In reply to FactorXXX:
> That's down to the fact that there could be thousands of electronic components in a modern device and it only takes one to fail and the whole device fails as well.
No, it's down to the fact that silicon is actually very reliable, compared to capacitors.
In the early days of integrated circuits, MIL-STD 217 (the reliability standard) penalised integrate circuits, on the basis of containing lots of transistors. Whereas the reality is that the silicon rarely fails; what fails is the interconnect within the package (unless the device is subject to ESD). This has eventually been recognised...
The most common capacitors to fail are in particular locations, and it's nothing to do with random failure of thousands of devices. It's down to the fact that capacitors are chemical devices, wet electrolytics especially so, and their MTBF is much lower than the surrounding components. This is exacerbated by using a device close to its maximum ratings of voltage and power, because bigger, higher voltage parts are slightly more expensive. The input capacitors on a mains-powered SMPS are particularly bad, because they have to deal with 400V DC, at a significant current. As a result, they often degrade, losing electrolyte until their capacity falls so far that the cheap regulator saturates the associated inductor, and blows the switch off the board, because it hasn't been designed to protect itself from overcurrent...
You can pick a six sigma device, but if its MTBF is still low, or it's over-stressed, it won't help you. If just means that it's a well-manufactured part; six-sigma is a manufacturing quality measure, not a basic failure mechanism measure.