I bet there’s gonna be a banger of a Mac Studio announced in June.
Apple really stumbled into making the perfect hardware for home inference machines. Does any hardware company come close to Apple in terms of unified memory and single machines for high throughput inference workloads? Or even any DIY build?
When it comes to the previous “pro workloads,” like video rendering or software compilation, you’ve always been able to build a PC that outperforms any Apple machine at the same price point. But inference is unique because its performance scales with high memory throughput, and you can’t assemble that by wiring together off the shelf parts in a consumer form factor.
It’s simply not possible to DIY a homelab inference server better than the M3+ for inference workloads, at anywhere close to its price point.
They are perfectly positioned to capitalize on the next few years of model architecture developments. No wonder they haven’t bothered working on their own foundation models… they can let the rest of the industry do their work for them, and by the time their Gemini licensing deal expires, they’ll have their pick of the best models to embed with their hardware.
> But inference is unique because its performance scales with high memory throughput, and you can’t assemble that by wiring together off the shelf parts in a consumer form factor.
Nvidia outperforms Mac significantly on diffusion inference and many other forms. It’s not as simple as the current Mac chips are entirely better for this.
You don’t need it if you use llamacpp on Windows, or if you compile it on Linux with CUDA 13 and the correct kernel HMM support, and you’re only using MoE models (which, tbh, you should be doing anyways).
You can still buy used 3090 cards on ebay. 5 of them will give you 120GB of memory and will blow away any mac in terms of performance on LLM workloads. They have gone up in price lately and are now about $1100 each, but at one point they were $700-800 each.
Where are you gonna find Apple hardware with 128GB of memory at enthusiast-compatible price?
The cheapest Apple desktop with 128GB of memory shows up as costing $3499 for me, which isn't very "enthusiast-compatible", it's about 3x the minimum salary in my country!
Jeff Geerling doing that 1.5TB cluster using 4 Mac Studios was pretty much all the proof needed to demo how the Mac Pro is struggling to find any place any more.
> Apple really stumbled into making the perfect hardware for home inference machines
For LLMs. For inference with other kinds of models where the amount of compute needed relative to the amount of data transfer needed is higher, Apple is less ideal and systems worh lower memory bandwidth but more FLOPS shine. And if things like Google’s TurboQuant work out for efficient kv-cache quantization, Apple could lose a lot of that edge for LLM inference, too, since that would reduce the amount of data shuffling relative to compute for LLM inference.
To me there is a fundamental difference. Even if PC hardware costs slightly more (now because of the RAM situation, Apple producing his chips in house can get better deals of course), it's something that is worth more investing in in.
Maybe you spend 1000$ more for a PC of comparable performance, well tomorrow you need more power, change or add another GPU, add more RAM, add another SSD. A workstation you can keep upgrade it for years, adding a small cost for an upgrade in performance.
An Apple machine is basically throw away: no component inside can be upgraded, you need more RAM? Throw it away and buy a new one. You want a new GPU technology? You have to change the whole thing. And if something inside breaks? You of course throw away the whole computer since everything is soldered on the mainboard.
There is then the software issue, with Apple devices you are forced to use macOS that kind of sucks, especially for a server usage. True nowadays you can install Linux on it, but the GPU it's not that well supported, thus you loose all the benefits. You have to stuck with an OS that sucks, while in the PC market you have plenty of OS choices, Windows, a million of Linux distributions, etc. If I need a workstation to train LLM why do I care about a OS with a GUI? It's only a waste of resources, I just need a thing that runs Linux and I can SSH into it. Also I don't get the benefit of using containers, Docker, etc.
Mac suck even hardware side form a server point of view, for example it's not possible to rack mount them, it's not possible to have redundant PSU, key don't offer remote KVM capability, etc.
> ...making the perfect hardware for home inference machines.
I really don't get why anybody would want that. What's the use case there?
If someone doesn't care about privacy, they can use for-profit services because they are basically losing money, trying to corner the market.
