The failure of GT6 (and GT5 to a certain extent) is a testament to how outdated the model is for "classic" GT games. I say this as a big fan of the GT series, the fact that none of current mainstream racing games are coming out with these big elaborate "career" modes is also a telling sign that the racing genre has shifted away this career-focused model. I would welcome career-focused games if AI was significantly improved to act more like humans (ei, making mistakes, taking dynamic lines and overtaking/defending at the appropriate times), however these improvements have never come about (in all sims, not just GT), probably due to the elevated resources it would take to program such a capable AI.
IMO the next GT game will feature the past (career mode similar to GT Sport), the present (online racing in GT Sport), and the future (improved physics engine, tire/collision model, larger races, FIA, etc.).
After being introduced recently to iRacing and playing it over the last couple weeks, I realized how big of a money-grab that game is (and poorly developed considering all of the parallel programs you need to run to get the same features as GT Sport), and how good of value GT Sport is to provide competitive online racing (with Daily and FIA races, a more "approachable" version of iRacing). The next GT7 game is very well positioned to disrupt iRacing IMO, and they've made great strides to try to emulating it in their first attempt on a console with GT Sport (obviously there's room for improvement but the game was solid for its first iteration). If GT7 can continue to concentrate on online racing, and improve the physics engine and improve the tire/collision model, add dynamic times of day and evolving track conditions, multi-class endurance races, etc., it really can reign supreme over other sims.