BlockBeats news: On June 5, the renowned investment research firm SemiAnalysis released a research report yesterday stating that NVIDIA’s next-generation AI server cluster, Rubin NVL72, will make significant adjustments to its memory configuration. To address supply chain constraints and ensure timely delivery of Rubin racks, the capacity per rack has been drastically reduced from the originally planned 55 TB to 28 TB. Additionally, most Rubin systems will now use 96 GB SOCAMM modules instead of the originally planned 192 GB modules.
Although Dylan Patel, founder, CEO, and chief analyst at SemiAnalysis, later stated that “this is not a disaster,” and the report itself did not intend to alarm the market with sensational headlines, the market reacted strongly: Micron’s stock plunged 7.7% yesterday, and SK Hynix opened down 8.32% today.