If they care about privacy, they can rent cloud instances in order to setup, run, close and it will be both cheaper, faster (if they can afford it) but also with no upfront cost per project. This can be done with a lot of scaffolding, e.g. Mistral, HuggingFace, or not, e.g. AWS/Azure/GoogleCloud, etc. The point being that you do NOT purchase the GPU or even dedicated hardware, e.g. Google TPU, but rather rent for what you actually need and when the next gen is up, you're not stuck with "old" gen.
So... what use case if left, somebody who is both technical, very privacy conscious AND want to do so offline despite have 5G or satellite connectivity pretty much anywhere?
I honestly don't get who that's for (and I did try a dozens of local models, so I'm actually curious).
PS: FWIW https://pricepertoken.com might help but not sure it shows the infrastructure each rely on to compare. If you have a better link please share back.
I'm not a big fan of reducing computing as a whole to just inference. Apple has done quite a bit besides that and it deserves credit. Mac Pro disappearing from the product line is a testament to it, that their compact solutions can cover all needs, not just local inference, to a degree that an expandable tower is not required at all.
CUDA 13 on Linux solves the unified memory problem via HMM and llamacpp. It’s an absolute pain to get running without disabling Secure Boot, but that should be remedied literally next month with the release of Ubuntu 26.04 LTS. Canonical is incorporating signed versions of both the new Nvidia open driver and CUDA into its own repo system, so look out for that. Signed Nvidia modules do already exist right now for RHEL and AlmaLinux, but those aren’t exactly the best desktop OSes.
But yeah, right now Apple actually has price <-> performance captured a lot of you’re buying a new computer just in general.
Agreed. I’m planning on selling my 512GB M3 Ultra Studio in the next week or so (I just wrenched my back so I’m on bed-rest for the next few days) with an eye to funding the M5 Ultra Studio when it’s announced at WWDC.
I can live without the RAM for a couple of months to get a good price for it, especially since Apple don’t sell that model (with the RAM) any more.
I do love the Mac Studio. I had a 2019 Mac Pro, the Intel cheesegrater, but my home office upstairs became unpleasant with it pushing out 300W+. I replaced it with the M2 Ultra Studio for a fraction of the heat output (though I did had to buy an OWC 4xNVMe bay).
> I bet there’s gonna be a banger of a Mac Studio announced in June. Apple really stumbled into making the perfect hardware for home inference machines.
This I'm not actually as sure about. The current Studio offerings have done away with the 512GB memory option. I understand the RAM situation, but they didn't change pricing they just discontinued it. So I'm curious to see what the next Studio is like. I'd almost love to see a Studio with even one PCI slot, make it a bit taller, have a slide out cover...
As to better or cheaper homelab: depends on the build. AMD AI Max builds do exist, and they also use unified memory. I could argue the competition was, for a long time, selling much more affordable RAM, so you could get a better build outside Apple Silicon.
The typical inference workloads have moved quite a bit in the last six months or so.
Your point would have been largely correct in the first half of 2025.
Now, you're going to have a much better experience with a couple of Nvidia GPUs.
This is because of two reasons - the reasoning models require a pretty high number of tokens per second to do anything useful. And we are seeing small quantized and distilled reasoning models working almost as well as the ones needing terabytes of memory.
Once you start adding RAM and storage, the price of the Apple explodes. There is zero expandability and options outside of what Apple provides. So you can easily build a better PC and still at a lower price point for higher end configs. So your argument is a fanboy argument and not based on reality.
Apple abandoned the pro market long before ever releasing the current iteration of Mac Pro. I doubt they care about getting it back considering its a smaller niche of consumers and probably significantly more investment on the software side.
