mac studio: which M3 Max features are worth waiting for?

[CWallace]

With M3 out, and now M4, Studio sales have to be cratering.

I would not be surprised if most people who wanted a Mac Studio bought an M1 or M2 already and the significant majority are not interested in replacing it with an M3 even if it was available, so arguably it makes sense to wait until WWDC 2025 to announce an M4 model.

I would not at all be surprised if the Mac desktop line (Mini, iMac, Studio and Pro) will go on an "every other" generation upgrade cycle since most owners would be looking at a 3-4 year (minimum) replacement window. I certainly am not using anything near the maximum performance of my M2 Max Studio so it will be many years before I consider upgrading.

 

[Gloor]

Yeah, true. I didn't get M2 Studio as I need raytracing so I'm waiting for that. I really hope that M4 studio will come this WWDC (or this year) and not next. Its embarrassing how Apple is again neglecting its pro users whilst giving useless chips to devices that don't really need it or don't have the software to take advantage of that.

Apple needs to do what others are doing. Flip the release and start with the most powerful chip first and then go with lower stuff. It won't put much strain on the manufacturing and it will work great around iPhones releases. If the M4 studion indeed comes mid 2025 then Apple lost the plot.

It will be 2013 Mac Pro all over again (ie. feeling like 6 years neglect)

I would not be surprised if most people who wanted a Mac Studio bought an M1 or M2 already and the significant majority are not interested in replacing it with an M3 even if it was available, so arguably it makes sense to wait until WWDC 2025 to announce an M4 model.

I would not at all be surprised if the Mac desktop line (Mini, iMac, Studio and Pro) will go on an "every other" generation upgrade cycle since most owners would be looking at a 3-4 year (minimum) replacement window. I certainly am not using anything near the maximum performance of my M2 Max Studio so it will be many years before I consider upgrading.

 

[secondsky]

What might speak for M4s this WWDC is the expected focus on AI for iOS and MacOS..
So if Apple wants to push its own platform for AI development, that the Mac Studio or even the MacPro would make sense. As only the M4 got upgraded Neural Engines.

And a side note: M4 will also mean OLED Tandem for MacBook Pro at end of year. My guess at least.

 

[CWallace]

What might speak for M4s this WWDC is the expected focus on AI for iOS and MacOS. So if Apple wants to push its own platform for AI development, that the Mac Studio or even the MacPro would make sense. As only the M4 got upgraded Neural Engines.

Except Apple is using M2 Ultras for the "AI servers" they are putting into their data centers.

We have to see whether or not the M4 Max has an interposer or not. If there was an M3 Ultra and it was a custom SoC and it was not a viable product, Apple might have to fall back on adding the interposer to M4 Max and continuing with the Ultra being two Max welded together, even if the performance gains do not scale in such a configuration.

 

[secondsky]

Except Apple is using M2 Ultras for the "AI servers" they are putting into their data centers.

We have to see whether or not the M4 Max has an interposer or not. If the M3 Ultra was a custom SoC and it was not a viable product, Apple might have to fall back on adding the interposer to M4 Max and continuing with the Ultra being two Max welded together, even if the performance gains do not scale in such a configuration.

Purpose of servers and desktops are different and as the apple servers are not a product publicly known but rather rumors, i take this with a grain of salt. But i would expect M2 Ultra as M3 is dead and M4 is just ramping up production.

Given that Qualcomms ARM SOC has way better Neural Engine (M4 38TOPS vs. QC 45 TOPS) I see Apple a bit behind on the race if they dont push their SOCs quickly into their products.

 

[CWallace]

Given that Qualcomms ARM SOC has way better Neural Engine (M4 38TOPS vs. QC 45 TOPS) I see Apple a bit behind on the race if they dont push their SOCs quickly into their products.

I don't see Apple "panicking" since they're much more conservative about AI than Google and Microsoft are. So I think they will be perfectly happy with being a bit slower in raw AI performance considering the best Qualcomm SoC is overall not much better than the base M4 and Apple still has M4 Pro and M4 Max waiting in the wings for this Fall.

Maybe we will even see "MHz Myths" make a return with Apple claiming their TOPS are better than Qualcomm's.

