Meta signs deal to use millions of AWS Graviton chips

On Friday, Amazon announced that Meta has signed a deal to use millions of AWS Graviton chips to power its growing AI needs. The agreement marks a significant expansion of Meta’s relationship with Amazon, securing a major customer for Amazon’s homegrown processors.

AWS Graviton is an ARM-based CPU—a central processing unit that handles general computing tasks, utilizing a specific processor architecture. Unlike graphics processing units (GPUs), which remain the primary choice for training large models, CPUs like Graviton are increasingly used for inference. Inference is the stage after an AI model is trained, when it actively processes prompts. This stage is growing more complex with the rise of AI agents, which are AI systems designed to perform multi-step tasks. These agents create compute-intensive workloads, including real-time reasoning, writing code, search, and coordinating multi-step tasks.

The deal helps Amazon capture a larger share of Meta’s infrastructure spending, which has recently been split among competitors. Last August, Meta signed a six-year, $10 billion deal with Google Cloud, though Meta has also used Microsoft Azure. Meanwhile, AWS has locked in other massive commitments. Earlier this month, AI research company Anthropic agreed to spend $100 billion over 10 years to run its workloads on AWS. As part of their partnership, Amazon agreed to invest $5 billion into Anthropic, bringing its total investment in the company to $13 billion.

The Meta deal allows Amazon to showcase a huge AI customer as a proving point for its homegrown CPUs as they compete with Nvidia’s new Vera CPU, which is also ARM-based and designed to handle workloads for AI agents. While Nvidia sells its chips and AI systems to enterprises and cloud providers, AWS only sells access to its chips through its cloud service. Amazon CEO Andy Jassy has signaled a clear intent to challenge competitors on cost, stating that he intends to win deals on the basis of better price-performance ratios for AI.

Why it matters

The Meta deal serves as a critical validation for Amazon’s homegrown CPU strategy, positioning its Graviton chips as a viable alternative to Nvidia’s offerings for AI agentic workloads.