When to Expect New M4 MacBook Air, Mac Studio, and Mac Pro Models

[bradman83]

Isn’t the memory bandwidth already superior due to SoC? I’m doing AI development also, and the dependency on CUDA sucks, but if the Silicon architecture is superior I’m hoping it will encourage the development of alternatives.

Compared to conventional DIMM main system memory SoC LPDDRX RAM is much faster, but not compared to GPU cards being used for AI tasks.

The memory bandwidth of an Nvidia 4090 is about 1008 GB/s versus 800 for the M2 Ultra. The rumored specs on the upcoming 5090 would push it to the range of 1500 GB/s.

The M4 Max ups the memory bandwidth to 546 GB/s. Depending on how Apple structures the Ultra chip that could put it in the vicinity of 1092.

 

[HobeSoundDarryl]

You can see this from many angles. If Apple starts with Utra and Extreme, if they manage to build one, they could hamper sale of their very succesfull line up, Macbook Pro. All will be waiting for next CPU.

I only foresee the "hamper" being those who must have "most powerful" now having an ability to kick the can for up to 6 months and spending less money on the next gen MAX chip in Fall releases. This would put some pressure on that thinking, "pulling up" what would probably be a small number of those would would buy M-next MBpro to buying Studio or Pro with which they would enjoy "most powerful" Mac status for at least the next year or longer. This wouldn't be ALL of them as many want a laptop vs. desktop or other FABS... but there's probably a slice revolving around "most powerful" and that's the slice possibly moved by this switch.

Conceptually, anyone who is "pulled up" is probably buying only ONE Mac anyway. So if they are tempted into paying up for these, Apple makes what they covet most: more money... instead of even the "most power"-hungry crowd knowing they can delay only a few months and scratch that itch while spending less money.

Would this "hamper" MBpro sales? I would guess that a few solely in pursuit of "most powerful" would be "lost" from giving Apple less money for that laptop to more money for a desktop. That's not really a loss. However, I would still expect the Fall releases to sell in similar volume. Why? Because they would be doing the SAME thing to BASE chip buyers: pulling a slice of them up to PRO & MAX vs. waiting for "cheapest" and being at the tail end of a M-series generation.

So, this flipping of the schedule, might steal a few buyers of MBpro and make them buyers of Studio or Ultra... but then do the same in the Fall by stealing a few BASE Mac buyers to pay the extra for the Fall releases. In all cases, everyone gets the Mac they want and Apple makes MORE revenue and presumably MORE profit. 💰💰💰

Else, at least through my lens, while I paid up for ULTRA once, I won't do that again knowing that a comparable powered MAX is only up to 6 months away in the next generation. So the added revenue & profit from selling ONE Mac to someone like me is on the line. Does Apple care about any one person? No. But would this kind of thinking really mean I'm the only one? I would think not.

In short: if I'm Apple, I'm not worried about the impact on the Fall releases if the loss is replaced by what would now be these spring/summer releases (at higher revenue and profit). Besides, I would expect the Fall releases to take some of the sales from BASE releases, so they would be more revenue and more profit too... probably with at least the "same great volume" of sales.

As to this making Fall release Mac buyer wait, they'd be waiting for a whole year again as this idea doesn't alter the Fall releases.

I don't KNOW all of this with any certainty- just sharing my own thinking about this topic. To me, this makes more sense than BASE... then PRO/MAX... then MAX/ULTRA by basically flipping that order.

 

[Darren.h]

Bring on the M4 Max Mac Studio. Thats what I Want.

Too many Broken MacBook Pro's in My Day

Bad Keyboard
Bad Battery
Screen going out.

Tired of it all.

 

[Rick_2020]

I just picked up a MBA 13" with 16 GB and 256 GB for $1,249 Canadian. It was $200 off. I don't think the M4 release will be much of a jump and I wanted to trade-in my current 2021 Mac while the trade value was still decent. I ended up trading it in to Amazon (via Phobio) for $200 more than Apple would have given me.

