Apple confirms the Mac Pro is over

As the article states, the writing was on the wall. I suppose they released an M2 Ultra variant to confirm it, but without an Extreme chip variant which would’ve been probably too expensive to make, the Mac Pro just didn’t make a lot of sense. It’s use cases over the Studio, while technically not nothing, were extremely narrow.
 
I used to own one. If I remember correctly, a 2008 model. It was clearly more than what I needed, but I was intrigued by what its potential was, especially with PCI slots available for different applications. When Apple introduced a rack mount on slides horizontally oriented Mac Pro in 2019, that opened up some interesting possibilities for different aerospace/defense and industrial applications. But at that point in time I pretty much lost interest.
 
I've never owned one but it's easy to understand why they're doing away with it. Looks like the Studio is the future, they're excellent machines.
 
I think it's a mistake for them to discontinue Mac Pro, but it's the culmination of a lot of mistakes. Apple never seemed to fully understand all the missteps they made, and after so many years of this they've probably driven away almost all their original tower Mac customer base, so I suppose it was inevitable.

I'm thinking about the progression from iMac Pro to 2019 Mac Pro in particular. Remember how they did a big mea culpa press event admitting the 2013 trashcan was a mistake, and about a year later shipped the 2017 iMac Pro, positioning it as an interim thing they could get out the door while working on the real Mac Pro update? While the 2013 MP was the original sin, that comeback plan also sucked. It challenged people who'd already been waiting for years for Apple to pull its head out of the ground to wait even longer, and those who were patient enough to stick around were presented with an overpriced overdesigned thing that was really lacking in some areas (e.g. out of the box internal storage expansion).

All those customers really wanted was a boxy Mac with a lot of internal expansion, not a halo prestige jewel. If they'd chosen to, I'm certain Apple could've delivered a non-blingy straightforward Xeon-powered tower Mac on the same schedule as the iMac Pro, and then could've iterated on it a couple times before 2019. This would've demonstrated to workstation customers that Apple actually had come around to understanding their needs, and was prioritizing giving them the essentials as soon as possible. Instead they got more confirmation that Apple didn't really understand how to design for them any more.
 
I love the Mac Pro industrial design, such a cool design and engineering solution. But Apple silicon was made for a different era, one of the future, not the past. We are done with 1.5 KW power supplies that take up half the desk. A better way forward is M chips. Mac Pro was always a stop gap to the inevitable, which the Mac Pro 2013 was: a small, light, expansive Thunderbolt based Mac. Mac Pro was always a very low volume, specialist, boutique solution in the Mac line up. Basically it's a mistake to call this a mistake.
Simplicity and focus are the values of Apple. Not to say "steve would have done this," but Steve would have done this. Apple silicon is the future and the Mac Pro is built for an older era of engineering. I love it, I love the design and care of it, but it's time to embrace future design and engineering.
 
General purpose computer platforms need to be inclusive of a wide variety of users and use cases, or they're on the road to not being general purpose anymore. Apple has now cut the Mac platform off from all users and developers and applications dependent on truly high performance IO, and that's not good.

Sure, that market's smaller than it used to be and may not represent a huge amount of revenue for Apple, but it didn't have to be expensive for Apple to keep the Mac relevant here. All Apple ever had to do was continuously make boxy Macs with lots of expansion slots and storage and IO and sell them for a reasonable price. Wouldn't have cost much engineering time, and it would have been money well spent. Had they done this instead of replacing the entire product line with the misguided moonshot attempt that was the 2013 MP, we wouldn't be having this conversation today - I bet Mac Pro would still be a lucrative and influential product, and they wouldn't have decided to put it out of its misery.

Deciding that it's fine is spin. The truth is something like "They carefully aimed the gun at their own foot, pulled the trigger, let gangrene set in before going to the doctor, tried desperate measures, but eventually had to have the whole leg amputated" and you're making that into "They've chosen to go monoleg and not even bother with a prosthesis and that's a great thing, actually!".
 
