Meta Platforms has signed a multi-year agreement with Amazon Web Services to use tens of millions of AWS Graviton CPU cores as part of its artificial intelligence infrastructure, a move that deepens the companies’ cloud partnership and adds major computing capacity for Meta’s next wave of AI systems. The companies said the first deployment will bring tens of millions of Graviton cores into Meta’s compute portfolio, with room to scale further as demand grows.
The Meta AWS chip deal centers on CPU-heavy workloads tied to agentic AI, a fast-growing area focused on systems that can reason, plan, and carry out complex, multi-step tasks. The agreement also positions Meta among the largest Graviton customers globally and gives AWS a prominent customer win in the race to supply AI infrastructure.
Deal details
Amazon’s Graviton line is the company’s in-house family of server processors used through AWS, and Meta said expanding to Graviton fits its broader push to diversify the compute sources behind its AI ambitions. Santosh Janardhan, Meta’s head of infrastructure, said diversifying compute sources is a strategic imperative as the company scales the systems behind its AI plans, and he said Graviton will help Meta run CPU-intensive agentic AI workloads with the performance and efficiency it needs at scale.
AWS has been developing Graviton processors since 2018, and the latest Graviton5 chip delivers up to 25% better compute performance than Graviton4, according to AWS statements cited by multiple reports. The chips are manufactured by Taiwan Semiconductor Manufacturing Company, adding another major technology supplier to the chain supporting Meta’s growing AI infrastructure.
Why CPUs matter
While graphics processors still dominate the training of large AI models, this agreement highlights how the rise of AI agents is shifting some demand toward CPUs that handle general-purpose computing tasks. TechCrunch reported that Graviton is an ARM-based CPU rather than a GPU, and said agentic AI creates compute-heavy workloads such as real-time reasoning, writing code, search, and coordinating multi-step tasks.
That distinction matters because the infrastructure needed after a model is trained can look different from the hardware used during training. AWS said its latest Graviton generation was designed to support AI-related computing needs, giving Amazon another way to compete as cloud providers and chip companies push to capture more AI spending.
Meta’s infrastructure strategy
The deal reflects Meta’s portfolio approach to AI infrastructure, combining its own data center investments, custom hardware efforts, and cloud partnerships that offer different capabilities. TahawulTech reported that Meta described AWS as a trusted cloud partner and said partnerships like this will help support Meta AI and agentic experiences serving billions of people worldwide.
News9 and The Economic Times both said the first wave of deployment involves tens of millions of cores, with the option to expand further if Meta’s requirements increase. That gives Meta a very large pool of cloud compute capacity at a time when AI systems are becoming more resource-intensive and more complex to operate at a global scale.
Broader competition
The agreement also strengthens AWS’s case for custom silicon as the cloud industry fights for AI workloads. Nafea Bshara, an Amazon vice president and distinguished engineer, said the effort is not just about chips but about providing the infrastructure foundation, along with data and inference services, to build AI that can scale efficiently to billions of people.
TechCrunch said the new deal brings more of Meta’s spending back to AWS instead of competitors, noting that Meta signed a six-year, $10 billion agreement with Google Cloud in August while also using Microsoft Azure. The Economic Times also noted that Amazon has previously made large chip supply agreements with Nvidia and Advanced Micro Devices and has worked with Arm Holdings on CPU development, underscoring how crowded the AI infrastructure market has become.
For Meta, the immediate takeaway is straightforward: the company is adding a massive number of AWS Graviton CPU cores to support agentic AI workloads and to broaden the mix of infrastructure powering its AI roadmap. For Amazon, the agreement gives its homegrown chip program one of its biggest public endorsements yet from a major AI customer.
