I am not sure what exactly Micron has planned. As others have pointed out, paying $2.5 billion to acquire fabs and then mothballing them seems unwise. Plus, Micron reportedly gave assurances that it would keep at least Elpida's main fab open. The idea of converting to NAND seems possible, especially considering that Micron is reportedly planning to invest $3.8 billion in the Elpida fabs over the next five years.
Commodity memory is a risky business and Micron is used to betting the ranch at every new technology node. I'm glad to see that they are still willing to be aggressive in the marketplace. Is this what Steve Appleton would have done? I think so.
Considering the different tooling of Elpida vs. Micron, for the first year or two of internally disruptive integration Hynix may take 30% while Micron/Elpida goes under 20%. They may not recover in time, though. So the outcome they sought to avoid got accelerated.