Nvidia's upcoming Rubin Ultra GPU has reportedly seen a significant downgrade in its High Bandwidth Memory (HBM) capacity. Initial previews suggested 1TB of HBM, but the production version is now expected to ship with 192GB of HBM4, a reduction that is also lower than the standard Rubin GPU's 288GB. This downgrade is attributed to changes in the number of stacked memory dies and a reduction in stack height, alongside a shift to HBM4 dies with lower capacity compared to HBM4E. AI
IMPACT Reduced HBM capacity on Nvidia's Rubin Ultra could impact AI training and inference performance, potentially affecting the cost-performance ratio for AI workloads.
RANK_REASON Report on a significant change in specifications for an upcoming GPU from a major hardware vendor.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →