Meta Platforms said on October 27, 2026 that it raised the size of its planned equity fundraising to $97.5 billion to pay for AI infrastructure expansion, according to Reuters coverage carried by TradingView. The package bundles a $25 billion common stock sale, a preferred-style depositary share offering, and a separate at-the-market program of up to $60 billion that has not started yet.
The Meta equity raise lands as one of the largest share-sale plans a technology company has put on the table. It also shows where the AI spending race is heading. Big Tech once paid for data centers out of cash flow. Now it is selling stock, carving out a new share class, and signing chipmakers as shareholders.
One of those chipmakers already has a name. Nvidia said it will put $3 billion into the offering, a move that builds on the relationship we covered in Nvidia’s $3B stake and the Muse agent pact. This analysis walks through the structure, the unanswered questions, and what the numbers mean for investors, rivals, and the wider AI infrastructure market.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Meta Announced on October 27
Meta Platforms increased its planned equity fundraising to $97.5 billion, per Reuters and CNBC reporting on the day. The company framed the money as support for the expansion of its AI infrastructure. It said it had launched public offerings of common and preferred stock totaling approximately $35 billion, then upsized the common portion before pricing.
Headline totals varied by outlet, and readers will see all of them. The Straits Times headline reads US$95 billion, while the Bloomberg Law headline says almost $98 billion. Reuters and Profit by Pakistan Today both use $97.5 billion. The spread reflects an offering that was upsized while the news cycle ran, so the $97.5 billion figure is the one to trust for the final package.
One caution before going further. Meta has announced a plan, not a completed raise. The $97.5 billion total includes a future at-the-market program whose terms the company has not finalized. Any article claiming Meta has already collected $97.5 billion misreads the filings.
How the $97.5 Billion Breaks Down
The cleanest way to read the Meta equity raise is as two buckets. The first bucket is the near-term public offerings, which have priced. The second is the at-the-market program, which waits on a new share class that does not trade yet. The table below lays out each piece, using the figures Meta and Reuters published and simple arithmetic where noted.
| Component | Size | Key term | Status |
|---|---|---|---|
| Common stock offering | $25 billion (up from $22.5 billion) | $625.00 per share | Priced |
| Implied common shares (our math) | About 40 million | $25B divided by $625 | Derived |
| Depositary shares | 250 million shares, about $12.5 billion (our math) | $50 per share | Priced |
| Near-term offerings combined (our math) | About $37.5 billion | Launched at about $35 billion before the upsize | Priced |
| At-the-market program | Up to $60 billion | Starts after Class C shares trade, expected Q1 2027 | Terms not finalized |
| Nvidia investment | $3 billion | Part of the equity offering | Announced by Nvidia |
| Nvidia, AMD, Broadcom interest | Up to $5 billion | Class A shares | Reported potential interest |
The arithmetic checks out against the headline. The $25 billion common sale plus the $12.5 billion in depositary shares gives $37.5 billion. Add the $60 billion program and the total reaches $97.5 billion. Roughly 61.5% of the package, then, depends on a program that has no start date beyond “after Class C shares begin trading.” Only about 38.5% has priced.
The Common Stock Offering at $625 a Share
Meta lifted the common stock portion from $22.5 billion to $25 billion, an increase of $2.5 billion or about 11%. It priced the shares at $625.00 each. At that price the company needs to issue roughly 40 million new shares to raise $25 billion. That is simple division, and Meta has not published a share count in the material we reviewed.
Upsizing a deal before pricing usually signals strong demand from the order book. We cannot confirm that reading from the reporting, because neither Reuters nor CNBC quoted the underwriters. What we can say is that Meta chose to take more money at the same time it announced the larger plan. That choice tells investors the company sees no shortage of buyers, or at least wants to test that claim in public.
For existing holders, the dilution math is the question. Forty million new shares against Meta’s total share count is a small slice, but the common offering is only one leg. The at-the-market program, at up to $60 billion, would add far more shares over time if executed in full. Investors should model the full $97.5 billion, not just the headline offering, when they estimate ownership dilution.
The $50 Depositary Shares and the Preferred Tranche
Meta priced 250 million depositary shares at $50 per share. Depositary shares usually represent a fractional interest in a preferred stock, which is why Meta describes the launch as common and preferred offerings. Multiplying the share count by the price gives $12.5 billion, which is our calculation and not a figure Meta quoted in the reports we reviewed.
