Mac Pro Reportedly on 'Back Burner' and 'Largely Written Off' at Apple

[Kidseeghost]

…

 

[zorinlynx]

I actually wonder why an Apple Silicon Mac Pro was released at all. There seemed to be no point to it. The number one reason for a Mac Pro was using a big beefy GPU, and the Apple Silicon version of the Mac Pro doesn't even support that.

There is literally no point to the machine. You can do anything it does with a Mac Studio and Thunderbolt.

The only market remaining for big expandable PC towers is gaming PCs, another market Apple doesn't care about.

Just let it die.

 

[maflynn]

As mentioned earlier, I work in the ProAudio field. One usecase is live mixing when massive amounts of inputs /outputs with minimal I/O latency is needed. (Live Performances of classical orchestras and choirs for example).

Honest question - this defined usage, is the Mac Pro the best cost effective tool do achieve this?

Wouldn't a threadripper  9980X (64‑core) wouldn't be more cost effective and it has PCIe 5.0

 

[theluggage]

But the naming scheme of Apple is not very good today, why the Mac Studio isn’t just the Mac Pro and the Mac Mini the Mac ? Why MacBook Air and not just MacBook ? Why Apple Watch Utra and not Apple Watch Pro ?

When the Mac Studio came out, Apple were still selling another product called Mac Pro.

The 2013 Mac Pro was a low point for Apple - calling the Studio "Mac Pro" (and effectively announcing the demise of the when it came out would have associated it too strongly with the 2013 debacle.

If you were going to call anything just "Mac" then it would be the iMac, which is closer to the original all-in-one Mac concept... but iMac has already become an iconic name. The Mac Mini name has a pretty clear lineage all the way back to the G4 Mini - at which time the alternative d was a Power Mac G5.

Likewise, when the MacBook Air first came out there was (or had recently been) a "just MacBook" product, and since then "MacBook Air" has become a strong brand in itself. Since the last "just Macbook" is now about 10 years ago I guess it's possible that the rumoured A18 MacBook will be a "just MacBook".

Apple Watch Ultra/Pro? Maybe they just though "ultra" made more sense for a watch... I mean, "pro" is actually a silly product label, since it implies your other products are somehow "amateur" but here we are...

More seriously, what's dumb is using the same pro/max/ultra labels for CPUs and complete products so you end up with "pro" machines that don't have "pro" chips and non-"pro" machines that do. Plus not having a distinct brand for the "base" M-series chip.

Don't ask me to defend the naming of the iPad range or I might start to whimper 🙂

The professional software industry bears the primary responsibility for the failure to support dual GPUs in the 2013 Mac Pro.

I'd say that Apple bears the responsibility for mis-reading the way that the industry was headed and/or over-estimating how much influence they still had over the industry. They tried to force things, and failed.

A more traditional Mac Pro design could quite easily have taken dual GPUs without ruling out the possibility of a single large one. Instead, they gave in to form-over-function and built the entire computer around a particular range of Xeon-W CPUs and AMD FireGL GPUs with no clear future roadmap. They also bet the farm on OpenCL instead of CUDA because NVIDIA wouldn't take back what they'd said about Apple's mom (or something).

Anyway, there was more to it than the whole "thermal corner" thing - dropping internal PCIe and internal storage expansion was a major stumbling block for many, when Thunderbolt was far less well-developed. "Ahead of it's time" isn't really a defence - they should have seen that some of the things that it depended on weren't ready.

The Mac Studio is a more credible take on the idea - after nearly 10 years of Thunderbolt/USB4/USB3.2 refinements, a massive advance in what integrated GPUs can do, and with Apple able to guarantee a future upgrade path for the SoC.

I think the Trashcan might even have succeeded if they'd also kept the Cheesegrater up to spec and ran the two lines side-by-side for a few years, instead of forcing a lot of people to make a change that they weren't ready for.

 

[digimc]

This. I doubt that Apple is selling many Mac Pros at all now.

The Mac Studio is powerful enough for local content creation. Very powerful.

And as the commentator above says, with so much cloud compute - and most AI networks running on the cloud - you can just farm off your work off there.

Very small niche of high end content creators who the need the benefits of ultra high bandwidth transfer speeds between large scale video and graphic setups. Also useful for those running large virtual environments that require extremely low latency. Some people are willing to pay through the nose to get that kind of scalability. Running as many virtual linux apps at those speeds, probably a nice thrill for those who can afford it. Not for me though.