All the above being said, I am happy to see Copilot Plus PCs and Qualcomm raising the bar, since that will encourage Apple to improve, as well. So we will hopefully see 12GB or 16GB becoming standard on the M5 MacBook Air in Spring 2026.

 

[purplerainpurplerain]

What might speak for M4s this WWDC is the expected focus on AI for iOS and MacOS..
So if Apple wants to push its own platform for AI development, that the Mac Studio or even the MacPro would make sense. As only the M4 got upgraded Neural Engines.

And a side note: M4 will also mean OLED Tandem for MacBook Pro at end of year. My guess at least.

It’s going to be a tough decision for people who waited for a purchase. Studio or Pro M4 Ultra or MBP M4 Max with tandem OLED and nano option.

 

[CWallace]

It’s going to be a tough decision for people who waited for a purchase. Studio or Pro M4 Ultra or MBP M4 Max with tandem OLED and nano option.

A new 16" MBP every year is the best way to stay on top of the performance curve and if you make your money with it, even taking the trade-in value hit every year is easily absorbable.

Apple has generally been slower with updating the desktops, so not really expecting that to change now so an 18-24 month upgrade cycle sounds "reasonable", even if not necessarily "desirable".

 

[drrich2]

A new 16" MBP every year is the best way to stay on top of the performance curve and if you make your money with it, even taking the trade-in value hit every year is easily absorbable.

Good point, and quite a paradigm shift from years past. Way back, desktops (especially tower units) offered multiple RAM expansion slots, multiple internal storage drive bays and larger, roomier enclosures enabling better air flow for hotter running, more powerful processors.

Apple has so undermined the benefits of non-notebook Macs that at this point, it's mostly about saving hundreds by not having to buy a built-in notebook display. I still wonder how much different in average lifespan a Studio has over a MacBook Pro given hopefully better internal thermals, but that's speculative.

One could argue the 'Ultra option in the Mac Studio' angle, but the performance gain to price ratio of going with the Ultra (M1 or M2, at least) is well beyond the point of diminishing returns for many.

 

[secondsky]

Good point, and quite a paradigm shift from years past. Way back, desktops (especially tower units) offered multiple RAM expansion slots, multiple internal storage drive bays and larger, roomier enclosures enabling better air flow for hotter running, more powerful processors.

Apple has so undermined the benefits of non-notebook Macs that at this point, it's mostly about saving hundreds by not having to buy a built-in notebook display. I still wonder how much different in average lifespan a Studio has over a MacBook Pro given hopefully better internal thermals, but that's speculative.

One could argue the 'Ultra option in the Mac Studio' angle, but the performance gain to price ratio of going with the Ultra (M1 or M2, at least) is well beyond the point of diminishing returns for many.

Well Apple is clearly aware that Mac Studio is cannibalising Mac Pro. Thats why there are rumors for a new Ultra Chip which itself has an interconnect to be able to double it to make an Extreme.
MacBooks are for most people more than enough.
BUT if you edit RAW 8k or higher resolutions, you want as many GPUs and as much CPU power as you can get.
Same for stuff in blender.
M2 Ultra can handle that well, Max models not really.

Cant speak for M2 but M1 was really bad with h265 and 8K, the media encoders delivered such a bad quality, you could throw 100Mps on it and still got way worse image than h264 4K.
So can only hope Apple did fix this since than. At least the M4 got seemingly a bump to 8K.

 

[CWallace]

Well Apple is clearly aware that Mac Studio is cannibalising Mac Pro.

Honestly it is the reverse. There is absolutely zero reason to buy a Mac Pro over a Mac Studio except you want/need PCIe slots for audio/network/storage cards. So every Mac Pro sold would have been a Mac Studio except for people not wanting to use external PCIe bays connected via ThunderBolt.

Thats why there are rumors for a new Ultra Chip which itself has an interconnect to be able to double it to make an Extreme.

If Apple did create a custom M3 Ultra SoC not based on two M3 Max fused together, my guess is that it was also going to be the basis for an M3 Extreme with two of them connected via an interposer. That we now apparently are never going to see either (per Mark), was it because the yields were too poor on N3B that it just was too expensive to manufacture and sell both SoCs? Or was it something like the "M1 Extreme" where the performance scaling was not providing enough value to make it worth buying in any kind of volume?