Preferred stock sits between debt and common equity. It typically pays a fixed return and ranks ahead of common shareholders in a payout, but behind bondholders. For a company that wants to fund AI infrastructure without leaning only on dilution or only on debt, a preferred tranche widens the investor base. Income-focused buyers who skip growth stocks may still buy a $50 depositary share.
The reporting we reviewed did not give the dividend rate, the redemption terms, or the conversion features of the depositary shares. Those details decide how expensive this capital is. Until Meta publishes the prospectus supplement terms, any claim about the cost of the preferred tranche would be a guess, and we are not making one.
The $60 Billion At-the-Market Program
The biggest single piece of the Meta equity raise is also the least defined. Meta plans a separate at-the-market share-sale program of up to $60 billion. In an at-the-market program, a company sells shares gradually into the open market at prevailing prices through agents, rather than in one priced block. That gives management flexibility on timing, and it avoids the one-day price pressure of a giant secondary.
Why the program waits for Class C shares
Meta said the larger program is expected to begin after its new non-voting Class C shares start trading. The company expects that to happen in the first quarter of 2027. So the earliest the at-the-market sales can begin falls roughly three to five months after today’s announcement, and the timing depends on a listing event outside the offering itself.
What Meta has not finalized
Meta stated that the timing and terms of the $60 billion program are not final. That matters for anyone building a model on the headline number. “Up to” $60 billion is a ceiling, and the company can sell less, sell slower, or change the structure before it starts. Sixty billion dollars of potential supply also hangs over the stock for months, which is a different kind of market pressure than a one-time offering.
Why Class C Non-Voting Shares Matter
The Class C shares may prove the most consequential piece of the announcement, even though they carry no votes. Meta already runs a dual-class setup with Class A and Class B stock. Adding a non-voting class means Meta can sell a very large amount of equity without handing out any additional voting power. Whoever controls the vote today keeps it after the sale, because Class C holders get no say in elections or proposals.
That design carries a trade-off. Non-voting shares are easier for a company to issue and harder for governance-focused investors to love. Index providers and large asset managers have debated how to treat non-voting listings for years, and some prefer to hold shares that carry a vote. Meta is betting that the demand for exposure to its AI buildout outweighs that preference.
Investors will also want to know how Class C shares trade against Class A. A non-voting class sometimes trades at a discount to the voting class, though the size of any gap varies by company and market. Meta has not given guidance on pricing parity in the reporting we reviewed. The first-quarter 2027 debut will answer the question with real trades, and the answer will shape how cheaply Meta can run the $60 billion program.
Nvidia, AMD and Broadcom: Suppliers Become Shareholders
Nvidia said it will invest $3 billion in Meta’s equity offering. Reports also described potential interest from Advanced Micro Devices and Broadcom, with the three chip companies together totaling up to $5 billion in Class A shares. Only Nvidia’s $3 billion is a stated commitment. The AMD and Broadcom interest, and the $5 billion combined ceiling, remain reported and not confirmed.
The circular nature of this money deserves a plain description. Meta buys GPUs and custom silicon from these vendors. The vendors then buy Meta stock with some of the revenue. Meta uses the proceeds to place bigger chip orders. The loop is legal and common in capital-intensive industries, but it ties the fortunes of customer and supplier closer together. If AI demand slows, both sides feel it at once.
The chipmaker’s valuation and margins have been under the microscope, as we noted in our look at Nvidia’s forward P/E and the custom-chip threat. A $3 billion equity stake in a top customer is one way to lock in demand while custom chips from rivals gain ground. For Meta, the stake brings a vote of confidence from the company that supplies a large share of AI accelerators.
