Meta $META has signed a multibillion-dollar agreement with Amazon $AMZN Web Services to deploy tens of millions of AWS Graviton processors, making the social media company one of the largest Graviton customers in the world, the companies announced.
According to CNBC, the agreement spans at least three years, starting with tens of millions of Graviton cores and leaving room for further expansion. Amazon did not disclose the financial terms of the deal.
Unlike Nvidia $NVDA's graphics processing units, which are used to train large AI models, Graviton is an Arm-based central processing unit designed to handle a broad range of computing tasks. Agentic AI — autonomous systems capable of reasoning, planning, and executing complex tasks — tends to generate CPU-heavy demand across areas such as real-time inference, code generation, search queries, and orchestrating sequences of actions across multiple steps.
The latest version, Graviton5, features 192 cores and a cache five times larger than the previous generation, which reduces communication delays between cores by up to 33%, according to Amazon. The chip is built on 3-nanometer technology and delivers up to 25% better performance than its predecessor, Amazon said.
"As we scale the infrastructure behind Meta's AI ambitions, diversifying our compute sources is a strategic imperative," Santosh Janardhan, Meta's head of infrastructure, said in a statement. "AWS has been a trusted cloud partner for years, and expanding to Graviton allows us to run the CPU-intensive workloads behind agentic AI with the performance and efficiency we need at our scale."
Nafea Bshara, an AWS vice president and distinguished engineer, told CNBC that Meta had previously used Graviton chips at a small scale and that the company has rented Nvidia GPUs from AWS since 2017.
The Amazon partnership is one of several major compute commitments Meta has made in recent weeks. CNBC reported that agreements with CoreWeave and Nebius — two companies that lease access to Nvidia GPUs — together totaled $48 billion. Meta has additionally signed a multibillion-dollar deal to use tensor processing units from Google $GOOGL, and has agreements with Nvidia and AMD $AMD, according to Bloomberg.
Bloomberg also reported that Meta is advancing four versions of its MTIA chip simultaneously and recently deepened its relationship with Broadcom $AVGO, which is assisting with the design and manufacturing of that in-house silicon.
The spending push is unfolding alongside significant workforce cuts: CNBC reported that Meta disclosed plans to eliminate roughly 8,000 positions, amounting to about one in ten of its employees.
On the day of the announcement, Amazon's share price climbed 1.8% in morning trading, while Meta's shares saw little movement.
Correction: An earlier version of this article misstated details of a deal between Meta and Amazon. Meta signed an agreement with Amazon to deploy tens of millions of AWS Graviton processors, not hundreds of thousands.