At best we probably get a chassis to awkwardly daisy chain a bunch of Mac Studios together
The interesting question is whether they'll lean into it intentionally (better tooling, more ML-focused APIs) or just keep treating it as a side effect of their silicon design
I don't think Apple just stumbled into it, and while I totally agree that Apple is killing it with their unified memory, I think we're going to see a pivot from NVidia and AMD. The biggest reason, I think, is: OpenAI has committed to enormous amount capex it simply cannot afford. It does not have the lead it once did, and most end-users simply do not care. There are no network effects. Anthropic at this point has completely consumed, as far as I can tell, the developer market. The one market that is actually passionate about AI. That's largely due to huge advantage of the developer space being, end users cannot tell if an "AI" coded it or a human did. That's not true for almost every other application of AI at this point.
If the OpenAI domino falls, and I'd be happy to admit if I'm wrong, we're going to see a near catastrophic drop in prices for RAM and demand by the hyperscalers to well... scale. That massive drop will be completely and utterly OpenAI's fault for attempting to bite off more than it can chew. In order to shore up demand, we'll see NVidia and AMD start selling directly to consumers. We, developers, are consumers and drive demand at the enterprises we work for based on what keeps us both engaged and productive... the end result being: the ol' profit flywheel spinning.
Both NVidia and AMD are capable of building GPUs that absolutely wreck Apple's best. A huge reason for this is Apple needs unified memory to keep their money maker (laptops) profitable and performant; and while, it helps their profitability it also forces them into less performant solutions. If NVidia dropped a 128GB GPU with GDDR7 at $4k-- absolutely no one would be looking for a Mac for inference. My 5090 is unbelievably fast at inference even if it can't load gigantic models, and quite frankly the 6-bit quantized versions of Qwen 3.5 are fantastic, but if it could load larger open weight models I wouldn't even bother checking Apple's pricing page.
tldr; competition is as stiff as it is vicious-- Apple's "lead" in inference is only because NVidia and AMD are raking in cash selling to hyperscalers. If that cash cow goes tits up, there's no reason to assume NVidia and AMD won't definitively pull the the rug out from Apple.
As someone who came from the SGI O2/Octane era when high-end workstations were compact, distinctive, and sexy, I’ve never really understood the allure of the Mac Pro, with the exception of the 2013 Mac Pro tube, which I owned (small footprint, quiet, and powerful).
For me, aesthetics and size are important. That workstation on your desk should justify its presence, not just exist as some hulking box.
When Apple released the Mac Studio, it made perfect sense from a form-factor point-of-view. The internal expansion slots in the M2 Mac Pro didn't make any sense. It was like a bag of potato chips - mostly air. And far too big and ugly to be part of my work area! I'm surprised that Apple didn't discontinue it sooner.
This would probably push some high-end audio professionals away from Logic. One of the niches Mac Pro has been popular is audio production. And with cheesegrader the ability to slot in many-many different audio interfaces into a box instead of dangling out to various PCIe enclosures has been a big win.
Apple really dropped the ball here. They had every ability to make something competitive with Nvidia for AI training as well as inference, by selling high end multi GPU Mac Pro workstations as well as servers, but for some reason chose not to. They had the infrastructure and custom SoCs and everything. What a waste.
It really could have been a bigger market for them than even the iPhone.
G5 was the thing. And companies were buying G5 and other macs like that all the time, because you were able to actually extend it with video cards and some special equipment.
But now we have M chips. You don't need video for M chips. You kinda do, but truthfully, it's cheaper to buy a beefier Mac than to install a video card.
Pro was a great thing for designers and video editors, those freaks who need to color-calibrate monitors. And right now even mini works just fine for that.
And as for extensions - gone are the days of PCIe. Audio cards and other specialized equipment works and lives just fine on USB-C and Thunderbolt.
I remember how many months I've spent trying to make Creative Labs Sound Blaster to work on my 486 computer. At that time you had to have a card to extend your system. Right now I'm using Scarlett 2i2 from Focusrite. It works over USB-C with my iPhone, iPad and Mac. DJIs mics work just as good.
Damn, you can buy Oscilloscope that works over USB-C or network.
It's not the Mac's or Apple's fault. We are actually live in the age where systems are quite independent and do not require direct installations.
The Ultra variants of the M series chips had previously consisted of two of the Max chips bonded together.
