MP All Models Apple sleeping on the mac pro is comically bad timing

[Omega Mac]

I agree with the idea, but Apple needed to do more than just keep a refreshed/reheated “updated” version of the M2 Pro. I think they should have and could have done more with the platform to capture this part of the market.

Imagine a Pro with adequate PCIE lanes to run blade-style M-Ultras in each PCI slot.

Or go back further in time and resurrect the idea of the PCI RAM disk, allowing the pro to utilize terabytes high-speed swap for model and context buffering.

That’s the type of hardware that would’ve differentiated the Pro platform from the Studio platform.

Back in Jan (2026) the concept and discussion of a new approach for the Pro using Mac Blades came up here and might be of interest https://forums.macrumors.com/threads/whats-happening-with-the-mac-pro.2475418/post-34362251

 

[leman]

Back in Jan (2026) the concept and discussion of a new approach for the Pro using Mac Blades came up here and might be of interest https://forums.macrumors.com/threads/whats-happening-with-the-mac-pro.2475418/post-34362251

We’ve been discussing this concept here ever since M1 got announced. Unfortunately, this is not happening - data routing would make it too expensive for a consumer product. For their own ML server farms, Apple uses a much simpler concept, which essentially boils down to having a bunch of Macs on a shared fast network interface, only packaged more efficiently.

 

[theluggage]

No one has claimed that it would still be unified memory. So that’s a straw man argument on your part.

Seriously?
You literally "corrected" my statement that "GPU expansion is incompatible with unified memory."

 

[teh_hunterer]

There is as new wave of people setting up computers locally (both personal and business) for AI inference.

It is striking to me that right before this happens, Apple discontinues the Mac Pro which was perfectly suited for this. We go decades with the Mac Pro having sluggish sales and then the year that it will actually start selling like hotcakes, it's gone!

I guess Nvidia captures the market share with RTX Spark, that they announced a couple of days ago.

For the uninitiated, running a local LLM requires enough VRAM to fully fit it in memory. Models get huge very quickly. Because of the unified memory, this means you can fit much larger models on a mac instead of a traditional machine that has a separate video card (the video card has its own RAM and video cards with a lot of RAM are very, very expensive)

What does the Mac Pro have to do with any of this? The Mac Studio and Mac Pro had the same SoC and memory options.

 

[teh_hunterer]

the much maligned 7 year old Mac Pro can stuff 1.5TB of RAM and/or 4xAMD GPU in them if economically justified for their work. There simply isn't an option for anything close with Mac right now.

How is that going to help you with AI? The benefit of unified memory for AI is that it's accessible by the GPU.

 

[ThomasJL]

It’s so typical how on previous threads (not this thread) about the Mac Pro, Tim Cook supporters, who probably none of have ever been in work-related situations that needed the power of a Mac Pro, pathetically tried to justify the previous Mac Pro not having user-upgradable RAM and user-upgradable main storage, and now pathetically try to justify the discontinuation of the Mac Pro. To them, Cook can do no wrong.

 

[leman]

It’s so misinformed how on previous threads (not this thread) about the Mac Pro, Tim Cook supporters, who probably none of have ever been in work-related situations that needed the power of a Mac Pro, pathetically tried to justify the previous Mac Pro not having user-upgradable RAM and user-upgradable main storage, and now pathetically try to justify the discontinuation of the Mac Pro. To them, Cook can do no wrong.

Datecenter solutions used to train largest language models in the world don't have upgradeable RAM. It really helps understanding the tradeoffs and limitations of technology when discussing wishes.

 

[ignatius345]

The bean-counter approach to anything that isn't for sale at Walmart. MBAs like to keep things simple to manage. That doesn't mean that companies like Apple are not making a profit on things like MBPs. Just a little more complex to manage.

Apple is a gigantic corporation. They pay the most attention to the things that make the most money. It’s always been this way. There was a time when desktop computers were their bread and butter, but that’s changed.