What the Money Is For, and What Remains Unconfirmed
Reports describe the purpose as expansion of Meta’s AI infrastructure. Nothing in the reporting ties the proceeds to a named model, a specific data center, or a single product. Headlines that promise a particular project are running ahead of the evidence. Infrastructure generally covers servers, networking, power, land, and the buildings that house them.
| Claim | Status | What the reporting supports |
|---|---|---|
| Meta plans $97.5 billion in equity funding | Confirmed | Reuters and CNBC report the announced plan |
| Meta has already raised $97.5 billion | Unconfirmed | The package includes a future at-the-market program |
| Common stock priced at $625.00 per share | Confirmed | Reported pricing for the upsized $25 billion sale |
| Nvidia invests $3 billion | Confirmed | Nvidia said it will invest in the offering |
| AMD and Broadcom will buy shares | Unconfirmed | Reported as potential interest, up to $5 billion with Nvidia |
| Proceeds fund a specific model or data center | Unconfirmed | Reporting names AI infrastructure only |
Physical buildout carries side effects that investors and communities both watch. Data centers draw heavily on power and water, a topic we dug into in our explainer on why data centers use 2M+ gallons of water daily. The more capital Meta pours into campuses, the more local permitting fights it will face. A large equity raise cannot buy its way past a zoning board.
Historical Context: How Big Tech Has Paid for AI So Far
For most of its history, Meta funded itself from advertising cash flow, and it returned surplus cash through buybacks. Large equity issuance was rare for the company. The AI buildout has changed the calculus across the industry. Large platforms have leaned on operating cash flow first, then added bonds, partnerships, and other outside capital as the bills grew.
The October 27 announcement moves equity from a footnote to a headline. A raise of this scale sits well beyond the follow-on offerings that mature tech companies typically run, and it pairs a share sale with two structural changes: a new non-voting class and strategic buyers drawn from the chip supply chain.
The pattern has precedent inside this very cycle. We tracked the early signs in our piece on how the AI infrastructure race echoes earlier build-outs, where heavy capital spending ran ahead of proven revenue. Earlier network buildouts, such as the fiber boom of the late 1990s, taught the same lesson: the buildout itself can prove worthwhile while the financing terms decide which companies survive the digestion phase.
Market Impact: Dilution, Supply, and Signals
Expect three channels of market impact. The first is share supply. The $37.5 billion in near-term offerings adds supply now, and the $60 billion program adds it later. Investors price in both. Investor’s Business Daily’s headline on the earlier $95 billion plan said Meta stock fell, a sign that the market read the news as a dilution event before it read it as a growth signal.
The second channel is the signal about capital intensity. A company that sells $97.5 billion of stock is telling the market that internal cash flow no longer covers its AI ambitions. Bulls read that as conviction in returns, since management would not sell stock at $625 unless it expected to earn more than the cost of that equity. Bears read it as proof that returns have not yet arrived.
The third channel is read-through to the chip supply chain. When a top buyer raises a record amount for infrastructure, orders at Nvidia, AMD, Broadcom, and memory and networking suppliers have a funded path. McKinsey’s value-curve analysis, which we covered in the $2.52T prize that shrinks by $850B if costs rise 30%, shows how sensitive the AI payoff is to cost assumptions. Equity funding lowers Meta’s financial risk, but it does not lower the cost of the hardware.
What Industry Data Says About the Spending Curve
Meta’s raise makes more sense against the industry forecasts. The numbers below come from IDC and Gartner, and each was published before today’s announcement. They describe the market Meta is financing its way into, not Meta’s own plans.
| Metric | Figure | Source | Year published |
|---|---|---|---|
| Worldwide AI infrastructure spending, Q3 2025 | $86 billion | IDC AI Infrastructure Tracker | 2025 |
| Projected full-year 2025 AI infrastructure spending | $334 billion | IDC AI Infrastructure Tracker | 2025 |
| Forecast AI infrastructure spending by 2029 | $758 billion | IDC AI Infrastructure Tracker | 2025 |
| Share of 2029 spending on accelerated servers | 94.3% | IDC AI Infrastructure Tracker | 2025 |
| Enterprise AI workloads on hybrid infrastructure by 2028 | 75% | IDC | 2025 |
| AI-optimized vs traditional server spending by 2027 | About 3x | Gartner | 2025 |
IDC’s tracker states that, in its words, “Worldwide spending on AI infrastructure reached $86 billion in the third quarter of 2025, and full-year 2025 spending was projected to reach $334 billion.” That run rate is the backdrop for a single company seeking $97.5 billion.
Looking further out, IDC wrote that “Global AI infrastructure spending was forecast to reach $758 billion by 2029, with accelerated servers representing 94.3% of total market spending.” If that forecast holds, accelerated servers (the GPU and custom-chip systems Nvidia, AMD, and Broadcom sell) take nearly the whole market, which explains why those three vendors want a seat at Meta’s table.
