Rumor

Meta and NVIDIA partner on multi-generation AI infrastructure featuring Blackwell and Rubin GPUs

Tuesday, February 17, 2026 at 09:49 PM

NVIDIA and Meta have announced a multi-year strategic partnership to build hyperscale data center infrastructure for AI training and inference. The collaboration includes the massive deployment of millions of Blackwell and Rubin GPUs, Arm-based Grace CPUs, and Vera CPUs starting in 2027. Meta will also integrate NVIDIA Spectrum-X Ethernet switches into its Facebook Open Switching System and adopt GB300-based systems to create a unified architecture across on-premises and cloud environments.

Context

NVIDIA and Meta have announced a multi-year partnership to deploy millions of Blackwell and Rubin GPUs, marking a massive expansion of Meta’s AI infrastructure. This collaboration integrates NVIDIA Spectrum-X networking and Arm-based Grace CPUs into Meta's hyperscale data centers to maximize performance for both training and inference. The deal focuses on energy efficiency through a unified architecture, with Meta adopting NVIDIA GB300 systems to streamline global operations and scale its recommendation engines. The roadmap includes the potential large-scale deployment of Vera CPUs by 2027 and the use of NVIDIA Confidential Computing to secure AI features within WhatsApp. By co-designing hardware and software, the two giants aim to deliver "personal superintelligence" to billions of users. This partnership represents a significant capital commitment, reinforcing NVIDIA’s dominance in the AI supply chain while securing the massive compute power necessary for Meta’s long-term generative AI ambitions.

Related Companies

Nvidia

NVDA