Meta Platforms is collaborating with Arm to create a new generation of central processing units (CPUs) specifically designed to handle the increasing demands of artificial intelligence (AI) workloads in data centers. This partnership will see them co-develop multiple generations of these cutting-edge CPUs, aimed at providing immense computational power efficiently within limited data center space. The goal is to support the AI-optimized data centers and large gigawatt-scale AI deployments crucial for Meta’s AI innovations.
The initial offering from this collaboration will be the Arm AGI CPU, which Arm describes as its first data center CPU tailored for the AI era, promising faster performance per rack with improved efficiency compared to traditional CPUs. Meta will act as the lead partner and co-developer for the Arm AGI CPU, ensuring its optimization for Meta’s family of applications and its seamless integration with Meta's custom MTIA silicon. The Arm AGI CPU will eventually be made available to the broader AI ecosystem through Arm, and Meta plans to release its board and rack designs under the Open Compute Project later this year.
This move by Meta aligns with its broader strategy to diversify its compute infrastructure and reduce reliance on single chip providers. The company has already committed to significant investments in AI, including spending up to $135 billion on AI this year. Beyond the Arm partnership, Meta has recently secured substantial deals with other chipmakers, including a multi-gigawatt deal with Broadcom for custom AI accelerators and a potential $100 billion deal with AMD for GPUs and CPUs, amounting to approximately six gigawatts of data center power demand. Meta also plans to deploy millions of Nvidia chips, demonstrating a comprehensive approach to building its AI computing foundation across various hardware solutions. These efforts are part of Meta's pursuit of "personal superintelligence," an AI system designed to deeply understand and empower individuals in their daily lives.
