Meta Pre-Commits $95 Billion of Equity With No Cash in Hand — and the Buyers Are Its Own Chip Vendors

Meta did not sell stock for cash on October 26. It sold a promise of stock: $22.5 billion of Class A common through forward sale agreements and $12.5 billion of depositary shares representing mandatory convertible preferred, alongside a $60 billion at-the-market program on a share class that does not yet legally exist. The $95 billion headline is a ceiling, not a receipt.

The third quarter of 2026 is where the arithmetic turned. Capital spending of $33.94 billion exceeded operating cash flow of $32.79 billion, so free cash flow printed negative $1.16 billion for the first time — with full-year capital spending guided to $135–145 billion against trailing operating cash flow of $133.1 billion.

Meta has already promised $454.52 billion it cannot walk away from, including $355.73 billion of data center, energy and colocation leases that have not even begun, with $121.41 billion falling due in 2027 alone. The $95 billion raise covers about a fifth of the committed book and leaves the remainder to be funded from operations, further leases, or more paper.

Moody's is explicit about why this matters. Its July 2026 review put hyperscaler capital spending at $785 billion for 2026 and about $1 trillion for 2027, tracked roughly $460 billion of direct debt across six large buyers, and counted $1.2 trillion of lease commitments as debt-equivalent. It placed Oracle on Baa2 with a negative outlook and CoreWeave at Ba3, while rating Meta's balance sheet among the strongest in the group. Equity protects a rating that debt would put at risk.

The timing matters as much as the choice. The 10-year Treasury yield reached 5.33% on October 8, 2026, levels last seen in 2002, then fell 17 basis points to 4.972% on October 22, the first sub-5% print since September 23. The buyers were not coming for growth; they were fleeing the sheer volume of AI-related bond issuance. The equity window opened because the credit window shut.

There is a second reason equity was tolerable: Meta's shares trade at 21.9 times trailing earnings, and the forward sale and at-the-market program let Meta choose the price and the date rather than accept one. Meta is selling equity like an option on its own share price, issuing into strength through November 2027 rather than at a fixed level today.

The headline detail is that Nvidia, AMD and Broadcom have each indicated interest in buying up to $5.0 billion of Meta's Class A stock at the same price and terms. On paper that closes a loop: the companies whose revenue depends on Meta's capital budget now hold Meta stock. In dollars, the loop is far too small to fund anything — $5.0 billion is 5.3% of the raise and 0.07% of the three vendors' combined $7.6 trillion market value.

That reframes the event. The $5 billion is not circular financing; it is the three companies with the clearest read on Meta's spending plans taking equity exposure to the returns on that spending, at a cost that is immaterial to their own balance sheets. The $5 billion signals confidence rather than filling a funding gap, which is why the market read the announcement as reassuring rather than alarming.

Class C is the reason this could be done without a governance fight. Meta's board approved a new non-voting Class C common stock on October 19, 2026 and shareholders consented on October 20; the restated charter takes effect January 19, 2027. Every Class A and Class B holder receives one Class C share as a stock dividend, which doubles the economic share count and halves the price — a two-for-one split delivered in a form that can be issued to vendors, used for employee compensation, or spent on acquisitions. The class turns a capital raise into a tradable instrument.

The cleanest way to see what Meta's equity buys is to look at the company doing the same work with debt. CoreWeave has $21 billion of contracted revenue from Meta running to December 2032 and another $14.2 billion signed in September 2025, on trailing revenue of just $7.59 billion. It funds the buildout with borrowings: net debt at 11.5 times EBITDA, debt at 10.3 times equity, and trailing free cash flow of negative $11.8 billion. Its shares closed at $55.19 against a 52-week high of $141.79. Meta funds the same buildout with equity while the debt-funded version sits 61% below its high.

Over the next two quarters, three dates matter more than the headline. Class C begins trading in late January 2027, the $60 billion at-the-market program opens in the first quarter, and the forward sale agreements can settle on any date Meta picks through November 19, 2027. Until then, none of the $95 billion is in the bank, and free cash flow has to service $121.4 billion of commitments falling due in 2027.

The one-to-three-year question is whether Meta's revenue can carry the balance sheet it is now building. Revenue growth ran 33.1% in the first quarter of 2026, 28.0% in the second and 26.1% in the third, while the capital budget was raised twice. The offsetting evidence is early but real: more than 10 million people in the US and Canada have set up Meta's Muse personal agents in the past month and a half, and Meta now charges developers $50 and $200 a month for access to its most capable Muse Wonder Full model. The equity raise is a bet that advertising and AI subscriptions outrun $454.5 billion of commitments.

Meta has done something no hyperscaler has done: it has asked the three companies that depend on its capital budget to hold its stock, and it has built a new non-voting share class to make that possible without surrendering control. The dollar amount is trivial and the information content is not. Meta is funding the AI buildout with equity because its operating cash flow can no longer do it, and the suppliers are telling the market they believe the returns will arrive.