What people seem though to fail mentioning is that sales are also up. It a game costs twice as much but sells twice as many copies the overall profit is the same (not including whatever costs are involved with manufacture and distribution).
Actually if that was the case the profits, too, would double
Example (using bad number of illustrations purposes only; obviously publishers don't net the full retail cost and it costs more than dev fees to make a game--you have to manufacture, distribute, advertise, etc)
Game 1
Dev Cost: $50M
Sales: $60M (1M @ $60/game)
Profit: $10M
Game 2 -- 2x Dev Cost, 2x sales - 2x profits
Dev Cost: $100M
Sales: $120M
Profit: $20M
Per this upcoming generation's costs: Art is the expensive part and the good news is developers already do multi-million vertex models and bake in normal maps. If a studio has a good workflow this is not a huge issue. In fact a big benefit of engines that have fully dynamic lighting (including GI) and shadowing solutions can benefit from near instance iteration.
The big potential cost will come from gamers wanting more: more content, more variety, and bigger worlds.
It is going to cost oodles to have an open city where every building can be accessed and looks unique. The "obvious" solution is middleware that generates unique content based on base models or simple variables and going with modular base models but the realtiy is these tools routinely generate sub-par assets.
The net sum is studios that plan smartly -- first in not tossing assets due to poor planning and secondly keeping art costs inline with the projects sales and find art choices that maximize their budgets -- will do well and studios that fail to adapt and cannot design and deploy reasonable budgets for their sales expectations won't be long for this world. Put another way: if you only plan on 500K sales you need to set your budget accordingly and THEN design the game to get the most out of that bang for buck. I get sick of hearing how dev costs are so crazy high--it is primarly "trying to keep up with the Jones" issue. If you don't have a $50M budget don't make a $50M game. The unwillingness of "mid-major" developers thinking a big budget game is what they need to do to leap into the big boys is the first sign there is trouble: they are trying to solve the design issue with people and money, not a better approach.
And the grim reality is also this: 70% of games are in the red and 30% in the black. This is why Publishers rule the roost these days: they hit a home run with AAA title which funds many other projects. The industry is "hit oriented" and everyone is looking for the next big hit.
If Publishers raise the cost they will back themselves into a corner where fewer and fewer unique titles will be made and "sure fire" titles will be the only titles worth developing. Long term this is bad for the industry because it diminishes the chances of floating new projects.
That said everything is cyclical: it will result in more unique ideas hitting the indie scene and alternative design efforts and more middle engines tailored toward ease of content creation and fast iteration.