MarkGrid

Meta's $95B Stock Sale Explained: Where the AI Money Goes and What It Means for Advertisers

Meta is raising about $95 billion in equity to fund AI data centers. Here is how the deal works, why investors balked, and what it means for ad budgets.

ST
Sunny TanwarAI Team, MarkGrid
Oct 28, 2026 7 min read
MarkGrid cover: Meta's $95B Stock Sale, Explained

Meta launched a plan on Monday, October 26, 2026, to raise roughly $95 billion in equity to pay for AI data centers and chips: about $35 billion in underwritten stock and preferred offerings now, plus a $60 billion at-the-market program from early 2027. Nvidia, AMD and Broadcom signaled non-binding interest in up to $5 billion. Investors were uneasy, and advertisers should expect even more AI inside Meta's ad system.

Key facts: Announced October 26, 2026, per Meta's SEC filing. About $35B of underwritten offerings priced that evening at $25B of common stock ($625 a share, upsized from $22.5B) plus $12.5B of 6.5% mandatory convertible preferred. A separate at-the-market program of up to $60B in new Class C shares is not expected before Q1 2027. Nvidia, AMD and Broadcom indicated non-binding interest in up to $5B. 2026 capex guidance is $135B-$145B, up from $130B-$145B. Q3 revenue was $64.6B, up 26%. Shares closed down 1.88% on October 26 after falling about 7% on October 22.

What exactly is Meta selling?

Meta is selling three different instruments, and only two of them are committed. Its preliminary terms filing set out the structure:

  1. Class A common stock, sold through forward sale agreements. The launch size was $22.5 billion. According to the final pricing terms, it was increased to $25 billion: 40 million shares at $625, slightly below Monday's $630.56 close. Meta receives the cash only when the forwards settle, on or before November 19, 2027.
  2. Mandatory convertible preferred stock. Meta sold 250 million depositary shares at $50 each, about $12.5 billion, with a 6.50% annual dividend. They convert into common stock around August 15, 2029.
  3. An at-the-market (ATM) program of up to $60 billion in newly created non-voting Class C shares. It is not expected to start before Q1 2027.

Is it $95 billion, $97.5 billion or "almost $100 billion"?

All three figures depend on which stage of the deal you count. The original Monday morning plan added $35 billion to $60 billion, giving the $95 billion headline used by the Irish Times/Financial Times and Reuters. After the common stock was upsized at pricing, Reuters' follow-up, carried by Business Recorder, put the plan at $97.5 billion.

Keep in mind that the $60 billion ATM is an intention, not a sale. Meta's filing says its timing, size and terms are undetermined.

Are Nvidia, AMD and Broadcom really investing?

They have expressed interest, but nothing binds them. Meta's filing says the three chipmakers indicated interest in buying up to $5 billion of Class A stock on the same terms as other buyers. One Reuters version reported that Nvidia alone would invest $3 billion. The SEC documents we reviewed do not break out amounts by company, so treat that split as unconfirmed.

All three are major Meta suppliers, which is why some critics, cited by 24/7 Wall St., call the deal circular financing: vendors help fund a customer that then pays them back through chip orders.

Where does the money go?

Officially, proceeds are for "general corporate purposes," but every source ties the raise to AI infrastructure. Meta's Q3 2026 results show what the money is meant to support:

  • 2026 capital expenditures of $135-$145 billion. The low end was raised from $130 billion.
  • Plans to keep investing aggressively through 2027, with no full-year 2027 number given.
  • About $825 billion in committed future spending, according to Reuters.

Spending has already overtaken cash generation. Meta reported negative free cash flow of about $2.1 billion in Q3, which Reuters calls the first negative quarter in more than a decade. According to the FT report carried by the Irish Times, Meta told investors it will need to raise "hundreds of billions of dollars" over several years.

Why did the stock fall, and how big was the drop?

