Meta's Muse AI Partnership With Nvidia To Expand Alongside $3 Billion Investment

Nvidia will invest $3 billion (£2.26 billion) in social media giant Meta as the companies expand their partnership on Muse artificial intelligence (AI) agents, combining a proposed equity investment with plans for greater computing capacity and closer technical collaboration.

The agreement covers Muse and its family of models, according to Nvidia's announcement on 27 October 2026. The companies aim to improve model training and inference performance, building on their existing infrastructure partnership.

The investment will form part of Meta's equity offering, placing the chipmaker's financial commitment alongside its role in supplying infrastructure for Meta's AI ambitions. Completion remains subject to applicable closing conditions.

Meta's Fundraising Plan Grows to $97.5 Billion

The partnership announcement follows Meta's plans to raise approximately $95 billion (£71.43 billion) for AI infrastructure, CNBC reported.

Meta initially outlined approximately $35 billion (£26.32 billion) in underwritten public equity offerings, followed by another $60 billion (£45.11 billion) through an at-the-market programme. Such a programme allows a company to sell shares into the market over time.

Per CNBC's report, Nvidia, Advanced Micro Devices and Broadcom had expressed interest in buying up to approximately $5 billion (£3.76 billion) in shares collectively.

Reuters subsequently reported that Meta had increased its planned equity offering to $97.5 billion (£73.31 billion), making it the largest fundraising in the company's history.

Nvidia's announced equity investment comes as Meta seeks capital to expand the infrastructure supporting its AI products. The fundraising figures describe planned share sales, rather than funds already raised.

Muse Reaches Into Users' Everyday Applications

Muse sits at the centre of Mark Zuckerberg's effort to bring what he calls 'personal superintelligence' to people using Meta's services.

Known internally as Hatch, the assistant was modelled on the open-source AI agent OpenClaw. It can connect to applications covering email, calendars, payments, health, shopping and smart-home functions.

Users can revoke that access at any time, according to Reuters. These connections give Muse access to services people use to manage everyday tasks and personal information.

'Muse is how we bring personal superintelligence to everyone,' said Zuckerberg, Meta's founder and chief executive.

Nvidia said Muse's models and agents already run on its infrastructure. It also reported strong benchmark performance and rising consumer downloads following their launch, although its announcement did not provide specific figures.

More Computing Power, With Energy Efficiency a Priority

Meta plans to expand its use of Nvidia's Vera Rubin infrastructure and Vera central processing units (CPUs) to support future AI agents.

The stated goal is to deliver greater computing performance per watt of electricity consumed. Alongside the processors, Meta plans to deploy Nvidia's Spectrum-X Ethernet networking and BlueField-4 technology to support network efficiency and data throughput.

The companies also intend to collaborate on technical improvements across the workflows used to run Muse. Nvidia said this work would target better performance and lower total cost of ownership.

Jensen Huang, Nvidia's founder and chief executive, described agents as software that can interpret goals, reason, use tools and carry out work for users.

'This changes how we use computers, from operating applications to working with intelligent agents,' Huang said.

Security Tools Remain Under Consideration

The companies are also considering further collaboration involving Nvidia's OpenShell Reference Architecture.

Potential work includes using OpenShell to support safer software deployment and Nvidia's Sentry tool to monitor AI agent behaviour. Reuters also identified these tools as areas the partners could explore.

These security discussions accompany the planned computing expansion but remain prospective. Nvidia described the OpenShell and Sentry work as opportunities under consideration rather than confirmed deployments.

The announcement also identifies Nvidia's investment as a proposed transaction, with completion subject to applicable closing conditions. The companies' stated plans for infrastructure expansion and technical collaboration should therefore be distinguished from completed deployments or guaranteed future results.