On topic: I think everyone who was going to dump Apple because of no tower has already dumped them by now. I also think that, given changes in technology in general, and given the improvements to the Studio and i/o, and given that there aren’t a lot of add-in cards that actually work in the Pro anyway, the remaining market for Mac towers is incredibly small.

That said, I think it would behoove Apple to make a real data center product. Something designed for racks mount, with processors appropriately tuned therefor. There are hacks to use minis and studios in racks, but bringing back xserve at this point would likely make Apple a fortune, given the AI boom going on, and would possibly result in more businesses switching to mac laptops/desktops too.
 
Last edited:
Simplicity and focus are the values of Apple. Not to say "steve would have done this," but Steve would have done this.
Rumor is someone showed him a prototype of the black thing and it was just too much for him to survive that insanity.
 
That said, I think it would behoove Apple to make a real data center product. Something designed for racks mount, with processors appropriately tuned therefor. There are hacks to use minis and studios in racks, but bringing back xserve at this point would likely make Apple a fortune, given the AI boom going on, and would possibly result in more businesses switching to mac laptops/desktops too.

I think Apple should bring back the 5U rack mount Mac Pro, but with an appropriate AS processor for management and communications. It had seven PCIe slots and a 1.4 kW power supply (should be enough to power a lot of AI).

As an aside, years ago I was contemplating developing signal processing products based on the rack mount MacPro.
 
I think Apple should bring back the 5U rack mount Mac Pro, but with an appropriate AS processor for management and communications. It had seven PCIe slots and a 1.4 kW power supply (should be enough to power a lot of AI).

As an aside, years ago I was contemplating developing signal processing products based on the rack mount MacPro.
what would you put in the PCI slots, though? Seems to me that times have changed and what you need is a lot of RAM and a lot of I/O to talk to other servers, and that’s about it.
 
I've never owned one but it's easy to understand why they're doing away with it. Looks like the Studio is the future, they're excellent machines.
Yeah i think for those who need more power than a studio, they’re in the market of “i need as much power as i can muster, at which point the workload runs in a server rack in an air conditioned room over the network. Whether its in their building, or elsewhere.
 
I think Apple should bring back the 5U rack mount Mac Pro, but with an appropriate AS processor for management and communications. It had seven PCIe slots and a 1.4 kW power supply (should be enough to power a lot of AI).

As an aside, years ago I was contemplating developing signal processing products based on the rack mount MacPro.

If you’re at that point - macOS and apple hardware are irrelevant; what you need is a cluster of nvidia hardware with liquid cooling, etc. You’ll be running Linux with whatever LLM host on it.

Something like this


Apple hardware is great for mobile, and desktop AI, but once you’re in the big leagues there’s no replacement for AMD or Nvidia at the moment - and its really not Apple’s core market to go chasing. They don’t need to, they’ll just partner with Nvidia, AMD, Intel or all three and win on the high margin user facing stuff while those three battle it out for minimum $/token.

Running a remote web UI to a chatbot or using an LLM via an API, there’s zero difference to the end user whether its on apple hardware or anything else.
 
Last edited:
what would you put in the PCI slots, though? Seems to me that times have changed and what you need is a lot of RAM and a lot of I/O to talk to other servers, and that’s about it.

If you’re at that point - macOS and apple hardware are irrelevant; what you need is a cluster of nvidia hardware with liquid cooling, etc. You’ll be running Linux with whatever LLM host on it.

Thanx for the heads-up. I know zip about AI server design.
 
Rumor is someone showed him a prototype of the black thing and it was just too much for him to survive that insanity.

Yet, the path to the Studio was paved with the corpse of the Cube.

I think the big difference here is that modern Apple is pursuing a level of system security that tends to exclude the sort of extensibility that a tower really wants. Back then, a tower was less in conflict with the system design goals.

what would you put in the PCI slots, though? Seems to me that times have changed and what you need is a lot of RAM and a lot of I/O to talk to other servers, and that’s about it.

