The 12VHPWR issue has been really inflated at Tom's. I think Steve at GN approximated 50 cases out of 130k+ GPUs sold. Johnny Guru was one of the first to figure out (at least publicly), and Steve then proved over testing. I wish Tom's would've cut down on the fear mongering with this whole issue, or at least the GPU team should've done some real experiments/testing like Steve before the incomplete picture. AMD did the same thing and imo at least looks really bad now for their classless shade thrown at the 12VHPWR and NVIDIA during RDNA 3 introductions without ever doing any of their own testing either or in reality having an understanding of the issue. The plug with the tiny sense pins are very difficult to know when plugged in fully unless looking to make sure flush or using an alternative cable such as CableMod's that had a more robust clip with confirming sound. You kind of have to wiggle the cable as you push it in when about 90% of the way plugged in to get fully seated. The CableMod's were also individually grounded and crimped, which while not the issue, at least doesn't cause the problem of a non-fully seated 12VHPWR to be compounded.
The 4090 is an incredible GPU. I'm still impressed how much of an unexpected upgrade it has been to me over my 3090.