Meta has poured hundreds of billions of dollars into artificial intelligence development, but investors and analysts remain skeptical that the spending will translate into meaningful returns anytime soon. Mark Zuckerberg has positioned the company as a serious contender in the AI race, yet Meta’s history is defined more by acquisitions and imitation than original breakthroughs — from WhatsApp and Instagram to Reels, which was modeled on TikTok.
A Pattern of Costly Experiments
Many of Meta’s self-driven projects have struggled to gain traction. Attempts to clone Snapchat, Houseparty, and Clubhouse all fizzled, while ventures like the Portal device, internet-delivery drones, and its cryptocurrency initiative were eventually shelved. The most expensive miscalculation was the metaverse, which reportedly cost Meta more than $80 billion before Zuckerberg’s attention shifted almost entirely to AI following the rise of ChatGPT.
Steep Odds of Recovering the Investment
The financial math behind Meta’s AI ambitions looks daunting. Even if the company generated $100 billion annually from AI-related subscriptions, it would still take over a decade to recoup its current spending. For context, Meta’s total 2025 revenue was just under $201 billion, with only a small fraction coming from sources outside advertising. That means AI would need to become a standalone business nearly as profitable as Meta’s dominant ad operations — a tall order given the company’s track record of relying on acquired or borrowed innovation rather than in-house breakthroughs.
Whether Meta can defy that pattern with AI remains an open question, one that will shape the company’s next decade as much as its ad business has shaped the last one.