 

[OverTheHill]

I actually wonder why an Apple Silicon Mac Pro was released at all. There seemed to be no point to it. The number one reason for a Mac Pro was using a big beefy GPU, and the Apple Silicon version of the Mac Pro doesn't even support that.

There is literally no point to the machine. You can do anything it does with a Mac Studio and Thunderbolt.

The only market remaining for big expandable PC towers is gaming PCs, another market Apple doesn't care about.

Just let it die.

Seriously, I think it was born out of Pixar needs and they kept it going as long as it was needed.

 

[Mac Fly (film)]

Eliminate:
Mac Studio
Mac Pro

Add:
iMac Studio (front-facing port ala Mac Studio—profile pic)
Mac Pro (for size and expansion, think: Mac Studio Plus)

Eliminate:
iPad Air
Apple Pencils (except Apple Pencil Pro, rename it Apple Pencil, lower price to $99)

Add:
More ports to iPad Pro

Eliminate:
iPhone 16
iPhone 16e
iPhone 17
iPhone 17 Pro
iPhone 17 Pro Max
iPhone Air

Eliminate:
Touch function on Camera Control (quick shutter button only)

Add:
iPhone in three sizes 5.7, 6.3, 6.9
iPhone Ultra in three sizes 5.7, 6.3, 6.9 (thicker than current Pro)

Add foldables:
Apple Fold 5.5/7.8
Apple Fold Ultra 6.7/9.5

 

[tonmischa]

Honest question - this defined usage, is the Mac Pro the best cost effective tool do achieve this?

Wouldn't a threadripper  9980X (64‑core) wouldn't be more cost effective and it has PCIe 5.0

I am not a live mixer, so I dont have practical experience in this particular use case. But I suspect that the MacPro is probably not the most cost effective tool. ;-)
Personally I would use a MacStudio with a Sonnet Thunderbolt Chassis.

Most professional ProTools users prefer Macs. So an AMD system would probably not be considered.
Also there is no need for PCIe 5. HDX cards are PCIe 1.1 x4.
Live Mixing is not about maximum bandwidth, but about minimal latency.

 

[ikramerica]

The concept is a relic of the Apple IIe and IBM PC days, where everything was big and bulky and power hungry.

It was novel that the original Macintosh did things without expansion cards, where the PC compatibles needed one card for video, one card for audio, and at first, another card for networking. So 3 slots were filled with necessities already. Many chassis came with no free slots at all. They were upgradeable by removing the video card for a better one, sometimes with onboard audio so you could free up a slot for other things, connected to the system via pins.

None of that is really necessary anymore.

 

[maflynn]

I'd say that Apple bears the responsibility for mis-reading the way that the industry was headed and/or over-estimating how much influence they still had over the industry. They tried to force things, and failed.

The problem is, this wasn't Apple's first time mis-reading. The trash can mac, failed precisely because Apple failed to see where the industry was moving.

 

[Omega Mac]

The iPad and iPad Air should merge into a single product. It's absurd how Apple artificially hold back the iPad by requiring the Apple Pencil to be charged via wire and not using a laminated display. The iPad should be available in three sizes (mini, normal, big) and the iPad Pro in two. The Mac Pro could also be discontinued. I would prefer if they use the Mac Pro name for the Studio so it matches the MacBooks and iPads.

You might see what is being done with iPhone Air will trickle down to into iPad Air to differentiate it even more and then it will make sense in the line up. Just a thought.

 

[JOLoughlin]

The idea of a Mac Pro I believe needs a rethink. A good first step and something Apple would do is utilize the PCIe slots with new Apple made modules (similar to the MPX models of Intel Mac Pro). Apple making more hardware accelerators for certain tasks. GPU rendering, AI training, physics simulation calculations cards, video input and processing specific, audio.

 

[theluggage]

It is certainly not necessary to have those PCIe cards inside the main computer, but for the sake of portability and reliabilty (and piece of mind) it might be better to have everything inside one sturdy rackmounted MacPro.

I see they offer a rackmount/TB3 enclosure. Ignoring the portability/reliability aspect, does Thunderbolt (even one Thunderbolt port per-card) with cut the mustard? If it really needs more than 4 lanes per card or newer PCIe versions, then something like the current Mac Pro is still needed.

