Meta's AI Narrative: Why the Stock Swings and the $16 Billion Scam Ad Problem

Meta just reported Q4 2025 earnings that beat estimates. The stock jumped 10% after hours. Mark Zuckerberg announced plans to spend up to $135 billion on AI infrastructure in 2026—nearly double last year’s already massive $72 billion. Wall Street applauded.

But beneath the AI hype lies a story most investors haven’t fully processed: according to internal documents obtained by Reuters, Meta projected it would earn $16 billion in 2024 from ads promoting scams, illegal gambling, banned pharmaceuticals, and fraudulent investment schemes. That’s roughly 10% of total revenue coming from what the company’s own teams classify as “high-risk” fraudulent advertising.

This isn’t a bug in the system. It’s a feature. And understanding this tension—between the aggressive AI spending narrative and the uncomfortable revenue realities—is essential for anyone trying to make sense of META’s violent stock swings.

The Volatility Pattern: Why META Moves So Aggressively

META shares have become a rollercoaster. In 2025 alone, the stock swung from a low of $479 in April to an all-time high of $796 in August—a 67% move in four months. Then came the Q3 earnings disappointment, which sent shares down over 11% in a single session.

What’s driving these swings? Three forces collide every earnings cycle:

1. The AI Capex Arms Race

Meta’s capital expenditure guidance has become the single most important number for the stock. Not revenue. Not earnings per share. Capex.

Here’s the trajectory:

2024: ~$38 billion
2025: $72 billion
2026 guidance: $115-135 billion

That’s a 255% increase in two years. The company is spending more on AI infrastructure than most countries spend on their entire military budgets. Zuckerberg has committed to at least $600 billion in U.S. data center spending through 2028.

When Meta guides capex higher than expected, the stock often sells off initially—investors worry about compressed margins and uncertain ROI. When earnings beat despite the spending, the stock rallies as the market concludes Zuckerberg might actually be building something valuable.

This creates a pendulum: fear of the spend vs. hope for the payoff. Every quarter, the pendulum swings violently.

2. The Narrative Shift from “Efficiency” to “AGI”

Remember 2023’s “Year of Efficiency”? Meta laid off 21,000 employees, cut costs aggressively, and the stock more than doubled. Investors loved the discipline.

Now the narrative has completely reversed. Meta is pursuing what Zuckerberg calls “the most aggressive capital deployment cycle in the history of the technology sector.” The stated goal is no longer efficiency—it’s building toward Artificial General Intelligence (AGI).

This creates cognitive dissonance for investors. The same management team that won back trust through cost-cutting is now spending at unprecedented levels on a technology whose payoff timeline is measured in years, not quarters.

Deutsche Bank analysts captured the tension: “Investor fears around the potential impact to earnings from the projected spend, as well as reduced financial flexibility from the elevated investments, could somewhat outweigh optimism around faster growth.”

3. The Reality Labs Black Hole

Meta’s Reality Labs division—responsible for VR headsets, AR glasses, and the metaverse—has lost over $50 billion cumulatively since 2020. In 2025 alone, Reality Labs burned approximately $17 billion.

Every quarter, analysts ask when this bleeding stops. Every quarter, Zuckerberg doubles down. For investors trying to value the company, this creates massive uncertainty. Is Reality Labs a visionary long-term bet or a $50 billion distraction?

The Scam Ad Revenue Problem: What the Documents Reveal

Now we get to the part Wall Street doesn’t like to discuss.

In late 2025, Reuters published a devastating investigation based on internal Meta documents spanning 2021-2025. The findings were damning:

The Numbers

Meta internally projected 10.1% of 2024 revenue—approximately $16 billion—would come from ads promoting scams and banned products
The company served as many as 15 billion “high-risk” fraudulent ads per day
China alone generated roughly $18 billion in ad revenue for Meta in 2024, with nearly $3 billion of that coming from scam-related advertising
Meta internally labeled China its top “scam exporting nation,” accounting for 25% of all fraudulent ads globally

The “Playbook”

Rather than aggressively cracking down on fraudulent advertising, Meta developed what internal documents describe as a “global playbook” for managing regulatory pressure while protecting revenue.

The strategy worked like this:

Ad Library Manipulation: After identifying keywords regulators used to search Meta’s public Ad Library (a transparency tool), Meta’s teams ran those searches repeatedly to identify and remove flagged ads from the searchable database—while barely reducing the actual volume of scam ads running on the platform.
Revenue Guardrails: Internal policies established “specific revenue guardrails” that blocked the company’s scam detection team from taking any action that would cut more than 0.15% of revenues. In other words, fraud enforcement had a budget ceiling.
95% Certainty Threshold: Meta’s internal policy required 95% certainty that an advertiser was fraudulent before banning them. Advertisers below that threshold were simply charged higher rates—turning suspected fraud into a premium revenue stream.
Protecting High-Spenders: When enforcement teams proposed shutting down fraudulent accounts, documents showed Meta sought assurance that “growth teams would not object given the revenue impact.” When asked whether Meta would penalize high-spending Chinese partners running scams, the documented answer was “No” due to “high revenue impact.”

The Targets That Tell the Story

Meta’s internal targets for scam ad revenue reveal how embedded this income stream has become:

Year	Scam Ad Revenue Target (% of total)
2024	10.1%
2025	7.3%
2026	6.0%
2027	5.8%

Notice what’s missing: there’s no target for zero. Meta isn’t trying to eliminate scam revenue—it’s trying to manage it down to an “acceptable” level. At 2027’s target of 5.8%, that still represents approximately $15+ billion annually based on projected revenue growth.

Why Meta Doesn’t Fix It

The Reuters investigation found that Meta estimated implementing universal advertiser verification—which would dramatically reduce fraud—would cost roughly $2 billion to implement and potentially reduce revenue by nearly 5%.

