Recently console games have switched away from using MSAA because post process techniques are giving typically better AA quality for the cost, e.g. approximating 4x4 MSAA quality. On 2160p, using 2x2 hardware MSAA may be just fine in terms of quality, and so the cost per pixel will be lower than using post-process AA. Either that or they use a post-process AA with smaller filter size... still reducing the cost per pixel.
To be honest, unless you look up close it's hard to see the 4k difference except on a very large format screen so in next console generation we may see 4k output but not full native 4k resolution, for example 1920x2160 scaled by 2x horizontally, or using hybrid AA.
In a way I think we can think of 4k as needing not 4x the power, but closer to 2x~2.5x because of the advantages of smaller pixels and visual limits. A comparison is 4k Blu-ray spec is not 4x the spec for 1080p Blu-ray but about 2.5x capacity and max 2.5x the bitrate.
Isn't it because MSAA only anti aliases the geometry? All the post process effects, transparancies and temporal aliasing effects still need to be dealt with.
Plus the 4K difference will be very noticeable in VR, next-gen 4K VR glasses should be very possible. OR is already aiming for 1440p and there are rumours Sony is too. And 4K smart phone screens are on their way.
Blu-ray 4k does use 4x the bandwidth, you're forgetting that h.265 is about twice as efficient as h.264 (and more than 4x as computationally expensive)
But yes 2560x2160p (2.6x 1080p) and 1600p (2.5x 1080p) upscaling is probably the max we'll get next gen.