Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MacRumors

macrumors bot
Original poster
Apr 12, 2001
63,835
31,356


Apple chipmaker TSMC has announced plans to produce highly advanced 1.6nm chips that could be destined for future generations of Apple silicon.

apple-silicon-1-feature.jpg

TSMC yesterday unveiled a series of technologies, including the "A16" process, which is a 1.6nm node. The new technology significantly enhances chip logic density and performance, promising substantial improvements for high-performance computing (HPC) products and data centers.

Historically, Apple is among the first companies to adopt new, state-of-the-art chip fabrication technologies. For example, it was the first company to utilize TSMC's 3nm node with the A17 Pro chip in the iPhone 15 Pro and iPhone 15 Pro Max, and Apple is likely to follow suit with the chipmaker's upcoming nodes. Apple's most advanced chip designs have historically appeared in the iPhone before making their way to the iPad and Mac lineups, and ultimately trickling down to the Apple Watch and Apple TV.

The A16 technology, which TSMC plans to begin producing in 2026, incorporates innovative nanosheet transistors along with a novel backside power rail solution. This development is expected to provide an 8-10% increase in speed and a 15-20% reduction in power consumption at the same speeds compared to TSMC's N2P process, alongside up to a 1.10x chip density improvement.

TSMC also announced the rollout of its System-on-Wafer (SoW) technology, which integrates multiple dies on a single wafer to boost computing power while occupying less space—a development that could be transformative for Apple's data center operations. TSMC's first SoW offering, which is already in production, is based on Integrated Fan-Out (InFO) technology. A more advanced chip-on-wafer version leveraging CoWoS technology is slated for readiness in 2027.

TSMC is also making progress toward manufacturing 2nm and 1.4nm chips that are likely destined for future generations of Apple silicon. Its 2nm "N2" node is scheduled for trial production in the second half of 2024 and mass production in late 2025, to be followed by an enhanced "N2P" process in late 2026. Trial production of the 2nm node will begin in the second half of 2024, with small-scale production ramping up in the second quarter of 2025. In 2027, facilities in Taiwan will start to shift toward production of "A14" 1.4nm chips.

Apple's upcoming A18 chips for the iPhone 16 lineup are expected to be based on N3E, while the "A19" for the 2025 iPhone models is expected to be Apple's first 2nm chip. The subsequent year, Apple will likely move to an enhanced version of this 2nm node, followed by the newly announced 1.6nm process.

Each successive TSMC node surpasses its predecessor in terms of transistor density, performance, and efficiency. Late last year, it emerged that TSMC had already demonstrated prototype 2nm chips to Apple ahead of their expected introduction in 2025.

Article Link: Apple Partner TSMC Unveils Advanced 1.6nm Process for 2026 Chips
 
  • Love
  • Wow
Reactions: gusmula and SFjohn

neilw

macrumors 6502
Aug 4, 2003
444
862
New Jersey
I remember when they were saying that the end of Moore's law was in sight, and the practicality of smaller nodes was doubtful... that was *decades* and many, many shrinks ago. The rate of change has gotten slower but they're still marching forward. Pretty amazing.

Presumably they'll hit a wall eventually, and it'll be mighty interesting to see what happens then.
 

Populus

macrumors 601
Aug 24, 2012
4,878
7,165
Spain, Europe
Wow, this goes faster than I expected! I didn’t expect the 2nm process until 2026 at the earliest. Maybe it’s gonna be a quicker transition than expected…
 

HobeSoundDarryl

macrumors G5
I remember when they were saying that the end of Moore's law was in sight, and the practicality of smaller nodes was doubtful... that was *decades* and many, many shrinks ago. The rate of change has gotten slower but they're still marching forward. Pretty amazing.

Presumably they'll hit a wall eventually, and it'll be mighty interesting to see what happens then.

Marketers always find a way. When the hard limit is reached- and it WILL be reached- they will just shift the "measure" to apply to something else, or split nanometers into a smaller unit of measure before the ultimate limit to restart the countdown. Never let the hard laws of physics get in the way of a good marketing blurb.

Analogy: work the cost of something down in dollars:
  • When it gets to only $1, what will we do? We can shift to cents for 99 more ticks.
  • When it gets to only one cent, what will we do? Work in tenths of a cent (like gas stations appending 9/10ths).
  • When it gets to 1/10th of a cent, what will we do? Shift the decimal point again and begin a fresh countdown.
  • Etc.
There's plenty of smaller measuring tags below "nano." Hello pico, femto, atto, zepto, yocto. Should marketers run out of yocto range some day, something will be invented to term fractions of yocto. None of us or our offspring will live long enough to see us reach a point where there is no longer a smaller measure for marketers to spin. Current kinds of tech products will be made obsolete first.
 
Last edited:

Siliconguy

macrumors 6502
Jan 1, 2022
275
409
I remember when they were saying that the end of Moore's law was in sight, and the practicality of smaller nodes was doubtful... that was *decades* and many, many shrinks ago. The rate of change has gotten slower but they're still marching forward. Pretty amazing.

Presumably they'll hit a wall eventually, and it'll be mighty interesting to see what happens then.
The Van Der Waals's radius of silicon is 210 picometers, so the diameter of an atom is 0.42 nm. There is your hard limit. Actually they won't get that close to it. The feature size TSMC (or Intel) quotes is not based on real physical features.
 

Chet-NYC

macrumors member
Aug 21, 2017
97
333
New York USA
To put this into some perspective: In April 2010, Apple introduced its first in-house processor, the A4. It was build on a 45nm node. Now we are looking at a 1.6 or even 1.4nm nodes. That's insane. to go from 45nm to 1.4nm in 16 years...that's about a 96.5% drop in node size.
 

blastdoor

macrumors newbie
Dec 29, 2022
29
107
how far can we go? will we get below 1nm?
Since it's just a marketing label that has little to do with the size of actual transistors, they can go as low as they want.


quote from article:

The term “3 nm” doesn’t accurately represent the actual size of a transistor. Process nodes still several generations away have a minimum feature size of 13 nm (see below). I would expect that the length of a complete transistor would be at least three times this size, if not longer.
 

blastdoor

macrumors newbie
Dec 29, 2022
29
107
So, 3nm for 2024
2nm for 2025 (and probably also 2026)
16A for 2027.

Meanwhile, Intel's plan is
20A (aka 2nm) for 2024
18A for 2025
14A for 2026
10A for 2027

In terms of roadmaps, Intel is projecting themselves to be ahead of TSMC. If that turns out to be true, then TSMC is going to have to beat their employees harder.
 
  • Disagree
Reactions: bergert

klasma

macrumors 603
Jun 8, 2017
6,020
16,897
I remember when they were saying that the end of Moore's law was in sight, and the practicality of smaller nodes was doubtful... that was *decades* and many, many shrinks ago. The rate of change has gotten slower but they're still marching forward.
The rate of change having gotten slower means precisely that Moore’s law doesn’t hold anymore.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.