The M5 generation Pro and Max chips have moved to a chiplet based architecture, with all the CPU cores on one chiplet, and all the GPU cores on another.
The latest Mac Pro really didn't make much use of its size, as there were too few useful things to put into. Especially as the GPU is now part of the package anyway. Also, the Mac Studio is the perfect workstation for the desk.
Still, there are a few things which could be improved relative to the current Studio. First, the ability to easily clean the internals from dust. You should be able to just lift the lid and clean the computer. Also, it would be great to have one Mac which you could just plug in a bunch of NVMe disks.
On the other side, they might replace the Mac Pro with a rack mountable machine as the demand for ARM servers in the cloud raises.
The 2013 trash can was the end of the Mac Pro. It was never the same after that. The 2012 and earlier Mac Pros were awesome. I had a 2010 model. Here's what I loved:
• Multiple hard drive bays for easy swapping of disks, with a side panel that the user could open and close
• Expandable RAM
• Lots of ports, including audio
• The tower took up no desktop space
• It was relatively affordable, starting at $2500. Many software developers had one. (The 2019 and later Mac Pros were insanely expensive, starting at $6000.)
The Mac Studio is affordable, but it lacks those other features. It has more ports than other Macs but fewer in number and kind than the old Mac Pro, because the Mac Studio is a pointlessly small desktop instead of floor tower.
The 2019 Mac Pro’s main purpose was to provide much needed reassurance that Apple cared about the Mac. In prior years the quality of the Macs had fallen over all product lines. And the question of does Apple care about the Mac at all was a legitimate one.
This Mac Pro was about resetting and giving a clear signal that Apple was willing to invest in the Mac far more than it was about ‘slots’.
Today, Mac hardware is the best it has ever been, and no one is reasonably questioning apple’s commitment to a Mac hardware.
So it makes sense for the Mac Pro to make a graceful exit.
While the trash can generation was somewhat present and around, I don't think I ever saw a cheese grater in the flesh. Did it have any users? Were there any actual useful expansion cards? Did anybody continue buying this at all, after it didn't get the M3 Ultra bump, that the Mac Studio got last year?
Still rocking a 2019 (Intel) Mac Pro here, all slots filled with various Pro Tools and UAD DSP cards, SSD, GPU, etc. I'm planning to get as much mileage out of it as I can. I'm sure a Studio would be more performant, but the Thunderbolt to PCIe chassis are not cheap.
They've been trying to kill the Mac Pro for over a decade. I wonder how long before they backtrack again? It seems like they should at least have a migration path for users who needed the expansion cards the Mac Pro supported. Pushing them to the PC seems pretty bad.
Apple's new "Pro" definition seems more like "Prosumer".
> Serviceable, repairable, upgradable Macs are officially a thing of the past.
Well, not exactly. Apple’s desktop Macs actually all have modular SSD storage, and third parties sell upgrade kits. And it’s not like Thunderbolt is a slouch as far as expandability.
I can see why the Mac Pro is gone. Yeah, it has PCIe slots…that I don’t really think anyone is using. It’s not like you can drop an RTX 5090 in there.
The latest Mac Pro didn’t have upgradable memory so it wasn’t much different than a Mac Studio with a bunch of empty space inside.
The Mac Studio is very obviously a better buy for someone looking for a system like that. It’s just hard to imagine who the Mac Pro is for at its pricing and size.
I think what happened is that the Studio totally cannibalized Mac Pro sales.
Not surprising, as the market has broadly moved on from add-in cards in favor of smaller form factors and external devices, absent some notable holdouts in specific verticals.
Gonna miss it, though. If they had reduced the add-in card slots to something more reasonable, lowered the entry price, and given us multi-socket options for the CPU (2x M# Ultras? 4x?), it could have been an interesting HPC or server box - though they’ve long since moved away from that in software land, so that was always but a fantasy.
At least the Mac Studio and Minis are cute little boxes.
I have three of the trash can ones. They are absolute pieces of art, as useless as they are computationally these days (energy-to-performance wise at least). I will never sell nor give them away.
