Steve is a mixed back. One can't argue with his numbers, but sometimes his choices for testing and his conclusions are a bit awkward. I'll use a GPU example... When he revisited the GTX 970 this year, he somehow decided to compare it to the R9 290 instead of the R9 390, despite him comparing it to the R9 390 all the years prior. When multiple people said he should have
included the R9 390 (not replaced the R9 290 with it), he called those people fanboys and that the R9 390 and 290 perform the same and there's no reason to test both. I think the reason for that article/video was more about getting views than really putting the 970 against its competition. He simply didn't want to put any bad light on the GTX 970. Oddly enough, he shortly after releases a video comparing two exact same nVidia RTX 2060 GPUs where only the brand is different, to see if branding matters....
For CPUs it's the same. It's not really consistent. So the best thing is to look at his numbers and draw your own conclusion, rather than following his conclusions and stories he wants to tell. He actually tries to gain more views by feeding the fanboy wars multiple times.