Karadjgne :
Personally, I'd RMA both units as defective/deficient/not working, because while I agree that it's entirely possible it's the psu at fault there is always a strong possibility that the gpu itself is defective. Would strongly advise thinking about the Corsair CXM 450w or other 450-550w reputable psu.
Well, if you send back a good card to EVGA for an RMA, they will test that card. If they find out it is good, they will ask for your credit card number and charge you to ship it back to you (I've had to RMA one of my EVGA GTX 970s). I don't remember the exact amount but it was something like $35 on the fine print of the RMA form (and this was after going through troubleshooting with a live chat EVGA rep). Don't forget that you have to pay to ship it to them initially on the RMA and they do not accept USPS. You do not get a UPS or FedEx label from them to slap on a box. I think I paid around $20 to ship it UPS ground + loss item insurance. over $100 since I declared it at $350. It was a strong deterrent to make damn sure your really have a bad GPU.
So based on that, and based on that PSU has about a 50% failure rate according to reviews (many of them being DOA out of the box), I'd put my return money on the PSU, cut my losses, and get another PSU elsewhere other than Amazon (like NewEgg). And I still think he can contest the return shipping charge because this PSU is not what it is advertised as. It is extremely rare for a GPU to fail out of the box. A cheap PSU, not so much.
And one thing you'll note about lower tier PSUs: generally (not always) the lower you go in rank, the larger their margin of difference between max output and average output. In mid-tier PSUs, the difference is nominal. The EVGA B2 series for example for a 500W has an average sustained output of 480W. I wouldn't worry about 20 watts. But 150 watts? The highest tier PSUs run their full rated wattage average or sustained within a watt or two (EVGA's G2 series 750W for example, what I have, runs at 748.8W sustained). You get what you pay for.