The reign of Moore's Law ... may be about to end. Or not; the end of Moore's Law has been predicted a few hundred times before, too :)
But suppose for a moment that it does end, that we really have reached the end of IC density increases. The linked article talks about other things that will almost certainly be explored to keep improving computers. One possibility I personally find intriguing: it may become more important again – like it was up until roughly 1990 – for programmers to know how to write efficient programs.
At the moment, the need for algorithmic efficiency has been nearly wiped out by the rapidly growing power of computers. The exceptions are relatively few and far between. When I started programming back in the '70s, the exact opposite was true – nearly every program needed to be written in a very efficient way, as the constraints of computer speed and limited memory were quite severe.
Maybe the pendulum will start to swing back the other way again...