 

[bradman83]

As a Studio ULTRA owner, I don’t buy it again if it is last in line for a generation of chips. Why? Because the M5 MAX will likely have comparable power for a much lower price, released only up to about 6 months later.

It's not a given that a future M5 Max will surpass or even match and M4 Ultra. And I think it's important to take a history lesson in how Apple designed their chips to see why.

The M1 and M2 generations did something unusual for the higher end chips. All of these chips are based on the exact same SoC die; M1 was called Jade and M2 was called Rhodes. The baseline/original die design is that of the Max. The Pro variants are Max dies with sections containing added GPU cores, memory and display controllers, and media encoders cropped out. The Ultra variant was just two of these dies stitched together.

That's why there's no CPU differentiation between the Pro and Max chips from those generations, aside from memory bandwidth and the binned low-end Pro variant. It's even evident in the code names: The M1 Max, Pro and Ultra were called Jade C-Die, Jade C-Chop, and Jade-2C Die, respectively; the M2 Max, Pro, and Ultra were Rhodes 1C, Rhodes Chop, and Rhodes 2C.

With the M3 and generations Apple did something different, each variant is its own unique die design. Doing so allowed Apple to scale the Max beyond the Pro to differentiate the two as true midrange and high end chips (Apple arguably made the M3 Pro a bit too midrange which they rectified with the M4 Pro). Apple essentially added 50% more performance cores into the M3 Max over the M2 Max, which is why it was such a huge leap in performance.

Now here's the important part. Had Apple released an M3 Ultra that was two M3 Max chips stitched together, the M4 Max would be nowhere near as fast (in multicore at least). That's because Apple kept the core counts in the M4 Max the same as the M3 Max. The M3 generation seems like a one time revision to its chip design strategy that allowed Apple to better differentiate their high end chips. Put another way, the M1 and M2 Max and Ultra chips were held back by having to share a die design with the Pro, and that's no longer the case.

 

[theorist9]

Apple actually seems to be de-prioritizing the base M chip for initial release; casting aside the M4 iPad Pro for a moment, both the M3 and M4 debuted in Macs alongside their Pro and Max versions, with the base chip landing first in a lower volume consumer desktop and were then launched in the Air a few months later. I really don't think we'll see the M5 in the spring in the Air; I think we'll see it next fall in a similar October launch that we saw in 2023 and 2024.

I suspect this happens for a few reasons:

• Apple needs to stagger releases to make sure TSMC's bleeding edge production nodes can keep up with demand. TSMC needs to supply enough chips for both the Mac launches but the iPhone as well.
• The MacBook Air gets de-prioritized because it's both a high volume seller but also a consumer-focused device which shouldn't be getting the latest chip before the MacBook Pros or desktop models.
• The Ultra chip is both the most complex to produce but also the lowest volume chip, so it goes last in keeping with industry tradition (Xeon, Epyc, and Threadripper chips are all released long after their microarchitecture siblings are).

The M4 launching in the iPad Pro was likely an aberration. Apple needed to use the M4 because the display controller was designed to handle the tandem OLED display (and is likely why M4s across the board support more displays than the M3), and didn't want to wait until 2025 to launch it.

Ah, you're right—thanks for the correction! I just double-checked, and the base and Pro/Max Macs were indeed released at the same time for both the M3 and M4. It was only with the M1 and the M2 that the base Macs came first (by 11 months for the M1, and 7 months for the M2).

So I should reformulate my question to ask why Apple doesn't release the Max Studio at the same time as the lower-end desktop Macs. As you say, the lower-end desktops (Mini, iMac) shouldn't be getting the latest generation of chip before the higher-end one. With the Ultra Studio, they have an excuse, because that's a different and more complicated chip.

But they don't have a chip-based excuse with the Max Studio, since the Max chip is already available. Unless they plan to redesign the Studio's case, there's no product design reason not to release the Max Studio at the same time as the Max MBP.

