NVIDIA has shown off its next-gen Vera Rubin Superchip for the first time at GTC in Washington, primed to spark the next wave of AI.

NVIDIA has shown off its next-gen Vera Rubin Superchip for the first time at GTC in Washington, primed to spark the next wave of AI. NVIDIA Has Received Its First Rubin GPUs In The Labs, Ready For Vera Rubin Superchip Mass Production Next Year, Around The Same Time or Earlier At GTC October 2025, NVIDIA's CEO Jensen Huang showcased the next-gen Vera Rubin Superchip. This is the first time that we are seeing an actual sample of the motherboard, or Superchip as NVIDIA loves to call it, featuring the Vera CPU and two massive Rubin GPUs. The motherboard also hosts lots of LPDDR system memory (32 sites in total), which will be combined with HBM4 memory featured on the Rubin GPUs. Jensen also stated that the Rubin GPUs are back in the labs, which means that these are the very first samples produced at TSMC (Taiwan). Each GPU is surrounded by lots of power circuitry, and from earlier posts, we know that each chip will feature 8 HBM4 sites and two Reticle-sized GPU dies. The Vera CPU will feature 88 custom ARM cores with 176 threads. Talking about the roll-out plan, Jensen revealed that he expects Rubin GPUs to enter mass production stage around the same time next year or earlier, so Q3 or Q4 2026. All of this is happening while NVIDIA's Blackwell Ultra "GB300" Superchip platforms are rolling out at full speed. NVIDIA Vera Rubin NVL144 System - Launching in 2H 2026 In terms of specifications, the NVIDIA Vera Rubin NVL144 platform will utilize two new chips. The Rubin GPU will make use of two Reticle-sized chips, with up to 50 PFLOPs of FP4 performance and 288 GB of next-gen HBM4 memory. These chips will be equipped alongside an 88-core Vera CPU with a custom Arm architecture, 176 threads, and up to 1.8 TB/s of NVLINK-C2C interconnect. In terms of performance scaling, the NVIDIA Vera Rubin NVL144 platform will feature 3.6 Exaflops of FP4 inference and 1.2 Exaflops of FP8 Training capabilities, a 3.3x increase over GB300 NVL72, 13 TB/s of HBM4 memory with 75 TB of fast memory, a 60% uplift over GB300 and 2x the NVLINK and CX9 capabilities, rated at up to 260 TB/s and 28.8 TB/s, respectively. NVIDIA Rubin Ultra NVL576 System - Launching in 2H 2027 The second platform will be arriving in the second half of 2027 and will be called Rubin Ultra. This platform will scale the NVL system from 144 to 576. The architecture for the CPU remains the same, but the Rubin Ultra GPU will feature four reticle-sized chips, offering up to 100 PFLOPS of FP4 and a total HBM4e capacity of 1 TB scattered across 16 HBM sites. In terms of performance scaling, the NVIDIA Rubin Ultra NVL576 platform will feature 15 Exaflops of FP4 inference and 5 Exaflops of FP8 Training capabilities, a 14x increase over GB300 NVL72, 4.6 PB/s of HBM4 memory with 365 TB of fast memory, a 8x uplift over GB300 and 12x the NVLINK and 8x the CX9 capabilities, rated at up to 1.5 PB/s and 115.2 TB/s, respectively. NVIDIA Data Center / AI GPU Roadmap GPU CodenameFeynmanRubin (Ultra)RubinBlackwell (Ultra)BlackwellHopperAmpereVoltaPascal GPU FamilyGF200?GR300?GR200?GB300GB200/GB100GH200/GH100GA100GV100GP100 GPU SKUF200?R300?R200?B300B100/B200H100/H200A100V100P100 MemoryHBM4e/HBM5?HBM4HBM4HBM3eHBM3eHBM2e/HBM3/HBM3eHBM2eHBM2HBM2 Launch202820272026202520242022-20242020-202220182016

Physicists at the University of Oxford show they can assemble Schrödinger’s cat-like superpositions from far more exotic quantum ingredients.

Selection marks the next phase of construction on Micron’s New York project, expected to generate 50,000 jobs.

OneAdvanced has launched what it bills as the UK’s first private, sovereign healthcare large language model trained on NHS primary-care data.