Well, I guess that this will be the first non-welcome post on this website and the topic of this piece is NVIDIA Corporation, specifically the next-gen server GPU and more broadly the NVIDIA Enterprise division.
Now please remember that what is written below are leaks, rumours, and speculation, so take this with an appropriate amount of caution however what is published in this post is believed to be truthful and accurate.
Next-Generation Server GPU
As kopite7kimi has already posted on Twitter on in this tweet (https://twitter.com/kopite7kimi/status/1336900538185572357), the next-gen server GPU is an MCM design but has been delayed. Now, kopite7kimi did not go into detail as to why the next-gen server GPU has been delayed, however, which is the topic of this article.
From the information we have gathered from sources, it appears that NVIDIA is experiencing issues with making the next-gen server GPU appear as a single unified processor, instead of multiple GPUs, to the operating system because of firmware and driver issues.
We have also received reports that the Enterprise Driver team at NVIDIA has told at least one company “not to expect a response on any issues” from them; this is apparently due to people being moved to a different team at NVIDIA.
It is possible that these two events are linked, and that people were moved to work on the next-gen server GPU’s drivers, however, this is speculation by us here at Chips and Cheese.
The Future of NVIDIA in the Enterprise Space
With the rumored issues that NVIDIA is having in the Enterprise Space and with the now resurgent AMD along with Intel trying to break into the Enterprise GPU as well, NVIDIA looks like it may hit a rough spot in the near future.