Meta’s Costly AI Gamble Faces Growing Doubts

Meta has invested hundreds of billions of dollars into artificial intelligence, but the market’s enthusiasm hasn’t matched Mark Zuckerberg’s own excitement for the technology, raising questions about whether the company can turn its massive spending into a viable, standalone business.

A History of Costly Experiments

Much of Meta’s success has come from acquisitions like Instagram and WhatsApp, or from replicating features pioneered elsewhere, such as Stories and Reels. Independent projects, however, have a spottier record — the Portal device, internet-delivery drones, a cryptocurrency initiative, and Instant Articles all failed to gain lasting traction. The company’s metaverse push alone reportedly cost more than $80 billion before Zuckerberg shifted his focus to AI following the rise of ChatGPT.

The Math Behind the AI Bet

Meta’s current AI spending would take more than a decade to recoup even with $100 billion in annual AI subscription revenue — a figure far beyond what the company currently earns outside advertising. In 2025, Meta generated just $4.8 billion in non-advertising revenue out of nearly $201 billion in total revenue. That gap underscores the scale of the challenge: AI would need to become almost as profitable as Meta’s core advertising business just to justify what’s already been spent.

Adding to the uncertainty, broader research suggests AI adoption isn’t delivering the productivity gains many companies expected. A National Bureau of Economic Research study of thousands of executives found limited operational impact from AI tools so far.

Meta’s deep pockets mean these bets won’t threaten its core business in the near term. But given the company’s uneven history with in-house innovation, it remains unclear whether this latest — and most expensive — wager will pay off.