Meta has expanded its AI infrastructure with the deployment of NVIDIA CPUs and millions of Blackwell and Rubin GPUs. The partnership also includes the adoption of NVIDIA Spectrum-X Ethernet for network efficiency and Confidential Computing for user privacy.

NVIDIA and Meta have entered a strategic partnership to build hyperscale data centers optimized for AI workloads. Meta will deploy NVIDIA CPUs and GPUs, as well as Spectrum-X Ethernet switches. This collaboration aims to support Meta’s long-term AI roadmap and enhance performance and efficiency.

Meta and NVIDIA are collaborating on deploying Arm-based NVIDIA Grace CPUs for Meta’s data center applications, delivering significant performance-per-watt improvements. The companies are also working on deploying NVIDIA Vera CPUs in 2027 to further enhance Meta’s AI compute footprint and advance the Arm software ecosystem.

Meta will deploy NVIDIA GB300-based systems and a unified architecture to simplify operations and maximize performance. The adoption of NVIDIA Spectrum-X Ethernet networking will provide AI-scale networking with low-latency performance and operational efficiency for Meta’s infrastructure.

Meta has adopted NVIDIA Confidential Computing for WhatsApp to enable AI capabilities while ensuring user data confidentiality. The collaboration will expand Confidential Compute capabilities across Meta’s portfolio to support privacy-enhanced AI at scale.

NVIDIA and Meta are codesigning next-generation AI models to optimize and accelerate AI workloads. The deep codesign efforts aim to drive higher performance and efficiency for new AI capabilities used by billions of users worldwide.

Read more at NVIDIA: Meta Builds AI Infrastructure With NVIDIA