It meana, its finely granulating of threads is a cache thrasher if true, which will slow it down with larger workloads, making it non competitive to GPUs in the markets where theyre to compete.
A smaller load level will suffice it appears.
It will come down to salesmanship, where they already have nVidia there, who does a good job, as most true usage is beyond the ease of use for the customer generally in whichever case they choose, be it GPU or x86 MiC.
Basically, they lose on the convenience factor having x86 at higher threading, it will come down to perf.