Meta is deepening its relationship with Amazon Web Services, signing a significant agreement to deploy AWS Graviton processors at scale for its burgeoning AI efforts. This deal marks a substantial expansion of their existing partnership, focusing on powering the agentic AI workloads behind Meta’s next generation of artificial intelligence.
The deployment kicks off with tens of millions of AWS Graviton5 cores, with built-in flexibility for future expansion as Meta’s AI capabilities evolve. This positions Meta as one of the largest customers for Amazon’s custom silicon.
This strategic move underscores a critical shift in AI infrastructure. While GPUs remain paramount for training massive models, the rise of agentic AI, systems capable of reasoning, planning, and executing complex, multi-step tasks, is driving immense demand for CPU-intensive processing. Workloads like real-time reasoning, code generation, and orchestrating autonomous agents are precisely where Graviton processors excel.
AWS designed its Graviton chips to offer a faster, cheaper, and more energy-efficient cloud computing experience. The latest Graviton5 boasts 192 cores and a significantly larger cache, improving inter-core communication by up to 33% for faster data processing and greater bandwidth, essential for the continuous reasoning required by agentic AI.
Meta’s expanded use of Graviton processors reflects a strategic imperative to diversify compute sources while optimizing for performance and efficiency at its massive scale. As Santosh Janardhan, Head of Infrastructure at Meta, stated, "AWS has been a trusted cloud partner for years, and expanding to Graviton allows us to run the CPU-intensive workloads behind agentic AI with the performance and efficiency we need at our scale."