Of course it wasn't defined; the point remains it's still not going to be very large if it's true SRAM. If they are using PS-RAM (Pseudo-Static RAM), though, they could go for 128 MB or so and keep it affordable, as PS-RAM is much cheaper than SRAM.
You can look at the prices for decent-quality SRAM at any wholesaler for their given capacities; even when you take out the wholesaler cut and MSRP mark up, it's very costly for even 32 MB of the better-tier SRAM.
So he's basically saying what some of us were saying in this thread months ago? That TFs aren't the end-all, be-all regarding the consoles and that there's more to the systems than just their TFs? How is this controversial, because John said it?
NX Gamer basically said about the same in his GPU video; throughput utilization (they provide this as a rough guess) would be about 67.5% of peak theoretical performance on both systems, aka usual TF use-cases of about 7.5 TF PS5, 9.5 TF XSX. But he didn't give that as a serious figure, just a rough idea.
We know both systems will have features that push utilization of their GPUs in ways the paper TF specs don't convey. Cache scrubbers, Geometry Engine customizations etc. in the PS5, increased push of executeIndirect and ML texture upscaling DLSS-type stuff in XSX. These are common-sense conclusions.
Don't know what type of hype or drama you tried making out of this but I'd say that was a fail
So