Yes,but even in the worst case scenarios,the power consumption calculations tend to be overrated at times and that is with very intensive FPS games,with no hard FPS cap which review sites use. Most reviews don't tend to test more than one game for power consumption let alone different parts of a game.
Very few sites go into detail about power draw anyway - at least TPU bothers to try and isolate card power consumption using a decent voltmeter and TH,is actually doing dynamic power measurements too. Interestingly they found that some of the newer boost systems actually do end up boosting factionally and massively over rated TDP(with high peak power consumption),but clock down quickly. This means that using a normal plugin meter might not be enough,as they cannot register these peaks quickly enough.
However I still think the power requirements for most systems are overrated anyway during normal gaming scenarios. For example - look at something like the Hexus at the wall power consumption measurements. Even with a highend card,and a overclocked Core i7 4770K,they are seeing under 400W at the wall.
Which is supported by what Valve was powering their prototype SteamBoxes off which uses a Geforce Titan and a Core i7 4770 - a SFF Silverstone 450W PSU. Not even a full sized PSU.
Even with my current system(Xeon E3 1230 V2 and a pre-overclocked GTX660),I am lucky to see beyond 200W peak power consumption(I used two different meters to verify). So slower cards like the GTX750TI and HD7850 will be even less,and they are since I had an HD7850 at hand to test for a few weeks.
But you can kind of tell companies know that - look at the prebuilt PCs from Lenovo,HP,etc with GTX660/R9 270 level cards - they are not shipping with massive PSUs themselves.
To put it in context the vast majority of highend single GPU cards,ie,the GTX780TI,GTX780,Geforce Titan,R9 290 and R9 290X tend to be lower than HD6970 level power consumption and lower than GTX570 and GTX580 cards at least going from the TPU figures.