Meta Signs Multibillion-Dollar Deal with AWS for Tens of Millions of Graviton5 Cores
April 24, 2026
Meta Signs Multibillion-Dollar Deal with AWS for Tens of Millions of Graviton5 Cores
Meta has entered into a significant, multiyear agreement with Amazon Web Services (AWS) to deploy tens of millions of Graviton5 processors, marking one of the largest CPU commitments in the hyperscale cloud sector. The deal, described by AWS as a "multibillion-dollar" arrangement, underscores the surging demand for CPU infrastructure tailored to support agentic AI workloads, a rapidly evolving segment of artificial intelligence that requires high-performance, low-latency compute resources.
Under the terms of the agreement, Meta will become one of the largest Graviton customers globally. While the exact financial details and deployment timeline remain undisclosed, AWS confirmed to DCD that the deal spans several years and represents a substantial investment in custom silicon. The initial deployment will involve tens of millions of Graviton5 cores, with Meta retaining the option to expand procurement to meet its growing AI compute needs. Each Graviton5 chip features 192 cores, meaning a deployment of 10 million cores would equate to roughly 52,000 chips; however, AWS declined to specify the exact chip count.
The announcement comes just two weeks after Amazon CEO Andy Jassy revealed in a shareholder letter that two major AWS clients had requested to purchase every available Graviton instance capacity for the entire year. “We can’t agree to these requests given other customers’ needs, but it gives you an idea of the demand,” Jassy wrote, highlighting the intense competition for CPU capacity in the AI era. AWS launched the Graviton5 processors in December 2025, touting an efficient design that reduces inter-core communication latency by up to 33 percent and features a five times larger cache than the previous generation, enabling customers to scale workloads while cutting infrastructure costs.
Meta’s infrastructure chief, Santosh Janardhan, emphasized the strategic importance of diversifying compute sources. “As we scale the infrastructure behind Meta’s AI ambitions, diversifying our compute sources is a strategic imperative,” he said. “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, vice president and distinguished engineer at Amazon, added that the partnership is about providing the foundation for AI that understands, anticipates, and scales efficiently to billions of users worldwide. “Meta’s expanded partnership, deploying tens of millions of Graviton cores, shows what happens when you combine purpose-built silicon with the full AWS AI stack to power the next generation of agentic AI,” Bshara said.
This deal is the latest in a series of CPU-focused investments by Meta. In February 2026, the company signed an agreement with Nvidia for the first large-scale Nvidia Grace-only deployment, with a commitment to additional deployments of Nvidia’s forthcoming Vera CPUs in 2027. In March, Meta partnered with Arm for the chip company’s first-ever data center CPU, designed specifically for agentic AI infrastructure. Earlier this month, Meta announced a collaboration with Broadcom to develop multiple generations of its MTIA chips for generative AI inference. The surge in CPU demand reflects a broader industry shift, as companies seek processors capable of handling serial computing tasks that underpin agentic AI, complementing the parallel processing power of GPUs.
Source: datacenterdynamics