Hard agree. Even outside the rack, a lot of the things you'd spend expansion slots on over the years have commoditized and been baked into the motherboard or even moved into software. I have a B&W G3 next to me with the slots all filled. Video, SATA, OrangePC, USB2 + FW Combo. My gaming rig has TB, USB3, SATA, and NVMe all baked into the board, and this has been the norm (minus TB) for what, almost a decade? The only PCIe cards I've bought in that time have been I/O cards and GPUs. Some folks might need M.2 carrier cards, but I'm not one of them. Audio and Video interfaces have been pushed to external devices, thanks in part to how capable laptops have gotten. All this erodes the market for towers.

But if you want a tower, you want a tower. /shrug

I mean, I'd love a rack-friendly Mac setup, but for now I'm just 3D printing rack mounts for the Mac Mini at home.
 
Hard agree. Even outside the rack, a lot of the things you'd spend expansion slots on over the years have commoditized and been baked into the motherboard or even moved into software. I have a B&W G3 next to me with the slots all filled. Video, SATA, OrangePC, USB2 + FW Combo. My gaming rig has TB, USB3, SATA, and NVMe all baked into the board, and this has been the norm (minus TB) for what, almost a decade? The only PCIe cards I've bought in that time have been I/O cards and GPUs. Some folks might need M.2 carrier cards, but I'm not one of them. Audio and Video interfaces have been pushed to external devices, thanks in part to how capable laptops have gotten. All this erodes the market for towers.
I loved that computer! Probably the only one I ever used a Dremel tool on (don't ask), LOL.
 
Could not agree more with the comments here. With modern fast interconnects like TB5 and TB4 I really don't see the need for PCIe ports.

I'd be very grateful if Apple was to make available modular upgradeable storage and memory for the studio, but that's unlikely as they'd lose the efficiency and performance metric benefits that are physically impossible with a modular design. But a guy can wish!

In all honesty the last 'Pro' Mac that I owned was the G5 dual from 2004. I had some I/O cards for that , but most have since moved to external Thunderbolt. I don't know about the rest of you, but I always lusted after the G4 cube from its initial unveiling. I could never justify the cost, but was delighted to see apple go the cube-ish route with the studio!
 
Could not agree more with the comments here. With modern fast interconnects like TB5 and TB4 I really don't see the need for PCIe ports.

It's not so much the ports. PCIe *boards* are great for special applications, whether purchased off the shelf, or designed from scratch. Music production, digital signal processing, high speed data acquisition via on-board A/D converters (properly shielded and on a separate power supply), digital filtering and processing, DDCs (digital down converters) for tuning and demodulating multiple RF signals/spectrums, DUCs (digital up converters - translating baseband data to RF) for signal simulation or communications, multipath simulators, high speed correlators, and on and on. A 1.4kW power supply was a great feature on the rack mount MacPro.

Apple's rack mount version of the MacPro was ideal for the above purposes.
 
It's not so much the ports. PCIe *boards* are great for special applications, whether purchased off the shelf, or designed from scratch. Music production, digital signal processing, high speed data acquisition via on-board A/D converters (properly shielded and on a separate power supply), digital filtering and processing, DDCs (digital down converters) for tuning and demodulating multiple RF signals/spectrums, DUCs (digital up converters - translating baseband data to RF) for signal simulation or communications, multipath simulators, high speed correlators, and on and on. A 1.4kW power supply was a great feature on the rack mount MacPro.

Apple's rack mount version of the MacPro was ideal for the above purposes.
All that can be done with Raspberry Pi or Arduino type boards and USB-C-connected back into PCs and Macs. The biggest demand for I/O bandwidth are desktop graphics cards and Thunderbolt 5 can give high performance even there, though short of maximum.

Everything on your list can communicate via USB 4, most with USB 3.2.
 
Back
Top