Are we just not going to talk about this claim? While I currently have more use for an Intel processor and thus this would be a potential benefit to me, it seems like a very odd move at this stage of the game.
No. It isn't very odd at all. Apple probably sells less than an 100K units of Mac Pro in a year. It is pretty unlikely Apple is going to design a chip with a less than 1M run rate.
The M-series design is primarily targeted at laptops and mobiles ( iPad Pro). That is the vast majority of what Apple sells. If Apple is completely dumping discrete GPUs from all laptops ( never were in iPad Pro) then they have a quite weak interest in composing large PCI-e I/O subsystems with huge bandwidth. Apple is going the best iGPU possible path ( Apple is about minimizing external high bandwidth .. SoC is a 'black hole' ; it is suppos to be pulling that inside the package) . So some high multiple x16 PCI-e 4 (or 5) lane bundles are probably not interesting to them at all.
Likewise with the memory system in their baseline design. LPDDR5 yeah sure... that aligned with their overwhelmingly laptop focus. DDR5 with module ECC ... nope. They have used it in the iMac Pro and Mac Pro because it was there in the basic Intel Xeon W class feature set. But are their laptops on the ECC module evolutionary path for the next 4-5 years ? Nope .... not even close. So Apple probably has very low motivation to persue that.
Apple is on the path of soldered on package RAM and iGPUs. They may go so some chiplets on package to get scale ( and keep about a 250-350mm cap on die size). Like the main iPhone and iPad offerings their primary strategy has been to make one SoC and use it in as many products as they can. That is a lower risk and higher profit path for them that they probably are not keen on getting off of. Making "Mac Pro" folks only buy Apple RAM probably has some folks inside of Apple salivating on the fat profit margins.
The large breadth of the Mac line up in terms of I/O bandwidth is going to make that "laptop baseline principles " hard to follow. The fully populated Mac Pro is far outlier from their laptop line up. The folks who at the upper end of the range they may just "write off" on M-series targeting for more than several years .
If Apple isn't using chiplets to get scale is an even bigger issue. Apple stays "ahead" by using bleeding edge process. That is much harder to do with large monolithic dies. If Apple has to wait until 5nm because a "much cheaper , bulks" process then could ( very probably 'will' given wafer/chip supply shortages ) take lots more time.
The upside of Intel is that they have their own fabs and available capacity. As a 3-4 year stop gap... it would make sense to do another Intel Mac Pro. If next gen Sapphire Ridge makes it out by early 202 into the Xeon W series then it wouldn't be all that strange.
Apple will do an I/O expansion chopped down "Mac Pro" and claim "Mac line up" transition victory.
Even more so if at WWDC 2021 that 3rd party GPU driver support in the ARM variant of macOS is still completely blocked. If Apple is engaged in some holy war against all 3rd party GPUs then that's even more credibility to do a substantive update of an Intel Mac Pro. [ If that opens up then yeah, another Intel Mac Pro gets more doubtful. Likewise if this new "half sized" M-series Mac Pro gets some slots. That would show Apple is willing to put some minimal effort into high end I/O in bulk. ]
There is a huge inertia of Intel Macs sold over the last 2-3 years. Those are going to require macOS Intel updates for many years. The average user is holding onto systems for longer amounts of time. ( Apple is still selling the non-Retina "edu iMac" with a quite old MBA class processor inside . Apple has some long term buy-and-squat customers. ) . So it isn't like they aren't going to be developing intel macOS updates anyway.
Apple could wait for 3nm ( or 2nm ). At some point they will probably stop adding high performance cores to keep Mac OS scheduler more tuned to the current data structure limitations ( < 64 processors ). At that point throwing in a chunk of PCI-e hub that may/may not use on the laptops will be much cheaper for them to do as a "throw away" on most systems the chip is used in (e.g., across imac , mini , mbp 16" , and Mac Pro ). If there are enough full sized Mac Pro who are still around then it would be a later move another design iteration substantively further down the road.