I'll note CoreWeave didn't commission the very first Blackwells until ~Jan 2025 so it hasn't been very long for RMAs. Furthermore, for training even when the NV

I'll note CoreWeave didn't commission the very first Blackwells until ~Jan 2025 so it hasn't been very long for RMAs. Furthermore, for training even when the NVL backplane can use the remaining GPUs a ~20% drop in performance means it isn't in the training cluster. I've personally heard this from several sources in the data centers (installers, training, network). It's not uncommon for 10% of racks to fail on delivery. I hear that's improved somewhat from GB200 to GB300, but the number of FW updates from the time they ship, until they're commissioned is >>10. If an HBM or GPU or backplane supply/cooling fails, it is basically not swappable or repairable. You have a "dead" rack, and deliveries are on allocation so you don't get a replacement for months (eg some "RMAs" for early delivered parts in late 2025 are still dead racks 9 months later). "Tray" swaps are technically possible, but still quite rare, perhaps because debugging takes as much time as commissioning a new rack. I don't want to out anyone, but these are similar comments: https://www.linkedin.com/posts/neelmaster1_aiinfrastructure-inference-nvl72-activity-7492641179134750721--lpJ https://www.linkedin.com/posts/neelmaster1_aiinfrastructure-... https://www.hostzealot.com/blog/news/nvidia-gb200-nvl72-is-not-yet-ready-for-training-advanced-ai-models https://www.hostzealot.com/blog/news/nvidia-gb200-nvl72-is-n... https://introl.com/blog/gb200-nvl72-deployment-72-gpu-liquid-cooled https://introl.com/blog/gb200-nvl72-deployment-72-gpu-liquid...