So is Apple making us (supposedly) wait until WWDC 2025 for at least an M4 Ultra because Apple is going back to the 2xMax joined with an interposer and they won't have enough Max SoCs available until then? (Since one can expect that there will be significant demand for M4 Max MacBook Pros at launch this Fall.) Or are they continuing with a custom M4 Ultra (and Extreme?) and can't dedicate TSMC production slots to it until next Summer because they need that capacity to fab A18 (Pros), M4, M4 Pro and M4 Max?

 

[secondsky]

Honestly it is the reverse. There is absolutely zero reason to buy a Mac Pro over a Mac Studio except you want/need PCIe slots for audio/network/storage cards. So every Mac Pro sold would have been a Mac Studio except for people not wanting to use external PCIe bays connected via ThunderBolt.



If Apple did create a custom M3 Ultra SoC not based on two M3 Max fused together, my guess is that it was also going to be the basis for an M3 Extreme with two of them connected via an interposer. That we now apparently are never going to see either (per Mark), was it because the yields were too poor on N3B that it just was too expensive to manufacture and sell both SoCs? Or was it something like the "M1 Extreme" where the performance scaling was not providing enough value to make it worth buying in any kind of volume?

At the time they had to decide wather M3 Ultra/Extreme i guess it was clear that the production process of TSMC was the contrary what was hoped for, expensive and very low yield rates. And TSMC wasnt able to fix that in months.

So is Apple making us (supposedly) wait until WWDC 2025 for at least an M4 Ultra because Apple is going back to the 2xMax joined with an interposer and they won't have enough Max SoCs available until then? (Since one can expect that there will be significant demand for M4 Max MacBook Pros at launch this Fall.) Or are they continuing with a custom M4 Ultra (and Extreme?) and can't dedicate TSMC production slots to it until next Summer because they need that capacity to fab A18 (Pros), M4, M4 Pro and M4 Max?

The current productions process, which the M4 is using has many customers jumoping on it this year. So i doont think that we have an issue with low capacity, low yield rates or high costs, at least compared to the M3 production process.
I just cant see apple leeting the Mac Studio on M2 for 2 years, you can do that with some iMac for a few bucks.

Even though Mark Gurman says no M4 Mac Studio this year... lets remember this guy also says just 4 weeks before last years Mac Studio M2 release that there wont be another Mac Studio release at all!

Also lets remind us to this news:

M3 Ultra Mac Studio Rumored to Launch in Mid-2024, But No Sign of New Mac Pro

A new Mac Studio model with the M3 Ultra chip—which could be more powerful than expected—will launch in mid-2024, according to a new report. The...

www.macrumors.com

Excerpt:
Interestingly, TrendForce claims that the M3 Ultra chip will be fabricated with TSMC's N3E node, just like the A18 chip that is expected to debut in the iPhone 16 lineup later in the year. This means it would be Apple's first N3E chip.

Now this is Interesting as technical it isnt possible to take M3 which was designed for N3B to N3E. So the claim is on that wrong. BUT, M4 would be that chip on the new N3E architecture.

Also keep in mind that the M3 Max seemingly doesnt have a interconnect interface on the die as you can see here (down on the chips). Which again speaks for a new type of chip design.

https://twitter.com/i/web/status/1740142759942750246

 

[Gloor]

Yeah, I think Gurman is wrong (he has to be) otherwise Apple really screws with Pro users again.

 

[Harry Haller]

Yeah, I think Gurman is wrong (he has to be) otherwise Apple really screws with Pro users again.

Why not?
What's Apple's punishment been in the last 11 years for screwing Pro users?

 

[secondsky]

Well another hint that there will be more M4 hardware coming very soon was the iPad event itself.
Did you get any details on the M4 during the keynote?
Exactly, not really.
Now how high are the chances that they wait till late fall to introduce something with M4 Pro/Max and explain the new M4 as they did with every single M generation? And they wont introduce something else with just the M4, they always did explanations with Macbooks when M Pro/Max was introduced. But Macbooks wont be updated in such short time again. So only products left are Studio/MacPro.

