The picture will never be entirely clear, since there is a whole bunch of legacy product being sold on legacy process nodes, all at the same time. There are a lot of transitions going on at the same time. For example, the very low end stuff is being migrated from fabs that were sold (JV1, JV2) to 110nm, some other legacy chips are being migrated to 90nm and 65nm.
Any estimates as to what portion of their Revs are coming from new design wins vs. legacy business?
Only SP1 is in production of pure 65nm parts. Fab 25, while it itself gets upgraded to 65nm (200mm) is mainly producing 90nm and probably even 110nm stuff. It is not because Spansion has a desire to do so, or that SP1 is not capable of producing more advanced chips. It is because clients want them for their existing designs. NOR market is very different from NAND market in this regard.
So while Spansion wants to hurry clients to start accepting chips made at latest process nodes, Spansion has a limited ability to do so. It is mainly the new design wins that get the latest 65nm output from SP1 (and at some point in the future, 65nm output from Fab 25, and eventually TSMC and/or SMIC).
There is no question NAND market is vastly different from NOR market and it seems like a significant portion of Spansion's business (or for that matter, the entire NOR industry's business) is coming from servicing legacy designs! This could be a major problem because in some sense it is slowing down the pace at which companies are trying to lower the cost per bit! |
