IMHO, sweetspot strategy isn't just about making the smallest, affordable chip possible, but rather how to make it as efficient as possible. Now, as long a dual chip card can still be made out of a single highend chip for a flagship, enthusiast HALO part while not breaking PCIE spec in the process, and the single highend chip actually becomes a king of the hill chip in the market (something that Cayman seems to be), i don't think ATi strategy is a flawed one. Last round, their Evergreen was (and still is) quite good but TSMC's 40 nm supply mess disrupted their short term strategy, forcing them to forgo any attempt of price war & aggressive marketshare grabbing and instead focusing on maximising profit. Ofcourse, TSMC problematic 40 nm process didn't help much either, it forced ATi to somehow bloat the chip design so the yield won't be in the toilet.

Now, as TSMC 40 nm process is maturing, and fact that 32 nm process isn't available, something needed to be done over Evergreen mArch that's getting long in the tooth. The added 40 nm wafer capacity is also creating an opportunity for ATi, that the rather neglected segment in the last round (200 US$) can & MUST be taken care of this time. ATi isn't in the subordinant position like R600 days of yesterday, they actually a market leader in DX 11 universe. Sweetspot strategy has to be applied flexibly to factor this current condition, you already control the market with Evergreen family launched much earlier than competition, now it's time to DOMINATE, and i think Northern Island family is the right answer for that call.