Meta and NVIDIA partner to boost AI infrastructure

The agreement supports Meta’s long-term AI roadmap and will span on premises data centers, cloud deployments and broader AI infrastructure. Meta plans to build hyperscale data centers optimized for both AI training and inference, aimed at supporting its core services and personalization systems used by billions of people.

As part of the collaboration, Meta will deploy millions of NVIDIA Blackwell and Rubin GPUs and expand its use of Arm based NVIDIA Grace CPUs in production data center applications. The companies said the Grace deployment marks the first large-scale implementation of Grace only systems, supported by joint engineering and software optimization efforts designed to improve performance per watt with each generation.

Meta and NVIDIA are also working on deploying NVIDIA Vera CPUs, with potential large-scale rollout in 2027. The effort is intended to further enhance energy efficient AI computing and strengthen the broader Arm software ecosystem.

“We’re excited to expand our partnership with NVIDIA to build leading-edge clusters using their Vera Rubin platform to deliver personal superintelligence to everyone in the world,” said Mark Zuckerberg, Founder and CEO of Meta.

To support AI workloads at scale, Meta will roll out NVIDIA GB300 based systems and adopt a unified architecture across its on-premises data centers and NVIDIA Cloud Partner deployments. The company has also adopted the NVIDIA Spectrum X Ethernet networking platform to provide low latency, predictable performance while improving operational and power efficiency.

In addition, Meta is using NVIDIA Confidential Computing for private processing on WhatsApp, enabling AI-powered features while protecting user data confidentiality and integrity. The companies plan to expand confidential computing capabilities to other services across Meta’s portfolio.

Earlier, Qazinform News Agency reported that Meta received patent for AI that could simulate users’ social media activity after death.