Otherwise, I was thinking that a future Mac Pro might look more like a 1U rackmount version of the Studio (rather than a honking great full tower case kludged into a rackmout) with plenty of Thunderbolt to interface with other rackmount devices. Maybe compatible with shallower/lighter-weight racks rather than full-blown data centre ones. E.g. the sort of thing you can get built in to flight cases etc.

Of course, the other silly problem with Thunderbolt/USB-C is that there's no standard retainer clip on the plug (although I note that Sonnet et. al. have created 3rd party solutions).

 

[HVDynamo]

I’d love a Mac Pro and I find it very attractive but I can’t afford one lol

This is the other part of the issue. They priced the average consumer that "would" buy one completely out of the market. $7k starting for 1TB storage and 64GB RAM is a ****ing joke. To be successful, it needed to provide more value for the price, OR be cheaper. I also don't need the Ultra chip. But what I do want is the ability to upgrade storage and RAM so it can remain viable longer without breaking the bank when I purchase it. If it started at $3500 with the Max variant/2TB storage/64GB RAM. They also needed to support other GPU's being used in the expansion slots. I think I'd buy one if they had done all that.

 

[woerden]

As mentioned earlier, I work in the ProAudio field. One usecase is live mixing when massive amounts of inputs /outputs with minimal I/O latency is needed. (Live Performances of classical orchestras and choirs for example).
A ProTools HDX system with 3 PCIe cards can do up to 192 physical inputs/outputs with 0.7 msec latency. Which is not achievable with any other Audio Interfaces.

DSP Audio System for Music and Post - Pro Tools | HDX

Experience the pinnacle of audio production with Avid Pro Tools HDX. Achieve studio-quality sound and unparalleled performance in your projects.

www.avid.com

It is certainly not necessary to have those PCIe cards inside the main computer, but for the sake of portability and reliabilty (and piece of mind) it might be better to have everything inside one sturdy rackmounted MacPro.

HDX might integrate uniquely with ProTools but i feel like their hardware solutions are not otherwise the only ones in the game with that capability. RME offer comparable performance and stability through their MADI stuff.

And if you go the audio networking route you can very easily achieve all of this, although not perhaps at 0,7 ms latency.

 

[deconstruct60]

For a non pro user can someone explain why the Mac Pro wasn’t refreshed with the M3 Ultra like the Studio? What was the benefit of not updating it versus updating it?

The benefit of not updating is should be clear.... it is less expensive. No R&D is always going to be cheaper than some effort that takes time, money , and resources.

If the upgrade user base is very small , then run the risk of doing the expense and not making the money back. Folks who just bought a > $8K Mac Pro very probably would not upgrade. The folks still on Intel MP who didn't upgrade are extremely likely content to just coast until the end of the Intel era on what they have got. (or switch to Linux/WinPC and stay on x86-64). If vast majority buyers are only upgrading on 5-7 year cycles then shorter product upgrade cycles are not going to generate more sales.

No upgrade is also a cheaper 'test' to see how much traction TBv5 could get.

Possible PCC coupling. There are some reports that the custom servers for Apple's Private Cloud Compute are M2 Ultra. Keeping the Mac Pro on M2 Ultra at least keeps a tiny income coming in for the chips, since PCC generates nothing.




The 'upside' benefit for doing an upgrade for a Mac Pro wouldn't be that much.

i. The M4 Max Studio would outclass it on single threan and lightweight-medium weight threaded apps.
( For the Mac Studio that isn't as much of a problem since that still results in a Studio sale. No M4 Max for Mac Pro would be no Mac Pro sale. Unit sales of Mac Pro would go down even more. )

ii. The M3 Ultra only is an advantage over the M4 Max in some corner cases. Very 'embarrassingly parallel' GPU work where must more cores that are a bit slower outpace a smaller number that are faster.
( a Mac Pro might actually 'dilute' a bit of the Studio Ultra sales which are very likely smaller this generation to begin with. ). The render upgrades were good , but that is probably not the folks more likely to get.

iii. If Apple did not upgrade the PCI_e backhaul of the M3 Ultra , then the PCI-e througput would have been the same. The Mac Pro would only really get additional Thunderbolt 5 ports as any I/O 'upgrade'. Tracking poeple into TBv5 is primiarly only tracking people off the Mac Pro. Apple seems to be have been looking at legacy PCI-e cards that were in the "faster than TBv3 but slower than TBV5" bandwidth range with the MP 2023.
( pretty good chance Apple threw any internal chip network bus increase allocated for I/O at TBv5 and not the backhaul.)

iv. Reputational help. If Apple got on to a predictable upgrade schedule then these gaps in Mac Pro upgrades would go over with less 'drama'. However, they are somewhat already crumbled that reputation. At this point, it is hard to sink any lower. It could be 4 years. Last iterations were 6 years , 4 years . It wasn't going to be less than 2 years. It would make the MP fan base happier to see quick cycle upgrades but would that generate substnatially any more sales. ( Probably not ).