For a company spending $135 billion on AI infrastructure, $2 billion is pocket change. But a 5% revenue hit on a ~$200 billion revenue base is a $10 billion annual problem. The math explains the inaction.

The AI Monetization Reality

So what is all that AI spending actually producing?

Advantage+: The $60 Billion Engine

Meta’s most successful AI monetization so far is Advantage+, its AI-driven advertising platform. Advantage+ reached a $60 billion annual run rate in 2025. The system uses AI to automate ad creative generation, targeting, and bidding—advertisers input a budget and goal, and Meta’s AI handles the rest.

This is genuinely impressive technology. It’s also technology that makes the scam ad problem worse, not better. When AI optimizes for engagement and conversion without human oversight, it’s optimizing for whatever works—including scams that prey on vulnerable users.

Reels Monetization

Instagram Reels now commands a $50 billion annual revenue run rate, up from essentially zero a few years ago. Meta successfully caught up to TikTok by deploying AI to optimize the content feed and ad delivery.

The Llama Question

Meta’s open-source Llama AI models have garnered significant developer attention, but direct monetization remains unclear. The company’s strategy appears to be using Llama to strengthen its ecosystem rather than licensing it for direct revenue.

Internal estimates suggest AI-generated content could contribute $12 billion to Meta’s ecosystem by 2026, but this remains speculative.

The ROI Problem

Here’s the uncomfortable question: if Meta is spending $135 billion in 2026 on AI infrastructure, what’s the return?

Advertising revenue growth has been strong—roughly 20-25% year-over-year in 2025. But advertising is also cyclical. The last recession cut Meta’s revenue growth dramatically. If a downturn hits while Meta is mid-way through a $600 billion spending program, the stock will crater.

Wall Street analysts are taking this on faith. Cantor Fitzgerald maintains a $920 price target, projecting “AI execution will lead to a sentiment reversal.” But the gap between current spending and proven returns remains vast.

What Investors Should Consider

The Bull Case

If you’re bullish on META, here’s the argument:

Advertising dominance is real: Meta, Alphabet, and Amazon will claim over 56% of global ad spend in 2026. Meta’s AI-optimized ad platform is genuinely best-in-class.
The spending will create moats: The $600 billion infrastructure commitment creates barriers to entry no competitor can match. If AI is the future, Meta is building the future’s plumbing.
Revenue is growing through the spend: Unlike previous tech bubbles, Meta is financing AI expansion with profitable operations, not debt. The company generated massive free cash flow even while investing $72 billion in 2025.
The scam problem is manageable: 10% of revenue from questionable sources sounds bad, but if Meta gradually reduces this while growing overall revenue, absolute dollar exposure stays flat while the percentage shrinks.

The Bear Case

If you’re skeptical, consider:

The scam revenue is a regulatory time bomb: The U.S. Virgin Islands has already sued Meta over scam ads. Two U.S. senators have urged FTC and SEC investigations. Meta assigned its scam handling the top possible score in internal risk rankings. The company knows this could blow up.
ROI on AI spending is unproven: $135 billion in 2026 capex needs to produce returns eventually. If AI monetization disappoints, the stock’s premium valuation collapses.
Reality Labs continues burning cash: $17 billion in losses in 2025 alone, with no clear path to profitability. At what point does Zuckerberg’s vision become an unjustifiable drag?
Cyclical exposure: Digital advertising collapsed during the 2022 mini-recession. A real recession while Meta is committed to $600 billion in spending could be catastrophic.
Ethical overhang: The Reuters investigation revealed a company that knowingly profits from fraud and has built systems to protect that revenue from enforcement. This isn’t a good-faith actor facing a hard problem—it’s a company that chose revenue over users.

The Regulatory Sword of Damocles

Meta assigned its scam advertising practices the highest possible risk score in 2025. This wasn’t paranoia—it was accurate assessment.

The regulatory exposure is significant:

U.S. Virgin Islands lawsuit: Filed December 2025, alleging Meta profits from scam ads
Congressional pressure: Two senators urged FTC and SEC investigations in November 2025
EU Digital Services Act: Requires platforms to address systemic risks, including fraudulent advertising
Australia and UK: Both pursuing aggressive platform accountability legislation

A major regulatory action could force Meta to actually implement universal advertiser verification—the $2 billion fix that costs $10+ billion in revenue. That’s a scenario the current stock price doesn’t fully reflect.

Where Does META Trade From Here?

Analysts remain broadly bullish:

Goldman Sachs: Price target $700+
Cantor Fitzgerald: $920 target
Consensus: ~$822, implying 27% upside from recent levels

But analyst targets are based on the AI narrative succeeding and the scam revenue problem staying contained. Both are assumptions, not certainties.

For investors, the question isn’t whether Meta’s AI technology is impressive—it clearly is. The question is whether a company that knowingly extracts billions from fraud, while demanding trust for $600 billion in speculative spending, deserves your capital.

The stock will continue swinging violently. Every earnings call is a referendum on the same tension: is Zuckerberg building the future, or is he just really good at extracting money—from advertisers, from users, and from investors who believe the narrative?

TL;DR

META stock swings violently because investors can’t agree whether $135B in AI spending (2026) is visionary or reckless
Internal documents show Meta projected $16 billion (10% of revenue) from scam/fraudulent ads in 2024
Meta built a “global playbook” to hide fraudulent ads from regulators while protecting the revenue stream
Company has “revenue guardrails” preventing fraud enforcement from cutting more than 0.15% of revenue
AI monetization (Advantage+, Reels) is real but unproven relative to the massive capex commitment
Regulatory risk is severe—lawsuits filed, congressional investigations urged, top internal risk score assigned
For investors: understand you’re betting on both AI success AND the scam revenue problem staying contained