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
64,165
31,898


Apple is said to be developing its own AI server processor using TSMC's 3nm process, targeting mass production by the second half of 2025.

Apple-Silicon-AI-Optimized-Feature-Siri-1.jpg

According to a post by the Weibo user known as "Phone Chip Expert," Apple has ambitious plans to design its own artificial intelligence server processor. The user, who claims to have 25 years of experience in the integrated circuit industry, including work on Intel's Pentium processors, suggests this processor will be manufactured using TSMC's 3nm node.

TSMC is a vital partner for Apple, manufacturing all of its custom silicon chips. The chipmaker's 3nm technology is one of the most advanced semiconductor processes available, offering significant improvements in performance and energy efficiency over the previous 5nm and 7nm nodes.

Apple's purported move toward developing a specialist AI server processor is reflective of the company's ongoing strategy to vertically integrate its supply chain. By designing its own server chips, Apple can tailor hardware specifically to its software needs, potentially leading to more powerful and efficient technologies.

Apple could use its own AI processors to enhance the performance of its data centers and future AI tools that rely on the cloud. While Apple is rumored to be prioritizing on-device processing for many of its upcoming AI tools, it is inevitable that some operations will have to occur in the cloud. By the time the custom processor could be integrated into operational servers in late 2025, Apple's new AI strategy should be well underway.

The Weibo user has a number of accurate previous claims, including that the iPhone 7 would be water-resistant and that the standard iPhone 14 models would continue using the A15 Bionic chip, with the more advanced A16 chip being exclusive to the iPhone 14 Pro models. These predictions were later corroborated by multiple credible sources and proved correct upon the products' release.



Article Link: Apple Reportedly Developing Its Own Custom Silicon for AI Servers
 

PeterKeller

macrumors member
Oct 7, 2021
50
385
Canada
This makes SO MUCH sense. The biggest long term cost of server farms is power and cooling. And with Apple's chip technology, they have an advantage here. Margins on these things must be excellent and a new line of Apple Servers with custom chips could be a big growth area - something that Tim Cook needs for the stock to continue to grow. Also, NVidia is the only game in town at the moment, so it's not like it's a crowded field.

This is a much better idea than a Car or even VR Goggles, IMHO. (Although they should be throwing a bunch of stuff at the wall to see what sticks)
 

jayryco

macrumors member
Oct 5, 2022
77
237
I asked this question about Apple developing their own server tech two months ago on these forums and few thought it makes sense for Apple when it actually makes so much sense when looking at it on a macro level and the vertical integration as well as ecosystem optimisation.

 

poorcody

macrumors 65816
Jul 23, 2013
1,325
1,570
Wow! That's exciting. I find it surprising since they seem to focus almost solely on consumer products (including professional of course). I would love a Mac AI workstation with a co-processor as good as or better than an Nvidia desktop GPU...

I wonder if their intense dislike of Nvidia had anything to do with it...
 

deconstruct60

macrumors G5
Mar 10, 2009
12,429
3,979
Is the EU going to let Apple use proprietary chips? It does seem to like companies using anything proprietary.

There is zero indication here that Apple is going to sell these chips. If they don't sell them, then the EU doesn't have squat to say in that case.

The hyperscale server farm vendors doing custom chips are not selling them. They are renting them ( Amazon , Microsoft, Google, Facebook , etc)

Apple is relatively a bit small to tag with "hyperscaler" status but the 'for sale to normal people' generic server market is shrinking.

https://www.nextplatform.com/2024/0...lers-and-clouds-buy-most-servers-and-storage/

Vast majority of Apple's server workload is Linux , not macOS. This isn't necessarily a "M-series" chip that will be sold at retail.
 

deconstruct60

macrumors G5
Mar 10, 2009
12,429
3,979
I asked this question about Apple developing their own server tech two months ago on these forums and few thought it makes sense for Apple when it actually makes so much sense when looking at it on a macro level and the vertical integration as well as ecosystem optimisation.



Most of Apple's server workload is Linux; not macOS. They don't have to build a customer CPU core for server work. They can just license it from Arm. ( just like Amazon , Microsoft , Nvidia , and initially Ampere did). Neoverse server cores are more than decent.

Apple could slap their own NPU/GPU cores on packaged die(s) if they wanted a unified one aimed at inference.
Pretty doubtful they are looking at building some soft of ML Training killer solution because the relative scale probably isn't all that big. This probably some 'spill over' cloud inference on there own hardware to reduce some long term costs.

It doesn't make much sense to build a macOS server chip. There is no Apple 'scale' there.
 
  • Like
Reactions: sos47 and Chuckeee

CWallace

macrumors G5
Aug 17, 2007
12,215
10,998
Seattle, WA
I am of the strong opinion this "server chip" will only be used in Apple's own data centers (as @deconstruct60 suggested) and it will not be sold to end-user customers.

Microsoft, Google, Amazon and others are developing their own "AI server chips" to power their AI data centers and this is just Apple following that trend, rather then depending on third-party solutions (as they currently do for their iCloud data centers).
 
  • Like
Reactions: NT1440

poorcody

macrumors 65816
Jul 23, 2013
1,325
1,570
I am of the strong opinion this "server chip" will only be used in Apple's own data centers (as @deconstruct60 suggested) and it will not be sold to end-user customers.
But if they develop their own AI chips, they are going to put them in their own servers, with their own M# CPUs. If they design such a server for the dataroom, why not put it in a different case and sell it as a Mac Pro?
 

twolf2919

macrumors 6502
Aug 26, 2014
454
765
Wow! That's exciting. I find it surprising since they seem to focus almost solely on consumer products (including professional of course). I would love a Mac AI workstation with a co-processor as good as or better than an Nvidia desktop GPU...

