Originally Posted by LAN_deRf_HA
And it only took what, like a year and a half?
It was always priced competatively.
I also don't think it's as easy as just shrinking the pre-existing GTX580 to 28nm, it would take a hefty amount of reworking. Kepler is intented to replace Fermi as a similar but much more efficient (in terms of power and heat generation) iteration. I don't see them wasting time on reworking that design when they already have the improved version in production.