Meta Launches $35 Billion Stock Sale in Drive to Raise Up to $95 Billion for AI Buildout
Meta Platforms launched about $35 billion of stock offerings on Monday and laid out a plan that could take its total equity fundraising to as much as $95 billion, as the Facebook parent turns to its own shareholders to pay for one of the largest industrial buildouts in corporate history. The move, announced on 26 October 2026, is a direct response to a question that has shadowed the company for weeks: how does a business that is already spending more than $130 billion a year on data centres, chips and energy actually pay for the artificial-intelligence future it is promising?
According to Reuters, Meta said on Monday it plans to raise up to $95 billion in equity to fund its costly expansion of data centres, cloud computing power and other infrastructure. The first stage, launched immediately, comprises roughly $22.5 billion of Class A common stock and roughly $12.5 billion of depositary shares representing mandatory convertible preferred stock. A second stage, a separate at-the-market share sale programme of up to $60 billion, would follow once Meta’s new non-voting Class C shares begin trading, which is expected in the first quarter of 2027.
The distinction between the two stages matters, and Meta itself drew it carefully. The $35 billion is underway. The $60 billion is a plan, not yet a signed programme. According to the company’s announcement, as reported by Dow Jones, Meta has not finalised the arrangement for the at-the-market programme, and its timing, size and terms have not been determined. That means the often-quoted $95 billion figure is a ceiling on what Meta could raise if both stages proceed in full, not cash already in the bank.
Even the first stage is structured in a way that spreads the money over time. According to reports of the company’s filing, the Class A shares are being sold through forward sale agreements, which means Meta will receive the cash when those agreements settle, on dates it chooses on or before 19 November 2027, rather than all at once on Monday. The preferred shares, by contrast, bring in proceeds at closing and are due to convert automatically into common stock around 15 August 2029. Meta has also granted underwriters options that could add up to $3.375 billion of Class A stock and 37.5 million additional depositary shares, according to reports of the offering terms.
One detail in the announcement drew particular attention on Wall Street: three of the chip companies that supply the very hardware Meta is buying have signalled they may buy some of the stock themselves. According to Reuters, Nvidia, Advanced Micro Devices and Broadcom have expressed interest in purchasing up to $5 billion of Meta’s Class A common stock in aggregate, on the same terms as other investors. That interest is non-binding, and none of the three is committed, but the symbolism is hard to miss. The suppliers of the AI boom are being invited to become part-owners of one of its biggest customers. Morgan Stanley and Goldman Sachs are leading the offerings as bookrunners, according to reports of the deal.
Meta said it will fund its AI investments through a mix of operating cash flow, investment-grade debt, strategic partnerships and equity capital, according to Reuters. That sentence, bland as it sounds, is the heart of the announcement. Until now, Meta had leaned heavily on debt. The equity route is new, large, and, for existing shareholders, permanent in a way that borrowing is not.
The timing is also striking. Only days earlier, on its third-quarter earnings call, Meta’s finance chief told analysts there was no specific financing plan to announce, describing the new share class as a way to reach a diverse pool of funding sources, according to reports of the call. Five days later, in the words of one report of the announcement, Meta put a plan worth up to $95 billion on the table. The speed of the reversal tells its own story about how quickly the company’s funding needs have hardened from a theoretical possibility into an immediate programme.
The numbers behind that urgency are formidable. According to Reuters, Meta raised the lower end of its annual spending forecast last week and now expects capital expenditure of between $135 billion and $145 billion this year, up from a previous range of $130 billion to $145 billion. That is spending on a scale that dwarfs almost every other company’s investment programme, and it comes on top of future commitments that Meta itself puts at about $825 billion, primarily driven by data centres, AI infrastructure and computing power, according to Reuters. That commitment figure compares with nearly $700 billion reported in July, a jump that shows how fast the buildout is accelerating.
At the same time, Meta’s own accounts show why the question of funding can no longer be deferred. According to Reuters, the company reported a $2.12 billion cash burn in the third quarter, its first negative free cash flow in more than ten years, compared with positive cash flow of $10.63 billion a year earlier. Its third-quarter revenue of $64.60 billion beat Wall Street estimates of $63.37 billion, according to Reuters, and it projected fourth-quarter revenue of between $73 billion and $76 billion. The business, in other words, is still growing strongly. It is simply spending faster than it is generating cash, and the gap is now measured in tens of billions of dollars.
Meta is not short of a balance sheet to start from. According to Reuters, it held cash, cash equivalents and marketable securities of $82.45 billion at 30 September, slightly more than its long-term debt of $80.92 billion. But $82 billion of cash does not go far against $135 billion to $145 billion of annual capital spending, let alone $825 billion of future commitments. Something had to give, and on Monday Meta chose equity.
