Meta’s $600bn Bet on AI Tests Wall Street’s Nerves

In the middle of the biggest artificial-intelligence build-out in history, Meta is spending more than almost anyone else. The company is constructing two vast data centres and, according to TechCrunch and other reports, total U.S. infrastructure investment in AI could top $600 billion over the next three years. That number might sound routine in Silicon Valley, but it is starting to make Wall Street nervous.

The tension came to a head during Meta’s latest earnings call. The company’s operating expenses jumped by about $7 billion year-on-year, while capital expenditure soared close to $20 billion, a result of relentless hiring and data-centre construction to power future AI products.

When pressed for details, Mark Zuckerberg told analysts that the company was “accelerating” investment to secure enough compute for both research and its core social-media business. He promised “frontier models with novel capabilities,” describing a “massive latent opportunity” that would soon pay off.

Investors were unconvinced. Meta’s shares tumbled 12 % by the end of the week, erasing more than $200 billion in market value. The results themselves were solid (about $20 billion in profit), but this was the first time the scale of Meta’s AI spending had a visible impact on its bottom line.

Analysts noted that aside from the Meta AI assistant and its experimental video and smart-glasses projects, it remains unclear what immediate revenue these billions are buying.

Zuckerberg has urged patience, saying new “superintelligence” models are on the way. But with rivals like OpenAI and Nvidia already monetising their AI ecosystems, Meta’s vast investment looks more like a wager than a strategy. For now, Wall Street is left hoping that the biggest infrastructure bet in AI will eventually build more than just data centres.