PulseAugur
EN
LIVE 23:05:02

NVIDIA AI infrastructure roadmap hit by major delays and cancellations

By PulseAugur Editorial · [6 sources] ·

NVIDIA is facing significant delays and cancellations across its high-performance computing infrastructure, impacting its AI accelerator roadmap. The Kyber NVL144 rack architecture has been pushed to 2028 due to manufacturing challenges with its PCB midplane, and the NVL576 configuration is also likely delayed. Furthermore, NVIDIA's proposed NVL72x2 back-to-back rack architecture has been cancelled following pushback from hyperscalers. The cancellation of the 4-compute-die Rubin Ultra leaves only a less powerful 2-compute-die version, potentially opening doors for competitors like AMD and TPUv8i Broadfly to gain market share. AI

IMPACT Delays in NVIDIA's high-performance interconnects and server architectures could slow down the deployment of large-scale AI models and impact the supply chain for AI hardware.

RANK_REASON The cluster details significant setbacks and cancellations in NVIDIA's AI infrastructure roadmap, impacting multiple product lines and potentially opening opportunities for competitors.

Read on X — SemiAnalysis →

AI-generated summary · Google Gemini · from 6 sources. How we write summaries →

NVIDIA AI infrastructure roadmap hit by major delays and cancellations

COVERAGE [6]

  1. X — SemiAnalysis TIER_1 English(EN) · SemiAnalysis_ ·

    NVIDIA will sell significantly more Oberon Rubin racks and Oberon Rubin “Ultra” racks to make up for this shortfall.

    NVIDIA will sell significantly more Oberon Rubin racks and Oberon Rubin “Ultra” racks to make up for this shortfall. We discuss the implications of these mass NVIDIA delays and cancellations for the memory, PCB, and ODM supply chains in our Core Research and AI Accelerator

  2. X — SemiAnalysis TIER_1 English(EN) · SemiAnalysis_ ·

    This news also comes as the 4-compute-die Rubin Ultra has been cancelled, leaving only the smaller 2-compute-die Rubin Ultra, which will deliver roughly half th

    This news also comes as the 4-compute-die Rubin Ultra has been cancelled, leaving only the smaller 2-compute-die Rubin Ultra, which will deliver roughly half the real-world performance of the 4-die Rubin Ultra. 5/6🧵

  3. X — SemiAnalysis TIER_1 English(EN) · SemiAnalysis_ ·

    As the NVIDIA roadmap indicates, CPO NVSwitch will not be available until Feynman. As a result, NVIDIA currently has no proven solution to expand the scale-up w

    As the NVIDIA roadmap indicates, CPO NVSwitch will not be available until Feynman. As a result, NVIDIA currently has no proven solution to expand the scale-up world size for Rubin Ultra, leaving a gap for competitors like AMD MI500X or TPUv8i Broadfly to gain scale-up advantages

  4. X — SemiAnalysis TIER_1 English(EN) · SemiAnalysis_ ·

    NVL72x2 back-to-back rack architecture was the new proposed architecture NVIDIA was developing as an alternative to Kyber. It was designed to increase the pure-

    NVL72x2 back-to-back rack architecture was the new proposed architecture NVIDIA was developing as an alternative to Kyber. It was designed to increase the pure-copper NVLink scale-up world size by placing two Oberon racks back-to-back. However, it has since been cancelled due to …

  5. X — SemiAnalysis TIER_1 English(EN) · SemiAnalysis_ ·

    Kyber NVL144 rack architecture has been delayed to 2028 as the PCB midplane remains challenging from a manufacturability standpoint. NVL576, which connects 8x O

    Kyber NVL144 rack architecture has been delayed to 2028 as the PCB midplane remains challenging from a manufacturability standpoint. NVL576, which connects 8x Oberon racks over CPO between the NVSwitches, is also likely delayed or restricted to small volumes given the current htt…

  6. X — SemiAnalysis TIER_1 English(EN) · SemiAnalysis_ ·

    MASSIVE DELAY: Just 3 months after Jensen demoed Kyber NVL144 at GTC, it has faced major setbacks and has been delayed by more than 12 months, pushing it back t

    MASSIVE DELAY: Just 3 months after Jensen demoed Kyber NVL144 at GTC, it has faced major setbacks and has been delayed by more than 12 months, pushing it back to 2028. Below, we explain why Kyber has faced massive delays and why NVIDIA’s NVL72x2 back-to-back rack architecture was…

RELATED ENTITIES

RELATED TOPICS