If the M2 Mac Pro had not been so late then perhaps there was uplift for a M3 generation ( e.g., if had gone M1 -> M3 ).



The M3 Ultra arrived relatively very late. Basically after the run for M3 Max was done. Some reports were that the M3 Max used in the MBP 14/16" didn't have UltraFUsion. That would make the M3 Max-v2 a special case chip. (and if a special case chip , then pretty likey it has a lame or no update on the PCI-e backhaul . In that case, the plain M3 Max probably dumped the PCI backhaul function to save die space. ). On top of TSMC N3B to boot and very large ( so also more expensive packaging). It wouldn't be too surprising if it were a money loosing chip. Apple decided to take they lumps for the Mac Studio , but placing it in MP would be just digging a deeper hole. [ They may be an option for Apple PCC since that is a money-pit anyway. Web services for zero fees. ]


The M3 generation was a bit of a transitional chip that probably wasn't worth 'betting the farm' on. Ultra's are too big and expensive to throw in the 'trash can' every year; so likely not coming every generation. I think the original Apple plan was M1 -> M3 -> M5 , but that got scrambled by pandemic and TSMC N3 quirks. If M5 Ultra comes and no Mac Pro that would be the odd-duck situation.


P.S. Since Apple has been putting together some custom server nodes , there is a decent chance parts of the MP team that did the rack version got 'hijacked' into doing non-customer facting work. Until Apple gets there PCC situation stabilized and sorted out the MP would get starved of some resources. If Apple diverts a large chunk of M5 Ultras into PCC then MP might just get left out.

 

[HVDynamo]

I really don't understand why the Studio doesn't have socketed, replaceable storage, as an actual touted feature.
The "Pro" demographic is exactly who that should be targeted at.

Sell overpriced first party modules too!
Some will still only but from Apple and it's nice for them to have the option to upgrade storage later.

It also gives base spec machines more longevity, as they can be upgraded down the line as well.

Even having Apples special modules for the Main drive is fine. But not having additional M.2 slots is the real problem. I should be able to drop in an aftermarket 4TB drive as an extra drive without replacing the main OS drive storage.

 

[tonmischa]

does Thunderbolt (even one Thunderbolt port per-card) with cut the mustard? If it really needs more than 4 lanes per card or newer PCIe versions, then something like the current Mac Pro is still needed.

TB3 is enough for 3 HDX cards.
As I said above: (Live) Mixing is not about maximum bandwidth, but about minimum latency.

This is the official TB chassis compatibilty chart from AVID:

https://resources.avid.com/SupportFiles/PT/Pro_Tools_Thunderbolt_Chassis_Compatibility.pdf

 

[Quu]

The biggest issue with the Mac Pro is it doesn't support the most important add-in card that users would want to install, graphics cards. Their architecture they say doesn't even allow it. People want to run multiple GPU's in this thing for AI compute.

So at that point you may aswell just get a Mac Studio. At-least it gets updated to the latest SoC every so often.

 

[Happy_John]

Not surprised. You can solve need for PCI-slots with breakout boxes. If you need serious power to train ML-models that is way beyond a Studio, then cloud services or a Nvidia cluster is the answer. It is mainly nostalgics like myself that still want to hold on to a big box built like a tank. To cheer me up, I bought a MP 6.1 as a hobby project.

Yep, that's it. As I hobbiest I love the Mac Pros 5,1 and 6,1 I have. It's embarrassing to say, but I have taken apart and reconfigured the 5.2 more times that I can remember just for the fun of it, and, even more embarrassingly, probably the most quietly satisfying few hours I've had over the past few years was completely tearing down a 6.1 and rebuilding it, ostensively to upgrade the CPUY and repast the GPUs, but really because it's a really enjoyable machine to do that to. IU;ve also boult a ridiculous amount of old Macs quite cheaply just to take apart and upgrade. It makes me happy.

