Login to Continue Learning
NVIDIA is reportedly redesigning its next-gen Rubin lineup to better compete with AMD’s Instinct MI450 series. According to Fubon Research, which is part of Taiwan’s largest banking group, this redesign will likely reduce production volumes in 2026 by about four to six months. The reason for this delay is linked to AMD’s upcoming IF64 and IF128 configurations with significant HBM capacity—up to 432 GB per GPU.
NVIDIA’s decision appears driven by the need to counter these advancements, although specific details of the redesign are still uncertain. The company aims to introduce newer Rubin configurations but faces challenges due to AMD’s modular design approach facilitated by Infinity Fabric over Ethernet, which simplifies rack deployment.
The next-gen AI hardware segment is expected to see a shift in market share from NVIDIA to AMD. This competition has prompted NVIDIA to reassess its product strategy and potentially delay the release of Rubin Ultra to align with the timing of AMD’s MI450 debut. The exact timeline for Rubin Ultra remains uncertain but might be pushed ahead, similar to how Blackwell Ultra faced delays.
For more detailed analysis on this topic, refer to our previous article: [Is NVIDIA Leading Itself Into an AI Supply Chain Exhaustion?](https://wccftech.com/is-nvidia-leading-itself-into-an-ai-supply-chain-exhaustion/)
