Meta and AWS sign multi-year deal to scale AI compute infrastructure

Meta Platforms has entered into a multi-year strategic agreement with Amazon.com’s cloud division, Amazon Web Services (AWS), to significantly expand its artificial intelligence computing capacity. The deal will see Meta leveraging tens of millions of AWS Graviton chip cores to power its AI agents and broader artificial intelligence initiatives across its infrastructure ecosystem.

While the companies have not disclosed the financial terms or the full duration of the partnership, AWS executives confirmed that the agreement spans approximately three to five years. According to Nafea Bshara, vice president and distinguished engineer at AWS, the majority of these Graviton CPU cores will be deployed within the United States, strengthening domestic cloud and compute infrastructure.

Expanding compute demand for AI workloads

The partnership reflects the intensifying global competition among technology leaders to secure large-scale computing resources required for advanced AI development. Meta has been aggressively expanding its infrastructure footprint through multiple chip partnerships this year, including agreements with Nvidia, Advanced Micro Devices, and Arm Holdings, highlighting a diversified strategy to meet rising AI workloads.

AWS noted that Meta selected its Graviton5 processors, built on a 3-nanometer architecture, due to their cost-performance efficiency. Industry analysts say the move underscores the importance of heterogeneous chip strategies as AI systems become increasingly complex.

CPUs regaining relevance in AI era

Although GPUs have dominated AI training in recent years, CPUs are regaining significance due to the rise of AI agents and post-training workloads in large language models. Experts note that CPUs and GPUs complement each other, with CPUs handling orchestration, application logic, and workflow management that support GPU-intensive operations.

This is not the first collaboration between Meta and AWS, with their partnership dating back to around 2016, initially focused on cloud infrastructure services and GPU clusters. However, the latest agreement marks a deeper shift toward large-scale AI compute collaboration.

Meta continues to expand its AI ambitions, including acquisitions such as AI startup Manus for over $2 billion. The company has also announced workforce reductions of around 10% as it restructures to support heavy AI investments, alongside recent model launches and upcoming AI product releases.

For AWS, the deal reinforces the growing relevance of its in-house Graviton chips, positioning them as a key component in next-generation AI infrastructure.