Gordon Moore - one of the Intel founders, famously proposed that over the history of computing hardware, the number of transistors in a dense integrated circuit will double approximately every two years.
For years everyone believed it, but it became harder and harder to do. You can't really make circuits in silicon that's smaller than a few nanometers, it gets harder to create photolithography, you have to deal with heat....
Now, at 10 nM, Intel finally stumbled and can't produce on schedule.
http://www.anandtech.com/show/9447/intel-10nm-and-kaby-lake
So yea, until we get quantum computing or can manage to multi-task out computers better, expect that you can keep your computer for a few years longer... Apropos of this thread (http://knucklecracker.com/forums/index.php?topic=18890.0). (which is in the wrong place, but I don't dare move it... :P
I still think that Moore made an embarassingly accurate prediction about tech development trends... I mean it's truly remarkable that this 'law' worked for several decades. How many other tech-related predictions are out there that haven't been falsified in a couple of years?
(However, it is totally possible that Moore wasn't a true visionary, but his 'law' was a self-fulfilling prophecy that set the 'goal performance' for the industry ...)
I do think the law still holds true in some sense. Only problem is that we have hit a wall that brought the doubling the CPU power every two years to a halt. Once we find a new method for improving CPU speed and there is no limit in sight, the law will become true again. The law more so represents how fast mankind can push the boundaries of technology. I would like to put the relationship between CPU size and speed on a graph though. It would be impossible to break into another quadrant on the graph, so you would find a limit to how far Moore's law would go.
You know, there are more than one way to uphold moore's law. Larger chips are a thing, just not in phones and tablets. In the future, stationary computers may have chips that are feet across. Plus, more layers of transistors will be stacked in the future.
Nonetheless, the size of an atom is fixed.