To me it makes very little sense to have much of any sentient drones at all in a robot society as they really serve a very little purpose aside for the ones you need to run things (intelligent ones). Machines really don't need much of any sentient beings at all... just sapient central computers and then specialised robotic vehicles (non sentient) to do all the work. The amount of sentient computers would depend on how much machinery they could control.
I also often have a problem what a Hive mind really is and how it works... how does the communication work and how independent is each "individual" in the hive?!?
Are they more like ants... so no real hive "mind" just ants randomly doing things based on instinct and following different scents... then how do they span the galaxy?!?
I know... this is just sci-fi and it is cool over logic anyway... but still...
Different Hive Minds work in different ways, according to the headcanon of the gamer.
Some will be on a model where there's essentially one "sapient directing intelligence" per pop, but all are subservient to the Hive as a whole - somewhat like the tyranid synapse beasts of earlier editions of 40k (kill the synapse beast, and the non-sapient drones basically go wild until another synapse beast can gather them up and make them useful. Presumably these are still using the psychic connection thing.
Some will be on a model where there's one controller per planet, answering to the sector brain bug, answering to the Hive, with the connections being psychic shenanigans that reaches across space. However, the individual planets here could be functionally ant colonies.
Some will be "one brain, many bodies", with every body directly being part of the Overmind.
Some will be more like a direct democracy, with everything the Hive must do as a whole being decided by an instant consensus of all parts. All drones in this model would be sapient, but dependent on the Hive for existence (early Borg), and might even be able to cope with the concept of "I" referring to a single drone, even if "I" is always subservient to "we".
For robot empires, I've generally considered each POP unit to be the machines subservient to a particular mid-tier command node which has been specialised for a particular purposes (whether overseeing manufacturing, mining, energy production, etc.), with the majority of the POP being non-sapient workers controlled by that command node (which might even have a "name"!). The planet or sector "governor" is then a high-tier command node that directs the POP command nodes.