> It has gone without an update since then, languishing at its $6,999 price point
What I find fascinating is how people pay so much for Apple-related products. Perhaps the quality requires a premium (I don't share that opinion, but for the sake of thinking, let's have it as an option here), but this seems more deliberate milking by Apple with such price tags. People must love being milked it seems.
I may be in the minority but I liked the cheese grater, it was a machine I could upgrade and use as a powerful workstation. The trashcan really turned me off of the Mac Pro series. I think Apple really missed an opportunity here, but hope Springs eternal.
It totally makes sense. Mac Studio basically ate the Mac Pro's lunch. But it's still kind of sad. The Mac Pro used to represent this idea that Apple cared about the absolute high-end, no-compromises workstation crowd
I never understood the point of the Mac Pro for the last decade or so - especially after the Mac Studio was released, Apple should have worked out what professionals actually want - basically a Mac Studio but with three or four PCIe slots and a few SSD slots. That’s literally all it should be!
The Mac Pro was at the same time bizarrely over the top while also weirdly limited in some ways - while also being way to expensive…
Reading comments, I don’t think people are being completely fair here. For Intel and AMD to approach what Apple has accomplished they’re making many of the same compromises with Panther Lake and Ryzen AI Max. Apple chose to put disk controllers on their SoP rather than having them on the storage module. This shaves a tiny bit of latency. Worth it? No idea. I’m shit at hardware design.
As for not having a Pro or otherwise expandable system? It’s shit. They make several variations of their chips, and I don’t think it would hurt them to make an SoP for a socket, put a giant cooling system in it, and give it 10 or 12 PCIe slots. As for what would go in those slots? Make this beast rack mountable and people would toss better network cards, sound/video output or capture, storage controllers, and all kinds of other things in there. A key here would be to not charge so much just because they can. Make the price reasonable.
I am incredibly saddened that the inevitable finally happened. The OG 5,1 cheese grater sparked so much joy. I added and expanded so much over the years before I finally donated it to a computer museum and moved on to Apple Silicon. I did everything from scientific computing, ripping movies, serving files, running websites, and everything in between.
With the popularity of mac mini (and macbooks for that matter) for doing ML/AI work, I would have thought Apple could make a Mac Pro that could make for a good workstation for doing in-house ML/AI stuff.
I bought a GPU maybe a decade ago for this, and it's not worth the hassle (for me at least), but a nice out-of-the box solution, I would pay for.
Sad. I had this pipe dream of an Apple Silicon system made as a PCIe endpoint, so a Mac Pro could be a coordinator and host to like 4 of such systems in a cluster with very fast interconnect. Imagine the possibilities.
I guess A/V pros are used to getting screwed constantly, but it must be really irritating to face the prospect of eventually having to move PCI add-in cards to TB5 enclosures that cost $1000 per slot.
This makes sense, for that kind of money you could always build a beastly workstation in a real ATX case with standard components. Install Linux and the Mac looks like an expensive toy in comparison.
A Mac Pro without external GPU support was always a dumb idea. They just made this to shut up the hard core fans who were complaining about the outdated Mac Pro in 2018.
This honestly saddens me a little. From the PowerMac's to the MacPro I always loved them when having the opportunity to work with them. Plus I loved the expandability they offered.
I don't find the external GPU houses for Mac Studio as appealing to use.
If I remember correctly, the maximum configuration was something like $35k back in the day. I wonder what those people feel like now. On the other hand, if they have $35k to burn, probably they don't even think about it.
650 comments
Apple really stumbled into making the perfect hardware for home inference machines. Does any hardware company come close to Apple in terms of unified memory and single machines for high throughput inference workloads? Or even any DIY build?
When it comes to the previous “pro workloads,” like video rendering or software compilation, you’ve always been able to build a PC that outperforms any Apple machine at the same price point. But inference is unique because its performance scales with high memory throughput, and you can’t assemble that by wiring together off the shelf parts in a consumer form factor.
