Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I agree. Either this "rumor" is garbage or this is supply chain induced craziness.
I don't give this particular rumour much credence, but there's nothing inherently unfeasible about Apple skipping straight to 3nm for their next Pro/Max/Ultra products, considering that the 14/16" MacBook Pros that would likely be the first to get them (a) aren't even 12 months old yet and (b) have already had a post-Apple-Silicon re-design, so there's no desperate need for Apple to update them before well into 2023.

Arguing about whether they're going to be labelled M2 or M3 is just irrelevant.
 
I'm not sure I believe this rumor. Apple has been quite consistent with their release cadence baring any delays from TSMC. "A" processors on 12 month cycles and "M" processors on 18 month cycles. I don't see them speed running the process nodes because they will hit the wall more quickly in terms of TSMC's capabilities. 2nm is already having delays with mass production late 2025 / 2026. And who knows what issues will crop up with even smaller nodes. It's better to sell a consistent 20% per year performance improvement rather than 40% one year and then 0% the next.

Here are my predictions for Apple Silicon:

A14 - Late 2020 - TSMC N5 (Process)
A15 - Late 2021 - TSMC N5P (Optimize)
A16 - Late 2022 - TSMC N4 (Optimize+) (Rumored intermediate node because of delays with N3)
A17 - Late 2023 - TSMC N3 (Process)
A18 - Late 2024 - TSMC N3P (Optimize)
A19 - Late 2025 - TSMC N3? (Optimize+) (There will probably be an intermediate node here from delays with N2)
A20 - Late 2026 - TSMC N2 (Process)

M1 - Late 2020 - TSMC N5 (Process)
M2 - Mid 2022 - TSMC N5P (Optimize)
M3 - Late 2023 - TSMC N3 (Process) (Rumored 15" Macbook Air)
M4 - Mid 2025 - TSMC N3P (Optimize)
M5 - Late 2026 - TSMC N2 (Process)

M1 Pro/Max - Late 2021 - TSMC N5 (Process)
M2 Pro/Max - Mid 2023 - TSMC N5P (Optimize)
M3 Pro/Max - Late 2024 - TSMC N3 (Process) (I expect some form of design refresh here since the MacBook Pro Retina and Touch Bar were around 3 years before their refresh. The rumored OLED screen is possible.)
M4 Pro/Max - Mid 2026 - TSMC N3P (Optimize)
M5 Pro/Max - Late 2027 - TSMC N2 (Process) (Again, probably some form of design refresh)
Actually you make a very good point which points to another possibility.

Maybe the base M2 is based off of the 5np A15 while the M2 Pro/Max is based on the A16 and therefore uses 4nm. So not quite the full jump to 3nm but still an improvement over the 5np silicon that can push the higher end chips a little further.

Then we go full on 3nm for the entire M3 family.
 
I'm not sure I believe this rumor. Apple has been quite consistent with their release cadence baring any delays from TSMC. "A" processors on 12 month cycles and "M" processors on 18 month cycles. I don't see them speed running the process nodes because they will hit the wall more quickly in terms of TSMC's capabilities. 2nm is already having delays with mass production late 2025 / 2026. And who knows what issues will crop up with even smaller nodes. It's better to sell a consistent 20% per year performance improvement rather than 40% one year and then 0% the next.

Here are my predictions for Apple Silicon:

A14 - Late 2020 - TSMC N5 (Process)
A15 - Late 2021 - TSMC N5P (Optimize)
A16 - Late 2022 - TSMC N4 (Optimize+) (Rumored intermediate node because of delays with N3)
A17 - Late 2023 - TSMC N3 (Process)
A18 - Late 2024 - TSMC N3P (Optimize)
A19 - Late 2025 - TSMC N3? (Optimize+) (There will probably be an intermediate node here from delays with N2)
A20 - Late 2026 - TSMC N2 (Process)