 

[Pressure]

Well another hint that there will be more M4 hardware coming very soon was the iPad event itself.
Did you get any details on the M4 during the keynote?
Exactly, not really.
Now how high are the chances that they wait till late fall to introduce something with M4 Pro/Max and explain the new M4 as they did with every single M generation? And they wont introduce something else with just the M4, they always did explanations with Macbooks when M Pro/Max was introduced. But Macbooks wont be updated in such short time again. So only products left are Studio/MacPro.

The Mac mini is still on M2 as well.

 

[djinn]

I can confidently say that Apple can do whatever they want and us Apple users will gripe, complain, threaten, but we always come back and purchase their products. It is what it is....

 

[Gloor]

Not really. Seeing how others are catching up etc. its easier to make a different choice than it was in the past.

Also, Apple's main revenue is the iPhone which is actually the easiest thing to replace. There are so many great phones out there. Few missteps with that and Apple will be less arrogant.

The more competition we get the better it is for customers. Its only a matter of time before Apple gets slapped. No company can be on the 'top' forever.

And if AI is truly the future and Apple is so behind then maybe that could be the catalyst for a shift. We don't know what it will be but something will eventually come. And then its up to Apple to do the right thing and innovate/improve or be stubborn and lose.

We have had tons of examples in the past.

I can confidently say that Apple can do whatever they want and us Apple users will gripe, complain, threaten, but we always come back and purchase their products. It is what it is....

 

[djinn]

Not really. Seeing how others are catching up etc. its easier to make a different choice than it was in the past.

Also, Apple's main revenue is the iPhone which is actually the easiest thing to replace. There are so many great phones out there. Few missteps with that and Apple will be less arrogant.

The more competition we get the better it is for customers. Its only a matter of time before Apple gets slapped. No company can be on the 'top' forever.

And if AI is truly the future and Apple is so behind then maybe that could be the catalyst for a shift. We don't know what it will be but something will eventually come. And then its up to Apple to do the right thing and innovate/improve or be stubborn and lose.

We have had tons of examples in the past.

Don't get me wrong, I agree that competition is good. But the change can be painful, at least from my own experience. I tried switching to the Samsung S24 Ultra, amazing phone, but started missing how well Apple product interface either one another.

 

[Gloor]

Yeah, I get that. I'm still using Apple stuff but I'm starting to be open to competition as I'm just getting bored with their arrogance.

$999 for a monitor stand is one of the few examples :-D

Don't get me wrong, I agree that competition is good. But the change can be painful, at least from my own experience. I tried switching to the Samsung S24 Ultra, amazing phone, but started missing how well Apple product interface either one another.

 

[CWallace]

At the time they had to decide whether M3 Ultra/Extreme i guess it was clear that the production process of TSMC was the contrary what was hoped for, expensive and very low yield rates. And TSMC wasn't able to fix that in months.

Apple and TSMC knew N3B was not a long-term process. Apple went to it because they needed to validate 3nm SoCs and TSMC needed to validate "larger scale" N3B production.

The current productions process, which the M4 is using has many customers jumping on it this year. So i doont think that we have an issue with low capacity, low yield rates or high costs, at least compared to the M3 production process.

But that general demand could be an issue. Apple was able to buy (effectively) all initial N3B production because nobody else was ready (or willing) to go to it. N3E is the process "everyone" wants to be on and while Apple can buy a majority of it due to their "Best Customer" status, they likely cannot buy all of it so they will need to balance capacity TSMC is able/willing to sell them with their product portfolio. And we know A18 and A18 Pro are going to be the most-important followed by the M4 family, starting with the base M4 and then M4 Pro and M4 Max.

I just cant see apple letting the Mac Studio on M2 for 2 years, you can do that with some iMac for a few bucks.

Yeah, I think Gurman is wrong (he has to be) otherwise Apple really screws with Pro users again.

Why not?
What's Apple's punishment been in the last 11 years for screwing Pro users?

What @Harry Haller said.

Apple has traditionally been much slower to update their desktop line, partly due to lack of CPUs from partners (Intel) and more partly due to them making up a relatively small portion of Apple Mac sales (15-20%) due to the longer replacement cycles for them.

