Warning, wall of text
Do you have any real numbers instead of "huge"? Can that community pay for the development of an mMP that suits them? I do not think so. Look at the pattern: everyone screams for user upgradable systems when in reality user want to avoid Apples upgrade costs. Part of Apple expensive BTO options may be to cover the whole platforms development costs. Using systems for >3-4 years (which make sense) also dilute the contribution to sustainable development.
Well I'm going to assume the people I know directly is a representative sample. The permanent employee is a declining phenomenon. Know what's easier than leasing hardware - paying freelancers based on the work they deliver and letting them deal with hardware.
But this idea that the user expandable MP has to be a distinct product paid for by just "people who want user-upgrades" is just nonsense. Apple is a single company, and products cross-subsidise each other all the time (the iMac Pro is going to be heavily subsidised by the scale the regular iMac gives to display panel prices, for example). Does anyone think the nMP was paid for purely by sales to people who wouldn't have bought a cheesegrater? Why apply that standard to people wanting a user-upgradable solution?
This didn't matter in the cheesegrater era, because for the non-upgrading customer, the cheesegrater was the same as the nMP - buy a standard config, plug it in and go, except that it didn't treat its internal components like consumables.
Contrary to this "is there a market for a user-upgradable system", it's the nMP that had no actual market - it was the only choice for a new non-AIO "pro" computer running macOS, so nothing about its sales are evidence that it was actually a product the market wanted, vs one they tolerated / endured. Put it this way, the cheesegrater brought people to the Mac, it was a good workstation within the world of workstations, so even if you ran Windows, it was an unambiguously good piece of kit. The nMP, both the machine, and the strategy it represented, caused a migration away from Apple hardware & macOS, it was a bad workstation within the world of workstations. Just like the iPad is a bad computer for educational customers, its fundamental "virtues" from Apple's perspective, were tied to greater liabilities in the eyes of the people who were the primary market (a million edge cases who need customisability over a machine's lifetime).
The amigos said it themselves - the nMP was
too focussed, to the extent that its inflexibility excluded users. Even if it hadn't been a failure-prone lemon, it was still a failed product idea. The MPs job is to be all things to all users - that's it's only significant role, the machine you buy when none of the focussed machines fit your need.
The cheesegrater did and could do everything the trashcan could do, with the exception of being put in carryon luggage, and when talking abut a vanishingly small numbers for a user case, it's the need for a small workstation one should point their attention towards. The next machine has to be able to hit 100% of users, not be good for some, but a compromise for others. The "nuts & gum" iMac Pro is the only machine Apple was intending to replace the nMP, THAT's the mac pro for the nMP mindset, so no matter what, nMP fans are getting the machine they were going to get, based on the design philosophy they seem to think is valid.
However, the non-upgradable appliance brings no user advantages
by virtue of being non-upgradable - literally, none. It wasn't made faster, more reliable, less chord spaghetti or cheaper. Literally all you got in return, was arguably less noise, but that was only achieved through a failed idea in the cooling design.
A workstation should be able to last a decade, high turnover on a machine like that is just ridiculous, especially from an environmental point of view that Apple likes to trumpet - the embodied energy lost in recycling / landfilling the old machine and manufacturing the new will always exceed the energy efficiency savings of the new, and the interconnect infrastructure between components that can be upgraded, simply doesn't progress at a speed that makes a meaningful impact on overall performance in those sort of timescales. Reuse is the first rule of sustainability, ahead of Recycle, for a reason.
As for avoiding Apples upgrade costs - here's the thing, that's unavoidable. Apple is losing the content creation market, formerly mac-only developers are jumping ship to hedge with windows versions, so their options are going to narrow to "make a machine that is upgradable for market rates" or "lose a sale entirely". macOS is an increasingly irrelevant differentiator.
Apple put a lot of time into VR at WWDC, does anyone honestly think VR developers are going to base themselves on a machine, immersed for hours a day, with second-rate graphics (which everything Apple sells has, and will have, compared to standalone GPU products), or in which they can't upgrade the GPU
every single time a vendor releases a new top of the range? Immersive environment quality, driven by GPU power, is going to be the new "I find macOS more comfortable to work in than Windows".
When talking about 3-4 year refresh cycles, GPUs are 8-12 months. We see this phenomenon in iOS development - developers rush to grab every device as soon as it's released, because they have to have it to make sure they're able to experience the cutting edge that their customers will use. VR will be no different - developers will want the same quality of experience their customers can get, and their customers can buy whatever new shiny team red or team green are selling as a standalone retail product.
The only product that makes no sense, frankly, is a VR development workstation, with non-upgradable GPUs.
