For shooters it has become obvious that you use "game engines" like Unreal or Crytek. Why shouldn't there be a "sim engine" as well? Publishers could focus on making planes and maps, but don't have to make biggest task of all anymore, the sim itself.
You raise a good question, but in part, I think you already answered it as well.
While the idea of a global terrain rendering engine is certainly appealing, it represents a massive amount of work, for a middleman project (by that, I mean, your "customer" isn't you or me, the end-user, it's the middleman publisher like 777, or ED, or LM, etc.).
While it would be nice to think that many publishers would be interested in licensing such a product, the picture isn't quite that rosy, I think.
For starters, many shooters, as you say, have their own "fantasy" settings, and have no need for an "Earth Engine". Racing games? Only need a few specific track locations (so far). Flight Sims? There's some value here, but many/most of the combat ones feature a particular Theater of Operations (or perhaps a few). That leaves flight sims like FSX, XP, P3D etc. who, for the most part have already built their own. Naval simulations are also a possibility, but again, the various submarine games or games like Naval Action have focused on a much smaller theater historically.
So, if you're the Business Manager from the "Earth Engine" Project, I should think it would be a pretty nervous business plan wondering who you'd market your product to (although, there are, of course, other opportunities beyond gaming), to cover the costs of producing that engine.
One avenue that might be interesting is licensing and leveraging the various satellite mapping services (ie. Google Earth, etc.) to populate your terrain, but I'd bet that wouldn't come cheap either.
I think it would be awesome to have an "Earth Engine" featuring the entire globe of high quality terrain, to build your particular virtual playground in, but I'm not sure who would be willing to undertake the risk to build it. Perhaps someone like Orbx, who is already highly invested in photoreal scenery might consider taking that next step, but I suspect, that it's probably unlikely from a cost / benefit standpoint. Particularly, when they can already sell bite-sized areas, rather than the entire (more costly for all) globe.
Anyway, we're getting a bit of the OP's track here, so while it's an interesting sidetrack, it might be better in its own thread. Sorry for the thread-jack OP.