Meta launched approximately $35 billion in stock offerings on October 26, with a further $60 billion share-sale program planned for later. The Meta AI funding plan puts a striking relationship in view: Nvidia, AMD and Broadcom have indicated interest in buying shares in a company that also buys their chips.
The three suppliers’ interest totals up to $5 billion collectively. Meta’s announcement explicitly says those indications are nonbinding: purchases could be larger, smaller or absent. None of that amount should be treated as committed cash.
What the Meta AI funding plan actually includes
The initial offerings combine about $22.5 billion of Class A common stock with $12.5 billion of depositary shares representing preferred stock that must eventually convert into common shares. The larger, future program would sell Class C stock into the market over time. Meta expects that program no earlier than the first quarter of 2027 and says it has no agreement establishing it yet.
Even the initial headline amount has a timing wrinkle. The preliminary prospectus describes forward sales in which intermediaries borrow existing shares and sell them. Meta generally receives proceeds when those arrangements settle, rather than immediately at closing. Expected settlement runs through November 19, 2027, subject to the agreements’ conditions.
More capital, familiar control
El País explains that the planned nonvoting Class C shares let Meta raise capital while preserving Mark Zuckerberg’s voting control. Buying an economic stake and gaining influence over corporate decisions are different propositions.
Meta forecasts $135 billion to $145 billion of capital expenditure this year. Its financing announcement presents equity alongside operating cash flow, debt and partnerships as ways to support that investment. Those spending plans remain forecasts, and fundraising alone cannot establish that the resulting infrastructure will earn adequate returns.
Reuters highlights the concern about circular financing: chip suppliers may help finance a major buyer’s expansion. That overlap does not itself prove improper conduct or artificial demand. It does make the distinction between sales, investments and binding commitments consequential.
The next test is execution: whether investors subscribe, whether the future program launches, and whether the computing capacity produces sustainable revenue. For readers tracking the AI race, the funding structure now deserves as much attention as the chips.
