NVIDIA invests $3 billion in Meta as the companies deepen their AI partnership
Meta will expand Vera Rubin infrastructure and work with NVIDIA to improve the Muse model family, agent security and inference economics.
NVIDIA and Meta are expanding their AI partnership, with NVIDIA committing $3 billion to Meta’s equity offering and Meta planning a broader deployment of NVIDIA’s next-generation infrastructure.
The companies said they will jointly optimize Meta’s Muse family of models and agents for training and inference performance, with a focus on lowering total cost of ownership. They will also consider integrating NVIDIA’s OpenShell reference architecture for safer agent deployment and NVIDIA Sentry for monitoring agent behavior.
On the infrastructure side, Meta intends to increase its use of NVIDIA Vera Rubin systems and Vera CPUs. It also plans to deploy Spectrum-X Ethernet networking and BlueField-4 data-processing units to improve network efficiency and throughput for AI workloads.
The announcement ties a strategic investment to a deeper technical dependency. Meta gets access to NVIDIA’s newest compute and networking stack as it pushes Muse toward mass-market personal agents; NVIDIA gains both a large customer commitment and an equity position in one of the biggest AI platform companies.
For the wider market, the deal shows that competition in frontier AI is increasingly about the full stack — models, chips, networking, security and operating cost — rather than model quality alone.
Meta
NVIDIA