M1 - Late 2020 - TSMC N5 (Process)
M2 - Mid 2022 - TSMC N5P (Optimize)
M3 - Late 2023 - TSMC N3 (Process) (Rumored 15" Macbook Air)
M4 - Mid 2025 - TSMC N3P (Optimize)
M5 - Late 2026 - TSMC N2 (Process)

M1 Pro/Max - Late 2021 - TSMC N5 (Process)
M2 Pro/Max - Mid 2023 - TSMC N5P (Optimize)
M3 Pro/Max - Late 2024 - TSMC N3 (Process) (I expect some form of design refresh here since the MacBook Pro Retina and Touch Bar were around 3 years before their refresh. The rumored OLED screen is possible.)
M4 Pro/Max - Mid 2026 - TSMC N3P (Optimize)
M5 Pro/Max - Late 2027 - TSMC N2 (Process) (Again, probably some form of design refresh)
I'm curious. Why are you specifically predicting N4 vs N5P for A16?
 
I don't think we'll see a Pro Mac mini personally because it brings it too close to the Studio. I think they'll keep a regular M2 in there or M3 if it waits that long for another refresh.
 
I am totally in. I'm often bringing down my M1 to its knees.
But I'm not like the majority, I totally know I need a powerhouse of a computer.
 
That's great but where is my iMac Pro??? 🧐
10 core Intel iMac Pro, 32GB RAM, 1TB SSD: $5000
vs.
10 core Mac Studio Max, 32GB RAM, 1TB SSD: $2200 + Studio Display = $3800

18 core Intel iMac Pro, 64GB RAM, 1TB SSD: $7800
vs
20 core Mac Studio Ultra, 64GB RAM, 1TB SSD: $4000 + Studio Display = $5600

The only fly in the ointment is if you need more than 128GB RAM or if there's a debate about what constitutes an equivalent GPU (which won't change just because you build the computer into the display).
 
I'm curious. Why are you specifically predicting N4 vs N5P for A16?

Based on my recollection of this MacRumors Article

The article has since been updated and now say A16 could be N4P.

I also don't think Apple would stay on the same node in order to have some marginal improvement to market.
 
Even more reason why the recent M2 should've really just been called M1S. If theres an M2 Pro that uses 3nm thats a totally different chip. Trying to mimic the iPhone annual chip update really wont work for desktop when the performance gains over a 1 year period are going to be tiny, and in some cases a backwards step (like the recent revelation that the I/O is slower on M2).
 
Based on my recollection of this MacRumors Article

The article has since been updated and now say A16 could be N4P.

I also don't think Apple would stay on the same node in order to have some marginal improvement to market.
N4P seems way too optimistic for A16. N4P is realistically a 2023 product.

N5P or N4 would make sense for A16 but not N4P. However, some of suggested that N4 doesn't really add much over N5P, whereas N5P is tried and true (and perhaps could be cheaper).


It's no news that Apple's naming is an utter mess. "So is this M2 chip 3nm or 5nm?" "Well, it's complicated..."
? Those chip names have never referred to processes. Hell, in the past even the same chip has been made on different processes.

Plus those process names are marketing names anyway.
 
All speculation, but I don't think they would jump from 3nm to 1nm. I assume the difficulty of decreasing sizes is all proportional. If you look at past transitions, I think we've gone:

20nm -> 16nm -> 10nm -> 7nm -> 5nm -> 3nm.

Moral of the story, 5nm -> 3nm is a big jump (40% reduction in size), going from 3nm to 1nm would be a giant reduction (67%). I think it'll be 3 -> 2 -> 1, and the jump from 2 to 1 might take a while!
These days the nm figures are just marketing terms as they stopped being relevant around 20nm mark with the move to FinFET transistors. The manufacturers, such as TSMC, use terms like N5 to refer to performance if the chip was 5nm even though it isn’t. This is why the M2 isn’t that much faster than M1 as both are built on TSMCs N5 technology when allowing for extra cores. There’s been differing reports on whether N3 will be available in volume this year, so wouldn’t be surprised if we only see M2 pro/ultra/max being N5 with N3 used for something lower sales volume, such as Mac Pro first.
 
