This guy is (pretty much) right. Firstly it's energy not power (power is more like the rate of flow of energy).. The embodied energy of a PC (ie, energy required to make it) is huge.
>life cycle energy use of a computer is dominated by production (83%) as opposed to operation (17%). The yearly life cycle cost of owning a computer is about 3,000 MJ/year, half again that of a refrigerator, a much larger appliance that uses far more electricity in operation. The short lifespan of computers and the variety of computing needs of users suggests that extension of lifespan, for example by promptly reselling to users who need less computing power, is a promising approach to mitigating environmental impacts.
This is from a study[1] from 2004 of a 1990 PC, inclusive of CRT... I suspect the figures are similar if not worse, as power consumption has come down a lot while chip sizes have remained similar... regardless, the point is so strong that it seems quite likely to remain true whatever the case -- unless your idea of a 'PC' is a rPI.2, which might be useful for some of the same scenarios, but certainly not all-- it's still far from the CPU power of a 2.8ghz p4. So, in many (most?) cases this is credibly a 'green' thing.
But I am talking about something like RPi2 on higher frequency, more ram and small ssd. I bet it will run better than this archaic machine with P4. Faster memory transfer rate, SSD, proper video acceleration makes a huge difference. P4 had terrible chipsets
Cost to build likely includes the cost of power, no? In other words, if building an ARM SoC costs $10, the electricity cost in the ARM SoC is under $10, so if you save $10 in electricity by running the ARM SoC, then you've offset the cost of the energy used in manufacturing.
Energy is dirt cheap in comparison to labor. Therefore cost of industrial goods is correlated to labor not to energy / power costs. (with few exceptions in heavy industries.)
That is not the point of the comment to which you're replying. They didn't claim that the purchase price provides a good estimate of the energy cost, but that it provides a strict upper bound. If the manufacturer is not selling at a loss, then they can't sell for less than their energy costs, let alone labor costs.
