Meta Unveils 2-for-1 Stock Split and New Share Class to Fund Its AI Spending Spree

Meta Platforms is getting creative to fund its massive investments in artificial intelligence. The company reported record third-quarter revenue on October 21, 2026 — and announced it will create a new class of nonvoting stock, effectively a 2-for-1 stock split, to raise the capital it needs for developing and serving AI models. The move lets Meta tap public markets for AI funding while ensuring founder Mark Zuckerberg keeps his grip on control.

The announcement, reported by The Wall Street Journal and Dow Jones Newswires, marks one of the most aggressive financial-engineering moves by a Big Tech company in the AI era — and a vivid illustration of just how expensive the race to build artificial general intelligence has become.

What Meta announced

Meta said it will create a new Class C common stock — publicly listed but nonvoting — to be issued as a one-time stock dividend to holders of common and supervoting shares, with an effective date of January 19, 2027. The distribution will function as an effective 2-for-1 stock split: shareholders will hold twice as many shares, with the new ones carrying no votes.

Crucially, the new stock "won't affect Chief Executive Mark Zuckerberg's voting control or the voting power of other shareholders," as Dow Jones summarized. "The creation of the Class C common stock will continue the founder-led model that has driven the company's success since its inception," Meta said in its earnings release.

The purpose is straightforward: the new shares give Meta a fresh currency to raise money, pay employees, and fund acquisitions — all without diluting Zuckerberg's control. In an era when AI talent and compute cost billions, that currency is strategic.

Record revenue, dented profit

The financial results behind the announcement were a study in contrasts. Meta brought in $64.6 billion in revenue for the third quarter, a 26% year-over-year increase and a new record — beating analyst expectations surveyed by FactSet. Daily active people across its app family grew 3% to 3.64 billion, underscoring the continued strength of the core advertising business.

But net income of $9.8 billion ($3.83 per share) came in considerably below analysts' forecast of $17.2 billion ($5.46 per share). The culprit: a $9.89 billion legal charge related to the company's August settlement with 52 US state attorneys general over teen mental-health and social-media litigation — a settlement providing for payments of up to about $18.59 billion. Excluding that charge, earnings per share would have been $7.40 and operating income $22.88 billion.

For the current quarter, Meta expects revenue of $73 billion to $76 billion, the high end above analyst expectations. It also raised the low end of its 2026 capital-expenditure range to $135 billion (from $130 billion), with the high end unchanged at $145 billion. Finance chief Susan Li declined to give a 2027 capex estimate but said leaders "expect to continue investing aggressively" in both the core business and AI.

Heavy data-centre spending pushed free cash flow into negative territory — negative $2.1 billion — for the first time since Meta became a public company, though that was better than analysts' projected negative $9.4 billion. Meta's shares rose 3-5% in after-hours trading, to around $726 from a $697.52 close.

Why a new share class?

The Class C structure solves a specific problem. Meta needs staggering amounts of capital for AI infrastructure — data centres, chips, and talent — but issuing ordinary shares would dilute existing holders' voting power and, eventually, Zuckerberg's control. Nonvoting shares let the company raise equity without touching the control structure.

Google set the precedent in 2014 with its own Class C nonvoting shares, and the structure has since become the standard tool for founder-controlled tech giants that want to spend big without ceding power. Critics argue it entrenches management and weakens shareholder accountability; supporters say it lets founders pursue long-term bets — like AI superintelligence — without quarter-to-quarter market pressure.

Zuckerberg framed the quarter in AI terms: "We hit major milestones this quarter toward delivering personal superintelligence for everyone — including launching Muse and Muse Wonder, our first frontier model," he said in a statement. The company's newly launched Muse AI assistant surpassed 10.9 million downloads within 40 days of its September launch, according to market data.

The AI spending race

Meta's move lands in a week when the scale of AI economics dominated markets. Anthropic's draft IPO filing revealed $80.7 billion in first-half losses and compute commitments that could reach $750 billion. Meta's own capex — up to $145 billion this year alone — shows the incumbents are spending at a similar order of magnitude.

The open question is whether any of this spending earns an adequate return. Meta's core advertising business is funding the AI build-out for now — $64.6 billion quarters do not happen by accident — but investors are watching closely for evidence that AI investments convert into revenue growth rather than just costs. The market's positive reaction to the earnings suggests investors, for now, believe the bet.

There is also a governance debate embedded in the Class C decision. By design, the new shares give public investors economic exposure without voice. As Meta's AI bets grow larger and more consequential — spanning frontier models, data centres, and potentially acquisitions — the question of who gets a say in those decisions matters more, not less.

What happens next

The Class C dividend takes effect January 19, 2027. Between now and then, Meta will need to file the relevant paperwork, and investors will parse the fine print of the new share class. The company's 2027 capex plans — still undisclosed — will be the next major signal of how far Zuckerberg intends to push the AI build-out.

For the broader market, Meta's quarter is a Rorschach test for the AI trade: record revenue funding record spending, legal overhangs denting profits, and financial engineering to keep the machine running. Whether that machine produces superintelligence — or just super-sized capital expenditures — is the trillion-dollar question. Related reading: Meta ordered to pay $25 billion in the New Mexico privacy case, Meta's Muse Wonder frontier AI model, and Meta pausing Muse chat sharing over exposed conversations.

Not every big stock move this earnings season was about results: Moderna lost a quarter of its market value after its cancer vaccine data leaked early.