Most of us had concluded that it's inherently harder to shrink the transistor nowadays. However, while that is undeniably true, concerning Intel CPUs it appears to be only part of the story. They have appeared to have shrinked the die size in latest releases hence giving them time to increase the die size in the future and therefore increase performance without necessarily advancing technology node.
That appears to point to me to the expected: That while they can still shrink the transistor, they have projected of possibly hitting a wall around 10 or 7nm, rendering them virtually incapable of increasing performance without multiplying cores or utilizing algorithmic advances. That can not only happen with simply running out of atoms' space but also with quantum mechanical effects that have already started creeping in and possibly render any shrinking on conventional silicon technology pointless after a point.
They can only escape that predicament with a new technological base. But Intel are smart boys, they have realized that this is at least partially wishful thinking and it's possible that around the 7nm node or around 2017 onwards to hit a wall with their main contingency saving a few years with increasing the die size or multicoring everything which everyone knows has its limits on interactive applications. I would not be surprised at all if investors at Intel start jumping ship from 2015 onwards if they start sniffing that, since that position would mean a dying company that can be replaced by any third rate chip manufacturer that just sells chips by the dozen on technology that is years old already.