FRAM memories - performance and cost comparision

What we need is for ferroelectric memory technology to be scaled to the sub 100-nm region, and for mainstream semiconductor manufacturers to overcome their horror of lead in their process lines. Then it would indeed show great promise for volume markets.

In the meantime, I would expect it to be restricted to applications where either the extremely large number of writes or radiation resistance is essential (the short write time looks to be a rather marginal advantage, as a decent cache can provide equivalent performance). If FRAM can be scaled without increasing the write energy density, low write power could also become an attraction at small geometries.