It's possible the reason is that initial production of the Max chip is limited, but I'm guessing the Max Studio isn't a high-volume item. My guess is that it's thus instead a business decision: That if they release the Max Studio well before the Ultra Studio, customers who might otherwise have gone for the higher-profit Ultra will instead purchase a Max.

OTOH, I suppose they could surprise us by putting an even more advanced gen-4 chip than the Max in the base Studio--e.g., maybe the Studio and Pro will be getting different versions of the Hidra.

 

[bradman83]

So I should reformulate my question to ask why Apple doesn't release the Max Studio at the same time as the lower-end desktop Macs. As you say, the lower-end desktops (Mini, iMac) shouldn't be getting the latest generation of chip before the higher-end one. With the Ultra Studio, they have an excuse, because that's a different and more complicated chip.

Well it would be a bit like releasing the MacBook Pros with just the Pro and not the Max chip. Apple views the Studio as a singular product with two CPU offerings so they won't release the update until the Ultra is ready, and the Ultra is likely the holdup.

 

[Frantisekj]

I don't KNOW all of this with any certainty- just sharing my own thinking about this topic. To me, this makes more sense than BASE... then PRO/MAX... then MAX/ULTRA by basically flipping that order.

Isnt it now BASE/PRO/MAX... then then Air and then MAX/ULTRA maybe extreme. No one buys Ultra as christmas present. Maybe MBP as well, idk.

 

[HobeSoundDarryl]

As is, BASE, then PRO/MAX (Fall) then (MAX)/ULTRA in these two the next spring or summer. Yes, we can mix in select Mac tags to frame it as BASE, then PRO/MAX, then more BASE and sometimes MAX/ULTRA but I'm focused on just chip releases which seems to be a BASE, PRO & MAX together and lastly ULTRA. And I propose flipping BASE & ULTRA to better fit the various goals of both consumers and Apple Inc. themselves.

Else, at least in my case, I don't pay up for ULTRA again... unless... ULTRA is truly ULTRA... meaning the next-gen MAX doesn't supersede it only up to 6 months later. ULTRA needs to be "king" (of power) for at least a year... else I'd suggest just about anyone wait for M-next MAX, which does get about a year+ most years. And that buyer can spend about HALF of the cost to buy the power "King" for that year+.

 

[theorist9]

Well it would be a bit like releasing the MacBook Pros with just the Pro and not the Max chip. Apple views the Studio as a singular product with two CPU offerings so they won't release the update until the Ultra is ready, and the Ultra is likely the holdup.

I don't think it's the same.

I agree the Pro and Max MBP are a singular product. However, the base price of the Ultra Studio is twice that of the Max Studio, which makes it an entirely different class of product (for comparison, the ratio of the Pro and Max base prices of the 16" M4 MBP is 1.4:1).

The Ultra Studio is more pro than prosumer, and is thus more in the class of the MP. Thus, functionally, it makes sense to me to release the Ultra Studio and MP at the same time.

Plus not doing so—i.e., waiting until the Ultra chip is available before releasing the Max Studio—creates egregious generational discrepancies between the Max Studio, which is still on M2, and the consumer desktops, which are on M4.

 

[developer13245]

I still don’t believe the Mac Pro was originally intended to top out with an Ultra. I think they were unable to fab an Extreme efficiently enough to sell at the price they were aiming for. So I believe they viewed selling the M2 Ultra model as the lesser of two evils, compared to releasing nothing.

I think there’s a market for a box with substantial cooling capacity, PCIE slots, and more compute. And the Pro should be the showcase for the best CPU they can deliver. I don’t need one but I hope they make one.