The two drops came from different events, and some coverage mixes them up. The roughly 7% decline happened on Thursday, October 22, the day after Meta's October 21 earnings. On the offering news itself, shares fell as much as 4% during the day and closed down 1.88% at $630.56 on October 26, according to 24/7 Wall St.

The skeptical case rests on three points:

  • Dilution. Estimates vary by method. 24/7 Wall St. cites roughly 5.5% if the whole plan is used, and others put the near-term hit from the forwards closer to 1.6%.
  • Repeat raises. William Blair analysts wrote, per Reuters, that investors would "continue to question the financing and potential for future capital raises."
  • AI bubble worries. To critics, capex that outruns cash flow and is backed by supplier money depends on AI returns arriving on time.

The other side of the argument: Q3 revenue rose 26% to $64.6 billion. Hargreaves Lansdown's Q3 review noted ad impressions up 10% and average price per ad up 14%.

Does Meta say AI is already paying off?

Yes. According to the Irish Times/FT report, Meta argued that its AI investments are producing tangible returns through higher engagement. Zuckerberg's line in the release was: "Our investments are paying off, the core business is strong." These are company claims. The ad metrics fit that story but don't isolate AI's contribution.

What does this mean for advertisers' budgets and AI ad tools?

For advertisers, the main effect is that Meta's ad system will rely even more on automation. The company is spending hundreds of billions largely so its models can choose audiences, placements and creative better than manual setups. That has three practical consequences:

  1. Rising CPMs are a real possibility. Meta reported a 14% increase in average price per ad. If AI improves conversion rates, Meta can charge more per impression. Your blended cost per result is the number to track.
  2. Automated campaigns will push for more of your budget. As Meta's tools get better, they will ask for broader targeting and more creative variety. A cross-channel layer such as MarkGrid Budget Optimization is designed to compare marginal returns across channels rather than relying on each platform's own reporting.
  3. Creative becomes the main lever. When the algorithm controls targeting, the inputs you still control are the ads themselves. Testing concepts before spending on them, for example with Neuro-Marketing predictive creative testing, helps you avoid paying Meta's AI to find out which ads fail.

Discovery is shifting too. As Meta's assistant and agent products such as Muse grow, brands will want to know how AI systems describe them. Model Share tracks how AI models mention and recommend brands, and our GEO guide covers how to adapt. MarkGrid runs 650+ AI agents across these workflows.

FAQ

How much is Meta raising in its stock offering?

Meta priced about $37.5 billion in underwritten offerings on October 26, 2026: $25 billion of common stock and $12.5 billion of mandatory convertible preferred. It also plans an at-the-market program of up to $60 billion.

Did Nvidia, AMD and Broadcom commit $5 billion?

No. Meta's SEC filing says the three chipmakers indicated interest in buying up to $5 billion combined and that the indications are not binding.

When does the $60 billion at-the-market program start?

Meta says it does not expect to start before the first quarter of 2027, after the new non-voting Class C shares begin trading.

What is Meta's 2026 capex guidance?

Meta now guides to $135-$145 billion for 2026, up from $130-$145 billion. It says it will keep investing aggressively in 2027.

Why are some investors worried?

They point to share dilution, the chance of more raises later, negative free cash flow in Q3, and "circular" financing involving Meta's own chip suppliers.

Will Meta ads get more expensive?

Not necessarily for every advertiser, but average price per ad rose 14% in Q3. Teams should track cost per result rather than CPM alone and should test creative before scaling spend.

Last updated October 28, 2026.

Model Share for this topic
ChatGPT
34%
Gemini
28%
Perplexity
41%

How often MarkGrid is named when AI models discuss this topic. About Model Share

ST

Sunny Tanwar

AI Team, MarkGrid

Sunny Tanwar is part of the AI team at MarkGrid, working on AI systems that support marketing automation and performance. He focuses on translating new AI capabilities into scalable, measurable marketing outcomes.

See it on your brand

Thirty minutes on where marketing is leaking for you.

Talk to Sales