A $200 Billion Company That Might Be Failing
Meta Platforms generated roughly $200 billion in revenue last year and serves close to 3.5 billion users across its apps — figures that, on their face, describe one of the most dominant companies in corporate history. Yet investigative journalist Julia Angwin, in a recent conversation on the podcast Factually with Adam Conover, argues the company is in the early stages of structural decline, comparable to what she witnessed firsthand covering AOL’s collapse as a Wall Street Journal beat reporter decades earlier.
The claim is counterintuitive enough to warrant scrutiny rather than acceptance at face value. This piece lays out Angwin’s evidence, tests it against the company’s own disclosed numbers, and examines where the argument holds up and where it’s more speculative — the kind of treatment the claim deserves given how much weight “is Big Tech actually vulnerable” claims tend to carry in current policy and academic debate.
Setting the Baseline: What “Decline” Means for a Company This Size
Before evaluating the claim, it’s worth being precise about what would actually count as decline for a company with Meta’s scale. Angwin isn’t arguing Meta is collapsing tomorrow — she explicitly draws a comparison to AOL and Yahoo, both of which survived for years in what she calls a “zombie state”: still operational, still technically profitable in narrow ways, but no longer central to how people use the internet. Under this framing, decline is a trajectory, not an event, and the relevant question is whether the underlying metrics are trending the wrong direction even while headline revenue looks fine.
User Numbers Turned Negative for the First Time
The single data point Angwin treats as most significant is that Meta reported a decline in daily active users across its properties for the first time in the company’s history last quarter. Because Meta has historically been defined by relentless user growth, a reversal — even a modest one — represents a break from the pattern investors have used to justify the company’s valuation for two decades.
Claim: Falling user numbers indicate genuine platform fatigue rather than a statistical blip.
Evidence: Angwin points to Meta’s own securities filings showing the number of ads shown per user increased by 27% in a single quarter — a move she interprets as the company compensating for shrinking engagement by extracting more revenue per remaining user rather than growing its base.
Interpretation: Rising ad density alongside falling users is consistent with a company managing decline rather than growth — squeezing more value from a shrinking audience is a classic late-stage monetization pattern.
Limitation/counterpoint: A single quarter of user decline is not, by itself, proof of a durable trend. User metrics for large platforms are volatile and subject to measurement methodology changes, seasonal effects, and one-off events. Angwin’s own framing acknowledges that companies work hard to prevent these numbers from dropping publicly, which means by the time a decline becomes visible externally, it may already reflect a longer internal trend — but that inference, while plausible, is not something the public data can fully confirm on its own.
Tens of Billions Spent on Ventures That Haven’t Paid Off
Angwin’s second pillar of evidence concerns capital allocation. She cites approximately $80 billion spent on Meta’s metaverse initiative — the project significant enough that the company renamed itself around it — with little commercial return. She further cites roughly $150 billion already invested in AI development, with the company signaling to investors that it plans to spend an additional $115 billion in the current year, even as its models continue to lag competitors like OpenAI and Anthropic on public benchmarks.
A specific illustration Angwin raises is Meta’s abandoned Llama strategy: an approximately two-year bet on an open-source AI model designed to run locally on users’ own hardware rather than through centralized servers. The approach failed for practical reasons — the model required substantial computational resources most users didn’t have, and its outputs didn’t meet competitive benchmarks — leading Meta to abandon the open-source framing it had spent two years publicly championing and pivot back toward a closed model architecture, arriving, by Angwin’s account, years behind rivals in a market Meta had spent significant capital and reputation trying to lead.
A Structural Mismatch: Consumer Company, Enterprise Market
Beyond spending totals, Angwin raises a strategic argument about why Meta’s AI investment may not pay off even with continued funding. She frames modern AI as increasingly an enterprise software market — companies selling AI tools to other businesses — rather than a consumer product market, and notes that companies rarely succeed at both models simultaneously. She points to Microsoft as an instructive parallel: a company with deep enterprise success that has never managed to build a lasting consumer hit, citing Bing’s negligible market share and Xbox’s persistently secondary position in gaming as evidence.
Claim: Meta’s core competency — consumer engagement products — doesn’t transfer to the enterprise AI market it’s now trying to compete in.
Evidence: Meta’s entire business (Facebook, Instagram, WhatsApp) is built on free consumer products monetized through advertising, a fundamentally different go-to-market motion than selling AI infrastructure or tools to enterprise buyers.
Interpretation: If this analysis holds, continued AI spending may not close the gap with competitors regardless of scale, because the mismatch is structural rather than a matter of resources.
Limitation/counterpoint: This is a strategic hypothesis rather than a settled fact. Large technology companies have occasionally made successful pivots across these categories over long enough timeframes, and it’s possible for a company to build enterprise capability through acquisition or talent hiring rather than organic product history. The claim is worth taking seriously as an analytical frame, not as a certainty.
Why This Spending Persists Despite Weak Returns: The Governance Angle
A recurring theme in Angwin’s analysis is that Meta’s ownership structure may explain why this level of unprofitable spending continues unchecked. Mark Zuckerberg holds a class of shares carrying disproportionate voting power, meaning the company’s board cannot remove him regardless of shareholder sentiment. Angwin traces this dual-class structure to a model pioneered by media conglomerates — she cites Rupert Murdoch’s companies as an early example — originally justified as protecting press organizations’ editorial independence from short-term market and government pressure.
She argues this justification doesn’t extend cleanly to Meta, since the company has simultaneously argued in legal and regulatory contexts that it functions as a neutral platform rather than a publisher exercising editorial judgment — while retaining a governance structure historically reserved for organizations claiming exactly that editorial role. This is a notable tension worth flagging for anyone examining platform governance or media law: Meta effectively occupies the legal privileges of both a neutral conduit and an editorially insulated publisher, without fully accepting the obligations that typically accompany either designation.