It’s simply not possible to DIY a homelab inference server better than the M3+ for inference workloads, at anywhere close to its price point.
They are perfectly positioned to capitalize on the next few years of model architecture developments. No wonder they haven’t bothered working on their own foundation models… they can let the rest of the industry do their work for them, and by the time their Gemini licensing deal expires, they’ll have their pick of the best models to embed with their hardware.
> But inference is unique because its performance scales with high memory throughput, and you can’t assemble that by wiring together off the shelf parts in a consumer form factor.
Nvidia outperforms Mac significantly on diffusion inference and many other forms. It’s not as simple as the current Mac chips are entirely better for this.
The cheapest Apple desktop with 128GB of memory shows up as costing $3499 for me, which isn't very "enthusiast-compatible", it's about 3x the minimum salary in my country!
I never liked apple hardware, but they are now untouchable since their shift to own sillicon for home hardware.
https://www.jeffgeerling.com/blog/2025/15-tb-vram-on-mac-stu...
> Apple really stumbled into making the perfect hardware for home inference machines
For LLMs. For inference with other kinds of models where the amount of compute needed relative to the amount of data transfer needed is higher, Apple is less ideal and systems worh lower memory bandwidth but more FLOPS shine. And if things like Google’s TurboQuant work out for efficient kv-cache quantization, Apple could lose a lot of that edge for LLM inference, too, since that would reduce the amount of data shuffling relative to compute for LLM inference.
https://marketplace.nvidia.com/en-us/enterprise/personal-ai-...
Maybe you spend 1000$ more for a PC of comparable performance, well tomorrow you need more power, change or add another GPU, add more RAM, add another SSD. A workstation you can keep upgrade it for years, adding a small cost for an upgrade in performance.
An Apple machine is basically throw away: no component inside can be upgraded, you need more RAM? Throw it away and buy a new one. You want a new GPU technology? You have to change the whole thing. And if something inside breaks? You of course throw away the whole computer since everything is soldered on the mainboard.
There is then the software issue, with Apple devices you are forced to use macOS that kind of sucks, especially for a server usage. True nowadays you can install Linux on it, but the GPU it's not that well supported, thus you loose all the benefits. You have to stuck with an OS that sucks, while in the PC market you have plenty of OS choices, Windows, a million of Linux distributions, etc. If I need a workstation to train LLM why do I care about a OS with a GUI? It's only a waste of resources, I just need a thing that runs Linux and I can SSH into it. Also I don't get the benefit of using containers, Docker, etc.
Mac suck even hardware side form a server point of view, for example it's not possible to rack mount them, it's not possible to have redundant PSU, key don't offer remote KVM capability, etc.
> ...making the perfect hardware for home inference machines.
I really don't get why anybody would want that. What's the use case there?
If someone doesn't care about privacy, they can use for-profit services because they are basically losing money, trying to corner the market.
If they care about privacy, they can rent cloud instances in order to setup, run, close and it will be both cheaper, faster (if they can afford it) but also with no upfront cost per project. This can be done with a lot of scaffolding, e.g. Mistral, HuggingFace, or not, e.g. AWS/Azure/GoogleCloud, etc. The point being that you do NOT purchase the GPU or even dedicated hardware, e.g. Google TPU, but rather rent for what you actually need and when the next gen is up, you're not stuck with "old" gen.
So... what use case if left, somebody who is both technical, very privacy conscious AND want to do so offline despite have 5G or satellite connectivity pretty much anywhere?
I honestly don't get who that's for (and I did try a dozens of local models, so I'm actually curious).
PS: FWIW https://pricepertoken.com might help but not sure it shows the infrastructure each rely on to compare. If you have a better link please share back.
But yeah, right now Apple actually has price <-> performance captured a lot of you’re buying a new computer just in general.
I can live without the RAM for a couple of months to get a good price for it, especially since Apple don’t sell that model (with the RAM) any more.
> I bet there’s gonna be a banger of a Mac Studio announced in June. Apple really stumbled into making the perfect hardware for home inference machines.
