Meta expands Nvidia deal to use AI data center chips (www.cnbc.com)

Meta has expanded its partnership with Nvidia through a significant deal that encompasses new standalone CPUs, next-generation GPUs, and Vera Rubin rack-scale systems. The agreement marks a pivotal moment for Meta as it aims to leverage millions of Nvidia chips in its AI data centers, including the deployment of Nvidia's Grace CPUs as standalone units for the first time. This development signals Meta’s commitment to scaling its AI infrastructure, with CEO Mark Zuckerberg emphasizing the goal of delivering "personal superintelligence to everyone in the world." The deal will also enhance AI features on platforms like WhatsApp through Nvidia’s networking technologies. This collaboration is noteworthy for the AI/ML community as it exemplifies the ongoing trend of major tech companies seeking to optimize AI workloads through advanced hardware. The integration of Nvidia's Spectrum-X Ethernet switches and the substantial investment—estimated in the tens of billions—reflects Meta's broader $600 billion commitment to enhancing its data center capabilities by 2028. Moreover, this partnership enhances Nvidia's strategic position within the AI space, particularly as Meta plans extensive AI model development, such as the new frontier model, Avocado. Analysts expect that this deal will significantly influence how AI services are deployed at scale, further cementing Nvidia's leadership in the computational resources required for cutting-edge AI applications.