I mean let's be honest, how many current M2 Mac Studio users are going to replace with M4 if it did drop at WWDC? These people spent $2000 to $9000 on their machines only one year ago and they are still pretty darn powerful, even if they are not the most powerful.

Well another hint that there will be more M4 hardware coming very soon was the iPad event itself.
Did you get any details on the M4 during the keynote?
Exactly, not really.
Now how high are the chances that they wait till late fall to introduce something with M4 Pro/Max and explain the new M4 as they did with every single M generation? And they wont introduce something else with just the M4, they always did explanations with Macbooks when M Pro/Max was introduced. But Macbooks wont be updated in such short time again. So only products left are Studio/MacPro.

Apple did do a relatively decent review of the M4 and they will probably talk more about it's AI-specific features at WWDC when the announce macOS 16. And then we will have the technical deep dive of the M4 Pro and M4 Max this fall when the MacBook Pro (and Mac mini?) launches.

 

[Gloor]

I don't think its about M2 Studio users but more for those that are still on Intel (like me) or M1 Studio users.

I need raytracing so I won't be buying M2 Studio and seeing that they are skipping M3 then I have to wait.

Few of my colleagues are in the same boat so as nice as it is to have raytracing in a laptop its the studio and pro users that need it the most.

So Mac Pro and Mac Studio should get updated at WWDC. Yes, Apple has a history to neglect but they also promised to do better.
As Studio is a new product I'm sure it sells way way more than the the Pro so there is probably bigger reason to keep it up to date. (unlike Pro that is now even more super niche)

Apple and TSMC knew N3B was not a long-term process. Apple went to it because they needed to validate 3nm SoCs and TSMC needed to validate "larger scale" N3B production.




But that general demand could be an issue. Apple was able to buy (effectively) all initial N3B production because nobody else was ready (or willing) to go to it. N3E is the process "everyone" wants to be on and while Apple can buy a majority of it due to their "Best Customer" status, they likely cannot buy all of it so they will need to balance capacity TSMC is able/willing to sell them with their product portfolio. And we know A18 and A18 Pro are going to be the most-important followed by the M4 family, starting with the base M4 and then M4 Pro and M4 Max.








What @Harry Haller said.

Apple has traditionally been much slower to update their desktop line, partly due to lack of CPUs from partners (Intel) and more partly due to them making up a relatively small portion of Apple Mac sales (15-20%) due to the longer replacement cycles for them.

I mean let's be honest, how many current M2 Mac Studio users are going to replace with M4 if it did drop at WWDC? These people spent $2000 to $9000 on their machines only one year ago and they are still pretty darn powerful, even if they are not the most powerful.



Apple did do a relatively decent review of the M4 and they will probably talk more about it's AI-specific features at WWDC when the announce macOS 16. And then we will have the technical deep dive of the M4 Pro and M4 Max this fall when the MacBook Pro (and Mac mini?) launches.

 

[secondsky]

Apple and TSMC knew N3B was not a long-term process. Apple went to it because they needed to validate 3nm SoCs and TSMC needed to validate "larger scale" N3B production.

They didnt know, the N3B process fall massivly behind the expected results and TSMC wasnt able to fix all issues with it. Thats why they push customers out of it (and customers pull out due to high costs - Only one is apple btw)

But that general demand could be an issue. Apple was able to buy (effectively) all initial N3B production because nobody else was ready (or willing) to go to it. N3E is the process "everyone" wants to be on and while Apple can buy a majority of it due to their "Best Customer" status, they likely cannot buy all of it so they will need to balance capacity TSMC is able/willing to sell them with their product portfolio. And we know A18 and A18 Pro are going to be the most-important followed by the M4 family, starting with the base M4 and then M4 Pro and M4 Max.

Not true, TSMC can handle all customers for N3E by the end of 2024. They build new fabs which will open soon. They already started production in December 2023.

Apple has traditionally been much slower to update their desktop line, partly due to lack of CPUs from partners (Intel) and more partly due to them making up a relatively small portion of Apple Mac sales (15-20%) due to the longer replacement cycles for them.

