Nvidia, AMD to invest $4 billion in Meta as tech giants deepen AI partnership

Nvidia and AMD agreed to buy Meta shares worth $3 billion and $1 billion, respectively. Together, their investments total $4 billion. The deal strengthens commercial ties between the chipmakers and Meta, whose AI ambitions require much more infrastructure.

Meta plans to raise $95 billion in total. It will first raise $35 billion through underwritten public equity offerings. It then plans to raise another $60 billion through an at-the-market program, which sells shares over time in the market.

The chipmakers’ $4 billion is therefore not Meta’s entire fundraising plan. It is part of the broader effort to obtain capital while deepening partnerships with companies supplying Meta’s AI systems. Meta also said Nvidia, AMD, and Broadcom were interested in purchasing up to $5 billion in shares. The money will support Meta’s AI infrastructure investments.

An equity offering is a company’s sale of new shares to investors. In exchange for money, investors receive ownership in the company. This can provide capital without taking on debt, but it also increases the number of shares outstanding and can dilute existing ownership.

An at-the-market program gives a company flexibility to sell shares gradually through the public market. Rather than selling the entire amount at once, the company can offer shares when market conditions and demand are suitable. Investors buy at prices available in the market when each sale occurs.

Meta said it would initially raise $35 billion through underwritten public equity offerings. It plans to raise another $60 billion through an at-the-market program. Together, those plans equal $95 billion. The source describes the fundraising structure but does not provide additional terms, pricing, or timing details.

Nvidia’s planned investment is $3 billion, and AMD’s is $1 billion. Combined, the chipmakers would put $4 billion into Meta’s equity offering. That amount is significant, but it represents only part of Meta’s larger capital plan.

Meta initially plans to raise $35 billion through underwritten public equity offerings. It then plans to raise an additional $60 billion through an at-the-market program. Adding those two stages produces the announced $95 billion total. Meta also said Nvidia, AMD, and Broadcom were interested in purchasing up to $5 billion in shares.

These figures show the scale of Meta’s AI infrastructure push. The $4 billion from Nvidia and AMD is investment from two strategic technology partners. The $95 billion figure covers Meta’s wider fundraising effort. The article connects that money to investments in AI infrastructure, while the partnerships add hardware and software support.

Nvidia and AMD are both suppliers and investors in Meta’s AI expansion. Supplying processors, racks, and related systems creates business today. Buying Meta shares can align the chipmakers with a customer whose demand for AI infrastructure may keep growing.

The article gives concrete examples of that connection. Nvidia said Meta would expand Vera Rubin and Vera CPU deployment and deepen work on Muse, Meta’s AI agent. AMD said its Helios racks were running at Meta and its EPYC processors were powering Muse. The investments therefore accompany active technology relationships.

This arrangement can help both sides coordinate infrastructure, software, and future systems. Meta gains deeper ties with major chip companies as it builds at gigawatt scale. Nvidia and AMD gain exposure to Meta’s ambitious AI models and services. The article also places these investments within a broader trend: both chipmakers are rapidly investing in companies involved in the AI boom.

The source identifies Vera Rubin and Vera CPU as Nvidia technologies that Meta will deploy more broadly. In general industry usage, Vera Rubin refers to an Nvidia AI computing platform, while Vera CPU refers to Nvidia’s central processing unit technology. The article does not provide technical specifications.

AMD’s Helios is described as a rack system running at Meta. A rack groups computing equipment into a deployable unit. EPYC is AMD’s processor family, and AMD says EPYC processors power Meta’s Muse AI agent. These technologies supply different parts of the computing foundation behind AI services.

Together, the examples show that Meta’s infrastructure depends on complete systems, not one chip alone. Nvidia contributes platforms and CPUs, while AMD contributes racks and processors. Meta said it would expand Vera Rubin and Vera CPU deployment. AMD said Helios and EPYC were already operating in its relationship with Meta.

Raising $95 billion gives Meta a much larger pool of capital for AI infrastructure. Deeper relationships with Nvidia and AMD can also connect Meta more closely with the companies providing chips, racks, software, and systems. That matters because Meta is pursuing large AI models and services.

The article points to specific results already planned or underway. Meta will expand Vera Rubin and Vera CPU deployment and deepen its partnership with Nvidia on Muse, its AI agent. AMD says Helios racks are running at Meta and EPYC processors power Muse. These partnerships link funding to physical computing capacity and AI products.

Meta and AMD describe their work as infrastructure deployment at gigawatt scale. The companies say it will support future silicon, software, systems, models, and services. The article does not quantify how much additional computing Meta will build or predict specific product outcomes. It does show a clearer direction: more capital and tighter hardware partnerships for Meta’s AI program.

Advanced AI systems perform enormous numbers of calculations. Training a model requires processing large datasets repeatedly, while serving an agent requires computing responses for users and handling tools or other tasks. That work needs specialized chips and many connected machines rather than one ordinary computer.

The article illustrates this layered infrastructure through Meta’s partnerships. Nvidia’s Vera Rubin and Vera CPU technologies are being expanded at Meta. AMD says its Helios racks are running there, while EPYC processors power Muse. Racks combine computing equipment, and data centers house and connect many such systems. Their operation also requires substantial electricity.

The article describes Meta and AMD as building infrastructure at gigawatt scale. A gigawatt is a unit of power, so the phrase signals an enormous physical and energy requirement. The source does not explain exact model-training needs or provide capacity figures. It does show why Meta is seeking $95 billion and closer hardware partnerships: its AI models and agents require large, coordinated systems.