Gartner offered a similar read on the server mix, stating that “AI-optimized server spending was expected to triple traditional-server spending by 2027.” And in a separate IDC brief on datacenter and cloud investment, the analyst firm said that “By 2028, 75% of enterprise AI workloads were forecast to run on hybrid, fit-for-purpose infrastructure.” Both projections point the same direction. Capital intensity keeps climbing, and the winners own or fund the most compute.
Competitive Comparison: How Meta’s Raise Stacks Up
Meta is not alone in reaching for outside capital, but its route stands apart. Anthropic chose the IPO path. Our coverage of Anthropic’s S-1 filing showed an $80.7 billion loss against $16.3 billion in sales, a profile that needs public-market money to keep training and serving models. Meta’s profile is the opposite. It is a profitable advertising giant selling stock because AI spending outruns even its cash flow.
Debt is the other lever. In Musk’s claim that SpaceX closed a $40B GPU debt deal, the financing came as borrowing, not a share sale. Debt keeps ownership intact but adds fixed payments. Equity costs ownership but adds no repayment schedule. Meta is using both, and its non-voting Class C design protects control in a way that a straight debt raise would not need to.
The contrast shows three paths to the same goal. Startups with heavy losses list on an exchange. Companies with strong collateral borrow against it. Cash-rich platforms with entrenched control sell non-voting stock and let chip vendors anchor the book. Each path has a different failure mode. IPO buyers can walk away, lenders can call a loan, and equity buyers can force a lower price if the stock sags during an at-the-market program.
Five Predictions for the Meta Equity Raise
These are our own assessments based on the structure announced today. They are not reported facts, and any of them could miss.
- The near-term offerings close without drama. The common and depositary shares have already priced, so closing is mostly mechanical. Watch the prospectus for the dividend terms on the depositary shares.
- The Class C debut sets the real price of the $60 billion program. If the new shares trade near Class A levels, the program gets cheaper to run. A wide discount would push Meta to slow or resize it.
- The program ends up smaller than $60 billion or slower than planned. “Up to” language gives Meta cover, and sustained selling into a falling stock would defeat the purpose.
- Other mega-cap spenders study the template. A non-voting class plus strategic chip-vendor buyers solves a control problem and a demand problem at once, so expect rivals to explore similar designs.
- Governance critics file objections. Pension funds and proxy advisers have long objected to non-voting structures. Expect public letters before the Class C listing, and possibly index-eligibility questions.
What to Watch Next
The next few days bring the closing of the priced offerings and the first trading reaction in full. A busy earnings calendar adds context, and we laid out the AI-related names in our preview of Big Tech earnings on October 28-29. Commentary from other hyperscalers on their own funding plans will show whether Meta’s equity route is a one-off or a trend.
Beyond that, track three documents. The prospectus supplement will spell out the preferred terms. The Class C registration will set the first-quarter 2027 timeline. And any updated Meta filing will say whether the at-the-market terms have been finalized. Our wider AI and machine learning coverage will track each step as it lands.
FAQ: Meta’s $97.5 Billion Equity Raise
How much is Meta trying to raise? Meta announced a planned equity fundraising of $97.5 billion on October 27, 2026. That figure combines near-term common and preferred offerings with an at-the-market program of up to $60 billion.
Has Meta already collected the full $97.5 billion? No. The reporting describes a planned package. Only the near-term offerings have priced, and the $60 billion program has not started.
At what price is Meta selling common stock? The upsized $25 billion common stock offering priced at $625.00 per share. Meta also priced 250 million depositary shares at $50 per share.
What are Class C shares? They are a new class of non-voting Meta stock. Meta expects them to begin trading in the first quarter of 2027, and the $60 billion at-the-market program is expected to start after that.
What is Nvidia’s role? Nvidia said it will invest $3 billion in Meta’s equity offering. Reports say AMD and Broadcom have potential interest too, with the three together totaling up to $5 billion in Class A shares, but only Nvidia’s amount is announced.
What will the money pay for? Reports say AI infrastructure expansion. No source we reviewed ties the proceeds to a named model, product, or specific data center.
Why does the headline number differ between outlets? The offering was upsized during the news cycle. Some headlines still show $95 billion, Bloomberg Law’s reads almost $98 billion, and Reuters uses $97.5 billion.