The current Mac Pro physical design was a "response" to the failed "Trash Can" MP, which we all said was a mistake from the beginning. Apple later confirmed our assessment when they admitted they designed themselves into a "thermal corner". Now it seems they have the opposite problem. The M2 Mac Pro seems like some sort of place holder, and a half baked one at best?
I was surprised to see the physical design survive the Apple silicon transition because the M series chips are all SoC. I had hoped Apple silicon would be more than just SoC, but I doubt that will ever happen. So any hopes of main SDD or memory upgradability are long gone.
The only hope I had left for the Mac Pro was some sort of reasonably priced "main board" full upgrade. Wouldn't it be great if someone with an M2 MP could drop it off at an Apple store and pick it up in a few hours with an M4 under the hood? The physical design would seem to have the flexibility to support this especially because M is SoC. It even makes a strong case for using SoC. But it's never going to happen.

 

[theorist9]

The current Mac Pro physical design was a "response" to the failed "Trash Can" MP, which we all said was a mistake from the beginning. Apple later confirmed our assessment when they admitted they designed themselves into a "thermal corner". Now it seems they have the opposite problem. The M2 Mac Pro seems like some sort of place holder, and a half baked one at best?
I was surprised to see the physical design survive the Apple silicon transition because the M series chips are all SoC. I had hoped Apple silicon would be more than just SoC, but I doubt that will ever happen. So any hopes of main SDD or memory upgradability are long gone.
The only hope I had left for the Mac Pro was some sort of reasonably priced "main board" full upgrade. Wouldn't it be great if someone with an M2 MP could drop it off at an Apple store and pick it up in a few hours with an M4 under the hood? The physical design would seem to have the flexibility to support this especially because M is SoC. It even makes a strong case for using SoC. But it's never going to happen.

Main SSD upgradeabilty has been available on the MP for a while:

Apple 8TB SSD Upgrade Kit for Mac Pro

The SSD kit for Mac Pro enables you to upgrade the internal SSD storage capacity of yourMac Pro. Installation required. Buy now at apple.com.

www.apple.com

RAM upgradeability is more problematic because of the design.

The only hope I had left for the Mac Pro was some sort of reasonably priced "main board" full upgrade. Wouldn't it be great if someone with an M2 MP could drop it off at an Apple store and pick it up in a few hours with an M4 under the hood? The physical design would seem to have the flexibility to support this especially because M is SoC. It even makes a strong case for using SoC. But it's never going to happen.

I agree, and have made this point myself. The entire board can slide out of the Intel MP, so the same should be possible on the AS MP (and perhaps it is). Given the price and nature of the MP, Apple should offer the ability to upgrade the board.

The MP's case, cooling system, PS, etc. should last for a decade. And any components that fail can be easily replaced. It would be a waste to force users who need to upgrade to the new generation to buy an entirely new box. Not green at all.

 

[richinaus]

From now on I am sticking with the pro [Max spec] laptops, and just got rid of my M1 Studio Ultra.
It was a great computer and silent but really it only really proved its worth for about 5% of what I do. Now I just farm that work off to a high spec PC box [access remote on Parsec] and do the majority of my work on an M1 Max MBP [which will be upgraded at M6 redesign] and leave the PC slogging away at renders like the donkey it is.

Has worked out very well so far and can just throw in a 5090 into the PC box when they are out.

 

[tenthousandthings]

The only hope I had left for the Mac Pro was some sort of reasonably priced "main board" full upgrade. Wouldn't it be great if someone with an M2 MP could drop it off at an Apple store and pick it up in a few hours with an M4 under the hood? The physical design would seem to have the flexibility to support this especially because M is SoC. It even makes a strong case for using SoC. But it's never going to happen.

Why won’t it happen? It could still be the case — everything is on one huge board, switching it out is easily done.

Mac Pro (2023) and Mac Pro (Rack, 2023) Exploded View and Orderable Parts - Apple Support

support.apple.com

An upgrade kit would just need the logic board and the Thunderbolt card. Internal SSD(s) could be brought over from the old board.