I really want 7,1. and I will get one when I have the spare silly money to do so. Then I'll spend more more on bits and pieces yo upgrade and modify it with.

But when I put my workout on, I'm working on AS Mac, which I will probably never open up, because there's no need to, and I need the work machine to be functional, reliable and have a good cost (price plus power consumption) to performance ratio.

As a professional machine, there are people out there who might need an AS Mac Pro, but it is incredibly niche. The Mac Studio is the de facto heart of a Mac-based desktop workstation setup now. (although the Mac mini with a pro chip is no slouch either).

 

[chucker23n1]

I think he'd mostly resigned himself to getting a Studio at some point anyhow.
Honestly, reports like this might be better for him as it nudges him closer to clarity and might get him to finally jump and move on.

It's kind of amazing an ATP host is still on an Intel Mac with 2026 upon us.

Fits with how increasingly out of touch the three have gotten.

 

[Happy_John]

I thought they were still working on their own chip for that? If they do succeed there, then the Mac Pro chassis isn't probably the ideal configuration. If Apple manages to break into the AI market, I'd expect a new form factor or the revival of the Xserve.

Maybe a rack mountable unit, but I'd think it'll just be more powerful Studio. Like PCIe breakout boxes, I imagine Apple will just leave it to third party to continue to produce rack mounts for the Studio form factor. It makes sense at this point for Apple to keep the Studio form factor as their standard for "high end boxes". It's why they kept the unibody Mac mini form factor unchanged for so long.

 

[Omega Mac]

Doing Ultra on the watch makes sense since the Watch is not for Pros. iPhone Pro doesn’t make sense. Its a phone not a pro tool. They should go with Pro = professional use and Ultra = high end.
iPhone and iPhone Ultra
Watch and Watch Ultra
iPad and iPad Pro or Ultra (it depends if they finally make it useful for Pros)
Macbook and Macbook Pro
Mac (mini) and Mac Pro (fka Studio)
I don’t know if the iMac should exist. It sells probably too well to get rid of it but its not environmentally friendly enough. Have to trow away the whole thing instead of just buying a new box. Maybe offer a third display for like 799 for the Mac mini.

I get all the comments that the Studio is the defacto pro in this topic but I can not help but see a Mac Studio Pro on the horizon when the Mac Pro inventory clears out.

- Mac Mini
- Mac Studio
- Mac Studio Pro

The Mac Studio Pro chassis can increase in height, if it is not pro enough now, then for some insanely great super over specifications $1000 per extra inch, and/or expansion zones docking bays etc. etc.

The only way is UP and the trashcan never went away, it's the mini and mac studio pro. I dunno how no one sees this. Same form factor just got some straight edges, but essentially the same conceptual approach but done right/things finally come together (various techs) into the optimal availability at point in time and space.

This form factor was already established by the original Mini, and it all harks back to the NeXT cube.

They are riffin' on past designs and the riffs are catchy.

 

[theluggage]

The idea of a Mac Pro I believe needs a rethink. A good first step and something Apple would do is utilize the PCIe slots with new Apple made modules (similar to the MPX models of Intel Mac Pro).

Not a bad idea - you could have a tower into which you could plug Apple Silicon-based "compute modules". The snag is that Apple Silicon doesn't offer great PCIe bandwidth beyond what you get from the unused SSD interface on the second die of an Ultra. Although it's better than Thunderbolt 3/4, even the current Mac Pro still doesn't have the sort of PCIe bandwidth that the 2019 Mac Pro did & certainly doesn't compare to an up-to-date Threadripper (or suchlike) system.

Apple have bet the farm on Thunderbolt and - once that offers enough bandwidth - you could have rack-mounted modules interconnected with Thunderbolt, with the big advantage that the modules would also work with laptops. Where TB isn't enough... Apple would really need a new Apple Silicon chip that emphasised PCIe.

 

[theluggage]

The Mac Studio Pro chassis can increase in height, if it is not pro enough now, then for some insanely great super over specifications $1000 per extra inch, and/or expansion zones docking bays etc. etc.

Where's the PCIe coming from? Only the super-expensive Ultra chips have significant PCIe to spare.

There are already various third-party hubs/docks that "stack" with the studio - and with short Thunderbolt cables they can be quite neat.