Yep. And you have to start asking why is the base ISO 1.3 stops higher than the R3?
I did a quick search, while specifically regarding the A9 III, I didn't find that, but Global Shutters in general, it seems that the "solution" allowing a Global Shutter isn't that the data is all processed instantly (if you think about it, 24 million readings/calculations/adjustments truly instantly does not seem likely). The "solution" is to have on-chip and even in-pixel memory. So the data from 24 MP is instantly moved to memory which is they read sequentially by the processor. If that memory is located adjacent to the pixel well, it is taking up surface area on the sensor, so you actually are using less than the full sensor surface area (apparently this has been the case with several sensors), smaller pixel wells (which might lead to a smaller ISO range?). The other option seems to be having the memory behind the pixel well, so the sensor surface area is unimpeded, but you have more distance between a heat sink at the back of the sensor and the pixel wells. Either way, by having on chip memory, you are adding a step to the process and each step increases noise.
Or so I read/watched.
In a way I am glad Sony did this: 1) push the envelope; but 2) mentally, I am now only really looking for faster read speeds.