I agree that it's a bit of a problem to read those who are happy about the discontinuation of the Mac Pro when they don't own one.
Coming to a forum for enthusiasts of a certain machine, on the day the line gets cancelled, to tell them to stop overreacting, borders on trolling. Albeit the writing's been on the wall for some time.
I have a 7.1, and it's still one of the best computers ever made. Seeing this workstation line completely abandoned by the world's largest computer manufacturer is quite disheartening.
This is the thing. If Apple had just dribbled out a half-hearted effort, long-suffering MP fans would have read between the lines. But going all-out with a completely custom design, a new slot standard (MPX), a 1.5TB RAM ceiling and a doubling of the starting price, customers could be forgiven for believing Apple's public statements. When they said they had re-committed to the Mac Pro, there seemed little reason to doubt them. Especially following an almost unheard of (from Apple) admission of error regarding the previous model.
Given the previous model had come out in 2013 (2017 if you count the iMac Pro), there was pent up demand for a new workstation. Previous MP towers had also enjoyed exceptionally long lifespans (albeit probably not intentionally), so some invested heavily in what they considered a '10 year' machine.
Reading grand pronouncements used to justify all this is a bit strange.
You know, Apple could very well sell a whole range of products with Apple Silicon but also continue making professional workstations with Xeon or Threadripper, and that would be perfectly relevant.
From Apple's point of view, though, would it make any (real) money? They'd be selling in tiny numbers compared to e.g. HP and Lenovo. Unlike with Windows PCs, these machines would share nothing hardware-wise with the rest of the Apple ecosystem. Nvidia / AMD Windows GPU drivers are common to swathes of PCIe cards, from budget to workstation. This effort would need to be done just for a handful of Mac Pros. There's no iMacs or MBPs using them anymore.
Apple's CPU cores are industry-leading. They would only need a version that supports lots of PCIe lanes. But this would be pointless unless they also supported (Nvidia) PCIe GPUs. Which would be a major departure from the unified memory model used by every other Apple device.
Apple consider the Mx Ultra GPU good enough for most purposes. And if it isn't, they're OK with losing you to Windows / Linux. They gain more from not needing to serve that section of the Mac userbase.
What, maintaining Mac OS for x86 would be a waste of time for such a small market? Well, maybe not if Apple had truly supported this entire market and maintained Nvidia. Whether it's for AI or 3D, we need these kinds of workstations.
Even if Apple had maintained Nvidia support over the past 15 years, the above issue would still exist - driver development would be borne by a single niche model, not e.g. MBPs.
Although it's widely assumed that Apple fell out with Nvidia over some dodgy laptop GPUs in 2007 (of which I had one), I've always doubted this. AMD had similar issues, albeit they likely handled it better. I've always assumed Apple saw CUDA as a competitor to Metal, and felt that allowing it on their platform would result in cross-platform apps simply settling for the former, killing the latter.
The Mac Studio has many problems that I won't list, but to summarize, it's a disposable computer, too expensive in its high-end versions, and I prefer to invest in upgradeable PCs starting at a certain price point (above $5,000) rather than disposable machines that won't even be usable on the second-hand market because they're completely unrepairable.
I don't know how reliable Studio's are. I'd assume they do well, given their laptop chips are easy to cool with a comparatively massive heatsink, and there don't seem to be many reports of failure. Dust build up could be an issue, though, given they're not designed to be easily opened. Assuming they don't break, there should be a healthy second hand market, so they're not a dead loss when upgrading.
Yes, I also have some environmental concerns, and seeing a device completely dead because everything is soldered, proprietary, and repairing it requires a highly skilled technician and a lot of time for a result that's somewhat limited in terms of the operating system (if Linux isn't installable) is very frustrating.
Using proprietary flash blades is blatantly intended to steer purchasers towards expensive upgrades. Whatever the technical justification, given a choice, no one would choose them over being able to add / replace commodity M.2 blades themselves. A studio could easily accommodate, say, 4 slots.
