Quo vadis nano-CMOS ?

When I was a student, 25 years ago, one of the hypotheses was that intelligence would appear spontaneously once complexity and speed of a logic system exceed a given level. In the span of last 25 years the CMOS switching frequency has increased X50, the number of transistors per chip X1000, whereas the transistor feature size has decreased X32. In spite of that extraordinary progress, our computers seem all but intelligent. Does it mean that we are still below this magic complexity level? Maybe, but taking into account that CMOS is already today a genuine Nano-technology, there is little room left for improvement. Therefore, will intelligence appear within the remaining 3 or so generations before Nano-CMOS hits the atomic limit? Or maybe we should admit that transistor performance is no longer a key, and targets for CMOS technologies should be refined? If so, what is then THE nano-device we should seek? What is THE nano-technology we should target? In this paper we will deliberate on these and relevant questions (hardly answering ANY).