The question is not whether memory prices have risen enough to matter. They have. It is whether the cost is travelling through channels diffuse and slow enough that, by the time it becomes fully visible, the underlying price shock will be two years old.