This I'm not actually as sure about. The current Studio offerings have done away with the 512GB memory option. I understand the RAM situation, but they didn't change pricing they just discontinued it. So I'm curious to see what the next Studio is like. I'd almost love to see a Studio with even one PCI slot, make it a bit taller, have a slide out cover...
Your point would have been largely correct in the first half of 2025.
Now, you're going to have a much better experience with a couple of Nvidia GPUs.
This is because of two reasons - the reasoning models require a pretty high number of tokens per second to do anything useful. And we are seeing small quantized and distilled reasoning models working almost as well as the ones needing terabytes of memory.
Is the Mac Studio great? Yeah, for Apple users.
That's a pretty good deal I would think
https://frame.work/de/de/products/desktop-diy-amd-aimax300/c...
At best we probably get a chassis to awkwardly daisy chain a bunch of Mac Studios together
Seem odd that a computer from a decade ago could have more than a 1TB of incremental RAM vs what we can buy today from Apple.
> home inference machines.
The market for this use case is tiny
If the OpenAI domino falls, and I'd be happy to admit if I'm wrong, we're going to see a near catastrophic drop in prices for RAM and demand by the hyperscalers to well... scale. That massive drop will be completely and utterly OpenAI's fault for attempting to bite off more than it can chew. In order to shore up demand, we'll see NVidia and AMD start selling directly to consumers. We, developers, are consumers and drive demand at the enterprises we work for based on what keeps us both engaged and productive... the end result being: the ol' profit flywheel spinning.
Both NVidia and AMD are capable of building GPUs that absolutely wreck Apple's best. A huge reason for this is Apple needs unified memory to keep their money maker (laptops) profitable and performant; and while, it helps their profitability it also forces them into less performant solutions. If NVidia dropped a 128GB GPU with GDDR7 at $4k-- absolutely no one would be looking for a Mac for inference. My 5090 is unbelievably fast at inference even if it can't load gigantic models, and quite frankly the 6-bit quantized versions of Qwen 3.5 are fantastic, but if it could load larger open weight models I wouldn't even bother checking Apple's pricing page.
tldr; competition is as stiff as it is vicious-- Apple's "lead" in inference is only because NVidia and AMD are raking in cash selling to hyperscalers. If that cash cow goes tits up, there's no reason to assume NVidia and AMD won't definitively pull the the rug out from Apple.
For me, aesthetics and size are important. That workstation on your desk should justify its presence, not just exist as some hulking box.
When Apple released the Mac Studio, it made perfect sense from a form-factor point-of-view. The internal expansion slots in the M2 Mac Pro didn't make any sense. It was like a bag of potato chips - mostly air. And far too big and ugly to be part of my work area! I'm surprised that Apple didn't discontinue it sooner.
Here's a good video how it looks like: https://www.youtube.com/watch?v=kIQINCWMd6I&list=PLi2i2YhL6o... (at 1:40 Neil Parfitt shows Mac audio setup his before and after).
It really could have been a bigger market for them than even the iPhone.
G5 was the thing. And companies were buying G5 and other macs like that all the time, because you were able to actually extend it with video cards and some special equipment.
But now we have M chips. You don't need video for M chips. You kinda do, but truthfully, it's cheaper to buy a beefier Mac than to install a video card.
Pro was a great thing for designers and video editors, those freaks who need to color-calibrate monitors. And right now even mini works just fine for that.
And as for extensions - gone are the days of PCIe. Audio cards and other specialized equipment works and lives just fine on USB-C and Thunderbolt.
I remember how many months I've spent trying to make Creative Labs Sound Blaster to work on my 486 computer. At that time you had to have a card to extend your system. Right now I'm using Scarlett 2i2 from Focusrite. It works over USB-C with my iPhone, iPad and Mac. DJIs mics work just as good.
Damn, you can buy Oscilloscope that works over USB-C or network.
It's not the Mac's or Apple's fault. We are actually live in the age where systems are quite independent and do not require direct installations.
