Does Apple really have some secret sauce, or are they just consistently first in line for TSMC's latest process? Not saying that's better or worse, but it's a very different problem.
For example, it was a big deal when they moved from IBM/Motorola to Intel, and Macs started getting much faster right away. A decade later and Intel is struggling to keep up at the high end, woefully behind in low-power contexts, and Apple is again hamstrung in their plans.
So at some point they decided to make their own chips. Well, to design them, and to use their leverage with TMSC to get them made exclusively. It’s true that they have fewer dependencies than a more general-purpose manufacturer like Intel, but don’t discount the many years they’ve spent eating shit while waiting for their “partners” to catch up. In the CPU space, they’ll never do that again.
This is the key point. There used to be a lot of chip makers and phone vendors in the Android space (hell, even Sony had their own processors for their early phones), both ARM and the odd non-ARM effort. The way that the Android market has collapsed down to essentially "Qualcomm and Samsung" is what's hurting Android, and Qualcomm, in particular, are the reason that things like support lifecycles are such garbage. Google have been completely disinterested in solving this problem in any way, while Apple have.
I'd love to see them try again. And I feel like that's an area where their particular culture (of looking at vendors that make them resentful and trying to do something about it) might even put them in a place where they're willing to sell those chips or designs to other carriers because fuck Qualcomm.
https://www.bloomberg.com/news/articles/2023-01-09/apple-pla...
I wonder if they'll roll it out on the iphone first or start with ipads and possibly the apple watch if they can get the power draw lower. That might be a good way to increase the battery life of those devices, lowering the gap between regular and low power mode.
If that turns out to be the case, the iPad has the largest battery and can absorb more of a regression than any of their other devices. But there's enough of those sold to get some progress toward a tighter solution
I can't remember whether it was sourcing Intel modems or rolling their own, but yeah, either way they fell foul of Qualcomm's nest of patents: Qualcomm's real competitive edge is that they are experts at lobbying to have their patented tech built into government and NGO standards.
You don't need to make decent chips or support them for more than a couple of years if it's illegal not to use them.
If someone won't give you favorable patent terms but you can buy a company that got better ones...
edit: yes, they did (modem division):
https://www.apple.com/newsroom/2019/07/apple-to-acquire-the-...
One potential avenue here is if Apple, with a cellular modem of their own, just includes the modem in all of their devices and cut out a bunch of SKUs.
It's the difference between tuning a game for consoles and tuning it for a PC.
Apple's is funding TSMC's latest process by guaranteeing future orders.
The other manufactures could do this but there is more profit in using older processes.
Only a fraction of Android devices are using a process that isn't 3+ years old.
Apple has end dates for their process. They are aim for this and the other companies are not.
This isn't Apple Bad(tm)
This is Android manufactures have a different focus. Android devices are priced to be disposable.
What money? They already paid them. Building fabs has a massive upfront cost that you depreciate for decades. The economics have to make sense for them to move to TSMC
Apple has good designs, consistent long-term plans, and doesn't have to directly compete with Intel over domination in the x86 market. They offer a complete package outside the x86 market. This fits well with who a market leader like TSMC would want as a partner.
For TSMC, it would be an error to start large-scale production of chips for someone that still has their own manufacturing capability. TSMC would have to allocate less fab capacity to their existing customers, and those won't like that.
Also, in the old days when they weren’t the juggernaut they are today, they always were good and minimizing SKUs and economy of scale. In 2005, the #1 part on state contracts were iMacs, even though Apple market share was like 3%.
That level of control and planning means they know what they need and can adjust. Overages in Apple Watch CPUs are subsidizing the HomePod, etc. In the case of the iPhone, they put an overpowered chip in there because why not.
Also their silicon itself is just more advanced. I don't think anyone else is doing UMA the way they are, though I'm not 100% sure and would welcome correction.
My point is that Samsung is not there yet.