He was referring to Medfield and Clovertrail. Intel never mentioned the need for contra revenue for these chips. I don't know for certain, but they may not have the same "BOM issue".
By the way: How do you position yourself in the market again? Just to understand where you are coming from (I am long Intel, obviously).
Regarding Cyclone and Silvermont: I actually don't really get it why Intel doesn't just use a stripped version of Haswell instead of going through the hassle of developing dedicated Atom cores. The best user experience you could get would actually be a Haswell single core with hyperthreading with probably some tweaks, like smaller cache and shorter pipeline. This power optimized to the maximum would also benefit the laptop CPUs. Crippled Atom cores are too much of a hassle just to be able to say: Quadcore.
Probably customers don't get it, but on mobile OSs that don't even support multitasking, a quadcore is completely nonsense, regardless of any benchmark results you get. For the few occasions where stuff is performed in the background by the OS itself, hyperthreading is absolutely sufficient. I do own a Razr I with Medfield and it performs better than most midrange quadcore phones in real life situations since those only use one lower performing Cortex A9 with three additional cores that just wait for jobs.
Apple reversed this technological failure by introducing a very fast dual core instead of quadcore and NVIDIA, together with others probably soon, start to adpot that. One hyperthreading core would even be more efficient in my opinion and Intel has that covered with patents I guess.
I mean, they could continue with Atom, but they still should give that a try, maybe as lowest end Core-I incarnation - I think this could be a real winner.
Besides, Intel really needs to push Core forward. The 10% gains like Haswell may not be sufficient to charge the premium they are used to. I hope they don't get that out of sight while focusing on making Atom similar to Core.
Market Data 
Markets 
AI