That's great but where is my iMac Pro??? 🧐
I'd imagine the holdup on that is a decent monitor panel. They cant realistically keep pushing the same LCD panel that they've been using since 2012 (yep, its that old) as a new amazing screen when all they've done since is tweaked the backlight and splashed a new anti-glare coating on it.
 
Actually you make a very good point which points to another possibility.

Maybe the base M2 is based off of the 5np A15 while the M2 Pro/Max is based on the A16 and therefore uses 4nm. So not quite the full jump to 3nm but still an improvement over the 5np silicon that can push the higher end chips a little further.

Then we go full on 3nm for the entire M3 family.
That's definitely a good point. It would also make for great marketing to go to 4nm. I can already hear Johny talking about how many more transistors M2 Pro/Max at 4nm compared to M1 Pro/Max at 5nm.

[edit]
Actually, the more I think about it, the more I think you may have the correct prediction. It would definitely help control the maximum die size. With the M1 to M2, the die size increase didn't matter much, but if you starting from the M1 Pro/Max and increased the die size further, there may be some issues with economics/production. So going down to 4nm would help offset the die size so the M2 Pro/Max can stays relatively the same size as the M1 Pro/Max.
 
Last edited:
It doesn't have any sense to have different production processes within the M2 silicon family.

Apple's original plan may very well have had M2 on the 3nm process, but TSMC is not ready yet and Apple doesn't want to keep waiting. Same reason A16 is not going to 3nm this summer.

Since the higher-end M2 SoCs are shipping later, TSMC likely has told Apple that they will be ready when Apple is.



I'm not sure I believe this rumor. Apple has been quite consistent with their release cadence baring any delays from TSMC.

You answered your own question. :)

TSMC is not ready to produce A16 and the base M2s on 3nm so Apple is "stuck" keeping them on 5nm.
 
Last edited:
Makes perfect sense. Apple had delays in shipping Avalanche and Blizzard (A15ish) based M2 chips. TSMC will start cranking out 3nm/N3 wafers this year to allow vendors to sell them next early next year and Apple will want to increase their cadence.

In this increasingly competitive market, Apple doesn't want to sit on those perf-throne-capturing (A17ish?) chips, especially when they can more easily drop them into already redesigned Airs/Pros/iMacs/Studios.

My question - why not just call those chips 'M3' (regular/pro/max) and have the M2 as only a 'regular' for this year. There's no reason Apple has to ship new Ms every year, and there's no reason they need to ship the full product spectrum (pro/max/ultra/???) every year.

If they do ship an "M2 Pro" based on TSMC N3, it will probably cause at large OCD.
 
Last edited:
  • Like
Reactions: Tagbert
I think the mac pro threw a spanner in the gears...
The M2 lineup is shaping up to be a little confusing; The only reason I'd see a 3nm M2 lineup is to provide the mac pro with an M2 Ultra before the end of the year.
To be fair it would be very depressing after having spent thousands on an M1 Ultra to have its successor ready only 7 months later...
 
It's no news that Apple's naming is an utter mess. "So is this M2 chip 3nm or 5nm?" "Well, it's complicated..."

thats more TMSC than apple. why should Apple's name expose the manufacturing process used? Its fine. M2 > M1. M2 pro > M2, etc. one can argue that Pro, Max, Ultra are confusing (is Max > Ultra? Is Ultra > Pro?) but its not that hard to grasp, there is the base chip and three upgrades.
 
  • Like
Reactions: Tagbert
Makes perfect sense. Apple had delays in shipping Avalanche and Blizzard (A15ish) based M2 chips. TSMC will start cranking out 3nm/N3 wafers this year to allow vendors to sell them next early next year and Apple will want to increase their cadence.

In this increasingly competitive market, Apple doesn't want to sit on those perf-throne-capturing chips, especially when they can more easily drop them into already redesigned Airs/Pros/iMacs/Studios.