 

[rp2011]

Love my M1 Max Mac Studio, I tell everyone it is by far my favorite computer I have ever owned. I look forward to the M4 to replace it and hopefully that it will also be finally able to replace my Windows machine for rendering. And I think this time around I may go for the Ultra if it can fully replace an RTX 4090

 

[ampleclarity]

My only bottleneck is RAM. I would love to see an MBA with 64gb or even 128gb of ram. I'd pay the premium!

 

[developer13245]

Why won’t it happen? It could still be the case — everything is on one huge board, switching it out is easily done.

Mac Pro (2023) and Mac Pro (Rack, 2023) Exploded View and Orderable Parts - Apple Support

support.apple.com

An upgrade kit would just need the logic board and the Thunderbolt card. Internal SSD(s) could be brought over from the old board.

Sooo.. has anyone upgraded memory via logic board swap?

Main SSD upgradeabilty has been available on the MP for a while:

Apple 8TB SSD Upgrade Kit for Mac Pro

The SSD kit for Mac Pro enables you to upgrade the internal SSD storage capacity of yourMac Pro. Installation required. Buy now at apple.com.

www.apple.com

RAM upgradeability is more problematic because of the design.

Not green at all.

Forgot about the SSD kit. Thanks.

It's actually very green, but a different type: $$$$ for Apple.

I agree both memory and even processor upgrades seem possible via a logic board swap, but I've lost confidence in Apple to think that way. They completely miss what small businesses have to deal with. For tax accounting purposes, an $8000 desktop is a capital asset that must be depreciated over 5 years. Purchase of upgrades over its lifetime can probably be written off in the year of purchase. I don't know how generous Apple is with trade in values. Does anyone believe Apple will give you $6000 for the $8000 M2 Mac Pro you bought 2 years ago?

 

[Bearxor]

It's not a given that a future M5 Max will surpass or even match and M4 Ultra. And I think it's important to take a history lesson in how Apple designed their chips to see why.

The M1 and M2 generations did something unusual for the higher end chips. All of these chips are based on the exact same SoC die; M1 was called Jade and M2 was called Rhodes. The baseline/original die design is that of the Max. The Pro variants are Max dies with sections containing added GPU cores, memory and display controllers, and media encoders cropped out. The Ultra variant was just two of these dies stitched together.

That's why there's no CPU differentiation between the Pro and Max chips from those generations, aside from memory bandwidth and the binned low-end Pro variant. It's even evident in the code names: The M1 Max, Pro and Ultra were called Jade C-Die, Jade C-Chop, and Jade-2C Die, respectively; the M2 Max, Pro, and Ultra were Rhodes 1C, Rhodes Chop, and Rhodes 2C.

With the M3 and generations Apple did something different, each variant is its own unique die design. Doing so allowed Apple to scale the Max beyond the Pro to differentiate the two as true midrange and high end chips (Apple arguably made the M3 Pro a bit too midrange which they rectified with the M4 Pro). Apple essentially added 50% more performance cores into the M3 Max over the M2 Max, which is why it was such a huge leap in performance.

Now here's the important part. Had Apple released an M3 Ultra that was two M3 Max chips stitched together, the M4 Max would be nowhere near as fast (in multicore at least). That's because Apple kept the core counts in the M4 Max the same as the M3 Max. The M3 generation seems like a one time revision to its chip design strategy that allowed Apple to better differentiate their high end chips. Put another way, the M1 and M2 Max and Ultra chips were held back by having to share a die design with the Pro, and that's no longer the case.

I’m super interested to see what the M4 Ultra looks like even though I’ll never buy one. I’m leaning towards it being a unique design and not just two M4 Max chips interconnected.

 

[tenthousandthings]

Sooo.. has anyone upgraded memory via logic board swap?

Interesting question, but just to be clear, I was talking about a kit for replacing an M2 board with an M4 board. Since the memory is on the logic board, it would be built into the kit.