The Legal Exposure Angwin Sees as an Emerging Risk
Separate from user and spending metrics, Angwin points to litigation as a growing financial threat. Section 230 of the Communications Decency Act generally shields platforms from liability for third-party content, and has historically blocked most lawsuits alleging platform-enabled harm. However, Angwin describes a wave of individual tort claims — reportedly numbering in the hundreds of thousands — that sidestep Section 230 by arguing direct harm to a specific individual rather than liability for third-party content itself.
She points to a recent bellwether case, where a court ruled in favor of a plaintiff alleging harm from platform-promoted content, as a potentially significant precedent. Angwin draws a direct comparison to the tobacco litigation of the 1990s, in which individual lawsuits were eventually consolidated into a mass tort settlement exceeding $200 billion. Her reasoning is that a favorable early bellwether ruling often signals how courts will treat the broader pool of similar claims, potentially opening the door to comparable consolidated liability for Meta.
Claim: Meta faces litigation exposure comparable in scale to historic tobacco settlements.
Evidence: The parallel legal mechanism (individual tort claims outside a liability shield, consolidated via bellwether cases) and the sheer volume of pending claims support the comparison structurally.
Interpretation: If courts continue ruling against the company in subsequent bellwether cases across different alleged harms (eating disorder promotion, bullying, exposure to graphic content), a mass tort consolidation becomes increasingly plausible.
Limitation/counterpoint: This remains a single case at an early stage. Legal outcomes in bellwether litigation don’t guarantee consistent rulings across the full docket, and companies often successfully differentiate or settle individual claims before broader consolidation occurs. The tobacco comparison is illustrative rather than predictive.
Data Points Worth Citing Directly
- First-ever reported decline in daily active users across Meta’s platforms in a single quarter
- Approximately $80 billion spent on metaverse-related initiatives
- Approximately $150 billion already invested in AI, with an additional $115 billion in planned spending disclosed to investors
- A 27% quarter-over-quarter increase in ads shown per user, per company filings cited by Angwin
- Federal Trade Commission reporting that a majority of reported fraud and scam cases — cited as over 60% — occur on Meta’s platforms
- An emerging body of individual tort litigation, described as numbering in the hundreds of thousands of filed cases
What This Means Beyond One Company
For students or researchers examining platform accountability, three implications extend well past Meta specifically. First, dual-class share structures originally designed for press independence are now widely used by technology companies that simultaneously disclaim editorial responsibility — a governance contradiction worth further legal and regulatory scrutiny. Second, Section 230’s traditional shield is being tested by a tort-based litigation strategy that could, if successful at scale, establish a new liability pathway for platform-enabled harm distinct from existing content-moderation law.
Third, Angwin’s broader argument — developed further in her forthcoming book on resisting authoritarianism — connects platform decline to a media-consolidation concern: that a small number of privately controlled information gatekeepers now hold outsized influence over public discourse, a concern with implications for democratic theory and media policy well beyond any single company’s balance sheet.
Where the Analysis Is More Speculative Than Proven
Several parts of Angwin’s argument are best treated as informed hypothesis rather than established conclusion. The claim that Zuckerberg’s continued high-risk spending reflects a psychological or status-driven motivation — a desire for legitimacy among AI-focused peers — is offered as interpretation rather than something verifiable through public data.
Similarly, while the AOL and tobacco comparisons are analytically useful, historical analogies of this kind are illustrative rather than deterministic; Meta’s specific competitive position, regulatory environment, and product portfolio differ from both precedents in ways that could change how (or whether) the pattern repeats. Finally, some of the discussion — including comparisons to competitors like Google’s relatively neutral political posture — reflects a snapshot of a specific and rapidly shifting competitive and political landscape that may look different within a short time.
So Is Meta Actually Dying, or Just Aging?
The more defensible version of Angwin’s argument isn’t that Meta is on the verge of collapse, but that several structural indicators — user erosion, unprofitable capital allocation at massive scale, mounting legal exposure, and a governance structure that removes normal market discipline — are moving in the same direction simultaneously for the first time in the company’s history.
Whether that constellation of signals produces the kind of prolonged, AOL-style fade Angwin describes, or proves temporary as the company’s AI investments eventually mature, is an open empirical question rather than a settled one. What’s clear is that the metrics worth watching going forward are less about revenue, which remains strong, and more about user trends, ad density, litigation outcomes, and whether Meta’s enterprise AI ambitions can overcome the structural mismatch Angwin identifies.
Frequently Asked Questions
Is Meta’s revenue actually declining?
No — Meta’s revenue remains strong at roughly $200 billion annually. The decline being discussed is in daily active users and long-term structural indicators, not current revenue.
Can Mark Zuckerberg be removed by Meta’s board or shareholders?
Effectively no. Zuckerberg holds a class of shares with disproportionate voting power, giving him controlling influence over governance decisions regardless of broader shareholder sentiment.
What is Section 230 and why does it matter here?
Section 230 of the Communications Decency Act generally protects platforms from liability for content posted by third parties. Recent individual tort lawsuits reportedly sidestep this shield by alleging direct harm to a specific plaintiff, rather than liability for the third-party content itself.
Source: Julia Angwin, interviewed by Adam Conover on the podcast Factually with Adam Conover, July 2026. Additional context drawn from Angwin’s reporting for her New York Times opinion piece “Meta is dying. It’s about time,” and her forthcoming book, On Courage: How to Be a Dissident in an Age of Fear (narrated by Rachel Maddow).