The M5 generation Pro and Max chips have moved to a chiplet based architecture, with all the CPU cores on one chiplet, and all the GPU cores on another.
https://www.wikipedia.org/wiki/Apple_M5
So what will the M5 Ultra look like?
If you integrate two CPU chiplets and two GPU chiplets, you're looking at 36 CPU cores, 80 GPU cores, and 1228 GB/s of memory bandwidth.
Still, there are a few things which could be improved relative to the current Studio. First, the ability to easily clean the internals from dust. You should be able to just lift the lid and clean the computer. Also, it would be great to have one Mac which you could just plug in a bunch of NVMe disks.
On the other side, they might replace the Mac Pro with a rack mountable machine as the demand for ARM servers in the cloud raises.
• Multiple hard drive bays for easy swapping of disks, with a side panel that the user could open and close
• Expandable RAM
• Lots of ports, including audio
• The tower took up no desktop space
• It was relatively affordable, starting at $2500. Many software developers had one. (The 2019 and later Mac Pros were insanely expensive, starting at $6000.)
The Mac Studio is affordable, but it lacks those other features. It has more ports than other Macs but fewer in number and kind than the old Mac Pro, because the Mac Studio is a pointlessly small desktop instead of floor tower.
This Mac Pro was about resetting and giving a clear signal that Apple was willing to invest in the Mac far more than it was about ‘slots’.
Today, Mac hardware is the best it has ever been, and no one is reasonably questioning apple’s commitment to a Mac hardware.
So it makes sense for the Mac Pro to make a graceful exit.
Apple's new "Pro" definition seems more like "Prosumer".
> Serviceable, repairable, upgradable Macs are officially a thing of the past.
Well, not exactly. Apple’s desktop Macs actually all have modular SSD storage, and third parties sell upgrade kits. And it’s not like Thunderbolt is a slouch as far as expandability.
I can see why the Mac Pro is gone. Yeah, it has PCIe slots…that I don’t really think anyone is using. It’s not like you can drop an RTX 5090 in there.
The latest Mac Pro didn’t have upgradable memory so it wasn’t much different than a Mac Studio with a bunch of empty space inside.
The Mac Studio is very obviously a better buy for someone looking for a system like that. It’s just hard to imagine who the Mac Pro is for at its pricing and size.
I think what happened is that the Studio totally cannibalized Mac Pro sales.
Gonna miss it, though. If they had reduced the add-in card slots to something more reasonable, lowered the entry price, and given us multi-socket options for the CPU (2x M# Ultras? 4x?), it could have been an interesting HPC or server box - though they’ve long since moved away from that in software land, so that was always but a fantasy.
At least the Mac Studio and Minis are cute little boxes.
> It has gone without an update since then, languishing at its $6,999 price point
What I find fascinating is how people pay so much for Apple-related products. Perhaps the quality requires a premium (I don't share that opinion, but for the sake of thinking, let's have it as an option here), but this seems more deliberate milking by Apple with such price tags. People must love being milked it seems.
https://www.macrumors.com/2026/03/26/mac-pro-wheels-kit-disc...
The Mac Pro was at the same time bizarrely over the top while also weirdly limited in some ways - while also being way to expensive…
As for not having a Pro or otherwise expandable system? It’s shit. They make several variations of their chips, and I don’t think it would hurt them to make an SoP for a socket, put a giant cooling system in it, and give it 10 or 12 PCIe slots. As for what would go in those slots? Make this beast rack mountable and people would toss better network cards, sound/video output or capture, storage controllers, and all kinds of other things in there. A key here would be to not charge so much just because they can. Make the price reasonable.
Would be a killer local AI setup...for $40k.
I bought a GPU maybe a decade ago for this, and it's not worth the hassle (for me at least), but a nice out-of-the box solution, I would pay for.
I don't find the external GPU houses for Mac Studio as appealing to use.
Believe t-shirts at WWDC were not enough.
Thus the workstation market joins OS X Server.