I'm really struggling to imagine there's more than a handful of users out there who want a rack mount machine that can also support a large number of direct-attach displays.
I don't think every Mac Pro user had to use every single one of those features. Some witll want more than others of those outlined. It is that the other Macs allow far fewer permutations of system configurations. The group that needs all of the list features is going to be smaller aggregate group that use a wider set of permutation. I also don't belief that maximum number of permutation gets to be a very large group , but it is bigger than than group that needs all of them at the same time.
A set of direct attached SDI monitors isn't going to work with just Thunderbolt ports.
It s also not necessarily rack mount only. In the past in order to make the Mac Pro as 'desktop beauty queen" Apple added stuff to make it rack hostile. That is more so drop the beauty queen stuff. The point is to get to a box that can be purposed either way. In part I'm saying 'rack focused' because low confidence Apple stop drifting into the beauty hole that doesn't 'buy' the Mac Pro anything in 2026. The folks left are going to buy it for utility ; not as a set prop.
It is probably the case that doing two chassis makes both options more expensive. If slot a rack system into a vertical case then it functionally complete as a 'tower'.
Also skeptical that there's any need at all for local spinning disks,
I wasn't pitching spinning disks. If there room for a J2-like backet and folks buy an add-in card to provision the connection , why block it? I not saying putting SATA connectors on the motherboard. I just saying don't block adding a card , bracket , and wires and setting it up. 1 or 2 drives for local Time Machine backup . Apple spent more than a decade encouraging folks to do that why block it?
1 or 2 PCI-e slots could be used for M.2 back up also. The general slots themselves provision both. So leave it user optional.
The whole hypermodularity is #1 priority for all components crowd is gone. Chasing away the 1-2 disk crowd is just adding fuel to the fire.
macOS stoftware wise there is still going to be thunderbolt docks/devices with 3.5 drives in them. The software to talk to a SATA drive via PCI-e is there because Thunderbolt provisions that through the rest of the line anyway. If it supported in software why block it?
especially in your hypothetical box that also has "better than 100GBE" networking.
Again not everyone buying a Mac Pro is going to buy that. But if you block it then nobody gets it.
If Apple contiues to do their own data center AI server thing then agoin there is probably going to be higher network speed drivers somewhat in the software stack that could be pulled over. ( unless crazy implementation specfic in the SoC. ) . But if doing the software work on those system why block it from the MacOS systems even if don't use exactly the same SoC. [ If Apple goes full outsource to Google/AWS/etc on that front then yeah this is a problem. ]
I have deep doubts Thunderbolt is going to get any faster. (Intel has bigger forest fires burning than that and USB-IF likely can't get consensus. ) And v5 doesn't do better than 100GbE. There is the asymetrical hack in v5 for display outbound so likely done there too.
Even at 10gbit networking you've completely removed any incentive to have slow local storage, ignoring the lack of large filesystem support in macOS.
At the enterprise level, yes. But I think there still are some reasonable number of folks with 'one man band" set ups that don't want a NAS or cloud drive storage. Apple internal drives only go up to 16TB. At Apple's SSD pricing per TB the more common configuration present is likely 4 or less. Don't have to get into exotic HDD drive territory to cover that with one backup image (e.g., with a 16-20TB drive). Yeah, there may be a network one also , but good policy is to have two.
It's hard for me to see much of any market for the machine you're describing here, much less view it as the "core Mac Pro market."
What I meant by core is a core aggregate group. I think the way the Mac Pro is oriented now is a "narrow subset of of large , high cost tolerant companies " that will buy these in chunks. If the 'targeted' customer all have large LAN networks of 10-100GbE and large budgets of even more than aggregate Mac Pro costs on SAN/NAS set ups. Structure IT departments that need 'roadmaps' . This highly indented group buys Mac Pro (Final Cut Pro and Logic) and that is the 'core'.
It is tough because Mac Pro is priced high. It isn't going to be a very large group relative to MacBook Air unit sales. The tipping point of just how big is big enough only Apple has the raw data.
If what folks are looking for is data center nodes scaled down to desktop level, then that is a dead end.
I think there is a substantive collection of smaller groups that are as easily to ID as a homogenous group. Just like Apple's Studio M3 Ultra Thunderbolt 5 4-node clusters don't really need a switch if point to point 100-200GbE nodes together in same size cluster. I think smaller groups will have more leverage going forward ( and getting to point where there will be 'rebound' from cloud everything. Just like local servers went too far away from centralized compute. It is a slow motion yo-yo that gets caught in hype trends. )
I think the Mac Pro helps Apple and macOS keep in touch with stuff that Apple is not doing (or at least not paying attention to). Linux/Unix/Windows 'discovered' RDMA about a decade ago. If Apple is primarily interested in building a bigger moat around macOS then doesn't need a Mac Pro . The Mac Pro has problems when software path is mostly moat building around Apple only software.
Thanks for sharing your thoughts, either way. Always interesting to hear what other people are dreaming up.
You are welcome. After years and years of reading the content of Mac Pro forum ... the Mac Pro is somewhat of a 'blind men grabbing different parts of the elephant" syndrome. Or at least it has been until about 2023-24.