The market’s first reaction was unsettled, which is itself informative. Reports of premarket trading showed Meta shares rising by more than 1 per cent, and in some accounts by more than 2 per cent, as investors welcomed the clarity of a funding plan. The Wall Street Journal, however, reported that the stock fell by about 1 per cent as the market opened on Monday. The two accounts are not necessarily contradictory: a stock can rise before the open and fall once trading begins, and early trading in a $1.7 trillion company on the day of a $35 billion offering is rarely tidy. What is clear is that Monday’s move was modest compared with the reaction the week before, when the share-class plan itself was unveiled.
That earlier reaction was severe. According to reports of the October trading session, Meta shares fell by nearly 7 per cent after the company announced the creation of the Class C shares alongside its third-quarter results, erasing approximately $124 billion of market value in a single session. The Nasdaq fell 2.6 per cent the same day, its biggest one-day fall since the start of June, according to one report, as the sell-off spread to other technology stocks. Nvidia fell 4.4 per cent that day, according to the same report, and the Philadelphia semiconductor index dropped 5.6 per cent. The message from investors was unmistakable: strong revenue growth was no longer enough on its own. Investors wanted to know how the AI buildout would be paid for, and what it would cost existing shareholders.
To understand Monday’s announcement, it is necessary to understand the Class C shares that make the second stage possible. According to Barron’s, Meta plans to issue non-voting Class C shares to holders of its Class A and Class B stock in what amounts to a two-for-one stock split, due to take effect in January 2027. Investors will receive one Class C share for every share they already hold. The distribution itself does not dilute anyone’s proportional ownership, and it does not affect founder and chief executive Mark Zuckerberg’s voting control. That is the point.
According to Barron’s, Zuckerberg holds a roughly 14 per cent economic stake in Meta but a 60.6 per cent voting interest, through his ownership of 342 million super-voting Class B shares carrying ten votes each. The Class C shares carry no votes at all. That means Meta can sell large quantities of Class C stock to the public in future, through the planned at-the-market programme and in other offerings, without reducing Zuckerberg’s control by a single vote. It is a structure modelled on the one Google, now Alphabet, created in 2014, when it issued non-voting Class C shares through a similar split to preserve the control of founders Larry Page and Sergey Brin, according to Barron’s.
For ordinary shareholders, the trade-off is real and deserves to be stated plainly. Debt has a maturity date and can be repaid. Equity dilution, once new shares are sold, is permanent unless the company later buys those shares back with surplus cash, which Meta is not currently generating. Every Class C share sold in the at-the-market programme will reduce existing shareholders’ proportional claim on Meta’s future earnings, even though it leaves their voting power, such as it is, unchanged. That is why some analysts described the Class C creation as the moment shareholders realised they were no longer just riding the AI buildout, but financing it.
There are limits on Zuckerberg’s own sales of the new shares, according to a report of the company’s securities filing. He can freely sell 20 per cent of the Class C shares he receives. To sell more, he generally must also sell a Class A or Class B share, or convert a Class B share, for each one, although a block equal to 2.5 per cent of his Class C shares is released from that rule for each quarter he remains in a full-time leadership role. The arrangement is designed to reassure investors that the founder cannot quietly cash out of the new share class while asking them to fund the company through it.
Monday’s announcement also lands in a week when the whole market is asking the same funding question. According to Reuters, global shares edged up on Monday ahead of a week of central bank meetings and earnings from Alphabet, Microsoft and Amazon that will further test confidence in the AI boom. The Federal Reserve is widely expected to hold its key interest rate steady on Wednesday, according to reports, while traders price in a roughly 80 per cent chance of a rate rise in December. Oil eased from six-month highs, according to Reuters, after reports that Qatari mediators were in Tehran on Sunday as part of an effort to de-escalate tensions. Against that backdrop, a $95 billion equity plan from one of the world’s largest companies is not a side story. It is one of the main events of the week.
It is also part of a pattern. According to TheStreet, citing Bloomberg and the Financial Times, Meta’s full $95 billion, if executed, would top Alphabet’s $90 billion June offering, which lead bookrunner JPMorgan called the largest multi-tranche equity transaction in history. At about $654 a share, according to TheStreet, the full $95 billion represents roughly 5.7 per cent of Meta’s market value. The comparison matters because it shows that Meta is not an outlier. The largest technology companies are all discovering, at roughly the same time, that the AI buildout cannot be funded from cash flow and debt alone. Equity, the most permanent form of capital, is now on the table.
One further detail in the reports is worth noting for what it says about the pace of events. According to one report, citing Semafor, Monday’s financing marks the first time Meta has issued new stock since 2013. Whether or not that precise claim survives closer scrutiny, the broader point stands: a company that has spent more than a decade funding itself from advertising profits and debt has now decided it needs to ask the stock market for tens of billions of dollars. That is a structural change in how Meta finances itself, not a routine treasury operation.
