I dati alla base di questo evento
4-
Chip
NVIDIA Rubin
The 2026 architecture: two-die GPUs with 288 GB of HBM4, paired with the Vera CPU in NVL72 racks.
Rubin is NVIDIA's 2026 data-center GPU architecture, named for astronomer Vera Rubin. Each package joins two reticle-size dies with 288 GB of HBM4, and NVIDIA's spec table rates it at 50 PFLOPS of NVFP4 inference (sparse) and 35 PFLOPS of NVFP4 training (dense). It ships mainly in Vera Rubin NVL72 racks (72 GPUs, 36 Vera CPUs), which NVIDIA briefly styled…
Apri scheda → -
Chip
NVIDIA Vera CPU
Grace's successor: 88 custom Arm cores, sold inside Vera Rubin and as a standalone server CPU.
Vera succeeds Grace and moves NVIDIA from off-the-shelf Arm Neoverse cores to its own design: 88 custom Olympus cores with 176 threads, LPDDR5X memory at up to 1.2 TB/s, and NVLink-C2C at up to 1.8 TB/s to the GPU. It ships as half of the Vera Rubin superchip, and since May 2026 NVIDIA also sells it as a standalone server CPU aimed at AI-agent workloads.…
Apri scheda → -
Sistema
NVIDIA Vera Rubin NVL72
The third-generation NVLink rack: 72 Rubin GPUs and 36 Vera CPUs in the Oberon frame.
Vera Rubin NVL72 is NVIDIA's 2026 rack-scale system and the successor to GB300 NVL72. It keeps the Oberon rack layout of 72 GPU sockets and 36 CPU sockets, filled with Rubin GPUs, Vera CPUs, NVLink 6 switches, ConnectX-9 SuperNICs and BlueField-4 DPUs. NVIDIA's spec table rates the rack at 3,600 PFLOPS of NVFP4 inference (sparse) with 20.7 TB of HBM4. First…
Apri scheda → -
Azienda
NVIDIA
The company the AI build-out is denominated in: GPUs, systems, networking and CUDA.
NVIDIA turned a graphics-chip business into the platform underneath modern AI: Hopper, Blackwell and Rubin GPUs, NVLink rack-scale systems, InfiniBand and Spectrum-X networking, and the CUDA software layer. Data-center revenue was $89.0B in Q2 FY2027 (quarter ended July 2026), up 117% year over year, and Vera Rubin production shipments began in August 2026.…
Apri scheda →
