Option Millionaires

[uranium-pinto-beans]

Wedbush says Penguin/SGH not mentioned in Meta's GenAI GPU clusters presentation
Wedbush notes that Meta (META) released a blog detailing the makeup of its two new AI clusters, both of which include 24,576 Nvidia (NVDA) H100 GPUs. The AI clusters are noted as building "upon the successes and lessons learned from RSC," though are able to "support models larger and more complex" and "pave the way for advancements in GenAI product development and AI research." As such, the firm believes it's logical to view the new infrastructure as the successor to Meta's Research Super Computer, or RSC. Wedbush further notes that there was no reference to Penguin/SGH (SGH), which was mentioned as managing Meta's original RSC deployment, and no mention of Supermicro (SMCI). Additionally, the firm points out that the clusters use a Linux Filesystem backed by a version of Meta's "Tectonic" distributed storage solution and as with servers, this architecture shift "seems to move Meta from third party sourced" Pure Storage (PSTG) to internally designed hardware. Wedbush has an Outperform rating on Pure Storage and a Neutral on Supermicro.