AI BriefingMeta signs deal with AWS to power agentic AI with Graviton chips
Key point
Meta is scaling its agentic AI infrastructure by adopting tens of millions of AWS Graviton cores.
Details
Meta is adopting AWS Graviton processors at scale. The deployment starts with tens of millions of Graviton cores and can expand further to match Meta's AI demand. Meta was already using Amazon Bedrock at scale, and this deal deepens the long-term partnership with AWS even further.
The spread of agentic AI is driving a surge in CPU-intensive workloads such as code generation, real-time inference, search, and multi-step task orchestration. AWS emphasizes that Graviton5 is designed for exactly these kinds of tasks.
- Graviton5: 192 cores, 5x larger cache than the previous generation
- Reduces inter-core communication latency by up to 33%, enabling faster processing
- Built on the AWS Nitro System, supporting bare metal, ENA, EBS, and EFA
- 3nm process combined with design-to-server integration boosts efficiency, delivering up to 25% improvement over the previous generation
AWS describes Meta as one of the world's largest Graviton customers, and believes the higher performance and energy efficiency can support large-scale AI operations.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Meta Releases Methodology for Attributing Incremental Revenue from AI Infrastructure Investments Using Causal Inference
Meta Admits AI Assistant 'Muse' Was Heavily Inspired by OpenClaw
DeepSeek Releases DSec: Elastic Sandbox Infrastructure for Large-Scale Agentic Training
AWS Introduces Enhanced Custom Event Buses for Organization-Wide Event Sharing