New Delhi: Meta Platforms has signed a major multi-year agreement with Amazon Web Services (AWS) to power its growing artificial intelligence infrastructure, marking a significant shift in how the tech giant is scaling its compute capacity. It will result in Meta utilising tens of millions of AWS Graviton CPU cores, which will make it one of the biggest implementers of the in-house chip technology at Amazon.
The move comes as AI workloads become more complex and resource-intensive, especially as agentic AI systems, which are able to plan and perform multi-step tasks, emerge. Meta is making bets on the diversification of its sources of compute to enhance performance and efficiency as it continues to expand its AI ecosystem around the world.
Meta bets big on AWS Graviton chips
As per a formal AWS blog post, Meta Platforms will roll out tens of millions of Graviton CPU cores in the first wave and has the ability to expand even more depending on demand. Graviton processors, developed by AWS since 2018, are designed to deliver high performance at lower costs for cloud workloads.
AWS says that its new generation, Graviton 5, has a compute performance of up to 25% higher than the previous generation. These chips are produced by Taiwan Semiconductor Manufacturing Company (TSMC), which is among the top chipmakers in the world.
Focus on CPU-heavy AI workloads
Meta indicated that the new infrastructure will be used mainly to perform agentic AI-related tasks, a quickly developing area dedicated to systems capable of performing complex operations on their own.
“As we scale the infrastructure behind Meta’s AI ambitions, diversifying our compute sources is a strategic imperative,” said Santosh Janardhan, Meta’s head of infrastructure, in an official statement. “Expanding to Graviton allows us to run these workloads with the performance and efficiency we need at our scale.”
AWS strengthens its position in the AI infrastructure race
The alliance also underscores the increasing drive that AWS has towards custom silicon in order to compete in the AI infrastructure market. This is more than chips, said Nafea Bshara, a vice president and distinguished engineer at Amazon, observing that AWS is focused on offering a complete stack of AI solutions such as data and inference services.
The recent acquisition is on top of its existing ties with other chipmakers like Nvidia and Advanced Micro Devices, although it has also been partnering with Arm Holdings in developing CPUs.
Through this deal, Meta is substantially increasing its cloud-based compute power, which will put it in the position to support the next generation of AI innovation on a global scale.