I was already wondering about the "return to base (finding nothing to bomb)" thing. Bases seem eminently vulnerable to not being in a fit state to return to, in a showdown such as this.
Would it be too complicated* to potentially separate bases from cities. Initially, bombers are presumed to come from cities, though you can build bases away from the cities (at some cost, maybe 2 turns to become operational, maybe decreased chances of targetting the first turn?). An attacker gets a list of bases as well as cities to target. Consider them to be air-bases, in the following, but the resources assigned to them could be not actually planes at all (rocket base, defence base... I was wondering if you need home-ports for the naval elements, which could be cities or dedicated/composite naval bases..? Naw, too far, and little useful added.)
Cities, destroyed, impact buying power and any military power (offensive/defensive) still based there. Cities destroyed that deploy bombers also force the on-mission bombers to return to another (random, unless pre-arranged with such a contingency/fall-back plan) surviving bomber-supporting place.
Bases, destroyed, only lose the (non-deployed) military assets there. But the various defined defences are more effective (less need to worry about masses of squishy civilians) and there are some hardened elements that can bring them back up from destroyed (skip a turn, less cost than new base, e.g.) and perhaps recover some 'lost' weapons/capabilities.
You lose if you have no cities left, but you can assume that each city has an integrated base so losing separated bases is a weakening blow (militarily) but not a crippling one (financially, in terms of still-spendable income, and ultimately by dying your final death).
I think there's a bit of balancing to be done there, and it'd be a biggish change (alongside other balancing that would interact with every other bit).
I'm also wondering about the squishy-civvies. Aside from the loss-of-cities bit, they seem missing from the equation (changing ease/difficulty of recruitment, not directly linked to cost/presumed revenues), but let's hold off on all those. It'd be an even biggisher change.
* Yes, it probably would be.