Comment by reissbaker
21 hours ago
The U.S. restrictions on Nvidia compute were clear and extremely widely reported on for years prior to Supermicro's sales. That's why Nvidia created special Chinese-market versions of their GPUs, e.g. the H800 and H20, which Supermicro was free to sell at the time (although eventually those were banned as well, which was similarly widely reported on and well known). The rest of your post is, similarly, basically nonsense.
You mean the widely reported and well known industry confusion around TPP revision and timeline of ban a800/h800, first 30 days then immediately i.e. it was moving goal posts / wiffle waffled and anything but extremely clear. US used presumption of denial to stop NVidia from selling A800 and H800 when US gov had no solid idea what TPP threshold to set, Nvidia had to wait months for BIS to specifically determine denial bucket for H20 (notification requirement). This directly analogous to Beijing hinting to Manus to hit the breaks for clarity... which you know Nvidia did in absence. Nvidia did not ILLEGALLY sell POTENTIALLY/UNDER EVALUATION export controlled goods, which Manus did. Nvidia HALTED ~3B in PRC orders and ate like ~5B inventory tax due to regulatory fog. If Nvidia sold, they would have done a supermicro, this comparison of Nvidia correct behavior vs supermicro and manus incorrect behavior, and you can bet your ass someone at Nvidia was going to get the supermicro/manus treatment.
Your grasp of subject matter and timeline is actually, complete nonsense.