My question - why not just call those chips 'M3' (regular/pro/max) and have the M2 as only a regular for this year. There's no reason that have to ship new Ms every year, and there's no reason they need to ship the full product spectrum (pro/max/ultra/???) every year.

If they do ship an "M2 Pro" based on TSMC N3, it will probably cause at large OCD.
Especially after Apple has already established the M1 family naming conventions, it doesn't make sense to call an A15-based M2 class chip with additional cores an M3.

Calling such a chip M2 Pro/Max makes much more sense. The next generation would be called M3, and M2 Pro/Max still will have advantages over M3.

Instead if you called M2 Pro/Max an M3, and called the next generation mainstream non-pro chip after that an M4, that would be really, really confusing, since M3 would be faster than M4. Naming it as M2 Pro/Max makes sense because it will remain faster than M3.

thats more TMSC than apple. why should Apple's name expose the manufacturing process used? Its fine. M2 > M1. M2 pro > M2, etc. one can argue that Pro, Max, Ultra are confusing (is Max > Ultra? Is Ultra > Pro?) but its not that hard to grasp, there is the base chip and three upgrades.
Apple has already largely established the meanings of Pro & Max with their iPhones. And Apple has established the meaning of Ultra with its pricing. The cheapest Ultra you can buy is US$3999, literally twice that of the Max.
 
Interesting if true. Definitely taking this with a grain of salt.

On one hand: Part of the efficiency of the M1 was its scalability. Each iteration just builds upon what already exist making manufacturing far more cost effective. So it would make more sense from that perspective to keep the M2 Pro and Max chips on the same 5np silicon as the base M2.

On the other hand: Moving the Pro/ Max chips to 3nm should give them a bigger lead over the base M2 and keep Apple competitive in terms of performance for laptops and computers at that price point. This would also line up with rumors of the VR headset possibly using a version of the M2 Pro and other rumors claiming the VR headset requires 3nm.

Pretty interesting for sure! As an M1 Max owner I’m probably waiting until the M3 generation to upgrade regardless but if the M2 Max makes the jump to 3nm that could mean Apple has some bigger things planned for the M3 generation to set it apart. Fingers crossed for GPU ray tracing.
I am on the M1 Max too but I usually keep my MacBooks for 5 years. I'll be M5 or M6 by then.
 
I don't think we'll see a Pro Mac mini personally because it brings it too close to the Studio. I think they'll keep a regular M2 in there or M3 if it waits that long for another refresh.

So then rationalize why an Intel Mac Mini remains in the Apple Store, configurable to $2999 by the way...

MacMiniLoaded.jpg


As a Studio Ultra owner, I'm towards 85% confident that at least a Mac Mini Pro will show up sooner or later base priced at about $1299-$1499, still about $500 less than a base Studio. And that will be the Mini that replaces this Intel Mini which, based upon price and key config options, is still positioned as the "premium" Mini in Apple's own store.

If not, I would guess a Studio Pro debuts at some point to get a Studio starting price DOWN to that "about $1400" range.

If I was betting though, I put my money on Mac Mini Pro.
 
  • Like
Reactions: freedomlinux
Arguing about whether they're going to be labelled M2 or M3 is just irrelevant.
Agreed. The process used to make a chip is unrelated to what Apple's marketing department calls it.

Brand new chip is better or faster or does more than previous chips? Call it something new and slap the new chip name on the boxes to thrill customers. What we're seeing, if this rumor is true*, is just the ability of a chip manufacturer to scale up the production of Apple's next design. That may not fit with what some thought to be a carved-in-stone timeline and naming convention. The name of a given chip made with a given process isn't what some people expected? Oh well.

*Big grain of salt here.
 
Calling such a chip M2 Pro/Max makes much more sense. The next generation would be called M3, and M2 Pro/Max still will have advantages over M3.
To me, this would be confusing, as you'd have Avalanche cores in the M2 and maybe the cores of an A16 Or A17 in an M2 Pro/Max...
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.