I mean let's be honest, how many current M2 Mac Studio users are going to replace with M4 if it did drop at WWDC? These people spent $2000 to $9000 on their machines only one year ago and they are still pretty darn powerful, even if they are not the most powerful.

Apple did do a relatively decent review of the M4 and they will probably talk more about it's AI-specific features at WWDC when the announce macOS 16. And then we will have the technical deep dive of the M4 Pro and M4 Max this fall when the MacBook Pro (and Mac mini?) launches.

Mac Studio makes like 1% all all Mac Hardware Sales and 4 % of all desktop Macs sales, so yes its a small market, but that market has a benefit for apple which goes beyond some little cash. These people are multiplicators, thats worth more for Apple than the cash they bring in with some small hardware sales.

And at least in last years we saw yearly updates. So something changed from Intel times clearly.
Tell me who would buy a Mac Studio for those high prices as M2 if you know exactly next year you get M4 and if you switch to M3 Max you get near to a M2 Ultra. Basically only a very very very small number of people who need the more GPUs in their workflows might be ok to somehow still buy a M2 Mac Studio.

 

[CWallace]

They didnt know, the N3B process fall massivly behind the expected results and TSMC wasnt able to fix all issues with it. Thats why they push customers out of it (and customers pull out due to high costs - Only one is apple btw)

I disagree based on all the reporting I have read about it, but not worth arguing over.

Not true, TSMC can handle all customers for N3E by the end of 2024. They build new fabs which will open soon. They already started production in December 2023.

Maybe, maybe not. Guess we will see.

Mac Studio makes like 1% all all Mac Hardware Sales and 4 % of all desktop Macs sales, so yes its a small market, but that market has a benefit for apple which goes beyond some little cash. These people are multiplicators, thats worth more for Apple than the cash they bring in with some small hardware sales.

Laptop sales are far more important to Apple and those laptops influence a heck of a lot more potential new sales than desktops. The replacement cycle for laptops is also usually shorter due to greater overall advantages per model generation (lighter weight, better displays, longer battery life, etc.) so laptops are updated on a more regular schedule than desktops.

And at least in last years we saw yearly updates. So something changed from Intel times clearly.
Tell me who would buy a Mac Studio for those high prices as M2 if you know exactly next year you get M4 and if you switch to M3 Max you get near to a M2 Ultra. Basically only a very very very small number of people who need the more GPUs in their workflows might be ok to somehow still buy a M2 Mac Studio.

People buy when they need to. If you do not need a Mac Studio until WWDC 2025, then you wait for the M4 model. If you cannot, then you buy the M2 model and plan an upgrade with M6 or M7. And if your Mac Studio is your primary source of income, then you buy M2 now and M4 in a year (and M5 a year after that, and so on...and so on...).

And if you must have what M3 or M4 offers "now", buy a 16" MacBook Pro with the M3 Max/M4 Max and run it in clamshell mode. Yes, you are paying much more for that unneeded portability, but again, if you make your living off of it, the machine pays for itself almost immediately.

 

[secondsky]

I disagree based on all the reporting I have read about it, but not worth arguing over.

just one report here, feel free to google yourself tons of more:

TSMC’s 3nm yield rate reportedly just 55% · TechNode

TSMC is struggling with its new 3nm process, with the semiconductor giant's yield rate reportedly far below the standard expected.

technode.com

55% yield is so bad, even if Apple would want to stay on N3B, TSMC would say, hell no it costs us tons of money.
Yeah they got better from that 55% but still not into number regions which makes it viable to produce this.

Laptop sales are far more important to Apple and those laptops influence a heck of a lot more potential new sales than desktops. The replacement cycle for laptops is also usually shorter due to greater overall advantages per model generation (lighter weight, better displays, longer battery life, etc.) so laptops are updated on a more regular schedule than desktops.

You got me wrong, this isnt about sales.
Apple still cares about the reputation in the creative space.
And that is something you cant buy.

And about sales of Mac Studio, again there is jsut nearly no reason from hardware specs to buy a Mac Studio even right now, the M3 Max is as fast as a M2 Ultra! Just in very very small use cases the more GPUs in the ultra play a role. But if that is worth the extra price from M3 Max Macbook Pro to M2 Ultra... i have doubts. I cant see anyone really buying that thing anymore.