NVIDIA to Invest $3B in Meta as Companies Expand Work on Muse AI Agents

NVIDIA will invest $3 billion in Meta’s equity offering as the two companies expand their work on Muse, Meta’s family of AI models and its consumer agent, according to an announcement from NVIDIA on Oct. 26.

The deal ties the world’s largest AI chip supplier more closely to one of its biggest customers. It arrives just weeks after Meta launched Muse as a personal AI agent that handles tasks autonomously, including browsing the web, filling out forms, making purchases and continuing to work after you close the app.

Inside the Expanded NVIDIA-Meta Deal

The companies said the expanded collaboration aims to improve inference and training performance for Muse and the broader Muse model family. According to the announcement, Meta’s Muse models and agents run on NVIDIA infrastructure and have posted frontier-level benchmark results and fast-growing consumer downloads since launch.

The work will focus on three areas:

• Workflow Optimization: Joint tuning across workflows to maximize performance and improve total cost of ownership

• Agent Security & Monitoring: Possible use of the NVIDIA OpenShell Reference Architecture, including NVIDIA OpenShell for safe deployment and NVIDIA Sentry for monitoring agent behavior

• Infrastructure Expansion: Broader deployment of NVIDIA Vera Rubin and Vera CPUs, plus Spectrum-X Ethernet networking and BlueField-4

The security piece is exploratory. NVIDIA said the companies will consider additional collaboration on OpenShell and Sentry, but did not say that Meta has committed to deploying them.

Where Muse Adoption Fits In

Meta has given investors reasons to expect Muse to drive compute demand. When Meta reported third-quarter results last week, CEO Mark Zuckerberg said Muse accounts had topped 10 million. In the same report, Meta said it now anticipates spending will range from $135 billion to $145 billion, compared with an earlier estimate of $130–145 billion.

“This [deal] 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, so Muse can empower billions of people to achieve more and pursue greater ambitions.”

Zuckerberg added that Muse is central to Meta’s long-term AI strategy. “Muse is how we bring personal superintelligence to everyone. We’re excited to keep working with NVIDIA to deliver the best personal agents and frontier models for billions of people,” he said.

The Hardware Behind the Deal

Meta said it will expand its use of NVIDIA’s Vera Rubin platform and Vera CPUs to support next-generation agents with a focus on performance per watt. It will also deploy Spectrum-X Ethernet networking and BlueField-4 data processing units to improve network efficiency and throughput.

Component	Role in Meta’s Deployment
Vera Rubin	Training and inference for next-generation agents
Vera CPU	Agent workloads with a focus on performance per watt
Spectrum-X Ethernet	Network efficiency and throughput
BlueField-4	Network efficiency and throughput

Security and Financing Questions Ahead

Consumer agents like Muse take real actions on users’ behalf, and NVIDIA is pairing its compute with tools for sandboxed deployment and behavioral monitoring. Organizations building their own agents face the same questions around containment and oversight.

The $3 billion equity investment also adds to scrutiny of NVIDIA’s financing ties with its customers. VKTR previously reported on concerns about circular financing in the AI sector, where NVIDIA has been linked to more than $750 billion in financing deals and partnerships in a single month. NVIDIA did not disclose further terms of the Meta investment or details of the equity offering.