The deal ties the world’s biggest AI chipmaker to Meta’s bet on consumer AI agents, as Meta raises the largest equity round in its history to pay for it.
Nvidia will invest $3 billion in Meta’s equity offering and expand their work together on Muse, Meta’s personal AI agent. Nvidia announced the deal in a joint press release.
The two companies say they will optimise workflows to improve Muse’s performance and total cost of ownership. Meta will also expand its use of Nvidia’s Vera Rubin and Vera CPUs, along with Spectrum-X Ethernet networking and BlueField-4.
The investment sits inside Meta’s $97.5 billion equity offering. Meta priced $25 billion of Class A common stock, upsized from $22.5 billion. It also priced 250 million preferred depositary shares at $50 each, around $12.5 billion.
Stocktwits reports that the rest is an at-the-market share sale programme of about $60 billion. It adds that Nvidia, AMD and Broadcom had agreed to buy up to $5 billion of the common shares between them. Reuters describes the raise as the largest fundraising in Meta’s history.
Nvidia has been writing bigger cheques across the AI market. EM360Tech has covered its potential $2.1 billion investment in IREN and its reported $12.9 billion move for Hugging Face.
Safety Tools Are Still Under Consideration
Nvidia and Meta have not agreed to everything in the announcement. The companies said they will “consider” further work on Nvidia’s OpenShell Reference Architecture. That includes OpenShell for safe deployment and Sentry for monitoring agent behaviour.
Nvidia CEO Jensen Huang described AI agents as a new computing paradigm. Meta CEO Mark Zuckerberg said: “Muse is how we bring personal superintelligence to everyone.”
What Muse Does
Meta launched Muse on 8 September. Meta says Muse runs on Muse Secure VM, a dedicated virtual machine with its own browser that holds both the agent and the user’s data. Users choose which apps Muse connects to and can disconnect a service whenever they want.
Modelled on the open-source agent OpenClaw, Muse can access email, calendar, payments, health, shopping and smart-home apps, Reuters reports. Reuters adds that Muse is known internally as Hatch.
Meta says a separate agent called Sentinel must approve anything Muse sends to the internet. Muse also checks with the person before sensitive actions, such as sending an email or making a purchase. Meta has also acknowledged that Muse can still make mistakes, even though it says the agent has reached the minimum safety threshold for release.
On 12 October, Meta introduced Muse Wonder and Muse Wonder Full, which it calls its most capable models to date for agentic tasks.
When Vector RAG Stops Working
Why retrieval strategy now hinges on matching Vector, Graph and hybrid RAG to the questions AI must answer across complex enterprise data.
Why Agent Oversight Matters
Agents with access to email and payments carry real operational risk. Axios noted that a Meta employee went viral after an OpenClaw agent deleted files from her Mac once she gave it access to her email.
Nvidia has been working on agent security beyond this deal. In July it launched the Open Secure AI Alliance after the Hugging Face breach. Nvidia said it plans to contribute open tools to the software that controls AI agents.
Meta has also faced scrutiny over its data collection practices before. Muse’s reach into health and payments data will keep that question live.
The Nvidia tools in the announcement target exactly this problem, but they remain at the consideration stage. Meta has not said when, or whether, OpenShell or Sentry will ship with Muse.
Comments ( 0 )