Meta to double down on Nvidia's AI stack, including CPUs

Meta is betting heavily on Nvidia's CPU, GPU and networking portfolio for its AI infrastructure buildout. 

The two companies announced a multi-year partnership that spans Nvidia's portfolio. Specifically, Meta will:

• Adopt Nvidia Grace CPUs. For Nvidia, Meta's CPU deployment is the first all Grace-only deployment.
• Collaborate with Nvidia on deploying Nvidia Vera CPUs "with the potential for large-scale deployment in 2027."
• Scale out AI workloads with Nvidia Spectrum-X Ethernet.
• Adopt Nvidia Confidential Computing for WhatsApp. 

And work with Nvidia to co-design models for Meta's workloads. The companies said: "Engineering teams across NVIDIA and Meta are engaged in deep codesign to optimize and accelerate state-of-the-art AI models across Meta’s core workloads."

• Nvidia touts Rubin platform production, hardware advances
• A look at how Mercedes Benz, Nvidia collaborated on autonomous vehicles
• Nvidia highlights its robotics momentum as Qualcomm makes its platform case
• Nvidia's Groq deal: Acquisition, acquihire or creative licensing deal?
• Nvidia launches Nemotron 3 open models to enable multi-agent systems

Not surprisingly, Meta will deploy "millions" of Nvidia Blackwell and Rubin GPUs and integrate Spectrum-X Ethernet switches with the Facebook Open Switching platform. The companies said Meta and Nvidia will create unified architecture for on-premises data centers and Nvidia Cloud Partner deployments. 

• Meta CEO Zuckerberg sees AI returns ahead, new models on deck
• Meta hopes to accelerate its AI agent plans with Manus acquisition

Constellation Research analyst Holger Mueller said:

"This is a major win for Nvidia as it solidifies the Meta workloads running on Nvidia architecture. It is not so good news for any of the other on-premises AI vendors. It's likely the stickiness of the Meta workloads will be enhanced with Meta adopting Nvidia's confidential computing. For Meta, the Nvidia deal makes design easier and the company likely got a good deal as well."

In a statement, Nvidia CEO Jensen Huang said Meta's scale is an example of a customer betting on the full Nvidia stack. "Through deep codesign across CPUs, GPUs, networking and software, we are bringing the full NVIDIA platform to Meta’s researchers and engineers as they build the foundation for the next AI frontier," said Huang. 

Meta CEO Mark Zuckerberg said the expansion of the Nvidia partnership helps it build leading-edge clusters with the Vera Rubin platform.