NVIDIA Opens Vera Rubin Architecture to Standardize Global AI Factories

NVIDIA is opening its Vera Rubin architecture to open standards through the OCP, facilitating the development of massive 'AI Factories' that require seamless software-hardware integration.

Share
NVIDIA Opens Vera Rubin Architecture to Standardize Global AI Factories

Architecting the Gigawatt AI Factory

As the demand for generative AI scales, the infrastructure required to support it is evolving from traditional data centers into massive, specialized 'AI Factories.' To accelerate this transition, NVIDIA has announced at the OCP Global Summit that it is opening key aspects of its upcoming Vera Rubin architecture to the Open Compute Project (OCP). This move is designed to standardize the physical and software interfaces required to build and manage AI clusters that are now reaching gigawatt-scale power requirements.

The Software-Defined Vehicle (SDV) philosophy—where hardware is decoupled from software to allow for continuous updates—is now being applied to the very data centers that train the models for those vehicles. By opening the Vera Rubin blueprints, NVIDIA is enabling a broader ecosystem of manufacturers to build compatible cooling systems, power delivery modules, and networking racks. This ensures that the underlying 'silicon-to-software' stack is unified, allowing for rapid deployment of new AI capabilities without total hardware overhauls.

Standardization through OCP is crucial for the industry's sustainability. As AI factories consume more power, the need for efficient liquid cooling and high-density power distribution becomes a software-defined challenge. NVIDIA’s commitment to open standards allows data center operators to use software-defined orchestration to manage power loads across thousands of GPUs, effectively treating the entire data center as a single, programmable entity. This infrastructure will serve as the backbone for the next generation of autonomous systems and digital twins.


Source: NVIDIA Blog