Analysis: Why It Matters
1. The AI race now has a price tag, and shareholders are being asked to pay it
For two years, the AI debate has been conducted mostly in the language of capability: bigger models, more computing power, smarter assistants. Meta’s announcement translates that debate into the language of capital, and the translation is sobering. A single company expects to spend $135 billion to $145 billion in one year, has committed itself to about $825 billion of future spending, and is now raising up to $95 billion in new equity to help pay for it. Those figures, taken together, suggest that the AI buildout is not a normal investment cycle that can be absorbed within a year’s profits. It is a multi-year claim on the savings of the entire stock market, and Meta has just made that claim explicit. The original analysis here is straightforward: investors can no longer evaluate Meta, or its peers, on revenue growth alone. They must now evaluate them as capital-raising machines, and judge whether the returns on data centres and chips will ever justify the dilution required to build them.
2. The Class C structure solves Meta’s control problem, not its shareholders’ dilution problem
Much of the commentary on the Class C shares has focused, understandably, on Mark Zuckerberg’s voting control. That is a real governance issue, and the comparison with Alphabet’s 2014 structure is instructive. But the more important economic question is different. Non-voting shares do not make dilution painless; they make it voiceless. Existing shareholders will own a smaller slice of Meta’s earnings after the at-the-market programme, and they will have no additional vote with which to object. The structure is elegant from the company’s point of view, because it separates the raising of capital from the exercise of control. From the shareholder’s point of view, however, that separation is precisely the risk. The lesson of Monday’s announcement is that governance design and economic cost can move in opposite directions, and investors should not mistake the former for a solution to the latter.
3. The chipmakers’ interest is a signal about the whole supply chain, not just Meta
Nvidia, Advanced Micro Devices and Broadcom have not committed to buying Meta stock, and their indicated interest of up to $5 billion is small beside a $35 billion offering. It would be a mistake, however, to dismiss it as a footnote. When a company’s suppliers offer to become its shareholders, they are expressing a view about the durability of demand for their own products. They are, in effect, saying that Meta’s data-centre spending is not a bubble that will vanish next quarter, but a programme worth having a financial stake in. That is a useful counterweight to the scepticism visible elsewhere in the market, where commentators have warned about debt-fuelled AI spending and questioned whether returns will match investment. The truth probably lies between the two views, but the suppliers’ signal deserves weight precisely because they have the most to lose if the buildout stalls.
4. Meta’s move raises the stakes for every other hyperscaler
According to TheStreet, Meta’s own filings flagged a desire to move before potential equity offerings from Alphabet, SpaceX, Anthropic and OpenAI. That competitive logic is worth taking seriously. If Meta can raise $95 billion, it can outspend rivals on data centres, energy contracts and chip supply at exactly the moment when those inputs are scarcest. Rivals that hesitate may find themselves paying more later, or settling for less capacity. Rivals that follow Meta into the equity market will, in turn, test how deep investors’ appetite for AI dilution really is. Either way, Monday’s announcement is likely to accelerate, not settle, the funding race. The next earnings reports from Alphabet, Microsoft and Amazon, due this week, should be read with that race in mind, not merely as reports on their own quarters.
5. The real test is not the offering, but what the money builds
Stock offerings are easy to announce and hard to judge. The $35 billion launched on Monday will be judged, in the end, not by how smoothly it is sold, but by whether the data centres, computing capacity and AI products it funds generate returns above their cost. That is a test that will take years, not quarters, and it will be complicated by the fact that Meta’s competitors are building similar capacity at the same time. If AI capabilities continue to improve rapidly and find paying uses at scale, today’s dilution may look cheap. If capacity outruns demand, or if model improvements slow, today’s dilution may look like the moment shareholders funded an overbuild. Neither outcome can be known on Monday. What can be known is that Meta has now committed itself, in public and at scale, to the view that building first and worrying about returns later is the right strategy. That commitment, more than any single figure in the offering, is the story.
What to Watch Next
- The pricing and demand for the $35 billion of offerings in the coming days, including whether the underwriter options are exercised and whether Nvidia, Advanced Micro Devices and Broadcom convert their indicated interest into actual orders.
- Whether Meta signs a formal agreement for the $60 billion Class C at-the-market programme, and what size, timing and terms it announces before the shares are expected to begin trading in the first quarter of 2027.
- The earnings reports from Alphabet, Microsoft and Amazon this week, and any signal they give about their own capital spending, funding plans or willingness to follow Meta into large equity offerings.
- Meta’s fourth-quarter results and its capital spending guidance for 2027, which will show whether the $135 billion to $145 billion forecast for this year was a peak, a plateau or merely a staging post.
- How the Federal Reserve’s decision on Wednesday, and the path of interest rates into December, affects the cost of debt for the whole technology sector, and therefore the balance Meta and its peers strike between borrowing and selling shares.
Sources
- Reuters: Meta to raise up to $95 billion in equity to fund costly AI expansion
- Dow Jones via Morningstar: Meta Plans $35 Billion Stock Offering to Power AI Buildout, Eyes $95 Billion Total Plan
- The Wall Street Journal: Meta Plans up to $95 Billion in Financing to Power Its AI Growth
- TradingView: Meta’s $35 Billion Stock Sale Draws Nvidia, AMD and Broadcom