Nvidia has officially unveiled its next‑generation AI data center platform, Vera Rubin, at the GTC 2026 event, bringing up to 40 million times more compute compared to systems from 10 years ago.
The Vera Rubin platform is built around a suite of seven different chips and six rack‑scale subsystems designed to support future AI workloads. At its core are the Rubin GPU and Vera CPU architectures, complemented by connectivity, networking, and storage components targeted at hyperscale AI infrastructure.
“Vera Rubin is a generational leap — seven breakthrough chips, five racks, one giant supercomputer — built to power every phase of AI,” said Jensen Huang, founder and CEO of NVIDIA. “The agentic AI inflection point has arrived with Vera Rubin kicking off the greatest infrastructure buildout in history.”
Each Rubin GPU features a massive 288GB of HBM4 memory, delivering up to 22TB/s of memory bandwidth and roughly 50 PFLOPs of NVFP4 compute performance per chip in standalone configurations. The GPUs are liquid‑cooled and are paired with the Vera CPU, which Nvidia says balances high single‑thread performance with energy efficiency using LPDDR5 memory.
According to Nvidia, the Vera Rubin NVL72 rack integrates 72 Rubin GPUs connected by NVLink 6 and 36 Vera CPUs, linked with high‑speed networking and storage tech. The platform promises dramatic efficiency gains, training large mixture‑of‑experts models with a quarter of the GPUs and achieving up to 10× higher inference throughput per watt at roughly one‑tenth the token cost compared with the prior Blackwell architecture.
The Vera Rubin platform also includes NVIDIA’s next‑gen interconnect and networking technologies, such as the NVLink‑6 Switch, ConnectX‑9 SuperNIC, BlueField‑4 DPU, and Spectrum‑X CPO optics to tie the components together in rack‑level deployments. These systems are aimed at lowering inference costs and improving training throughput for large‑scale AI models.
“Vera Rubin is a generational leap — seven breakthrough chips, five racks, one giant supercomputer — built to power every phase of AI,” said Jensen Huang, founder and CEO of NVIDIA. “The agentic AI inflection point has arrived with Vera Rubin kicking off the greatest infrastructure buildout in history.”
“Vera is arriving at a turning point for AI. As intelligence becomes agentic — capable of reasoning and acting — the importance of the systems orchestrating that work is elevated,” said Jensen Huang, founder and CEO of NVIDIA. “The CPU is no longer simply supporting the model; it’s driving it. With breakthrough performance and energy efficiency, Vera unlocks AI systems that think faster and scale further.”
According to Nvidia’s new roadmap, the Vera Rubin lineup is expected to begin shipping in the second half of 2026, with configurations that can scale from single‑chip systems up to large rack arrays designed for data centers and cloud providers.
