go with interconnects and vend multi-die SoC packages.
If the majority of the laptop user base isn't going to turn on the cellular modems it isn't a 'yield' thing, it is more so as a wasting wafers thing. Once the working die run rate gets up in the 10M range , they would be churning through lots of wafers. If the users are getting zero utility out of even small portion of that
The yields aren't going down. To compete with Apple they CPU/GPU/other 'uncore' die space usage would likely have to go up. Not really going to get a net decrease in die size here. Net decrease in package size but that isn't a die yield issue.
As I mentioned at the end. Decoupling the modem for the primary die is the issue. So yeah Qualcomm can do a multiple die package ( to improve Pref/Watt and ease of manufacturing an optional modem capable version). But they also need to able to dump the modem with the end user doesn't need it.
www.tomshardware.com
www.theverge.com
www.trustedreviews.com
techcrunch.com
