Quite simply the M1 is designed for the machines it was put into. That's why 3 machines were updated at the same time. The 3 lowest end Macs.
....
Actually not. The Mini wouldn't have to backslide on port allocation and display support if it wasn't an addition volume outlet for the others.
Apple primary approach with their processor design to date as been to use the same SoC as "hand me down" into several product. The mainstream iPad line up is entirely phone SoCs . Homepod .... really old phone processors. Apple TV really old iPad Pro processor. The Apple watch is now sharing efficiency cores design with iPhone SoC . etc. etc. (and all watches $300-3,000 all come with same SoC for a given generation. )
Apple isn't super customizing this to each every Mac product. They are probably working toward the minimal set they can get away with. Apple hasn't previously pushed the GPU envelope on the 21.5" class iMacs so pretty good chance whatever the MBP 16" gets will be dropped in an iMac 21-24" model also. ( along with a 4 port MBP 13" ).
Similarly the half sized Mac Pro rumors are indicative to Apple "walking back" the Mac Pro to something closer to an iMac 27" SoC. ( dumping most of the slots because major chucking of i/O bandwidth provisioning ).
On the whole the M-series is probably generally going to be closer coupled to laptop requirements than desktops ones. Apple will extend the performance so get some bump on the desktops eventually rolled out to but brings "power usage" savings to those system is likely going to be principle side effect.
I'm really tired of this constant moving of the bar and dooming about graphics. I'm sorry but graphics are not the untouchable, exclusive domain of 2 companies. Apple knows what they're doing and will deliver the graphics performance that each machine needs.
It wasn't two companies in the current Intel era. The larger GPU vendor in the current Mac space is Intel. Not the AMD vs Nvidia fan boy wars that commonly boil the forum threads here.
That is Apple's primary target. The dGPUs in the MBP 16" is next in line. ( as stated above that will probably have a happy side effect for Apple of covering the bottom half of iMac line up with a "good enough" solution also. ). Volume wise that is probably large double digit percentage of the dGPU component purchases that Apple does. An Unified Memory iGPU with bigger caches and lots more bandwidth could cover that with just incremental modifications.
But iMac 27 , iMac Pro , and Mac Pro. That's probably more so putting an iGPU common floor under all of those rather than replacing all of the dGPU options there. Replacing every single dGPU that Apple buys would be a dubious move for Apple. As they go up product line up , the volume goes way, way ,way down. [ The Mac Pro comes with a 580X or W5500 starting point. They'd only really need to just get within a reasonable distance of that to call it quits (provision a virtualized interface for native iPhone apps to run on) and just punt work on the rest of the range to third parties through the available slot(s). ]
Apple's super high priority on energy efficiency is probably going to keep them in the "Unified Memory" GPU zone. Which at some point in the super high , embarrassingly parallel computation space ... starts to loose traction. Just way more cores than going to get around a single centralized , limited size "System cache" just won't cut it.