It's not such an exaggeration in this case -- for every 10° Celsius (18° Fahrenheit) you lower the operating temperature of your chips, you double the MTBF (mean time between failures; i.e. lifespan). At stock speeds, this isn't a huge consideration, since the average user in Spreadsheetville will probably replace their system before hitting the MTBF. However, if you're going to put the screws to your components by asking them to do a lot of high-utilization processing for long stretches of time -- play games for hours on end frequently, engage in distributed computing projects, generate multimedia content, or other such computational heavy lifting.

Stock cooling solutions have come a long, long way, but they're still, especially in the case of the retail-package CPU or a system bought from an OEM (Dell, HP, etc), merely adequate and not a whole lot more. This is why there's a thriving aftermarket cooling industry, especially for processors -- anyone trying to squeeze a bit more performance out of their CPU is going to need a better solution, or risk some potentially expensive failure in a big hurry. Video card manufacturers have typically been somewhat better about this. That said, the current state of the GPU art, with upwards of 150-200 watts of heat to cope with, means that aftermarket cooling is going to remain in robust demand here as well.