I could only, due to family commitments, sporadically watch the 4.5 hour presentations of "AMD Financial analyst day 2022" and following Q&A. As marketing infospeak goes, what I saw was as impressive and entertaining as AMD's presentation at Computex, a few weeks back, was lackluster. We've "known" for some time that Zen-4 would have AVX-512 but "known" is not the same as known. Now we know that some form of AVX-512 is part of Zen-4 and unless AMD fumble's this ball it will be available uniformly across Zen-4, mobile, desktop and server. Details of which branch of Intel's gnarly AVX-512 tree (or perhaps a new branch) are not as yet known. AI acceleration in Zen-4 appears to mean (from Robert Hallocks comments ~2 weeks ago) AVX-512 VNNI or some such.