Purported AMD Epyc Genoa info points to 96C/192T chip

[trillo_del_diavolo]

AVX-512 has no practical use on desktops. In fact, it has an extremely limited uses in general seeing how its nowhere near in widespread use. Its rarely (if ever) actually used.
Its also a power hog.

It "has no use" because Intel decided you need a several thousand dollar Xeon platform to have it. Same with ECC and many-core CPUs before Ryzen.

I was under the impression that AMD invented x86/x64 and has patents on techniques used in AM64 which have to be licensed from AMD... and that actually seems to be the case.

AMD64 was the only time AMD had the upper hand, and only because nobody wanted anything to do with Itanic. The rule is: AMD develops an extension (e.g. 3DNow), Intel makes another that does the same thing (SSE), AMD gets shafted and has to go along with Intel. Same with FMA3/4 hodgepodge: AMD's developed FMA4 before FMA3, Intel settled on FMA3, AMD had to drop FMA4 for FMA3 with Zen. If to reach more enterprise/HPC customers means wasting silicon with AVX-512, they will certainly do it. Intel and AMD have extensive cross-licensing agreements anyway.

(Ok I misread your final lines to sneer at Intel. Let's try again.)

GPU's seem to have those capabilities, but most software isn't written to take advantage of hw acceleration and requires older methods.
Aka, it could be a software stagnation issue.

Again, without the hardware we'll never know. Vectorization could speed up some tasks in games, for instance, or you could combine CPU+GPU vectorization for more speed in rendering tasks, or crypto.

Just look at how long it takes devs to incorporate support for AMD gpu's in terms of hw acceleration in pro software, despite the fact that Open CL and other open standards work just as good (or better) than CUDA.

IIRC CUDA came first, and GPU computing wasn't in ATI's radar until they were bought by AMD, which made the GPGPU market 100% NVIDIA until OpenCL was mature enough.