Cracks are emerging in the AI spending boom

Investor patience with the AI infrastructure boom is fraying. Over two weeks, Alphabet, Amazon, Meta, and Microsoft unveiled capital expenditure plans exceeding $740 billion for this year — spending Wall Street now reads as AI infrastructure bets on data centers and chips rather than a sign of confidence. Anxiety spilled into markets on Friday: the Nasdaq fell 2.9%, Nvidia dropped 4%, Broadcom shed more than 5%, and Qualcomm slid 11%. Alphabet, which lost $500 billion in market value in a 12% plunge Thursday — its worst day on record — fell again.

The sharpest concern is the disappearance of the free cash flow that once buffered these companies. Alphabet reported negative free cash flow of $9.1 billion in the third quarter, its second straight quarter in the red, as quarterly capex climbed to $51.4 billion. It expects to spend as much as $205 billion this year, with more increases coming in 2027. Compounding the tension, the popularity of AI Mode — its AI chatbot built into Search — is pressuring the ad business that funds all that spending: Search ad revenue growth slowed to 12%, its weakest in five quarters. Amazon posted negative free cash flow of $42.4 billion over the trailing 12 months, versus positive $14.8 billion a year earlier, and plans $220 billion in capex; even with AWS revenue up 39%, its stock fell 7% after Thursday's earnings. Meta reported negative free cash flow of $2.1 billion and is guiding to capex as high as $145 billion, with CEO Mark Zuckerberg acknowledging its new free AI agent, Muse, is "very compute-intensive to run." Microsoft alone stayed positive, with $18.6 billion in free cash flow against $45.2 billion of capex — below its $50 billion projection, which it attributed to third-party capacity delays rather than any pullback, dragging down shares of AI compute rental firms.

Financing is also shifting. Funding is arriving not just from profits but from capital markets: Microsoft, Alphabet, Amazon, Meta, and Oracle raised a combined $326 billion in debt in 2026, three times last year's total, per S&P Global Market Intelligence. Demand itself isn't the problem — Google, Amazon, and Microsoft all posted cloud growth — but the cost of serving it is spooking investors.

The turbulence arrives at an awkward moment for Anthropic, which filed to go public earlier this week and depends on vast compute access for growth. The Financial Times reported Friday that Blackstone has warned investors Anthropic might not qualify for an investment-grade rating post-IPO, potentially increasing its reliance on hyperscalers and third parties to fund infrastructure.

What to watch: whether hyperscalers can demonstrate returns that justify capex growth through 2027, whether debt markets keep funding the buildout at triple the prior year's pace, and whether Anthropic's IPO pricing reflects compute financing risk. The core open question is whether strong compute demand can outrun the cash burn it requires.