Intel stated that TriGates manufacturing cost are 2-3% more while providing 37% more performance and 50% less power in active state.
R&D isn't cheap but everyone is spending the money on it. How do you think more important breakthroughs occur? ...by R&D money and time. Non-planar silicon has been an idea for decades now.
Where do you think that Intel is going to pass the cost of that R&D to? They certainly aren't going to eat it.
I read the whole article and of course, a 22nm chip should have better performance and use less power than the 32nm chips they compared it to in every graph.
I will admit that it sounds nice right now, but I'll still wait for real world performance, just the same as I have made no firm commitment to Bulldozer.
Sounding nice and being nice are 2 different things.