News
NVIDIA and OpenAI partner to deploy 5GW of power for Vera Rubin AI infrastructure
Thursday, February 26, 2026 at 07:38 PM
NVIDIA and OpenAI have announced a strategic partnership to deploy 5GW of power capacity using the Vera Rubin architecture for large-scale AI model training.
Context
NVIDIA and OpenAI have entered a landmark strategic partnership to deploy 5GW of power for next-generation AI infrastructure. This massive buildout will utilize the newly launched Vera Rubin platform, which integrates the Rubin GPU and Vera CPU to support the training and inference of frontier models. This "gigascale" project is a critical component of a broader roadmap to scale compute capacity toward achieving artificial general intelligence.
The Vera Rubin architecture delivers a significant technological leap, offering up to 5x the inference performance and 3.5x the training speed of the previous Blackwell generation. Featuring NVLink 6 and HBM4 memory, the platform is designed for AI factories utilizing liquid cooling and high-efficiency 800-volt DC infrastructure. To support the deployment, NVIDIA plans to invest up to $100 billion in OpenAI progressively as infrastructure milestones are met.
Production shipments for the Vera Rubin systems are scheduled to begin in the second half of 2026. The collaboration focuses on co-optimizing hardware and software to slash model training times and reduce inference costs by up to 10x. This multi-year initiative cements NVIDIA as the primary engine for the AI economy, ensuring a steady pipeline of high-performance compute for the world's most advanced AI labs.
Related Companies
Nvidia
NVDA