Meta Signs Multi-Year NVIDIA Deal for Millions of AI Chips, Including First Large-Scale Grace CPU Deployment

First Large-Scale Standalone Grace CPU Deployment

NVIDIA and Meta announced a multiyear, multi-generational strategic partnership on February 17, 2026. Meta will deploy millions of NVIDIA GPUs across its AI data centers and will become the first company to deploy NVIDIA Grace CPUs as standalone chips at scale—a significant milestone for both companies.

Deal Highlights

• Standalone Grace CPUs: Meta will use NVIDIA Grace CPUs independently from GPUs in its AI data centers, marking the first large-scale standalone Grace deployment in NVIDIA's history
• Vera Rubin Systems: Meta plans to deploy NVIDIA's next-generation Vera Rubin CPU and GPU systems in 2027
• Multi-generational Commitment: Long-term supply agreement spanning multiple GPU and chip generations

Industry Context

Meta continues to invest heavily in AI infrastructure to support its research programs (Llama 4, Code Llama), advertising recommendation systems, and AR/VR hardware. For NVIDIA, this deal validates standalone Grace CPU deployments at hyperscale—an important proof point for its CPU ambitions beyond data center GPU sales.

NVIDIA CEO Jensen Huang has also teased that the company will announce a new chip at GTC 2026 on March 16.

Sources: NVIDIA Newsroom, CNBC