At this point it should be obvious Ariel and Oberon were the PS5 chips (Oberon C0 being the PS5 chip) and even back when the leaks happened there was more than enough proof of it i.e the regression testing modes with PS4 BC (why would an Xbox chip need that? How would they even do it, it'd be illegal xD). The reason so many people wanted to dismiss them was because they painted "low" TF numbers for PS5, by association, and also didn't list certain things like RT being present.
Yes, having an air of skepticism on those chips and the leaks themselves in healthy doses was always welcomed but a contingent of folks were ADAMANTLY DENYING everything about the leaks, not just for PS5 but also Series X, and whenever they did they always found ways to bump the PS5 specs higher (sometimes much higher) than XSX's on TF grounds. The obsession with TFs was real, and some of us even back then were trying to say that TFs didn't mean everything, but no one was listening.
Even when analysis from quite well-informed people on places like B3D and on Twitter through Komachi & Rogane were able to figure out some things to explain why Oberon had listing data the way it did (Ariel iGPU profile lists for regression testing, Ariel being RDNA1 therefore no RT built in while Oberon likely had the RT hardware but it was disabled/not needed for regression testing, etc.), others still dismissed it. Also never mind that the timing of the leaks generally coincided well with the timing of PS4/XBO leaks before the start of that generation as well, and I don't remember people being so adamant that the chips and specs eventually uncovered there was in fact in no way related to either of those systems, especially as time narrowed down.
There are still people who think just because Sony's GPU clock is higher than what the last Oberon revision showed, or that MS's CU count is lower in active units than what the last Arden revision showed, means (at least moreso in Sony's case) that Oberon was not the PS5 chip and is not the PS5 chip. But it is. Because since then we've not had a single other viable chip emerge with data on it, and by this point that would have absolutely happened. The truth is the last Oberon revision was still showing regression tests, and in MS's case, they have always had more CUs enabled (but lower-clocked) on devkits vs. the retail units (less active CUs but higher-clocked), which is what they did with the One X.