Apple's upcoming A16 chipset might stick to the 5nm TSMC process node

Tudor Cibean

Posts: 182   +11
Staff
TL;DR: The A16 SoC that will be used on the iPhone 14 will probably not bring massive performance and efficiency gains if it's going to use the same process as the A15 Bionic. Apple will reportedly rely on architecture improvements and a switch to LPDDR5 memory and use the new chip only in their iPhone Pro series.

In a series of tweets today, analyst Ming-Chi Kuo indicated that Apple's upcoming A16 chipset might continue to use TSMC's N5P process node, the same as its predecessor, the A15. The decision was reportedly made because the Taiwanese foundry won't be able to produce enough chips on the new N3 node this year to fulfill demand. Meanwhile, N4P, another improved variant of the N5 process, will only start mass production next year.

This leaves Apple with a choice between N4 and N5P, with the latter being the better-performing (and probably more expensive) node. It's unlikely that the company will switch over to Samsung's GAAFET 3 nm process, as the last A-series chip that wasn't manufactured exclusively by TSMC was the A9, which came out in 2015.

As a reminder, Apple is expected to only equip higher-end iPhone 14 Pro models with the new chip, with the vanilla phones shipping with the A15 instead.

Kuo also reiterated that the upcoming MacBook Air refresh might keep using the M1 chip instead of launching with the new M2. The laptop is rumored to come with an all-new design, something Apple seems to believe is a much better selling point than new internals (and we tend to agree, the M1 is pretty good in terms of efficiency).

The Cupertino company would opt to debut the M2 SoC next year on a MacBook Pro lineup update. This would also allow them to take advantage of the more advanced N4P or N3 node.

Permalink to story.

 
Might be an unintended consequence of how crowded TSMC is: Even if Apple has first serve they might not be able to reserve *all* of the 3nm node for just them.

 
Might be an unintended consequence of how crowded TSMC is: Even if Apple has first serve they might not be able to reserve *all* of the 3nm node for just them.

The 3nm has already been reserved for many others including Intel. Despite that, their 5nm M1 and an update has already a couple of years advantage, so no worry here.

Apple's issues are mainly on the software front, so they should focus there.
 
To be honest a MacBook Air with a M1 running LDDR5 and a 10% clock increase would probably give enough of a performance uplift to suit a generational refresh. M1 is only clocked at 3.2 GHz, a bump to 3.5 GHz would likely be easily achievable, and given the maturity of TSMC 5nm would likely not increase power consumption, or be offset by LDDR5 over LDDR4 on the current Air.

We know A15 didn't offer much of a performance uplift from A14 (which the M1 is based on), so it makes sense that M2 might be delayed until after A16 is released with the iPhone 14.
 
I might be in the minority here, but I'd actually be fine with Apple de-emphasizing CPU upgrades. I'm on the A9, and it's plenty fast enough for me; I don't know that I'd see any benefit from upgrading to the A15. So this move is more appealing to me than trying to up the CPU specs at a higher cost.

If feasible, what I'd like is a low-end that emphasizes the opposite of what the SE 3rd-gen does. Instead of the same CPU as the high-end and lower specs elsewhere, use an older, relatively cheap CPU, but keep some of the other features such as a 2X zoom camera, or bump the storage.

It might not be their best possible strategy since it might result in fewer high-end sales, but I probably would have bought an SE upgrade if that had that option.
 
Back