Yep, I do. The risks he was talking about is the fact that EPYC up until Zen 2 hadn’t been certified with several certifications (that take years of steady supply to gain). Not being certified by, say, CISCO can be an automatic percieved high risk to a good chunk of companies and a complete disaualifier. Zen 2 is going to be the first set of EPYC chips to be certified by such organizations.
The thing you have to understand as well is that these corporations do not care about initial cost. They care about density and cost of ownership, both of which AMD have huge advantages in:
Density: AMD can fit 128 cores and 256 threads on a single Motherboard with current-gen (next gen is pretty much garunteed to be 256/512, and according to leaks maybe more…), with a whack ton more memory (4TB per socket x 4 sockets) and a monsterous amount of more PCIe lanes at 128 x 4(something VERY important to data-centric services, such as Dropbox… that have now completely switched over to EPYC). Intel can only fit 96 cores and 192 threads on a single board, with 3TB of memory per socket x 4 sockets and 36 PCIe lanes each (other xeons have higher amounts of PCIe lanes, up to 48, but don’t have as many cores).
Cost of ownership: AMD’s processors as-is are already a pretty significant power savings. The higher density allows for less servers… which leads to lower electricity usage and lower HVAC cooling bills. On top of that even at the same amount of physical processors AMD processors are still more efficient, with the highest end current EPYC processor having a 180w TDP and the Intel processor above having a 220w TDP. Add on top of this 7nm having a 60% per-clock power efficiency advantage over 12nm LP (current EPYC is on 14nm), and suddently Zen 2 is posed to be an insanely efficient and cool chip, with insane density and insane short and long term cost savings. Insane!
As a bonus…
Security: Now that EPYC is being certified with the same certifications as Intel, you now have a more even playing field where processors can compete by the content of their character rather than the color of their advertising. Intel has several low-level exploits that cost performance to fix, whereas AMD is now the inherintly more secure chip (even with Intel’s fixes. AMD has some of these exploits too, but they require you to have local in-person admin access to the machine. By that point the hacker has already won and done more scary stuff than Spectre or meltdown. That means several organizations are going to chose AMD over Intel simply because they’re more secure. Intel isn’t able to completely fix this issue without creating an entirely new architecture, which they haven’t done I’m over a decade. That’s why so many chips spanning so far back are affected. And remember how long it takes to create a new architecture? 6-8 years. Meaning if Intel started when this stuff broke, they have 5-7 years to go.