AMD and Meta Expand Strategic Partnership for AI Infrastructure
AMD has announced an expanded strategic partnership with Meta to support the social media company’s AI infrastructure, spanning processors, GPUs, networking, systems and software.
The collaboration includes AMD Helios rack-scale solutions running at Meta, EPYC processors supporting Meta’s Muse AI agent, and Day-0 software support for Meta AI models, including Muse Spark Open and Muse Glimmer.
The expanded partnership builds on years of joint engineering work involving AMD EPYC processors, AMD Instinct GPUs, AMD Pensando networking and AMD ROCm AI software.
“Meta is a strategic partner for AMD, and we have built this partnership through years of deep technical collaboration – from EPYC processors powering Meta’s infrastructure to Instinct GPUs and open software supporting its AI workloads,” said Lisa Su, Chair and CEO, AMD. “With Helios now running at Meta, we are taking that partnership to a new scale in the superintelligence era. Together, we are building and deploying infrastructure at gigawatt scale and shaping future generations of silicon, software and systems to enable Meta’s most ambitious superintelligence models and services.”
“AMD has been a key partner in building out the compute that’s running our frontier models,” said Mark Zuckerberg, founder and CEO, Meta, “We’re making big investments in the computing power that will enable us to bring personal superintelligence to everyone, and AMD plays an important role in that.”
AMD technologies support Meta’s infrastructure across several areas, including computing, networking and software. Initial AMD Helios rack-scale systems have been installed at Meta and are undergoing testing ahead of a broader deployment.
AMD EPYC processors power dedicated computing environments supporting Meta’s Muse personal AI agent. Meta has deployed millions of EPYC processors across multiple generations and is a lead customer for AMD’s sixth-generation EPYC CPUs. The companies are also collaborating on future processor generations, including the codename Verano.
AMD Instinct GPUs are used in Meta’s production infrastructure for AI training, ranking and recommendation systems. Meta was an early customer involved in defining the AMD Instinct MI450 architecture and plans to deploy a customised Instinct GPU optimised for its workloads. The collaboration is also informing future AMD platforms, including Instinct MI500-series GPUs and next-generation rack-scale systems.
The companies have also worked together to improve support for AMD GPUs through ROCm and PyTorch, enabling developers to use AMD hardware with established AI software frameworks. AMD has provided Day-0 support for Meta’s Muse Spark Open models on Instinct GPUs and Muse Glimmer on AMD Ryzen AI PCs and Radeon GPUs.
Their software collaboration has expanded to large-scale AI training, including the optimisation of recommendation models on AMD Instinct MI350X GPUs. The work has also contributed improvements to PyTorch, Triton and ROCm for use by the wider developer community.
As Meta expands its AI capabilities, the companies are coordinating engineering work and product roadmaps across silicon, software and systems to address the demands of running AI infrastructure at scale. The collaboration covers current and future Meta workloads, with both companies working towards integrated AI systems designed to support infrastructure deployments at gigawatt scale.
As part of the announcement, AMD also said it would invest USD 1 billion in Meta’s equity offering.
