And that's fair enough. For me, based on Microsoft's
spec, and nVidia and AMD's releases on the DX12 Ultimate utilisation, I'm more inclined to believe their claims - AMD and nVidia both quite literally spent millions building hardware for it.
I actually highlighted Sony's potential to replicate SFS. As you keenly pointed out - hardware is foundational. Microsoft's APU design incorporates hardware specially designed for DX12U implementation. That includes the CU count and the optimal VRAM speeds. Sony will be hard-pressed to replicate Microsoft's patented technology's performance using hardware that wasn't designed for the same technology. As for Kraken, it's a third party solution designed for general purpose compression. It's good stuff, no question. Let's ignore that your 22gb/s figure doesn't match Cerny's own presentation, which touted the SSD as moving around 9gb/s compressed.
Microsoft invented a new compression system called BCPack. This system was specially designed for texture compression - textures, as I highlighted in my original post, are the vast majority of a game's assets - and can reportedbly outpace Kraken for texture compression. Combined with SFS, the amount of texture data that would clog the IO throughput of Sony's custom-built SSD would shrink significantly, freeing up Microsoft's IO throughput for... well, everything else. Combined with Microsoft's optimal VRAM configuration, and they have a distinct edge in loading, storing, recalling, and rendering texture data. Sony didn't bother trying to outpace Kraken, their hardware wasn't designed for SFS, and Microsoft own both pieces of technology. Sony's hardware is well designed, no questions there, but Microsoft's software engineers are the best in the world; I wouldn't discount their efforts so easily.