Precisely (the first part of your post, I totally disagree with the second half which I cut), it is a consequence of the small nm / bleeding-edge nature of the CPU manufacturing process, which based on most responses in here I think almost no one has any clue how it works. The "design being done" doesn't mean a damn thing about how yield will be at TSMC.
The more complex the architecture (higher transistor count), the more difficult it is to get good yields. This is not a design challenge it is a physical consequence of how processors are manufactured, you will get defective parts, full stop, and that's how we get regular m1, M1 pro, M1 pro with more GPU cores, M1 Max, M1 Max with more GPU cores, and finally M1 Ultra. They literally can't launch the high end first unless they have been in production for quite a while unless TSMC's 3nm process has the best yields ever seen in modern fab history – and the opposite has been happening as the nm gets smaller and smaller.
It's not limited to Apple either, Intel starting with their CPUs circa ~2010-11 (complex Xeons always come late, sometimes YEARS late), AMD with Zen core (Threadripper always behind by 6-18 months, lately 24 months because the ~1 year later slot is being taken up by high-margin EPYC processors). It's the nature of how CPU manufacturing works now.
I believe the last Mainstream HEDT desktop that launched with bleeding-edge technology at the highest-end was in 2008 with the Intel Nehalem processor line, you can maybe make the case for Westmere in 2009/10 but that used the same chipset so there wasn't a whole massive re-engineering process going on. Since then they have *always* trailed by 6-24 months vs. a basic variant. Sometimes even longer in Intels case where they have canceled entire lines of HEDT CPUs.
Quick edit to keep it on topic: I actually hope they hold out and don't even release a new Mac Pro this year, I think they should start with M3 given the rumored raytracing capability that was cut from the M2 (which cores could possibly somewhat useful for Machine Learning). If they start making these M3s in high quantity in the late fall, we can get new Macbook Pros and then they'll have built enough stock of 'perfect' silicon to use in some M3 Ultra or Extreme in an M3 Mac Pro in early/mid 2024. IMO it is worth the wait because launching some $6,000+ desktop 6 months before you debut a vastly improved core seems crazy to me, but we'll see what happens.
I think the plan WAS for Apple to leapfrog the Mac Studio / Mac Pro, but it wouldn't shock me if we get an M2 Ultra Mac Studio and they save the M3 for the Mac Pro at this point given how long it's already taken. Another 6 months won't kill them, and it will give those very expensive machines a much longer lifespan. AND it will also allow them to build in the proper driver/architecture support for PCIe cards / maybe video cards / whatever add-in apple cards (if they exist) to the next OS which will release late this year. It all lines up better this way IMO, but We'll see what WWDC/this summer brings.