 

[teh_hunterer]

It’s so typical how on previous threads (not this thread) about the Mac Pro, Tim Cook supporters, who probably none of have ever been in work-related situations that needed the power of a Mac Pro, pathetically tried to justify the previous Mac Pro not having user-upgradable RAM and user-upgradable main storage, and now pathetically try to justify the discontinuation of the Mac Pro. To them, Cook can do no wrong.

Do you have such an example of something you did that needed the "power" of a Mac Pro that the Mac Studio is not powerful enough to do?

Don't say AI, because the user upgradeable RAM in the older Mac Pros can't be addressed by the GPU and is thus useless for AI.

 

[kwikdeth]

So, slightly off-topic, but do you think the MP has a chance at life again with Ternus at the helm?

 

[teh_hunterer]

So, slightly off-topic, but do you think the MP has a chance at life again with Ternus at the helm?

No. What did the Mac Pro even provide?

* User upgradeable RAM - no longer a thing with Apple silicon, and largely pointless in a world where we are moving towards unified memory that both the CPU and GPU can utilise

* PCIe GPU - no longer a thing with Apple silicon. The world of PCI GPUs is going away anyway. The advantages of having the CPU and GPU in the same SoC with memory that both can utilise is the future. Even Nvidia is moving towards that. All high performance computing is moving towards that. It's undeniable.

* PCIe expansion - also basically pointless now. Audio cards are moving to Thunderbolt. "Afterburner" type encoding cards are pointless now that those encoders are built into the SoC. Anything that people needed to put inside their desktop on a PCIe slot can be done via thunderbolt expansion.

* Performance - every Mac Pro ever created is utterly destroyed by the current Mac Studio.

I genuinely think the main people who are rallying against the death of the Mac Pro now are people who are just attached to the desktop tower form factor because it is how they LARP as some sort of elite super interesting tech genius.

 

[Schmedly]

There is as new wave of people setting up computers locally (both personal and business) for AI inference.

It is striking to me that right before this happens, Apple discontinues the Mac Pro which was perfectly suited for this. We go decades with the Mac Pro having sluggish sales and then the year that it will actually start selling like hotcakes, it's gone!

I guess Nvidia captures the market share with RTX Spark, that they announced a couple of days ago.

For the uninitiated, running a local LLM requires enough VRAM to fully fit it in memory. Models get huge very quickly. Because of the unified memory, this means you can fit much larger models on a mac instead of a traditional machine that has a separate video card (the video card has its own RAM and video cards with a lot of RAM are very, very expensive)

The problem is twofold: Nvidia fracture and bandwidth. The Apple Ultra UMA has great bandwidth (key for LLM), but an extra card obviates this. Nvidia cards have great bandwidth internally, but there aren't any official Apple drivers for it. It was a cleaner path to just go with UMA rather than partner with AMD for competitive hardware.

 

[seek3r]

It’s so typical how on previous threads (not this thread) about the Mac Pro, Tim Cook supporters, who probably none of have ever been in work-related situations that needed the power of a Mac Pro, pathetically tried to justify the previous Mac Pro not having user-upgradable RAM and user-upgradable main storage, and now pathetically try to justify the discontinuation of the Mac Pro. To them, Cook can do no wrong.

I’ll be honest, I work on SaaS data platforms these days, have for a long time, and long ago mostly HPC before that, so big tech corp america and govt/uni science before then, and I’ve rarely seen anyone actually even try to get their workstation upgraded post-purchase at work

Usually the company just writes off the hardware at the next refresh cycle and you get a new machine.

As expensive as, say, a fully loaded studio (before the ram shortage made them unavailable) is, it’s a rounding error in the cost of employment of anyone using it, so it makes way more sense to frontload the cost and deprecate and replace the machine in 3-5 years.

Corporate buyers, which is the target for professional machines, dont care whether the ram is upgradable because it’s likely never going to get upgrades even if it is.

Hell, at work right now IT’s been trying to get me to upgrade off my M1 Pro MBP right now and *I’m* resisting because I’m holding out for next year’s rumored redesign. So I’m actually past the refresh cycle and IT sends me a polite nastygram about it periodically.