 

[nathansz]

based on the m4 pro and max gpu,
m4 studio could make a pretty nice little gaming box if the extra gpu cores didn’t cost a fortune

With whiskey/gptk pretty most windows games work fine in macOS now, just the gpu horsepower is lacking

Been playing cities: skylines 2 onnny MacBook m4 pro and it’s entirely playable, just less fps than my former hackintosh 14700/6800 rig

 

[Tago]

The memory bandwidth of an Nvidia 4090 is about 1008 GB/s versus 800 for the M2 Ultra. The rumored specs on the upcoming 5090 would push it to the range of 1500 GB/s.

The M4 Max ups the memory bandwidth to 546 GB/s. Depending on how Apple structures the Ultra chip that could put it in the vicinity of 1092.

My impression is that Apple balances the hardware and won`t provide a lot of bandwidth beyond what the system can handle. Sufficient and then some to ensure it doesn`t become a bottleneck. That`s pretty much all which makes sense to provide. Adding massive unusable bandwidth will ruin that as a sales argument for a next generation or a level up, which CAN benefit from it. Makes sense for them to carry on as they are. Believe they added excess to one of the M1`s and brought it down in the M2 equivalent, but I wiped that story from my caveza.

With the M3 and generations Apple did something different, each variant is its own unique die design. Doing so allowed Apple to scale the Max beyond the Pro to differentiate the two as true midrange and high end chips (Apple arguably made the M3 Pro a bit too midrange which they rectified with the M4 Pro). Apple essentially added 50% more performance cores into the M3 Max over the M2 Max, which is why it was such a huge leap in performance.

Now here's the important part. Had Apple released an M3 Ultra that was two M3 Max chips stitched together, the M4 Max would be nowhere near as fast (in multicore at least). That's because Apple kept the core counts in the M4 Max the same as the M3 Max. The M3 generation seems like a one time revision to its chip design strategy that allowed Apple to better differentiate their high end chips. Put another way, the M1 and M2 Max and Ultra chips were held back by having to share a die design with the Pro, and that's no longer the case.

They surely must have gained some valuable experience from the 3 past generations of the various M`s, and I would assume both Apple and the manufacturer have better capabilities related to varying the puzzle and diversify between levels.

Unless there are serious leaks, I don`t believe the analysts (fortune tellers) can predict how Apple chooses to combine the various cores in a potential Ultra or Extreme. They could add more CPU and a Neural cores but "modestly", and add scores of GPU cores. Or not.

They probably have the ability to choose a desired performance level and stitch up the cores needed to make that happen. Would assume it becomes impressive, but not impressive enough to prevent significant progress for the next generation. Suppose only Apple knows, and the analysts don`t.

My wallet has vetoed any exploration of these seriously tempting machines anyway. My wallet has a taser, and I just got a level 3 warning shot just for considering the idea

 

[cocoua]

About the MS Max, is So sad and disapponment to wait 8 months longer than the MBP to get just the sane computer but even simplier to build and thermal wise .

marketing decisions affecting user experience at it highest

 

[Tago]

About the MS Max, is So sad and disapponment to wait 8 months longer than the MBP to get just the sane computer but even simplier to build and thermal wise .

marketing decisions affecting user experience at it highest

That would be a though one. Any brand who more or less frequently updates their products will have a launch strategy and release products when it makes sense to them.

Christmas ahead, they would launch and prioritize the manufacture of products more likely to sell like hotcakes approaching the holiday season. That would not be Mac Pros so to speak.

Anyway, it is just as much about expectation management I believe. The M4 MBPs hadn`t even arrived at users` desks before the talk about next generation started.

 

[gregmac19]

This article would be a lot more useful if it contained solid information on whether the M4 MacBook Air will be available with the Nano-texture display.

 

[SouthPalito]

and there will be many more as we will close to the Spring/WWDC events

Summary of the template for these stories: Apple plans to upgrade the __ models over the next 12 months. Apple released the M_ processor this year and intends to refresh the entire lineup next year. M_ processors deliver unparalleled performance according to _____ benchmarks ( insert several marketing images grabbed from Apple.com). Reword the same until 4 paragraphs are generated.