Meta plans to raise $95B to finance its AI expansion, including a $35B underwritten equity offering, backed by cornerstone investors Nvidia, AMD, and Broadcom (Tim Bradshaw/Financial Times)

Meta is planning to raise $95 billion through a stock offering to fund its AI expansion. That is the central development: the company wants a very large pool of capital for building out its AI capabilities, rather than treating the effort as a small project.

The proposed package includes a $35 billion underwritten equity offering. Nvidia, AMD, and Broadcom are identified as cornerstone investors, meaning chipmakers are expected to support the fundraising at its foundation. The article does not give a fuller breakdown of the remaining $60 billion.

The size of the planned raise signals how capital-intensive Meta’s AI expansion is. The supplied article does not specify whether the rest will come from debt, additional equity, or another source. It does establish the key purpose: financing Meta’s push to expand AI, with major chip companies involved in the offering.

An underwritten equity offering is a share sale supported by financial institutions or investors that commit to buying the securities, including any portion not sold to the wider market. This gives the company greater certainty about how much money the sale will raise. Meta’s planned $35 billion offering is described as underwritten.

The key mechanism is ownership, not repayment. Meta would issue shares and receive capital from buyers. Those buyers would own part of the company and could benefit if its value rises. Existing shareholders would still own shares, but their percentage stake could shrink unless they also buy new shares.

Borrowing works differently. A loan or bond creates a liability that normally requires interest payments and eventual repayment. Equity does not create that same debt obligation, though it can reduce existing owners’ relative control. The article identifies the offering but does not name its underwriter or terms.

The $35 billion equity offering is a substantial but minority part of Meta’s planned $95 billion fundraising. Dividing 35 by 95 gives about 0.368, so the equity component represents approximately 36.8% of the total. It is slightly more than one-third of the announced amount.

Subtracting the equity offering from the total leaves $60 billion. That is the amount not covered by the specified $35 billion offering. The remaining sum could involve other financing methods or additional elements, but the supplied article does not identify them.

This split matters because it shows that Meta’s financing plan is broader than one share sale. The $35 billion figure is large on its own, yet it does not explain the whole $95 billion target. The article provides the totals but no further breakdown of the remaining financing.

A cornerstone investor is a major participant that commits to support a new offering near its launch. Its involvement can help anchor the deal, signal confidence, and encourage other investors to take part. The article identifies Nvidia, AMD, and Broadcom as cornerstone investors in Meta’s planned equity offering.

For example, these chipmakers could agree to buy shares or otherwise provide early backing under the offering’s terms. Their participation would connect Meta’s fundraising directly with companies supplying important AI technology. The article does not state the exact amount each chipmaker would invest or the precise terms of their commitments.

Their possible motivation is strategic as well as financial. Meta’s AI expansion could require more chips and supporting technology, potentially creating business opportunities for Nvidia, AMD, and Broadcom. Supporting Meta could strengthen commercial relationships. Those benefits are reasonable possibilities, but the article does not state the chipmakers’ specific reasons.

Issuing new shares increases the number of pieces into which Meta’s ownership is divided. Existing shareholders would still hold the same number of shares if they did nothing, but those shares could represent a smaller percentage of the company. This effect is called dilution.

Suppose an investor owns 1% before the offering. If new shares expand the total share count and that investor buys none, the investor’s percentage would fall below 1%. The same principle can reduce voting power, because voting rights usually follow share ownership. The precise impact depends on the number and type of shares issued.

Share value is less certain. New capital could support Meta’s AI growth and encourage investors, but issuing shares can also pressure the price if markets view dilution negatively. The article gives no share price, valuation, voting structure, or shareholder response, so the financial outcome cannot be determined from it.

AI expansion requires infrastructure that can perform huge numbers of calculations and move data quickly. Specialized chips provide the computing power, while data centers house the equipment. Networking systems connect the machines, and electricity keeps them operating. Each part is necessary because computing capacity cannot work in isolation.

During training, an AI system repeatedly processes large datasets and adjusts its internal parameters. That demands many chips working together for long periods. After training, operating the system for users also requires repeated calculations, especially when many requests arrive simultaneously. Faster networks help coordinate the machines and transfer data between them.

The article lists chips, data centers, networking equipment, and electricity as areas requiring large investment, but does not provide technical or cost details. Meta’s planned $95 billion raise shows the scale of its stated AI expansion. The spending could create demand for chipmakers such as Nvidia, AMD, and Broadcom.

AI computing infrastructure is the physical and technical foundation behind AI. It includes specialized chips, data centers, high-speed networking equipment, storage, and the electricity that powers them. Together, these resources let companies train models and operate them for users. Meta’s fundraising is intended to support an expansion that depends on this foundation.

Training typically involves processing large datasets repeatedly while adjusting a model’s parameters. That work is divided across many processors to finish faster. Once a model is deployed, each user request requires more calculations. Many simultaneous requests can therefore demand a large, constantly available computing system rather than one ordinary computer.

The source article specifically identifies chips, data centers, networking equipment, and electricity as major investment needs. It does not describe Meta’s systems or technical specifications. The broader implication is clear from the planned $95 billion raise: expanding AI requires infrastructure at substantial scale, with Nvidia, AMD, and Broadcom positioned as cornerstone investors.