A couple variations on a central idea:
The central idea is a grand strategy game with a procedural world and equipment/visible technology . My biggest complaint about most GSG's is they are historical 'sims', so the scenarios are limited and the playing field is static. Technology doesn't need to be procedural, though it should be flexible (read: freaking complicated) and relatively unconstrained by 'invisible walls'. There would be no reason you couldn't try to discover and use gunpowder in your prehistoric nation-state, if you have access to the raw materials and a culture that allows systemic thought.
Freespace, for example, is a cool setting for this kind of thing. You generate a galaxy map and solar system maps and generate some fluffy faction traits (that still have some crunch) for your initial vasudan and human polities. You'd generate research based on what planets you hold (Ancient artifacts and records), what you build (practical experience), whether you capture enemy ships/supplies (reverse engineering), and deliberate scientific research. Research would both modify existing technology and unlock new ones. So, when you finished researching a Maxim or torpedo, you don't get +2 to anti-capship weapons; you get to start building and equipping maxims on your fighter wings. Expanding to other solar systems would carry a risk of insurgency/independence movements and finding Shivans. You would probably go minimal on micro-managing for this game, so you could focus on fleet movements and let the player spend lots of his time piloting/captaining/admiralling in battles and wars.
Or if space isn't your cup of tea, consider a planetary version. This is basically alpha centauri or civ with a different tech system and tactics beyond dice rolls. The civilization style of tech-tree is very focused on paradigm shifting technologies, with very little attention paid to refinements on existing technology. IMO, any kind of military strategy you can apply in those games is strategic, not tactical in scope. good AI is a big plus for this iteration; the AI needs to be good at research and flexible in expansion and warfare. The AI in civ5 flounders on water worlds and won't expand or war like a human would.
A separate idea:
More games in the style of ogre battle. I'm disappointed that FF:T became Square's go-to tactics game, after I played through ogre battle and bahamut lagoon. These styles of TRPG are basically unknown and they are quite fun, but show their ages quite badly. Hell, BL seems like it was probably dated at release. Presumably squeenix owns the rights to all of quest and square's IP, which includes stuff like secret of mana and ogre battle. The only thing squeenix has done with them is release March of the Black Queen on the wii shop. The only RPG franchises they do anything with anymore are FF and Dragon Quest.