Meta has expanded strategic partnerships with NVIDIA and AMD in moves that collectively signal a broader multi-vendor approach to building the infrastructure behind its Muse family of AI models and future superintelligence ambitions. The announcements highlight Meta’s efforts to scale AI training, inference, networking, and systems infrastructure as demand for AI agents continues to grow.

While announced separately, the two agreements reveal a common objective: strengthening the compute and networking foundation supporting Meta’s AI roadmap. Together, they position NVIDIA and AMD as key contributors across multiple layers of Meta’s AI stack, from processors and accelerators to software frameworks and networking technologies.

Expanding the Muse ecosystem

Under the expanded partnership with NVIDIA, the companies will increase collaboration around Muse and the broader Muse family of models. The effort focuses on improving training and inference performance while optimizing AI workflows to enhance efficiency and reduce operational costs. NVIDIA said the companies are also evaluating additional collaboration opportunities involving OpenShell Reference Architecture, OpenShell deployment capabilities, and NVIDIA Sentry monitoring technologies.

“This changes how we use computers, from operating applications to working with intelligent agents,” said Jensen Huang, founder and CEO of NVIDIA. “NVIDIA and Meta are expanding our partnership to optimize the computing infrastructure end to end for capable, efficient and secure AI agents.”

The announcement also includes plans for Meta to expand deployments of NVIDIA technologies, including Vera Rubin systems, Vera CPUs, Spectrum-X Ethernet networking, and BlueField-4 networking platforms. NVIDIA additionally disclosed plans to invest US$3 billion in Meta’s equity offering.

AMD expands infrastructure role

Separately, AMD outlined a broader role within Meta’s AI environment. The company said AMD Helios rack-scale solutions are running at Meta, while AMD EPYC processors support the infrastructure behind the Muse AI agent. AMD also noted that Meta has deployed millions of EPYC processors across multiple generations and is collaborating on future processor platforms.

AMD reported that its Instinct GPUs are currently used for production training, ranking, and recommendation workloads. The company also highlighted software collaboration through ROCm and PyTorch, including Day-0 support for Meta’s Muse Spark Open and Muse Glimmer models.

“With Helios now running at Meta, we are taking that partnership to a new scale in the superintelligence era,” said Lisa Su, chair and CEO of AMD. “Together, we are building and deploying infrastructure at gigawatt scale and shaping future generations of silicon, software and systems.”

AMD separately announced plans to invest US$1 billion in Meta’s equity offering.

Implications for the AI infrastructure market

Taken together, the announcements suggest that Meta is pursuing an increasingly diversified approach to AI infrastructure development. Rather than emphasizing a single technology stack, the company is expanding relationships across CPUs, GPUs, networking platforms, software frameworks, and integrated systems.

“Muse is how we bring personal superintelligence to everyone,” said Mark Zuckerberg, founder and CEO of Meta, in NVIDIA’s announcement.

For semiconductor suppliers, manufacturing ecosystems, and infrastructure investors, the developments underscore how future AI deployments may depend on broader supplier collaboration and increasingly integrated hardware-software environments capable of supporting large-scale AI services.

28 October 2026