I wonder if their intense dislike of Nvidia had anything to do with it...
Nowhere does it suggest that Apple will actually sell AI servers. Most likely, it'll have to use Google & others for server-side AI processing (for stuff that can't be done on-device) and, when these chips are ready, will do those things in-house on their server farms.
 

deconstruct60

macrumors G5
Mar 10, 2009
12,429
3,979
This makes SO MUCH sense. The biggest long term cost of server farms is power and cooling. And with Apple's chip technology, they have an advantage here. Margins on these things must be excellent and a new line of Apple Servers with custom chips could be a big growth area - something that Tim Cook needs for the stock to continue to grow. Also, NVidia is the only game in town at the moment, so it's not like it's a crowded field.

Nvidia leading in server chips? Not really. AI/ML training card that plug into a server being run by a server chip, yes Nvidia is dominant (but the field has multiple players) . But the main processor in the server, Nvidia is both relatively late to the game and clearly not the only player. Ampere, Amazon , Microsoft , Google. etc. Are all players. Arm has a very viable server core in the Neoverse family. Nvidia is using. Amazon, Microsoft, other hyperscalers are using it. Ampere Computing was/is using it ( transitioning to a custom core for future generation. That may or may not work out for them if all of their major customers just keeping buying Arm's version. )

In AI/ML inference, Nvidia absolute does not have exclusive hold on the market. Inference and training do not have to be done on the same hardware.

MobileEye does AI/ML inference to help automatic car safety features. ( millions of cars with no Nvidia ).

If peak minimal latency is required several inference workloads run solely on the CPU if possible. ( copying the data out to the Nvidia card and back is time. ). One reason why Intel has thrown highly skewed AVX-512 and "DL boost' at the Xeon SP processors to try to backstop some of there competitive losses in server space.

Similarly.
https://www.tomshardware.com/tech-i...-supply-dollar752-million-in-ai-chips-instead

Similarly,
" ... probably setting the stage for what we are calling the AmpereOne-3 chip, which is our name for it and which is etched in 3 nanometer (3N to be precise) processes from TSMC. We think this will be using a modified A2+ core. Wittich confirmed to us that a future AmpereOne chip was in fact using the 3N process and was at TSMC right now being etched as it moves towards its launch. And then he told us that this future chip would have 256 cores. He did not get into chiplet architectures, but did say that Ampere Computing was using the UCI-Express in-socket variant of PCI-Express as its chiplet interconnect for future designs. ...
...
There are a lot of possibilities, but we know one thing: Ampere Computing is hell bent on capturing as much AI inference on its CPUs as is technically feasible.
... "

With an UCI-express inference if Apple wanted to put some of their NPUs on a chiplet (also with a UCI-e interface) and package it together for some custom inference Apple wouldn't have to build a whole server chip and some narrow custom mods to software to invoke the accelerator to offload perhaps more custom inference workloads.
( Arm is using on Neoverse with UCI-express also. )

By second half 2025, Ampere Computing could be on their second generation N3 Arm server chip aimed at inference.
( I'm a bit skeptical they will keep that yearly cadence. )

Finally on the inference front, Google is rolling out Gemini Nano. Apple is doing tons of AI inference in the Vision PRo. The whole Apple line up is reported to be doing more local inference in next versions of iOS/iPadOS/macOS. That is 100's of millions of devices where there is zero Nvidia in sight. Nvidia having AI/ML inference in some kind of unilateral monopoly hold is a complete farce. The AI/ML inference market is far, far , far bigger than the 'largest memory footprint possible', LLM model.

This is a much better idea than a Car or even VR Goggles, IMHO. (Although they should be throwing a bunch of stuff at the wall to see what sticks)

The 'car' project never was much of a 'car' project as a autonomous driving project. That was the bigger problem. If wanted to just make a car that people could drive it would not have been as big of a money pit. The VR Google are largely AI/ML inference though. It is one of the key essential, differentiating features present. It just isn't sold as "AI".

Both of those were 'moving'/'mobile' AI/ML computation projects. Apple doing something that is static and goes no where; that isn't particularly unique or lacking in players tackling the problem.
 

deconstruct60

macrumors G5
Mar 10, 2009
12,429
3,979
I think you're going to be pretty surprised when you learn what Apple did for the M-Series!

There is a difference when Apple can make in tuning a solution for iOS and/or MacOS. This is not a MacOS problem and it is vastly far away from being a iOS problem.

Apple execs has said on multiple occasions that if there is a very good solution they can just buy , they'll buy it. Apple just killed their micro-LED project. (going to buy more OLEDs from LG and Samsung).

Arm's designed sever cores are very competitive. Amazon has replaced a large fraction of it x86 server workload onto Arm desgined cores. The raw boot Llnix and run sophisticated data center virtualization workloads just fine. The hardware, firmware, drivers , software is already there. ( Apple's workload that is still at Amazon (and the portion at Cloudflare) probably is already running on some Arm implementation. Likely will be true for Apple workload at Azure/Google over the course of late 2024/ early 2025 also ) And if they don't like Arms there is a Ampere Computing core options they can choose also that is also currently running real server production workloads at scale.

Yes, there are some hackers justing to cobbling together a complete Linux implementation on M-series. But that isn't production quality OS. It isn't done.
 
Last edited:
  • Like
Reactions: sos47
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.